Zea 2017 64 issue 1 by Hogrefe

Volume 64 / Number 1 / 2017

Experimental Psychology

Editor-in-Chief Christoph Stahl Editors Tom Beckers Arndt Bröder Adele Diederich Chris Donkin Gesine Dreisbach Andreas Eder Magda Osman Manuel Perea James Schmidt Samuel Shaki Sarah Teige-Mocigemba

The European standard and benchmark for education and training in psychology “This book is right now a landmark in current analyses of the profession.” Roger Lécuyer, Emeritus Professor in Developmental Psychology. University Paris Descartes, France

Ingrid Lunt / José Maria Peiró / Ype Poortinga / Robert A. Roe

EuroPsy

Standards and Quality in Education for Professional Psychologists 2015, xiv + 218 pp. US $34.80 / € 24.95 ISBN 978-0-88937-438-6 Also available as eBook EuroPsy has been accepted and adopted as the European standard for education and training in psychology by EFPA. This book, written by its initiator and leading members of the working groups that set EuroPsy up, is the only comprehensive text available about this European benchmark. It first reviews the development of EuroPsy in the historical context of psychology as science and profession and policies for higher education set by international bodies, and in particular the European Union. This handbook then goes on to address the curricula of university

www.hogrefe.com

courses and programmes following from the Bologna Agreement, the flexibility allowed to reflect diversity in Europe, licencing and accreditation, and benchmarking, as well as other prerequisites for meeting the EuroPsy standards. These include the use of a competence model to assure professional standards, supervision, continued professional development, supervision, and ethics. Finally, the authors examine the current and future role of EuroPsy in psychology in Europe, including practical examples of how it has been applied in practice.

Experimental Psychology

Volume 64, Number 1, 2017

Editors

C. Stahl (Editor-in-Chief), Ko¨ln, Germany T. Beckers, Leuven, Belgium A. Bro¨der, Mannheim, Germany A. Diederich, Bremen, Germany C. Donkin, Sydney, Australia G. Dreisbach, Regensburg, Germany

A. Eder, Wu¨rzburg, Germany M. Osman, London, UK M. Perea, Valencia, Spain J. Schmidt, Gent, Belgium S. Shaki, Samaria, Israel S. Teige-Mocigemba, Freiburg, Germany

Editorial Board

U. J. Bayen, Du¨sseldorf, Germany H. Blank, Portsmouth, UK J. De Houwer, Ghent, Belgium R. Dell’Acqua, Padova, Italy G. O. Einstein, Greenville, SC, USA E. Erdfelder, Mannheim, Germany M. Goldsmith, Haifa, Israel D. Hermans, Leuven, Belgium R. Hertwig, Berlin, Germany J. L. Hicks, Baton Rouge, LA, USA P. Juslin, Uppsala, Sweden Y. Kareev, Jerusalem, Israel D. Kerzel, Geneva, Switzerland A. Kiesel, Freiburg, Germany K. C. Klauer, Freiburg, Germany R. Kliegl, Potsdam, Germany I. Koch, Aachen, Germany J. I. Krueger, Providence, RI, USA S. Lindsay, Victoria, BC, Canada

E. Loftus, Irvine, CA, USA T. Meiser, Mannheim, Germany K. Mitchell, West Chester, PA, USA N. W. Mulligan, Chapel Hill, NC, USA B. Newell, Sydney, Australia K. Oberauer, Zu¨rich, Switzerland F. Parmentier, Palma, Spain M. Regenwetter, Champaign, IL, USA R. Reisenzein, Greifswald, Germany J. N. Rouder, Columbia, MO, USA D. Shanks, London, UK M. Steffens, Landau, Germany S. Tremblay, Quebec, Canada C. Unkelbach, Ko¨ln, Germany M. Waldmann, Go¨ttingen, Germany E. Walther, Trier, Germany P. A. White, Cardiff, UK D. Zakay, Tel Aviv, Israel

Publisher

Hogrefe Publishing, Merkelstr. 3, D-37085 Go¨ttingen, Germany, Tel. +49 551 99950-0, Fax +49 551 99950-425, E-mail publishing@hogrefe.com, Web http://www.hogrefe.com

Production

Regina Pinks-Freybott, Hogrefe Publishing, Merkelstr. 3, D-37085 Go¨ttingen, Germany, Tel. +49 551 99950-0, Fax +49 551 99950-425, E-mail production@hogrefe.com

Subscriptions

Hogrefe Publishing, Herbert-Quandt-Str. 4, D-37081 Go¨ttingen, Germany, Tel. +49 551 99950-956, Fax +49 551 99950-998

Advertising/Inserts

Melanie Beck, Hogrefe Publishing, Merkelstr. 3, D-37085 Go¨ttingen, Germany, Tel. +49 551 99950-0, Fax +49 551 99950-425, E-mail marketing@hogrefe.com

ISSN

ISSN-L 1618-3169, ISSN-Print 1618-3169, ISSN-Online 2190-5142

Ó 2017 Hogrefe Publishing. The journal as well as the individual contributions to it are protected under international copyright law. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, digital, mechanical, photocopying, microﬁlming or otherwise, without prior written permission from the publisher. All rights, including translation rights, are reserved.

Publication

Published in six issues per annual volume. Experimental Psychology is the continuation of Zeitschrift fu¨r Experimentelle Psychologie (ISSN 0949-3964), the last annual volume of which (Volume 48) was published in 2001.

Subscription Prices

Annual subscription (2017): Individuals: US $259.00/1185.00; Institutions: US $508.00/1398.00 (postage & handling: US $24.00/118.00). Single issue: US $85.00/166.50 (+ postage & handling).

Payment

Payment may be made by check, international money order, or credit card to Hogrefe Publishing, Merkelstr. 3, D-37085 Go¨ttingen, Germany, or, for customers in North America, to Hogrefe Publishing, Inc., Journals Department, 7 Bulﬁnch Place, 2nd ﬂoor, Boston, MA 02114, USA.

Electronic Full Text

The full text of Experimental Psychology is available online at http://econtent.hogrefe.com/loi/zea

Abstracting/Indexing Services

Experimental Psychology is abstracted/indexed in Current Contents/Social & Behavioral Sciences (CC/S&BS), Social Science Citation Index (SSCI), Medline, PsyJOURNALS, PsycINFO, PSYNDEX, ERIH, Scopus, and EMCare. Impact Factor (2015): 2.000.

Printed on acid-free paper

Experimental Psychology (2017), 64(1)

Ó 2017 Hogrefe Publishing

Contents Editorial

Experimental Psychology: Methodological Rigor and Transparency Christoph Stahl

Research Articles

Gestalt Effects in Visual Working Memory: Whole-Part Similarity Works, Symmetry Does Not Patrycja Kałamała, Aleksandra Sadowska, Wawrzyniec Ordziniak, and Adam Chuderski

Short Research Articles

Ó 2017 Hogrefe Publishing

The Role of Cognitive Load in Intentional Forgetting Using the Think/No-Think Task Saima Noreen and Jan W. de Fockert

Competent and Warm? How Mismatching Appearance and Accent Inﬂuence First Impressions Karolina Hansen, Tamara Rakic´, and Melanie C. Steffens

Automatic Retrieval of Newly Instructed Cue-Task Associations Seen in Task-Conﬂict Effects in the First Trial after Cue-Task Instructions Nachshon Meiran and Maayan Pereg

Individual Differences in the Flexibility of Peripersonal Space Samuel B. Hunley, Arwen M. Marker, and Stella F. Lourenco

The Role of Embodiment and Individual Empathy Levels in Gesture Comprehension Karine Jospe, Agnes Flo¨el, and Michal Lavidor

Experimental Psychology (2017), 64(1)

Editorial Experimental Psychology Methodological Rigor and Transparency Christoph Stahl Psychologische Methodenlehre & Experimentelle Psychologie, Department für Psychologie, Universität Köln, Germany

Editor’s Report In the past year, Experimental Psychology has continued to publish highest-quality experimental research in all areas of psychology, which has had great impact on psychology as a basic science (as reflected in continuously strong citation metrics, e.g., impact factor [IF] of 2, h5-index of 22, and h5-median of 31).1 Experimental Psychology also strives to be a particularly fast outlet: Despite the increasing challenge to find reviewers in a timely manner (which is due to an increasing total publication volume across all outlets), in the majority of cases, authors received editorial decisions – concerning initial submissions as well as revisions – within our 6-weeks deadline. To further advance both impact and speed of dissemination, authors are encouraged to make their manuscripts publicly available as preprints at the time of submission (e.g., https://osf.io/preprints/psyarxiv). Doing so will help authors reach out to their peers, and their research will have the chance to impact the field at a much earlier time, which may in turn increase the visibility of the published article.2

Methodological Rigor and Transparency Experimental Psychology published methodologically rigorous experimental psychological research. Since initially defined by Klauer (2002), the scope of Experimental Psychology is – and continues to be – defined by the

experimental method, and thus, papers based on experiments from all areas of psychology are welcome. As Meiser (2011) succinctly added, manuscripts considered for Experimental Psychology have to be motivated by a relevant research question that derives from – and contributes to – the theoretical foundation of psychology as a basic science. He further notes that these two core criteria – theoretical foundation and methodological rigor – typically overlap: Theoretically stringent hypotheses allow for making more specific predictions, with greater empirical content, that are therefore more likely to be falsified. Empirically evaluating such specific predictions in a methodologically rigorous manner amounts to strong tests of their theoretical foundations. Another aspect of methodological rigor is the choice of optimal statistical procedures for data analysis (Erdfelder, 2010), as well as their comprehensive reporting. Regarding the reporting of statistical results, Erdfelder highlighted the APA publication manual’s call for results sections to “include sufficient information to help the reader fully understand the analyses conducted and possible alternative explanations for the outcomes of those analyses” (APA, 2010, p. 33). He pointed out that (in the Neyman-Pearson framework) this includes alpha level and statistical power as well as effect size estimates. But in recent years it has become obvious that a wide range of additional information is also required: To be able to fully understand and critically assess a given analysis, not only that analysis must be reported in a fully transparent manner; in addition, the reader must be made aware of the context of all additional analyses that were conducted, as well as their confirmatory or exploratory nature (Gelman & Loken, 2014;

The impact factor (IF) is calculated by dividing the number of citations, in 2015, of articles published in 2013 and 2014 by the number articles published in that two-year period. Retrieved from Thomson Reuters, January 18, 2017. The h5-index is the h-index for articles published in the last five complete years. It is the largest number h such that h articles have at least h citations each. The h5-median is the median citation count for these articles. Retrieved from Google Scholar, January 18, 2017. 2 Publishing a finding as a date-stamped and citable preprint can also be helpful in claiming priority of discovery in case a finding has been simultaneously made by independent researchers. Ó 2017 Hogrefe Publishing

Experimental Psychology (2017), 64(1), 1–4 DOI: 10.1027/1618-3169/a000355

Simmons, Nelson, & Simonsohn, 2011). For instance, where multiple tests may be performed, the (lack of) alpha level correction can only be correctly interpreted if the family of tests associated with the a priori hypothesis is known (otherwise, the reader cannot assess whether the true alpha level at which a hypothesis has been tested is .05 or rather, say, .5). Lack of transparency – as reflected, for instance, in phenomena such as the selective reporting of significant results, the reporting of post hoc hypotheses as if they had been stated a priori (e.g., Kerr, 1998), and others – threatens the validity of the empirical foundation upon which our theories are built, and by which they are assessed. If they are to serve as the basis of a cumulative science, research results must be reported in a maximally transparent way. In my term as Editor-in-Chief, Experimental Psychology has so far focused on increasing transparent communication of all aspects of the research cycle, including a-priori hypotheses, methods and materials, raw data, analysis plans, and (confirmatory as well as exploratory) results. As one example, the Registered Reports article type (e.g., Klauer, Becker, & Spruyt, 2016) provides the reader with independent evidence about the a-priori nature of the hypothesis under investigation, it provides authors with a guaranteed publication regardless of direction and statistical significance of the results, and it thereby eliminates selective publication of significant results and provides unbiased input for meta-analytic synthesis.3 A (small but) growing number of Registered Report submissions indicates that researchers are beginning to make use of these benefits. In addition, Experimental Psychology encourages authors of all types of manuscripts to preregister their hypotheses as well as analysis plans by default, for all of their (confirmatory or exploratory) studies, at an independent repository or registry (e.g., osf.io, aspredicted.org), and to point to the registration document within their manuscripts. In addition to enabling the reader to make a valid distinction between confirmatory and exploratory parts of the reported research, this may require an explicit a-priori commitment to a specific statistical analysis that is otherwise often made only after data collection, and it may also serve as an important point of reference for the researchers (and help their future selves counter hindsight bias effects). In the long term, a clear distinction between confirmatory and exploratory results will help build Experimental Psychology’s reputation as a source of both solid and reliable findings as well as new discoveries.

3 4

Editorial

As another example, all articles in Experimental Psychology are accompanied by the raw data necessary to reproduce the reported results, and in many cases additional materials such as analysis scripts are also provided. Making raw data publicly available at the time of submission has proven to be practical; there are many options for publishing even large quantities of raw data (e.g., zenodo.org). Of course, ethical and legal considerations, especially the necessity of securing the anonymity of research participants, may impose constraints on data sharing in a given case (Schönbrodt, Gollwitzer, & Abele-Brehm, 2016).4

Outlook I am thankful to the authors for their active support of these first steps toward increasing transparency. At the same time, I encourage all of us to take further steps to advance transparent scientific communication in our various roles as authors, reviewers, editors, and readers (e.g., Morey et al., 2016). In particular, the peer review process is one of the core elements of scientific communication and quality control; while it is widely respected, much is not known about its efficiency and effectiveness as a mechanism of quality control because empirical data are typically not openly available. As scientists, we expect arguments to be supported by empirical evidence; we must apply this standard also to our discussions about peer review as one of the institutions at the center of the workings of the scientific method. Toward this end, I encourage all stakeholders to work together towards assessing and increasing transparency and quality of peer review: For instance, engaging in open commentary about a published preprint as an additional form of peer review may highlight ways in which a manuscript, and the research reported therein, may be improved upon before publication that can be helpful for the authors in revising their manuscript and increase the validity of the literature (Kriegeskorte, Walther, & Deca, 2012). As another example, assuming that all parties concur, authors may upload the decision letters and reviews (signed or not) along with their revised manuscripts to a preprint server, to broaden the basis for independent investigations of the determinants of peer review quality. To be sure, the debate about the quality of a published article does not end at the time of publication – yet, substantial parts of this debate are typically not reflected in the literature. As a final suggestion, I therefore

For details about article types, see author instructions at http://hogrefe.com/j/exppsy. An English version of the Deutsche Gesellschaft für Psychologie (DGPs) data management guidelines is available at https://www.dgps. de/fileadmin/documents/Empfehlungen/Data_Management_eng.pdf

Experimental Psychology (2017), 64(1), 1–4

Ó 2017 Hogrefe Publishing

Editorial

encourage researchers to engage in this debate by submitting (unsolicited) theoretical or methodological Comment articles to Experimental Psychology (see, e.g., Akiva-Kabiri & Henik, 2014; Gast, 2014; Grégoire, Perruchet, & Poulin-Charronnat, 2014; Moeller & Frings, 2014; Zakay, 2014). The increasing transparency of scientific communication across many journals and areas of study has already facilitated numerous discoveries, as one could recently observe across diverse fields, and it has great potential for accelerating scientific progress. Yet, to optimally benefit, we may need to adapt our goals and incentives: In an increasingly transparent environment, researchers can most effectively turn the available information into scientific progress if they rebalance their professional modus operandi toward more cooperation (Kliegl, 2016).

Editorial Team The support of a dedicated team of excellent researchers, in their roles as Action Editors and reviewers, lays the foundation for the high quality of Experimental Psychology’s publications. I am greatly indebted to all of them for their relentless effort toward securing the quality of the published research as well as a fast editorial process. In particular, I wish to welcome and thank three colleagues who have accepted my invitation to join the team of Associate Editors in the past year: Sarah Teige-Mocigemba (University of Freiburg; automaticity and implicit cognition); Andreas Eder (University of Würzburg; emotions and actions); and James Schmidt (University of Ghent; attention, learning, and cognitive control). Finally, two of Experimental Psychology’s Associate Editors have decided to accept new challenges: Kai Epstude, who has served on the team since 2012, has been named Editor-in-Chief of Social Psychology; and Klaus Rothermund, a long-term pillar of support since 2004, now heads Cognition & Emotion, where he has previously been Associate Editor. Both have excellently served this journal with their expertise and effort for long years; notably, Klaus Rothermund has served Experimental Psychology as Editorin-Chief in 2006. I thank both of them for their great, and continuing, contribution to securing the quality of research in the field.

Call for Special Issue Proposals Experimental Psychology continues to invite proposals for thematic special issues. Special issue topics must be within the journal’s scope, and articles must meet the journal’s Ó 2017 Hogrefe Publishing

primary criteria, namely the rigorous use of experimental methodology together with state-of-the-art statistical analysis, and a strong and innovative theoretical contribution to psychology as a basic science. A special issue would typically comprise a review of the topic under focus (Theoretical Article) as well as empirical papers (Short/ Research Articles or Registered Reports) or methodological contributions (within the Theoretical Article format). A target article (Theoretical or Research Article) might also be published together with one or more invited comments. Proposals can be submitted at any time (for detail see hogrefe.com/j/exppsy).

References American Psychological Association. (2010). Publication manual of the American Psychological Association (6th ed.). Washington, DC: American Psychological Association. Akiva-Kabiri, L., & Henik, A. (2014). Additional Insights. Experimental Psychology, 61, 75–77. doi: 10.1027/1618-3169/ a000208 Erdfelder, E. (2010). Experimental Psychology: A note on statistical analysis. Experimental Psychology, 57, 1–4. doi: 10.1027/ 1618-3169/a000001 Gast, A. (2014). What is learned, and when? Experimental Psychology, 61, 71–74. doi: 10.1027/1618-3169/a000206 Gelman, A., & Loken, E. (2014). The statistical crisis in science. American Scientist, 102, 460–465. doi: 10.1511/2014.111.460 Grégoire, L., Perruchet, P., & Poulin-Charronnat, B. (2014). Is the musical stroop effect able to keep its promises? Experimental Psychology, 61, 80–83. doi: 10.1027/1618-3169/ a000222 Kerr, N. L. (1998). HARKing: Hypothesizing after the results are known. Personality and Social Psychology Review, 2, 196–217. doi: 10.1207/s15327957pspr0203_4 Klauer, K. C. (2002). Tradition and change. Experimental Psychology, 49, 1. doi: 10.1027/1618-3169.49.1.1 Klauer, K. C., Becker, M., & Spruyt, A. (2016). Evaluative priming in the pronounciation task: A preregistered replication and extension. Experimental Psychology, 63, 70–78. doi: 10.1027/ 1618-3169/a000286 Kliegl, R. (2016). A vision of scientific communication. In P. Weingart & N. Taubert (Eds.), Wissenschaftliches Publizieren – Zwischen Digitalisierung, Leistungsmessung, Ökonomisierung und medialer Beobachtung (pp. 263–270). Berlin, Germany: De Gruyter. Kriegeskorte, N., Walther, A., & Deca, D. (2012). An emerging consensus for open evaluation: 18 Visions for the future of scientific publishing. Frontiers in Computational Neuroscience, 6, 1–5. doi: 10.3389/fncom.2012.00094 Meiser, T. (2011). Experimental Psychology: A place for innovative research and methodological developments. Experimental Psychology, 58, 1–3. doi: 10.1027/1618-3169/a000105 Moeller, B., & Frings, C. (2014). How automatic is the musical stroop effect? Experimental Psychology, 61, 68–70. doi: 10.1027/1618-3169/a000204 Morey, R. D., Chambers, C. D., Etchells, P. J., Harris, C. R., Hoekstra, R., Lakens, D., . . . Zwaan, R. A. (2016). The peer reviewers’ openness initiative: Incentivizing open research practices through peer review. Royal Society Open Science, 3, 150547. doi: 10.1098/rsos.150547

Experimental Psychology (2017), 64(1), 1–4

Schönbrodt, F., Gollwitzer, M., & Abele-Brehm, A. (2017). Der Umgang mit Forschungsdaten im Fach Psychologie: Konkretisierung der DFG-Leitlinien [Data management in psychological science: Specification of the DFG guidelines]. Psychologische Rundschau, 68, 20–35. doi: 10.1026/00333042/a000341 Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). Falsepositive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22, 1359–1366. doi: 10.1177/ 0956797611417632 Zakay, D. (2014). Can the “Musical Stroop” task replace the classical stroop task? Experimental Psychology, 61, 78–79. doi: 10.1027/1618-3169/a000211

Experimental Psychology (2017), 64(1), 1–4

Editorial

Christoph Stahl Psychologische Methodenlehre & Experimentelle Psychologie Department für Psychologie Universität zu Köln Herbert-Lewin-Str. 2 50931 Köln Germany christoph.stahl@uni-koeln.de

Ó 2017 Hogrefe Publishing

Research Article

Gestalt Effects in Visual Working Memory Whole-Part Similarity Works, Symmetry Does Not Patrycja Kałamała,1,2 Aleksandra Sadowska,2 Wawrzyniec Ordziniak,2 and Adam Chuderski2 1

Institute of Psychology, Jagiellonian University in Krakow, Poland

Institute of Philosophy, Jagiellonian University in Krakow, Poland

Abstract: Four experiments investigated whether conforming to Gestalt principles, well known to drive visual perception, also facilitates the active maintenance of information in visual working memory (VWM). We used the change detection task, which required the memorization of visual patterns composed of several shapes. We observed no effects of symmetry of visual patterns on VWM performance. However, there was a moderate positive effect when a particular shape that was probed matched the shape of the whole pattern (the whole-part similarity effect). Data support the models assuming that VWM encodes not only particular objects of the perceptual scene but also the spatial relations between them (the ensemble representation). The ensemble representation may prime objects similar to its shape and thereby boost access to them. In contrast, the null effect of symmetry relates the fact that this very feature of an ensemble does not yield any useful additional information for VWM. Keywords: Gestalt, visual working memory, change detection

The last 40 years of research has yielded substantial knowledge on the key role for human cognition of working memory (WM), a cognitive mechanism responsible for the active maintenance and manipulation of information within the current task (Baddeley & Hitch, 1974; Cowan, 2001). WM is involved in perception, language, long-term memory access, cognitive control, problem solving, reasoning, and many others. Substantial research has been devoted to one particular type of WM, visual working memory (VWM; Brady, Konkle, & Alvarez, 2011; Pashler, 1988; Phillips, 1974). Models of VWM generally assume that it maintains visual representations of objects (Luck & Vogel, 1997), features defining these objects (Alvarez & Cavanagh, 2004), and spatial relations among them (Clevenger & Hummel, 2014). The main function of VWM is to support continuity of perception, by integrating features of external objects into a coherent representation of these objects and the visual scene constituted by them (Brady et al., 2011; Rensink, 2000; Treisman, 1988). The key paradigm to investigate VWM is the change detection (or visual arrays) task (Luck & Vogel, 1997; Pashler, 1988). In each trial of this task, a visual array, filled with a certain number of objects (each possessing one or more distinctive features, such as shape, color, orientation, etc.), is shown briefly. Next, the array is replaced by a mask. Finally, it reappears either in an identical form or with one (or more) objects changed. The instruction is to memorize Ó 2017 Hogrefe Publishing

the first array (source) and then to decide whether the second array (target) differs from the source or not. The crucial fact about VWM is its very limited capacity with regard to the number of actively maintained objects (usually less than four objects; Luck & Vogel, 1997). However, the issue as to what the real building blocks of VWM capacity are (the very objects, their features, or some combination of them) has been debated (Alvarez & Cavanagh, 2004; Bays & Husain, 2008; Hardman & Cowan, 2015; Oberauer & Eichenberger, 2013; Vogel, Woodman, & Luck, 2001; Zhang & Luck, 2008). Recent research suggests that each factor may count, depending on the task properties and the individual’s preferences (Vergauwe & Cowan, 2015). Such a view of VWM capacity is compatible with studies that differentiate the VWM subsystem for maintaining individual features, located within the superior parietal lobule, from the subsystem responsible for binding the features into complete objects, which takes place within the inferior parietal lobule (Xu & Chun, 2009). Studies on VWM capacity have also shown that its actual value can be influenced by the global organization of perceptual scene. Specifically, particular VWM objects may be stored in some relation to other items, yielding substantial contextual effects (Brady et al., 2011). When the context of probed items (e.g., surrounding objects) changes or disappears between the source and the target, Experimental Psychology (2017), 64(1), 5–13 DOI: 10.1027/1618-3169/a000346

retrieval is worse, compared to when the context is constant (Jiang, Olson, & Chun, 2000). Moreover, the statistical distribution of features is important for retrieval. For instance, it is easier to reject a false probe if its features differ substantially from the dominant features in a scene (e.g., it is easier to reject a new cold-color probe in a warm-color scene; Brady et al., 2011). Context also influences how we recall individual items, as the recall of items that possess the extreme value of a particular visual feature is often biased toward the average value of that feature in a display (the above cold-color probe will be judged more warm-colored in the warm-color scene; Brady & Alvarez, 2011). All of these context effects suggest that people do not only encode particular items in VWM, but that they also (or – even – primarily) represent their ensembles, compressing redundant information from a display into the concise but very informative higher-level description of an ensemble. Such a description can then be used to predict features of individual items, enhancing the actual VWM capacity (Alvarez, 2011). In consequence, it seems that VWM represents information on multiple, hierarchical levels (e.g., the level of features, objects, and ensembles). One particularly interesting type of context effect in VWM regards the satisfying (or not) of the Gestalt principles of perceptual organization (Wertheimer, 1923/1938), which hold that people tend to see perceptual patterns as “wholes” rather than as separated, unrelated elements. According to the Gestalt approach, people use several rules or “laws” in order to group perceived items into a whole (Laws of Proximity, Similarity, Closure, Symmetry, Continuity, Integrity, etc.). Studies have demonstrated that satisfying such principles not only helps in perceiving visual objects in a particular way, but also facilitates their retrieval from VWM. That is, Gestalt principles “work” even when objects of interest are not perceptually accessible. For instance, objects grouped by proximity to a cued object were more likely to be reported than distant objects (Woodman, Vecera, & Luck, 2003). Moreover, the overall number of reported objects was larger if they were grouped than when they were separated (Xu & Chun, 2007). Analogous results were noted for grouping by similarity (Peterson & Berryhill, 2013). Another example of Gestalt effects pertains to the facilitating role of the symmetry of the layout of objects for their recall from VWM. Kemps (2001), using the Corsi blocks test (the task that requires subjects to manually tap objects in a 5 5 matrix in the same sequence as they were previously highlighted), has demonstrated that recall was better when the sequence was spatially symmetrical than when it was not. This result was later replicated by Rossi-Arnaud, Pieroni, and Baddeley (2006), who additionally showed that symmetry along the vertical axis was more effective than along the horizontal and diagonal axes. The three types of symmetry improved Experimental Psychology (2017), 64(1), 5–13

P. Kałamała et al., Gestalt in VWM

recall as long as the target items were highlighted simultaneously (as this facilitated symmetry detection). Again, all these results suggest that VWM contents are globally and hierarchically structured.

Overview of the Study The VWM effect of grouping by proximity and similarity, investigated by means of the change detection task, seems to be solid (Peterson & Berryhill, 2013; Woodman et al., 2003; Xu & Chun, 2007). However, existing studies on the symmetry effect used other paradigms than the change detection task, in particular, the Corsi blocks test, which has a pronounced motor component (sequential hand movements). Moreover, these latter studies were limited to memory for spatial locations and have not tested memory for objects. For some reason, symmetrical movement sequences might be easier to perform than asymmetrical ones, but the effect may have little in common with the VWM functioning. Thus, our first aim was to examine the symmetry effect within the standard change detection paradigm, which would require memorization of distinct visual objects (instead of bare locations) and simple yes/ no responding. To foretell the results, we found no effect of symmetry of the stimuli pattern (Experiments 1, 2, and 4). Our second aim was, thus, to look for other Gestalt effects that would “work” for VWM. We investigated the similarity effects beyond the simple grouping (as in Peterson & Berryhill, 2013) by testing the “whole-part” effect of the similarity between the shape of the stimulus probed and the shape of the pattern of stimuli (Experiments 3 and 4). In contrast to symmetry, we found a facilitating effect of the whole-part similarity on object recognition in VWM, supporting the hierarchical nature of VWM representations.

Experiment 1 Participants A total of 24 women and 18 men participated (42 people). All of them were recruited via emails or adverts on social networking webpages. Mean age was 23.0 years (SD = 5.9, range 18–46). For a one hour participation, each person received an equivalent of €5 in local currency. Each person had normal or corrected-to-normal vision and no history of neurological problems. Each filled a written consent to participate and was informed that she or he could stop and leave the laboratory at will. Participants were tested in a cognitive psychology laboratory in groups of a few people, under the supervision of an experimenter. Ó 2017 Hogrefe Publishing

P. Kałamała et al., Gestalt in VWM

Materials and Procedure Each trial of the change detection task consisted of a virtual array filled with nine stimuli (only some cells in the array were filled). The stimuli were presented in black, on a light gray background. They were selected from 18 available figures (a square, circle, rhombus, cross, triangle, teardrop, star, tetragon, trapeze, heart, mast, arrow, hexagon 1, hexagon 2, T, thunderbolt, cloud, blur), each approximately 3 3 cm in size (3.4° of the visual angle). The total size of the array was 10 10 cm (11.3° of the visual angle). Each trial began with the presentation of a fixation point in the center of the screen presented for 2 s. Then, the source array was presented for 2 s, and followed by a mask of the same size as the array, presented for 0.8 s. Finally, in a randomly selected 50% of trials, a target array was presented that was identical to the source, while in the remaining trials a target array was shown which differed by exactly one item at one position, compared to the source. If both the source and target arrays differed, then the new item was highlighted by a square gray border. If the arrays were identical, a random item was highlighted. Pressing one of two response keys was required, depending on whether the highlighted item differed (the “change” response) or not (the “no-change” response) between the source and target array. The target array was shown for 2 s. A response time window started at the target array presentation and ended when a participant responded or 8 s elapsed. The trials were self-paced. The PsychoPy code and stimuli for all four experiments presented in this paper can be downloaded from an open repository (https:// dl.dropboxusercontent.com/u/21397852/Gestalt_VWM_ experiments.zip). The score on this task is the estimated sheer capacity of VWM (Cowan, 2001; Rouder, Morey, Morey, & Cowan, 2011), which is based on the proportion of hits (H, the change responses for arrays with one item changed) and the proportion of false alarms (FA, the change responses for unchanged arrays). The capacity of VWM is estimated to be k items (out of N items of a memory load), on the assumption that a participant elicits a correct hit or avoids a false alarm only if a cued item is transferred to his or her VWM (with the k/N chance). If a non-transferred item is cued, then a participant is assumed to be guessing the answer. In consequence, the following formula evaluates the score on the task for each N: k = N (H – FA). In the present experiment, the total score of each participant was thus k = 9 (H – FA), and it was an estimate of how many items this participant had successfully memorized in VWM. Note that such a measure effectively corrects for response bias (i.e., a tendency of an individual for making only one type of response, either the change or no-change ones).

Ó 2017 Hogrefe Publishing

The sole independent variable was whether the stimuli in both arrays formed either a vertically and horizontally symmetrical or an asymmetrical pattern. The symmetrical patterns were a cross and a circle, while a trapeze and a tetragon made the asymmetrical patterns. See Figure 1 for an illustration of the sequence of events in the change as well as no-change trials of the symmetrical, as well as the asymmetrical, condition. In total, there were 8 training trials and as many as 128 experimental trials (64 trials per each condition), randomly intermixed.

Results and Discussion The raw data for the following analyses is provided in Electronic Supplementary Material 1. The mean k value was M = 2.74 (SD = 1.40), matching other studies that tested VWM capacity for shapes (e.g., Alvarez & Cavanagh, 2004; Chuderski, 2015). The individual k values ranged from 0 items (indicating the random performance) to 5.91 items (indicating the top performance). Most importantly, we found no significant difference in k values between the symmetrical (2.84 items, 95% CI [2.09, 3.59]) and the asymmetrical condition (2.64 items, 95% CI [1.89, 3.39]), t(41) = 0.54, p = .593. The Bayesian paired test of means difference (with the Cauchy prior width set to 1) yielded Bayes Factor of 5.13, suggesting moderate evidence in favor of the hypothesis stating no difference in the mean k between both conditions, compared to the hypothesis assuming a larger k value in the symmetrical condition. These data indicated that the symmetry of the stimuli pattern did not facilitate the maintenance of that pattern in VWM. However, one possible shortcoming of Experiment 1 was that too many stimuli were presented, in comparison to the mean estimated capacity of about three objects, and, thus, the symmetry of the source array could only be perceived at the encoding (i.e., when it was being presented), but not during the maintenance (i.e., after it disappeared). Although nine stimuli allowed for a welldefined symmetry of global shapes, in order to rule out the symmetry effect more definitely, this effect should also be tested using lower memory load. This was the goal of Experiment 2. Also, we doubled the sample in order to decrease the probability of type II error.

Experiment 2 Participants A total of 69 women and 31 men participated (100 people). All of them were also recruited via emails or adverts on

Experimental Psychology (2017), 64(1), 5–13

P. Kałamała et al., Gestalt in VWM

SYMMETRICAL CONDITION CHANGE TRIAL

Figure 1. Example sequence of events in the symmetrical versus asymmetrical condition of Experiment 1. The source array, presented for 2 s, was replaced by the mask of the same size for 0.8 s, and then the mask was replaced by the target array. In the change trials, one object in the source array was substituted with another object in the target array and highlighted with the gray border. In the no-change trials, the arrays were identical, and the random object was highlighted.

ASYMMETRICAL CONDITION NO-CHANGE TRIAL

0.8 s

until response or 6 s time

social networking webpages. Mean age was 24.3 years (SD = 5.8, range 18–41). Testing conditions and gratification were the same as in Experiment 1.

Materials and Procedure Experiment 2 had the same design as Experiment 1, with the same set of black figures presented on a gray background, 50% of changed and 50% of unchanged target arrays, the array size of 10 10 cm (11.3° of the visual angle), 2 s of the source array presentation, 0.8 s of the mask, and 8 s of the limit for responding. One exception was that 48 symmetrical and 48 asymmetrical trials were applied (together with eight training trials). Another exception was that either five or six stimuli (on random) were presented in each array. The symmetrical patterns consisted of a cross, plus, rectangle and letter T shapes, whereas the asymmetrical patterns consisted of either five or six elements scattered randomly throughout the background. Figure 2 illustrates the trials in both the symmetrical and asymmetrical condition of the task. The total score of each participant was her or his k = (5 [H5 FA5] + 6 [H6 FA6])/2, where bottom indices refer to the hits and false alarm rates in either set size 5 or set size 6.

Results and Discussion The raw data for the following analyses is provided in Electronic Supplementary Material 2. The mean k value in Experiment 2 was M = 2.66 (SD = 1.03, range 0.34 Experimental Psychology (2017), 64(1), 5–13

to 4.81). Again, the symmetrical condition yielded the k value (M = 2.70, 95% CI [2.49, 3.01]) that was not significantly larger than the k value in the asymmetrical condition (M = 2.62, 95% CI [2.41, 2.83]), t(99) = 0.76, p = .447). Bayes Factor equaling 6.12 indicated moderate evidence in favor of the hypothesis assuming no difference in k between the conditions, compared to the hypothesis assuming a larger k value in the symmetrical condition. The present experiment replicated, with increased power and improved design, the results of previous experiment, and both studies suggest that, at least within the change detection paradigm, symmetry of the overall shape of the stimuli pattern does not facilitate the encoding/recall of the particular objects from VWM. Indeed, it seems that previous positive results on the symmetry effect in the Corsi blocks test might have simply been caused by the substantial motor requirements of this task, but that they did not result from a better organization/compression of symmetrical pieces of information in VWM. On further reflection, there seems to be little reason for the symmetry to work. Seeing that a stimuli pattern is symmetrical as a whole does not yield any additional information about the probability of particular objects to occur in that pattern (the stimuli on the left side of the pattern are always completely different from the ones on its right side). In contrast, in the symmetrical spatial Corsi blocks, encoding one side of the pattern gives perfect information about what can be expected on the opposite side (i.e., a mirror image). Thus, once the spatial task becomes an object/feature task, and the motor component is eliminated, the lack of symmetry effect seems to be justified. Ó 2017 Hogrefe Publishing

P. Kałamała et al., Gestalt in VWM

SYMMETRICAL CONDITION CHANGE TRIAL

Figure 2. Example sequence of events in the change trials of the symmetrical versus asymmetrical condition of Experiment 2. The source array, presented for 2 s, was replaced by the mask of the same size for 0.8 s, and then the mask was replaced by the target array. In the change trials, one object in the source array was substituted with another object in the target array and highlighted with the gray border. In the no-change trials, the arrays were identical, and the random object was highlighted.

ASYMMETRICAL CONDITION NO-CHANGE TRIAL

0.8 s

until response or 6 s time

In contrast, the grouping by proximity/similarity, observed in previous change detection studies, probably worked due to a higher probability of encoding objects grouped around some target item, for instance, because more attention was paid to such items, or because more two-object bindings (e.g., “object A is a close left-side neighbour of the target item”) could be set in VWM. In two subsequent experiments, we aimed to improve our understanding of the similarity effects, by extending them to whole-part similarity. Specifically, we assumed that when the shape of one of the stimuli matched the shape of the overall pattern of stimuli, the latter pattern could somehow prime the encoding of the former stimulus, and that the stimulus would, thus, be more correctly recognized in VWM, in comparison to stimuli that were not primed by the whole-part similarity.

Experiment 3 Participants A total of 34 women and 26 men participated (60 people). All of them were recruited via emails or adverts on social networking webpages. Mean age was 22.5 years (SD = 5.3, range 18–46). Testing conditions and gratification were the same as in Experiments 1–3.

Materials and Procedure The same task was used as in Experiment 1, again with the same set of black figures, gray background, set size of 5 or 6 Ó 2017 Hogrefe Publishing

(on random), 50% of changed and 50% of unchanged target arrays, array size of 10 10 cm (11.3° of the visual angle), 2 s of the source array presentation, 0.8 s of the mask, and 8 s of the limit for responding. The sole independent variable was whether an item from a to-be-highlighted location in the source array had the same shape as the complete pattern of stimuli in the array. For example, the stimuli could form an X pattern, and the target could be either an X figure (the similar condition) or another figure (the dissimilar condition). See Figure 3 for illustration of the sequence of events in the change trials of the similar and the dissimilar condition.

Results and Discussion The raw data for the following analyses is provided in Electronic Supplementary Material 3. The mean k value equaled M = 2.60 (SD = 0.99, range 0.11 to 4.24). Most importantly, the similar condition yielded a significantly larger k value (M = 2.74, 95% CI [2.45, 3.02]) than the dissimilar condition (M = 2.47, 95% CI [2.19, 2.75]), t(59) = 2.28, p = .030, Bayes Factor = 2.24 in favor of the hypothesis assuming a larger mean k value in the similar condition, compared to the hypothesis assuming no difference in k between the conditions. However, the effect was weak, as indicated by Cohen’s d = .25. The results indicated that, although, on average, participants were able to effectively hold in their WM about two and a half objects, the cases of similarity between the target stimulus and the overall pattern of stimuli were most likely to be detected successfully, and, on average, it increased VWM capacity by a quarter object ( 10% of mean k).

Experimental Psychology (2017), 64(1), 5–13

P. Kałamała et al., Gestalt in VWM

SIMILAR CONDITION CHANGE TRIAL

DISSIMILAR CONDITION NO-CHANGE TRIAL

0.8 s

until response or 6 s

Figure 3. Example sequence of events in the change trials of the similar versus dissimilar condition of Experiment 3. The source array, presented for 2 s, was replaced by the mask of the same size for 0.8 s, and then the mask was replaced by the target array. In the change trials, one object in the source array was substituted with another object in the target array and highlighted with the gray border. In the no-change trials, the arrays were identical, and the random object was highlighted. In the change trial, a small gray arrow indicates the matching item that is probed (note that no arrows were present in the actual task).

time

The present experiment provides the first evidence for the Gestalt-like effect of similarity between the pattern of stimuli held in VWM and the shape of a particular stimulus (i.e., the whole-part similarity). Such a similarity increased the likelihood of effectively encoding/retrieving that stimulus in/from VWM. This fact suggests that people encoded not only single objects, but also some ensemble representation of the pattern constituted by these objects. The detection of the shape of a whole might have primed the encoding of the single shape in question. However, the observed effect of the whole-part similarity was only moderate in size. In Experiment 4, thus, we aimed to replicate the effect of similarity, as well as to compare it to the supposedly null effect of symmetry (i.e., we also aimed to replicate data from Experiments 1 and 2). This time, instead of testing each effect in a separate sample of participants, we examined both effects in parallel, in the within-subjects design.

Experiment 4 Participants A total of 27 women and 14 men participated (41 people). All of them were recruited via emails or adverts on social networking webpages. Mean age was 26.7 years (SD = 8.2, range 18–49). Testing conditions and gratification were the same as in Experiments 1–3.

Materials and Procedure Experiment 4 combined the designs of Experiments 2 and 3, but used lower set sizes of 3, 4, and 5. Again, black Experimental Psychology (2017), 64(1), 5–13

figures, gray background, 50% of changed and 50% of unchanged target arrays, array size of 10 10 cm (11.3° of the visual angle), 2 s of the source array presentation, 0.8 s of the mask, and 8 s of the limit for responding, and eight training trials were applied. Moreover, the set of available stimuli was limited to an equilateral triangle, right-angled triangle, square, diamond, cross, and plus shapes. Finally, in addition to random asymmetric/dissimilar patterns, “regular” patterns were used, in which there were equal distances between neighboring stimuli. In consequence, there were four task conditions (96 trials each, 384 trials in total), defined according to an increasing level of the Gestalt principles’ satisfaction. In the no-Gestalt condition, the pattern of stimuli was irregular/random (unequal distances between the neighboring stimuli), asymmetrical, and the probed stimulus in the source array was always dissimilar to the stimuli pattern. In the regular condition, the pattern of stimuli included equal distances between neighboring objects, but was still asymmetrical, bearing no similarity to any of its stimuli. In the symmetrical condition, the pattern was also regular (as regularity was the sine qua non condition for symmetry to be possible) and symmetrical. All symmetrical patterns had an overall shape of one of the stimuli out of the set of available stimuli (i.e., the equilateral or right-angled triangle for set size 3, the square or diamond for set size 4, and the cross or plus for set size 5), but the presented pattern always had a different shape than the shape probed in the source array. Finally, the whole-part similar condition included regular, symmetrical patterns that were similar to the stimulus probed. With such a design, we could test whether each consecutive level of Gestalt principle satisfaction yields Ó 2017 Hogrefe Publishing

P. Kałamała et al., Gestalt in VWM

NO-GESTALT CONDITION

IRREGULAR ASYMMETRICAL DISSIMILAR

REGULAR CONDITION

REGULAR ASYMMETRICAL DISSIMILAR

SYMMETRICAL CONDITION

REGULAR SYMMETRICAL DISSIMILAR

SIMILAR CONDITION

REGULAR SYMMETRICAL WHOLE-PART SIMILAR

Figure 4. Schematic illustration of stimuli patterns used in four conditions of Experiment 4. The large arrow reflects an increasing organization in the stimuli pattern across the conditions. Small arrows indicate the items probed (note that no arrows were present in the actual task).

INCREASED LEVEL OF CONFORMING TO GESTALT PRINCIPLES

an increase in k values or not. The schematic illustration of all experimental conditions can be found in Figure 4.

Results and Discussion The raw data for the following analyses is provided in Electronic Supplementary Material 4. Table 1 presents descriptive statistics for all conditions of the change detection task. The data were submitted to one-way analysis of variance (ANOVA), with four levels of the experimental factor (no-Gestalt, regular, symmetrical, and the whole-part similar condition). The effect was highly significant, F(3, 120) = 11.0, p < .001, and strong, η2 = .22. The post hoc test (Tukey’s HSD) informed that the mean k value in the similar condition was significantly larger than mean k values in each of the remaining conditions (ps < .001 for the regular and symmetrical condition, and p = .002 for the no-Gestalt condition), with mean d = .39. Also Bayes Factors equaling 65.42, 52,248.0, and 154.2 showed very strong evidence in favor of the hypothesis claiming a larger mean k value in the similar condition than in the no-Gestalt, regular, and symmetrical condition, respectively, compared to the respective null hypotheses. There was no significant difference between any pair of the latter conditions (ps > .294, Bayes Factors in favor of each of the null hypotheses exceeded 1.74). Comparing to the alternative conditions, k values in the similar condition increased almost by one third of item (Δk = 0.31, 14.7%). The overall pattern of results is summarized in Figure 5.

General Discussion Our initial hypotheses assumed that Gestalt-like effect of symmetry, observed in the Corsi blocks test, could also facilitate the maintenance and retrieval of information from VWM. However, in Experiments 1 and 2 we did not observe evidence for the symmetry effect (the same occurred later in Experiment 4). Upon further analysis, we realized that, Ó 2017 Hogrefe Publishing

in order to facilitate storage in VWM, a Gestalt effect needs to yield some useful information about the objects maintained. Symmetry suffices this condition in the Corsi block task (the sequence of movements will be identical on both sides of the scene), but does not suffice it in the change detection paradigm (different objects will be placed at symmetrically corresponding locations). However, the whole-part similarity fulfills this condition in the change detection task, and, indeed, in Experiments 3 and 4 we observed an average increase of above ten percent in actual VWM capacity for probed objects whose shapes were similar to the shape of the overall pattern of stimuli. It should be noted that our conclusions pertaining to significance of whole-part similarity for VWM performance, alongside the nonexistence of symmetry effect in that regard, are limited only to the change detection paradigm we have used, as well as to the figural stimuli that were employed. For a stronger support for the whole-part effect in VWM, studies using other types of perceptual features (colors, size, etc.) are needed. Although the present experimental design was relatively simple, and the results cannot reveal the precise form of VWM representation which was constructed during the similar condition, several possibilities can be considered. First, the ensemble representation (including the information on the ensemble shape) could be maintained independently from the particular objects. In consequence, when a participant noticed the similarity between the to-behighlighted shape and the shape of the whole pattern, the information on the shape was represented in two layers in parallel. So, even after losing this information from the object layer, it could be restored from the ensemble layer. That fact might yield improved performance, compared to the dissimilar condition, when the ensemble layer could not be used in order to restore the object shape’s representation. Second possibility assumes some form of interaction between the lower- and the higher-level layer, for example match between the object’s shape and the overall shape, the latter represented at the ensemble level, might lead to the top-down spread of activation from that level, and to the boost in the object activation in VWM, compared Experimental Psychology (2017), 64(1), 5–13

P. Kałamała et al., Gestalt in VWM

Table 1. Descriptive statistics for k values observed in each condition of Experiment 4 All NoWholeconditions Gestalt Regular Symmetrical part similar Mean

2.18

2.16

2.04

2.12

0.76

0.82

0.74

0.79

0.86

Minimum

0.87

0.58

0.57

0.73

0.53

Maximum

3.61

3.58

3.51

3.76

3.67

Low 95% CI

1.94

1.90

1.80

1.87

2.15

High 95% CI

2.42

2.27

2.37

2.68

k value

2.4

2.42

n.s.

n.s. 2.2

n.s.

2.0

No-Gestalt

Regularity

Symmetry Whole-part similarity

Figure 5. Mean number of objects (k value) maintained in VWM in the four conditions of Experiment 4, and the results of significance tests between the pairs of conditions.

to the dissimilar condition. However, it is also possible that the whole-part similarity “worked” solely during the encoding phase (i.e., when the source array was still available perceptually). For example, noticing the similarity between one of the objects and their pattern could lead to a larger allocation of attention to that object (and, thus, to its more robust encoding), or such a similarity could yield even some simpler priming effects. Nevertheless, all these alternatives implicate VWM’s ability to access ensemble representations of the overall pattern of objects, either during encoding or during maintenance, or both (what seems most likely). The present results are important for the existing debate on the building blocks of VWM capacity. It seems that models which assume only one level of representation in VWM, be it slots filled with objects (Vogel et al., 2001; Zhang & Luck, 2008), attentional resources devoted to the features making up the objects (Alvarez & Cavanagh, 2004; Bays & Husain, 2008), or some combination of both (Hardman & Cowan, 2015; Oberauer & Eichenberger, 2013), may be insufficient to explain such between-level effects as the whole-part similarity effect observed in the present study. As the number of objects and the amount of features did not differ between the similar and dissimilar condition, the valid explanation of mechanisms underlying

Experimental Psychology (2017), 64(1), 5–13

limits in VWM capacity may lie beyond the slot versus resource view. The whole-part similarity effect rather suggests that some hierarchical and relational representation of perceptual scene is formed in VWM, and the features of such a higher-level layout do affect a participant’s ability to encode a particular number of objects of particular featural complexity (Alvarez, 2011; Clevenger & Hummel, 2014; Jiang et al., 2000; see Brady et al., 2011). In consequence, VWM might rely substantially on relational processing that is typical for more complex cognitive functions, like thinking and reasoning (Holyoak, 2012). However, more studies are necessary to investigate this issue. Concluding, the finding of Gestalt effects spanning from low-level perception (Wertheimer, 1923/1938) to high-level thinking and creativity (Walas, 1926) has been one of the hallmark results in psychology. The present work demonstrated another solid Gestalt-like effect pertaining to human working memory (in addition to grouping effects that were known to exist in VWM). Although our experiments do not allow for the univocal identification of a particular form of representation underlying the effect found, overall they support the view that the human mind represents information in memory at multiple interrelated levels and in various interdependent formats. Acknowledgments This work was sponsored by the National Science Centre of Poland (Grant No. 2015/17/B/HS6/04152 awarded to A. Chuderski). A preliminary analysis of Experiment 3 was presented during EuroAsianPacific Joint Conference on Cognitive, Torino, Italy, in September 2015. The PsychoPy code for experimental procedures can be obtained from the following link: https://dl.dropboxusercontent. com/u/21397852/Gestalt_VWM_experiments.zip, or from the corresponding author upon request (via email to: adam.chuderski@gmail.com). Electronic Supplemental Materials The electronic supplementary material is available with the online version of the article at http://dx.doi.org/10.1027/ 1618-3169/a000346 ESM 1. Data file (csv). The raw data for Experiment ESM 2. Data file (csv). The raw data for Experiment ESM 3. Data file (csv). The raw data for Experiment ESM 4. Data file (csv). The raw data for Experiment

1. 2. 3. 4.

Ó 2017 Hogrefe Publishing

P. Kałamała et al., Gestalt in VWM

References Alvarez, G. A. (2011). Representing multiple objects as an ensemble enhances visual cognition. Trends in Cognitive Sciences, 15, 122–131. doi: 10.1016/j.tics.2011.01.003 Alvarez, G. A., & Cavanagh, P. (2004). The capacity of visual shortterm memory is set both by information load and by number of objects. Psychological Science, 15, 106–111. doi: 10.1111/ j.0963-7214.2004.01502006.x Baddeley, A. D., & Hitch, G. (1974). Working memory. In G. H. Bower (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol. 8, pp. 47–89). New York, NY: Academic Press. Bays, P. M., & Husain, M. (2008). Dynamic shifts of limited working memory resources in human vision. Science, 321, 851. doi: 10.1126/science.1158023 Brady, T. F., & Alvarez, G. A. (2011). Hierarchical encoding in visual working memory: Ensemble statistics bias memory for individual items. Psychological Science, 22, 384–392. doi: 10.1177/ 0956797610397956 Brady, T. F., Konkle, T., & Alvarez, G. A. (2011). A review of visual working memory capacity. Beyond individual objects and towards structured representations. Journal of Vision, 11, 1–34. doi: 10.1167/11.5.4 Chuderski, A. (2015). The broad factor of working memory is virtually isomorphic to fluid intelligence tested under time pressure. Personality and Individual Differences, 85, 98–105. doi: 10.1016/j.paid.2015.04.046 Clevenger, P. E., & Hummel, J. E. (2014). Working memory for relations among objects. Attention, Perception, & Psychophysics, 76, 1933–1953. doi: 10.3758/s13414-013-0601-3 Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. The Behavioral and Brain Sciences, 24, 87–114. doi: 10.1177/ 2F0963721409359277 Hardman, K. O., & Cowan, N. (2015). Remembering complex objects in visual working memory: Do storage limits restrict objects or features? Journal of Experimental Psychology: Learning Memory and Cognition, 41(2), 325–347. doi: 10.1037/ xlm0000031 Holyoak, K. J. (2012). Analogy and relational reasoning. In K. J. Holyoak & R. G. Morrison (Eds.), The Oxford handbook of thinking and reasoning (pp. 234–259). New York, NY: Oxford University Press. Jiang, Y., Olson, I. R., & Chun, M. M. (2000). Organization of visualshort term memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 683–702. doi: 10.1037/ 0278-7393.26.3.683 Kemps, E. (2001). Complexity effects in visuo-spatial working memory: Implications for the role of long-term memory. Memory, 9, 13–27. doi: 10.1080/09658210042000012 Luck, S. J., & Vogel, E. K. (1997). The capacity of visual working memory for features and conjunctions. Nature, 390, 279–281. doi: 10.1038/36846 Oberauer, K., & Eichenberger, S. (2013). Visual working memory declines when more features must be remembered per object. Memory & Cognition, 41, 1212–1227. doi: 10.3758/s13421-0130333-6 Pashler, H. (1988). Familiarity and visual change detection. Perception & Psychophysics, 44, 369–378. doi: 10.3758/ s13423-013-0460-x

Ó 2017 Hogrefe Publishing

Peterson, D. J., & Berryhill, M. E. (2013). The Gestalt principle of similarity benefits visual working memory. Psychonomic Bulletin & Review, 20, 1282–1289. doi: 10.3758/s13423-0130460-x Phillips, W. A. (1974). On the distinction between sensory storage and short-term visual memory. Perception & Psychophysics, 16, 283–290. Rensink, R. A. (2000). The dynamic representation of scenes. Visual Cognition, 7, 17–42. doi: 10.1080/135062800394667 Rossi-Arnaud, C., Pieroni, L., & Baddeley, A. D. (2006). Symmetry and binding in visuo-spatial working memory. Neuroscience, 139, 393–400. doi: 10.1016/j.neuroscience.2005.10.048 Rouder, J. N., Morey, R. D., Morey, C. C., & Cowan, N. (2011). How to measure working memory capacity in the change detection paradigm. Psychonomic Bulletin & Review, 18(2), 324–330. doi: 10.3758/s13423-011-0055-3 Treisman, A. (1988). Features and objects: The 14th Bartlett Memorial Lecture. The Quarterly Journal of Experimental Psychology, 40, 201–237. doi: 10.1080/02724988843000104 Wallas, G. (1926). The art of thought. New York, NY: Harcourt, Brace. Wertheimer, M. (1938). Laws of organization in perceptual forms. In W. D. Ellis (Ed.), A source book of Gestalt psychology (pp. 71–94) (Originally published in Psychologische Forschung in 1923). London, UK: Routledge & Kegan Paul. Woodman, G. F., Vecera, S. P., & Luck, S. J. (2003). Perceptual organization influences visual working memory. Psychonomic Bulletin & Review, 10, 80–87. doi: 10.3758/BF03196470 Vergauwe, E., & Cowan, N. (2015). Working memory units are all in your head: Factors that influence whether features or objects are the favored units. Journal of Experimental Psychology: Learning, Memory, and Cognition, 41, 1404–1416. doi: 10.1037/ xlm0000108 Vogel, E. K., Woodman, G. F., & Luck, S. J. (2001). Storage of features, conjunctions, and objects in visual working memory. Journal of Experimental Psychology: Human Perception and Performance, 27, 92–114. doi: 10.1037/0096-1523.27.1.92 Xu, Y., & Chun, M. M. (2007). Visual grouping in human parietal cortex. Proceedings of the National Academy of Sciences of the United States of America, 104, 18766–18771. doi: 10.1073/ pnas.0705618104 Xu, Y., & Chun, M. M. (2009). Selecting and perceiving multiple visual objects. Trends in Cognitive Sciences, 13, 167–174. doi: 10.1016%2Fj.tics.2009.01.008 Zhang, W., & Luck, S. J. (2008). Discrete Fixed-Resolution Representations in Visual Working Memory. Nature, 453(7192), 233–235. doi: 10.1038/nature06860 Received April 17, 2016 Revision received September 14, 2016 Accepted November 8, 2016 Published online February 20, 2017 Adam Chuderski Institute of Philosophy Jagiellonian University in Krakow Grodzka 52 31-044 Krakow Poland adam.chuderski@uj.edu.pl

Experimental Psychology (2017), 64(1), 5–13

Research Article

The Role of Cognitive Load in Intentional Forgetting Using the Think/No-Think Task Saima Noreen1 and Jan W. de Fockert2 1

Faculty of Life and Health Services, De Montfort University, Leicester, UK

Goldsmiths, University of London, UK

Abstract: We investigated the role of cognitive control in intentional forgetting by manipulating working memory load during the think/no-think task. In two experiments, participants learned a series of cue-target word pairs and were asked to recall the target words associated with some cues or to avoid thinking about the target associated with other cues. In addition to this, participants also performed a modified version of the n-back task which required them to respond to the identity of a single target letter present in the currently presented cue word (n = 0 condition, low working memory load), and in either the previous cue word (n = 1 condition, high working memory load, Experiment 1) or the cue word presented two trials previously (n = 2 condition, high working memory load, Experiment 2). Participants’ memory for the target words was subsequently tested using same and novel independent probes. In both experiments it was found that although participants were successful at forgetting on both the same and independent-probe tests in the low working memory load condition, they were only successful at forgetting on the same-probe test in the high working memory load condition. We argue that our findings suggest that the high load working memory task diverted attention from direct suppression and acted as an interference-based strategy. Thus, when cognitive resources are limited participants can switch between the strategies they use to prevent unwanted memories from coming to mind. Keywords: memory inhibition, suppression, think/no-think, cognitive load, working memory load

Ever since early theorists made the link between our apparent ability to suppress upsetting memories and the potential consequence of doing so for our physical and mental wellbeing (James, 1950), researchers and clinicians have tried to establish empirically, the extent to which we possess executive control over memory (e.g., Glucksberg & King, 1967; Rosenzweig & Mason, 1934). More recently, Anderson and Green (2001) have arguably come closer than most in identifying the possible cognitive mechanisms underlying this kind of motivated forgetting. Using the Think/NoThink (TNT) procedure, Anderson and Green found that participants who had been trained to “not think” or suppress target words associated with particular cues that had previously been learned, remembered fewer suppressed words than controls which had received neither “think” nor “no-think” instructions. The decrement in recall performance for suppressed items (i.e., suppression-induced forgetting) has been demonstrated with a range of materials (e.g., Depue, Curran, & Banich, 2007; Noreen & MacLeod, 2013, 2014; Noreen, Bierman, & MacLeod, 2014; but see Bulevich, Roediger, Balota, & Butler, 2006) and taken by some as evidence of an inhibitory mechanism that can reduce the Experimental Psychology (2017), 64(1), 14–26 DOI: 10.1027/1618-3169/a000347

subsequent availability of a memory (e.g., Anderson et al., 2004; Bergström, de Fockert, & Richardson-Klavehn, 2009a, 2009b; see Anderson & Huddleston, 2012, for a review). According to this account, suppression is due to an executive control mechanism that inhibits the representation of the unwanted memory, deliberately impairing its retention and rendering it inaccessible (Anderson et al., 2004). The notion that memory suppression is an active process requiring cognitive control is supported by research which has demonstrated a suppression-induced forgetting effect when recall is tested with the original cue, and with a novel independent probe (i.e., semantic category plus the first letter of the target; Anderson & Green, 2001; Anderson et al., 2004) suggesting that the memory representation of the unwanted information has been inhibited (but see Camp, Pecher, Schmidt, & Zeelenberg, 2009, and Anderson & Huddleston, 2012, for a detailed discussion on the independent nature of independent cues). Further support for this account comes from neuroimaging studies which have found increased activation in fronto-parietal regions, including the right dorsolateral prefrontal cortex (DLPFC) and reduced hippocampal activation during Ó 2017 Hogrefe Publishing

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

memory suppression (e.g., Anderson et al., 2004; Benoit & Anderson, 2012; Benoit, Hulbert, Huddleston, & Anderson, 2015; Depue et al., 2007). Furthermore, effective connectivity analyses have shown a top-down modulatory influence of DLPFC on the hippocampus (Benoit & Anderson, 2012; Gagnepain, Henson, & Anderson, 2014), with negative coupling from the DLPFC predicting the size of the suppression-induced forgetting effect (Benoit & Anderson, 2012; Benoit et al., 2015; Depue, Orr, Smolker, Naaz, & Banich, 2015). The finding that strong engagement of control-related brain regions and the degree of hippocampal activity during memory suppression are related to the size of the forgetting effect suggests that individuals can regulate activity in the hippocampus and disengage from conscious recollection which disrupts later memory performance. Additional support for the notion that memory suppression involves cognitive control comes from research with attention deficit hyperactivity disorder (ADHD) individuals who show significantly impaired suppression-induced forgetting, reduced engagement of the DLPFC, and diminished regulation of the hippocampus (Depue, Burgess, Willcutt, Ruzic, & Banich, 2010). Research has also found that the variability between individuals in the magnitude of the suppression-induced forgetting effect maps onto known individual age-related differences in cognitive control efficiency (Levy & Anderson 2008). For example, children and older adult groups that have previously been associated with reduced cognitive control ability have also shown a reduced suppression-induced forgetting effect (Paz-Alonso, Ghetti, Matlen, Anderson, & Bunge, 2009). Furthermore, the forgetting effect is greater following advance warnings of no-think trials (Hanslmayr, Leipold, & Bauml, 2010), suggesting the effect is under active strategic control. Some researchers have further suggested that inhibitory processes involved in this type of memory suppression might be analogous to inhibitory processes involved in motor response inhibition (e.g., Garavan, Ross, Murphy, Roche, & Stein, 2002; Menon, Adleman, White, Glover, & Reiss, 2001), such that the same basic mechanism may be involved in the control of both memory and behavior (e.g., Anderson, 2005; Anderson & Huddleston, 2012; Anderson & Weaver, 2009; Levy & Anderson, 2008). Electroencephalogram (EEG) research, for example, has revealed electrophysiological indices of cognitive control during memory suppression, including an enhanced frontal N2 component that was considerably greater for suppressed items that are later successfully forgotten (Mecklinger, Parra, & Waldhauser, 2009). The N2 has also been implicated in the inhibition of motor responses (Kopp, Rist, & Mattler, 1996). Furthermore, research has also found a significant relationship between measures of memory and Ó 2017 Hogrefe Publishing

behavioral inhibition, such as the stop signal task (Depue et al., 2010; but see Noreen & MacLeod, 2015). Although there is considerable support for the involvement of an inhibitory mechanism in memory suppression, there is also some research which suggests that forgetting effects may be due to a non-inhibitory mechanism, such as associative interference. Associative interference involves creating new associations with the cue word in order to avoid thinking about the target word. Bergström et al. (2009b), for example, had participants think of a substitute memory in order to help them “not think” about the original target or undergo a standard direct suppression condition. They found that while participants showed a forgetting effect on the same-probe test for both the direct suppression and thought substitution conditions, only individuals in the direct suppression condition demonstrated a forgetting effect on the independent-probe test. These findings suggest that there is a dissociation between inhibitory and non-inhibitory forgetting. More recently, however, an inhibitory account of thought substitution has also been put forward with research suggesting that when two items in memory are associated with the same cue they compete with each other for retrieval, and inhibition is recruited in order to resolve the competition and suppress the unwanted target memory. In line with this, research has found that participants can also show successful forgetting on both the same and independent-probe recall tests using a thought substitution strategy under conditions in which sufficient competition is created (Benoit & Anderson, 2012; Del Prete, Hanczakowski, Bajo, & Mazzoni, 2015). Taken together, these findings suggest that, while forgetting on both same-probe and independent-probe tests can be due to an inhibitory or a non-inhibitory mechanism, forgetting on same-probe tests but not independent-probe tests is more readily explained in terms of a non-inhibitory mechanism. The notion that we can forget by retrieving alternative memories is extraordinarily similar to the widely documented effect of retrieval-induced forgetting (RIF; Anderson, Bjork, & Bjork, 1994). Retrieval-induced forgetting has been studied using the selective retrieval-practice paradigm (Anderson et al., 1994) and refers to the notion that retrieving some information subsequently induces forgetting of non-retrieved competing information. Research has found retrieval-induced forgetting effects using both the same and independent cue technique providing support for an inhibitory account of forgetting (e.g., Anderson & Spellman, 1995; E.L. Bjork, Bjork, & MacLeod, 2006; MacLeod, 2002; MacLeod & Macrae, 2001; Shaw, Bjork, & Handal, 1995). Given the similarities between the thought substitution strategy of the TNT task and the retrieval-practice task, more recent research has attempted to distinguish between Experimental Psychology (2017), 64(1), 14–26

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

the inhibitory and interference accounts in thought substitution. Wang, Cao, Zhu, Cai, and Wu (2015) used a double-cue paradigm in which two cues were associated with the target during memory formation. Participants were then presented with one cue and underwent interference or suppression training. In the interference-based training, participants were presented with the substitute words. In the final test, both cues were used to retrieve the target memory. Wang and colleagues found that while interference caused memory impairment restricted to the cuetarget association, suppression-induced forgetting was observed with the independent cue-target association. Given that no retrieval was recruited in the interference condition of the study, these findings provide strong support that suppression and interference produce qualitatively different effects, with suppression inhibiting the target memory from conscious awareness. Currently, much of the existing research has indirectly investigated the role of cognitive control functions in intentional forgetting. We therefore decided to adopt a more direct approach and examine the impact of manipulating cognitive control on suppression. Specifically, we aimed to measure the forgetting effect under varying conditions of concurrent cognitive load. Previous research has found that efficiency of inhibitory control is related to working memory capacity (WMC), such that individuals with high WMC are significantly better at inhibiting taskirrelevant information compared to those with lower WMC (Redick, Heitz, & Engle, 2007). In line with this, measures of WMC are found to predict performance on tasks requiring controlled inhibition, such as the Stroop task (Kane & Engle, 2003), directed forgetting task (Aslan, Zellner, & Bauml, 2010), and the retrieval-practice task (Ortega, Gomez-Ariza, Roman, & Bajo, 2012). For example, research has found that while individuals may exhibit retrieval-induced forgetting effects in typical conditions, when retrieval practice is undertaken concurrently with a task that requires executive control resources, participants show impaired forgetting (Ortega et al., 2012; Roman, Soriano, Gomez-Ariza, & Bajo, 2009). We report two experiments in which we tested participants’ ability to intentionally suppress retrieval of unwanted memories in the TNT paradigm under varying conditions of concurrent cognitive load. In both experiments, participants learned a series of cue-target word pairs and were told to recall the target words associated with some cues (i.e., respond condition), and “not think” about the target word associated with other cues (i.e., suppress condition), as well as perform a modified version of the n-back task. The n-back task required participants to respond to the identity of a single target letter, present in the currently presented cue word (0 = back condition, low working memory load), or in either the previous cue word (1 = back condition, high Experimental Psychology (2017), 64(1), 14–26

working memory load, Experiment 1), or the cue word presented two trials previously (2 = back condition, high working memory load, Experiment 2). Thus, whereas in the high working memory load condition participants had to maintain the identity of the target letter, no such requirement was made in the low load condition. We used this task as a manipulation of cognitive control, since working memory is regarded as a key component of the cognitive control functions supported by the frontal lobes (De Fockert, 2013; Lavie, Hirst, de Fockert, & Viding, 2004). Our principal effect of interest was the magnitude of the forgetting effect on the same-probe and independent-probe tests as a function of working memory load during the TNT phase. We predicted that, if forgetting involves active cognitive control, then the forgetting effect should be greater during low working memory load, when cognitive control will be available to prevent processing of the learned associate in the “no-think” condition. The high working memory load task may compete for cognitive control resources with the process of memory suppression, leading to a reduction in the forgetting effect under high working memory load. Importantly, the load effect should be present on both the same-probe and the independentprobe tests, as the impairment of inhibition by load should leave the no-think item representation more highly activated under high working memory load.

Experiment 1 Method Participants Thirty-two healthy never-depressed undergraduate students (12 M; 20 F) from Goldsmiths, University of London (mean age = 22.0 years; SD = 3.4) volunteered to take part in the study in exchange for course credit. All participants were screened for depression using a screening questionnaire and the Beck Depression Inventory-II (Beck, Steer, & Brown, 1996). Ethical approval was obtained from Goldsmiths Research Ethics Committee. Materials The TNT task comprised 45 neutral noun-noun word pairs taken from Anderson and Green (2001). Word pairs were divided into five sets of nine words, with one set each assigned to the respond high load, respond low load, suppress high load, suppress low load, and a baseline condition. The assignment of each word pair set was fully counterbalanced across all participants. In addition, eight additional noun-noun pairs (also from Anderson & Green, 2001) were used for the practice TNT phase. Ó 2017 Hogrefe Publishing

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

Thought Intrusions Questionnaire Participants were asked to rate the degree to which they focused on looking at the word as it appeared on the screen (1 = never to 5 = all the time); how difficult they found it to “not think” about the original memory associated with the word (1 = not difficult at all to 5 = very difficult); how often the (original) associated memory came to mind (1 = never came to mind to 5 = always came to mind); how often other thoughts unrelated to the task came to mind when the cue was presented (1 = never came to mind to 5 = always came to mind); the extent to which they tried to actively stop the associated target word from coming to mind when they saw the cue word (1 = tried all the time to 5 = did not try at all) and the extent to which they focused their attention on the black letter to avoid focusing on the word (1 = never to 5 = all the time). Procedure Learning Phase Participants were initially presented with each word pair for 5,000 ms and were asked to study each word pair as they would be tested on these word pairs after the whole list had been presented. This was then followed by a 500 ms intertrial interval. All 53 word pairs were presented. Cued Recall Phase Following the presentation of the word pairs, participants were given a cued recall test. The recall test consisted of each cue word being presented for 4,000 ms. During this time, participants were asked to recall the associated target word. Following a delay of 300 ms participants were provided with feedback (i.e., the correct response) for 1,000 ms. This was then followed by an intertrial interval of 300 ms. In line with Anderson and Green (2001) participants were required to achieve 50% accuracy on the recall test in order to advance to the next stage of the experiment.1 Think/No-Think Practice Phase See Figure 1 for an example of TNT trials. Participants were informed that they would see some of the cue words, all of which they had seen previously. The cue words were presented in either a green or a red font. In addition, participants were informed that for each of the green or red cues, they would see one letter in black (i.e., the letter “tape” in red with the letter “e” in black). The letters in black were always one of five vowels (i.e., “i, o, u, e, a”) and differed during trials, such that no cue word always contained the same letter in black. Participants were told that for the 1

green cue words (i.e., “respond” condition) their task was to recall the associated word out loud. In line with the direct suppression instructions by Bergström et al. (2009b) and Benoit and Anderson (2012), for the red cues, participants were told to block the memory of the corresponding target word from coming to mind, but not by replacing it with other memories or thoughts (i.e., “suppress” condition). Low Load Think/No-Think Condition In the low load condition, participants were instructed to press the letter on the keyboard that corresponded to the letter in black for both respond and suppress cue words. Participants were told that they could press the key that corresponded to the black letter at any time during the 4 s the cue word was on the screen. For example, if the cue word was “dough” in green font with the letter “u” in black font, participants would press the letter “u” during that trial. High Load Think/No-Think Condition For the high load condition participants were instructed to press the key that corresponded to the black letter presented one trial previously, regardless of whether it was in red or green font. Participants were told that they could press the key that corresponded to the black letter at any time during the 4 s the cue word was on the screen. For example, if the cue word was “relax” in green font with the letter “a” in black font on the previous trial, and the current trial presented the cue word “radio” in red font with the letter “i” in black font, participants would press the letter “a” on the current trial, and “i” on the next trial during the 4 s the cue word was presented. Participants were informed that it was very important that they try to recall or “not think” about the green and red cue words, respectively, while also complying with high and low load task instructions. In order to ensure that participants understood the procedure for the main think/no-think phase of the TNT task, participants were given a training phase for the high load and low load TNT conditions prior to completing the main phase.2 Think/No-Think Main Phase In the main TNT phase, participants were given a total of 576 trials, which included all respond and suppress words being presented 16 times each from the four sets of words. The 576 trials were split into four blocks of 144 trials. Two of the blocks were allocated to the high load condition, while the other two blocks were allocated to the low load

Participants were given three attempts to reach this criterion. If participants failed to reach this criterion within the permitted number of attempts the experimental procedure was terminated. In our study, however, all of the participants reached this criterion. For each of the high and low load conditions, two filler cue words appeared in green (eight times each) and two cue words appeared in red (eight times each), thereby resulting in 32 practice trials.

Ó 2017 Hogrefe Publishing

Experimental Psychology (2017), 64(1), 14–26

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

Figure 1. An example of low and high working memory load think/nothink trials. The high load example is from Experiment 1 and required participants to report the black letter from the previous trial (1back). In the high load condition in Experiment 2, participants reported the black letter from two trials before the current trial (2-back).

condition. The order of instructions was counterbalanced, such that half of the participants completed the low load blocks first, while the remaining participants completed the high load blocks first. In addition, we had nine items for each vowel in black font, with subsequent blocks being counterbalanced across respond, suppress, and baseline instructions. We also ensured that no same letters in black font were presented in two consecutive trials. Each trial began with a small fixation cross appearing in the center of the screen for 200 ms. Subsequently, a cue word appeared for 4,000 ms. On respond trials, participants were instructed to recall aloud the target word. Incorrect responses resulted in the correct target being displayed for 1,000 ms in blue. This was then followed by an intertrial interval of 400 ms before the next trial began. On a suppress trial, participants were required to withhold their response and to prevent the corresponding word from coming to mind. In addition, for the low load condition, participants were instructed to press the letter on the keyboard that corresponded to the letter in black for both respond and suppress cue words. In the high load condition, participants were instructed to press the letter that corresponded to the letter presented one trial previously. Final Recall Phase The final recall phase consisted of participants being tested for all of the target words by completing both the sameprobe and an independent-probe test. The order of test administration was counterbalanced. In the same-probe test, participants were presented with all 45 cue words and were asked to recall all the target words associated with every cue. Each trial began with a central fixation cross being displayed for 200 ms. Subsequently, a cue word was presented for 4,000 ms. This was followed by an intertrial interval of 400 ms before the next trial began. For the independent-probe test, participants were presented with the first letter and the semantic category of each target word and were informed that all the target words had been seen by them previously. Each trial began with a fixation cross being displayed for 200 ms. Subsequently, the first Experimental Psychology (2017), 64(1), 14–26

letter and the semantic category of the target word were presented for 4,000 ms. Participants were asked to recall the target word out loud. This was then followed by an intertrial interval of 400 ms before the next trial began. Finally, participants were given the thought intrusions questionnaire to complete.

Results Performance on the n-Back Task We first investigated accuracy on the n-back task, by conducting a 2 (Working Memory Load: High Load vs. Low Load) 2 (Instruction: Respond vs. Suppress) Analysis of Variance (ANOVA) on the n-back results. Our analysis showed a main effect of Working Memory Load, F(1, 30) = 6.34, p = .02, ηp2 = .170, with subsequent analyses confirming that, overall, participants’ performance was significantly better in the low than high load condition (M = 98.17, SEM = .30 vs. M = 90.69, SEM = 3.0, respectively). There was no main effect of Instruction, nor an Instruction by Working Memory Load interaction; F(1, 30) = 0.26, p = .61; F(1, 30) = 3.65, p = .07, respectively. Recall Accuracy at Final Test We computed the percentage of correctly recalled target words in the respond and suppress conditions for each participant, as a function of working memory load, as well as in the baseline condition, on both the same-probe and independent-probe tests. In order to analyze all the data in a single factorial design, we first calculated the instruction effect for each experimental condition by subtracting the baseline score from the respond and suppress scores for high and low load, respectively. These difference scores were entered into a 2 (Instruction: Respond vs. Suppress) 2 (Cognitive Load: High vs. Low) 2 (Probe Type: Same vs. Independent) ANOVA. Our analysis revealed a significant main effect of Instruction, F(1, 31) = 81.42, p < .01, ηp2 = .724, with participants overall showing a facilitation effect in the respond condition and a forgetting effect in the suppress condition (M = 11.1, SEM = 2.38 vs. Ó 2017 Hogrefe Publishing

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

M = 7.5, SEM = 2.43). There were no significant main effects of Working Memory Load or Probe Type; both tests F < 1. There were significant two-way interactions between Instruction and Probe Type, F(1, 31) = 9.65, p < .01, ηp2 = .237, and between Working Memory Load and Probe Type, F(1, 31) = 4.49, p = .04, ηp2 = .127. These two-way interactions were qualified by a significant three-way interaction between Instruction, Working Memory Load, and Probe Type, F(1, 31) = 4.37, p = .04, ηp2 = .123 (see Figure 2). Subsequent pairwise comparisons revealed that for the respond condition, there was no effect of working memory load on the size of the facilitation effect on the same-probe (high load M = 12.15, SEM = 2.46 vs. low load M = 12.85, SEM = 2.87; t(31) = .39, p = .70) and the independent-probe test (high load M = 9.37, SEM = 3.50 vs. low load M = 10.07, SEM = 3.81; t(31) = .24, p = .81). Furthermore, results also revealed that working memory load also had no impact on the size of the forgetting effect for the same-probe test (high load M = 11.81, SEM = 3.42 vs. low load M = 9.73, SEM = 3.55; t(31) = .50, p = .62). Importantly however, there was a significant effect of working memory load on the size of the forgetting effect on the independentprobe test, with participants showing a forgetting effect of 9.4% in the low load condition and showing enhanced recall of suppress items in the high load condition (high load M = 1.04, SEM = 3.75 vs. low load M = 9.38, SEM = 3.60; t(31) = 2.17, p = .04, d = .50). See Table 1 for a breakdown of recall accuracy across experimental conditions. Recall in the Respond Condition of the TNT Phase We also investigated whether there were any differences in recall performance in the respond condition for the high and low conditions and found that participants showed no significant differences in the recall of respond target words in the high and low load conditions, (M = 95.85, SEM = .75 vs. M = 96.14, SEM = .90; t(31) = .27, p = .79. Thought Intrusions Questionnaire Our analysis of the thought intrusion scores revealed that participants showed no significant difference in their overall responses as a function of load, suggesting that participants focused on looking at the suppress cue words to the same extent in the high and low load conditions during the think/no-think phase (M = 3.09, SEM = .13 vs. M = 3.22, SEM = .12, respectively; t(31) = 1.11, p = .27). We also found that there were no differences as a function of load in the extent to which unrelated thoughts to the task came to mind (M = 2.52, SEM = .12 vs. M = 2.60, SEM = .13, respectively; t(31) = .72, p = .48) or in the extent to which participants actively tried to prevent the target word from coming to mind in the high and low load conditions Ó 2017 Hogrefe Publishing

Figure 2. The mean percentage of words recalled as a function of the experimental factors in Experiment 1, represented as the difference between each experimental condition and their respective baseline condition (same-probe baseline, independent-probe baseline). Error bars represent ±1 Standard Error of the Mean (SEM).

(M = 3.32, SEM = .16 vs. M = 3.16, SEM = .14, respectively; t(31) = 1.38, p = .18). Furthermore, there was also no difference in the extent to which participants focused their attention on the black letter to avoid focusing on the cue word itself in the high and low load conditions (M = 2.79, SEM = .20 vs. M = 2.84, SEM = .18, respectively; t(31) = .44, p = .66). We did, however, find that participants expressed greater difficulty in “not thinking” about the original memory associated with the cue word in the low than the high load suppression condition (M = 3.0, SEM = .15 vs. M = 2.58, SEM = .12; t(31) = 2.94, p < .01, d = 0.56). We also found that the target words associated with the cue words came to mind more often in the low than the high load condition (M = 3.19, SEM = .18 vs. M = 2.74, SEM = .12, respectively; t(31) = 2.46, p < .05, d = .52).

Discussion The results from Experiment 1 found that participants were successful at suppressing “no-think” words on the same and independent-probe test in the low load condition. In the high load condition, however, although participants showed a forgetting effect on the same-probe test, they failed to suppress “no-think” words on the independentprobe test. We predicted that if suppression of no-think items competes for resources with the working memory task, then forgetting of no-think items would be reduced under high working memory load on both the same-probe and the independent-probe tests. We found that while the Experimental Psychology (2017), 64(1), 14–26

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

Table 1. Percentage of words correctly recalled on the same and independent-probe tests as a function of instruction and working memory load High cognitive load mean (SEM)

Low cognitive load mean (SEM)

Respond

Suppress

Respond

Suppress

Baseline

Same-probe

96.18 (1.62)

72.22 (3.23)

96.88 (1.34)

72.22 (3.23)

84.03 (2.54)

Independent-probe

84.38 (2.28)

76.04 (2.69)

85.07 (2.52)

65.63 (3.29)

75.0 (2.44)

suppression effect for no-think items was indeed eliminated under high working memory load on the independentprobe test, it remained unaffected on the same-probe test. One possible reason why we found a forgetting effect on the same-probe test, but not the independent-probe test in the high load condition may relate to the nature of the working memory task that we used. As the n-back task involved participants focusing on retrieving the target letter in black (i.e., the correct letter from the previous trial in the high load condition) as well as recalling or suppressing target words, it is possible that the high load condition acted as an interference-based strategy by allowing participants to focus on retrieving the target letter in order to prevent the suppress words from coming to mind. Support for this comes from research which has found that suppression and interference produce qualitatively different effects. For example, Wang et al. (2015) used a double-cue paradigm and found that participants undergoing suppression training showed a forgetting effect on the same and independent-probe test, but participants undergoing interference training only showed a forgetting effect on the same-probe test. Support also comes from our thought intrusion questionnaire which found that participants reported that the target words came to mind more often in the low than high load condition, and expressed greater difficulty in “not thinking” about the target word in the low load condition. Furthermore, participants also reported no significant differences in the extent to which they experienced task-irrelevant thoughts for the high and low load conditions which further suggests that working memory load did not lead to an overall difference in additional strategy use or attentional focus (i.e., differences in the extent to which participants focused on the cue word on the screen). The increased tendency of suppress words coming into awareness under the low working memory load condition is likely to have necessitated greater inhibitory control in order to prevent the target memory from coming to mind. This is consistent with research which has found that intrusions of the unwanted memory are important in triggering inhibitory control in the TNT task. For example, Levy and Anderson (2012) found that intrusions triggered profound down-regulation of hippocampal activity relative to non-intrusions, and also predicted later forgetting. Furthermore, Benoit et al. (2015) also reported increased engagement of DLPFC during intrusions and showed that DLPFC coupling with hippocampus was related to intrusion Experimental Psychology (2017), 64(1), 14–26

reductions. Thus, these findings provide support the notion that suppress words under low load were successfully inhibited (e.g., Anderson & Huddleston, 2012; Bergström et al., 2009b). Our findings that high working memory load eliminated forgetting only on the independent-probe test, and not same-probe test, suggest that the high working memory load task may have worked as an interference strategy rather than create direct competition for cognitive control resources between working memory and intentional forgetting mechanisms. Thus, any forgetting following a strategy involving diverting processing away from the target word would be based on differential processing of the target word, rather than the to-be-avoided associate, and therefore would only lead to a forgetting effect on same-probe tests, and not on independent-probe tests where the target word doesn’t feature (see Bergström et al., 2009b). It is important to mention that in our study we found that working memory load had no significant effect on recall in the “think” condition. This is surprising given that research has found that when two concurrent tasks are performed simultaneously performance deteriorates (e.g., Kahneman, 1973; Navon & Gopher, 1979). The fact that we failed to find differences suggests that working memory was insufficiently loaded to lead to consistent changes in recall of the “think” and “no-think” words, which in turn could explain the finding that suppression of the “no-think” target words was significantly modulated by high working memory load only in the independent-probe condition. In other words, a higher load on working memory may prevent successful suppression of the “no-think” target words and eliminate the suppression effect on both the same-probe test and the independent-probe test. The aim of Experiment 2 was to test this possibility as well as replicate the results of Experiment 1.

Experiment 2 Given the limitations of the working memory load task, the aim of Experiment 2 was to increase the task demands for the high load condition. If working memory was insufficiently loaded which led to suppression of the “no-think” target words being significantly modulated by high working memory load only in the independent-probe condition, then we would expect that by increasing task demands for the Ó 2017 Hogrefe Publishing

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

high load condition should lead to reduced forgetting effects on the same and independent-probe tests. If, however, our post hoc interpretation that the high working memory load tasks acts as an interference-based strategy to avoid thinking of the to-be-suppressed word, then Experiment 2 should replicate the forgetting effects in the high load condition and find that high working memory load is associated with reduced forgetting only on the independent-probe test. Similar to Experiment 1, the low working memory load condition required participants to report the black letter from the currently presented word (0-back task). In Experiment 2, however, the high working memory condition used the 2-back task which required participants to report the black letter from the cue word presented two trials previously (2-back task).

Method Participants Thirty-three healthy students (27 F and 6 M; age 18–30 years) from Goldsmiths, University of London took part in the study and were reimbursed (£10) for their participation. All participants were screened for depression using a screening questionnaire and the Beck Depression Inventory-II (Beck et al., 1996). Ethical approval was obtained from Goldsmiths Research Ethics Committee. Materials and Procedure The materials and procedure for this experiment were identical to those used in Experiment 1, with one notable exception. In this experiment we used the 2-back task as the high load working memory condition. For the 2-back task, participants were instructed to press the key that corresponded to the black letter presented two trials previously. Participants were informed that it was very important that they try to recall or “not think” about the green and red cue words, respectively, while also complying with high and low load task instructions. A potential limitation of Experiment 1 was that we did not explicitly ask participants about the extent to which they engaged with using an interference-based strategy. Thus, it currently remains unclear whether participants did in fact focus on the black letter to avoid thinking about the target word. In order to determine if this was the case, we changed the wording of our final question on the thought intrusions questionnaire to determine the extent to which participants were using the black letter as a strategy to prevent the target word from coming to mind (i.e., “to what extent did you focus on remembering the black letter as a strategy to avoid thinking about the target word”; 1 = never to 5 = all the time).

Ó 2017 Hogrefe Publishing

Results Performance on the n-Back Task Performance on the n-back task was assessed by conducting a 2 (Working Memory Load: High Load vs. Low Load) 2 (Instruction: Respond vs. Suppress) ANOVA, which revealed a main effect of Working Memory Load, F(1, 31) = 102.06, p < .01, ηp2 = .76. Performance was better in the low compared to the high load condition (M = 90.58, SEM = 1.0 vs. M = 48.17, SEM = 4.36, respectively). There was no main effect of Instruction, nor an Instruction by Working Memory Load interaction; F(1, 31) = 2.91, p = .10; F(1, 31) = 3.35, p = .08, respectively. Recall Accuracy at Final Test As in Experiment 1, we calculated the instruction effect for each experimental condition by subtracting the baseline score from the respond and suppress scores for high and low load, respectively. These scores were then entered into a 2 (Instruction: Respond vs. Suppress) 2 (Cognitive Load: High vs. Low) 2 (Probe Type: Same vs. Independent) ANOVA. Our analysis found a significant main effect of Instruction, F(1, 32) = 109.42, p < .01, ηp2 = .63, with participants, overall, showing a facilitation effect in the respond condition and a forgetting effect in the suppress condition (M = 15.82, SEM = 2.29 vs. M = 4.19, SEM = 2.43). There was also a significant main effect of Working Memory Load, F(1, 32) = 5.24, p = .03, ηp2 = .208, with participants, overall, showing a larger facilitation effect in the high than low load condition (M = 7.92, SEM = 2.40 vs. M = 3.71, SEM = 2.20). There was no significant main effect of Probe Type; F(1, 32) = 0.04, p = .85. We also found that there were significant two-way interactions between Instruction and Probe Type, F(1, 32) = 17.02, p < .01, ηp2 = .21, and between Instruction and Working Memory Load, F(1, 32) = 4.29, p = .04, ηp2 = .06. These two-way interactions were qualified by a significant three-way interaction between Instruction, Working Memory Load, and Probe Type, F(1, 32) = 5.0, p = .03, ηp2 = .07 (see Figure 3). Consistent with Experiment 1, subsequent pairwise comparisons showed that for the respond condition, there was no effect of working memory load on the size of the facilitation effect on the same-probe test (high load M = 19.86, SEM = 3.09 vs. low load M = 18.85, SEM = 2.80; t(32) = 0.72, p = .48) and the independentprobe test (high load M = 11.78, SEM = 3.55 vs. low load M = 12.79, SEM = 3.87; t(32) = 0.41, p = .69). Furthermore, we found that working memory load had no impact on the size of the forgetting effect on the same-probe test (high load M = 8.39, SEM = 3.49 vs. low load M = 8.72, SEM = 4.12; t(32) = 0.07, p = .94). Importantly, we found a significant effect of working memory load on the size of

Experimental Psychology (2017), 64(1), 14–26

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

Suppress High Load

Respond High Load

Suppress Low Load

Respond Low Load

Mean Percentage of Words Recalled Minus Baseline

25 20 15 10 5 0

-5 -10 -15

cue word in the low, compared to the high load condition (M = 3.17, SEM = .21 vs. M = 2.27, SEM = .20; t(32) = 3.44, p < .01, d = 0.77). We also found that the target words associated with the cue words came to mind more often in the low than the high load condition (M = 3.06, SEM = .16 vs. M = 2.13, SEM = .18; t(32) = 4.43, p < .01, d = 0.97). Interestingly, we found that participants focused their attention on remembering the black letter in order to avoid thinking of the target word significantly more in the high than low load condition (high load M = 2 = 3.45, SEM = 0.28 vs. low load M = 2.63, SEM = 0.24, respectively; t(32) = 2.83, p < .01, d = 0.54).

Discussion

- 20 Same Probe Test

Independent Probe Test

Figure 3. The mean percentage of words recalled as a function of the experimental factors in Experiment 2, represented as the difference between each experimental condition and their respective baseline condition. Error bars represent ±1 Standard Error of the Mean (SEM).

the forgetting effect on the independent-probe test, with participants showing a forgetting effect in the low load condition but enhanced recall of suppress items in the high load condition (low load M = 8.08, SEM = 4.28 vs. high load M = 8.42, SEM = 4.91; t(32) = 2.95, p < .01, d = 0.62). See Table 2 for a breakdown of recall accuracy. Recall in the Respond Condition of the TNT Phase We investigated whether there were any differences in recall performance in the respond condition for the high and low conditions and found that participants correctly recalled significantly more target words in the low than high load condition, (M = 98.08, SEM = 0.36 vs. M = 88.15, SEM = 1.94; t(31) = 5.25, p < .01, d = 1.23. Thought Intrusions Questionnaire Consistent with Experiment 1, our analysis revealed that participants focused on looking at the suppress cue words to the same extent in the high and low load conditions during the think/no-think phase (M = 3.81, SEM = .19 vs. M = 3.0, SEM = .17, respectively; t(32) = 1.18, p = .25). We also found no differences as a function of load in the extent to which unrelated thoughts to the task came to mind (high load M = 2.18, SEM = .17 vs. low load M = 2.37, SEM = .16, respectively; t(32) = 1.96, p = .06) or in the extent to which participants actively attempted to prevent the target word from coming to mind in the high and low load conditions (M = 3.21, SEM = .22 vs. M = 3.33, SEM = .20, respectively; t(32) = 0.60, p = .55). Participants, however, expressed greater difficulty in “not thinking” about the original memory associated with the Experimental Psychology (2017), 64(1), 14–26

Consistent with the results from Experiment 1, our results from Experiment 2 revealed that participants undergoing the high load condition showed a forgetting effect on the same-probe test, but were unsuccessful at demonstrating a forgetting effect on the independent-probe test. These findings were contrary to the prediction that a higher working memory load would prevent successful suppression of the “no-think” target words and eliminate the suppression effect on both the same and independent-probe tests. Rather, we found that the forgetting effects on the same-probe test were almost identical across working memory load conditions. The fact that a high working memory load eliminated the forgetting effect on the independent and not the same-probe test, suggests that the effect of working memory load on memory suppression is due to an interference-based strategy rather than competition for cognitive control resources. This is further supported by the results of the thought intrusion questionnaire which found that participants focused on remembering the black letter as a strategy to avoid thinking about the target word significantly more under high compared to low working memory load. Interestingly, increasing the loading of the high load condition did not lead to significant changes in the final recall of respond words in the high and low conditions. Rather, we found that participants were reporting high levels of recall of the respond words across both conditions. This is surprising given that research suggests that when two concurrent tasks are performed simultaneously performance often deteriorates (e.g., Kahneman, 1973; Navon & Gopher, 1979). It is important to mention, however, that loading did have an impact on participant’s ability to recall the respond words during the TNT phase of the experiment as participants recalled significantly more respond words correctly in the low than high load conditions. One reason for these findings may relate to demands of the high load condition in this experiment and the nature of the TNT task. As participants are expected to recall each respond Ó 2017 Hogrefe Publishing

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

Table 2. Percentage of words correctly recalled on the same and independent-probe tests as a function of instruction and working memory load High cognitive load mean (SEM)

Low cognitive load mean (SEM)

Respond

Suppress

Respond

Suppress

Baseline

Same-probe

97.31 (1.28)

69.06 (3.09)

96.30 (1.15)

68.72 (3.66)

77.44 (2.96)

Independent-probe

82.83 (2.27)

79.46 (3.21)

83.84 (2.22)

62.96 (4.10)

71.05 (3.02)

target 16 times, it is possible that the while the high load condition impairs their ability to retrieve the target words correctly during the task, repeated attempts at recalling the target words lead to sufficient practice thus eliminating any recall differences at final test.

General Discussion The aim of the current work was to examine the impact of manipulating cognitive control on suppression. Specifically, we used the TNT task to examine suppression under varying concurrent cognitive load. Across two experiments our results revealed that participants successfully showed a suppression-induced forgetting effect in the low load condition. In the high load condition, however, although participants were successful at showing a forgetting effect on the same-probe test, they failed to demonstrate a forgetting effect for “no-think” words on the independent-probe test. These findings are consistent with our explanation that the high working memory load serves as an interferencebased strategy by enabling individuals to focus on retrieving the letter from a previous trial in order to prevent the target word from coming to mind. The fact that there was no evidence of forgetting on the independent-probe test for the high load condition is consistent with research which has found that participants engaging in interference fail to show below-baseline forgetting on the independent-probe test (Wang et al., 2015; Bergström et al., 2009b; but see Racsmány, Conway, Keresztes, & Krajcsi, 2012, for evidence for yet another suggested source of the forgetting effect). Given that research suggests that interference-based strategies, such as thought substitution, may rely on inhibitory processes with suppression as a function of retrieving alternative memories (Benoit & Anderson, 2012), it is important to mention that our working memory load task in both experiments was similar to the interference training of Wang et al. (2015). In their study, participants learnt to associate cues with the target memory and were then given interference training with substitute words being presented. The fact that participants were not required to retrieve any information eliminated any potential inhibition caused by retrieval practice. In our experiments, the high load

Ó 2017 Hogrefe Publishing

condition involved participants recalling the target letter from the previous trial or previous two trials while also engaging in suppression, thus, we were very careful to ensure that our memory load tasks minimized the development of alternative associations during the suppression condition, making it unlikely that forgetting was a function of retrieving alternative thoughts. It is, however, possible that as participants had to remember the letter from the cue word of the previous trials as well as prevent the target word from coming to mind, the letter may have been encoded with the cue word, and this association may have subsequently blocked access to the target word on the same-probe test, but not the independent-probe test. While we do acknowledge this possibility, it is important to mention that in our experiments all the target letters were vowels that were counterbalanced across trials, such that participants were exposed to each of the five vowels equally per cue word. Thus, it is unlikely that the vowels, featuring in the spelling of the cue words, would have been associated to such an extent that they could have interfered with the retrieval of the target word on the same-probe test. Our prediction that the high load cognitive task may consume processing resources that would otherwise be directed to memory suppression, and should result in reduced forgetting on both same and independent-probes under high load was not supported by the results of the current experiments. It is still possible, however, that the high memory load condition did in fact disrupt the ability to engage inhibitory control and prevent participants from actively suppressing intruding memories. Given the fact that the high memory load condition was indeed distracting, it is possible that adding a concurrent memory load abolished inhibitory resources that would otherwise have been directed at suppression. This idea is consistent with research on executive control and retrieval-induced forgetting which has found that while individuals may exhibit retrieval-induced forgetting effects in typical conditions, when retrieval practice is undertaken concurrently with a task that requires executive control resources, participants show impaired forgetting (e.g., Ortega et al., 2012; Roman et al., 2009). Therefore, it is possible that as the high load condition imposed greater processing demands, participants were unable to recruit inhibitory resources to actively suppress the unwanted memories form coming to mind,

Experimental Psychology (2017), 64(1), 14–26

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

and instead used the working memory task as a means of actively “not thinking” about the target words. The idea that individuals focused on remembering the target letter from previous trials instead of actively suppressing the target word in the current trial is also supported by our results from the thought intrusions questionnaires. Participants from both experiments reported that the target words came to mind more often in the low than high load condition, and expressed greater difficulty in “not thinking” about the target word in the low load condition. Moreover, when we asked participants more explicitly about a possible strategy involving them focusing on remembering the black letter in order to avoid thinking about the target word in Experiment 2, we found that they reported doing so more often under high compared to low working memory load. The increased tendency of suppress words coming into awareness under the low working memory load condition is likely to have necessitated greater inhibitory control in order to prevent the target memory from coming to mind. The fact that participants were successfully able to forget “suppress” words on both the same-probe and independent-probe tests under low load suggests that these items were successfully inhibited (Anderson & Huddleston, 2012; Bergström et al., 2009b). These findings are consistent with previous research which has found that hippocampal activity was down-regulated during memory intrusions and predicted later forgetting (Levy & Anderson, 2012). In the high working memory load conditions, however, participants reported that suppress words came to mind less often, and that not thinking of these items was significantly easier. This finding is not in line with the prediction that high working memory load would interfere with participants’ ability to inhibit the target word: in that case not thinking of these items should be more, rather than less difficult. Instead, it is consistent with our prediction that the high working memory load task acted as an interference strategy (Anderson & Huddleston, 2012; Wang et al., 2015). Importantly, participants reported no significant differences in the extent to which they experienced taskirrelevant thoughts for the high and low load conditions which further suggests that working memory load did not lead to an overall difference in additional strategy use or attentional focus. To conclude, we have presented evidence that manipulating working memory load during intentional forgetting leads to a distinct pattern of recall of the to-be-suppressed memories on the same and independent-probe tests. In doing so, the current research provides direct evidence that when cognitive resources are limited individuals can switch strategies to prevent unwanted memories from coming to mind. Experimental Psychology (2017), 64(1), 14–26

Acknowledgments We would like to thank Ayesha Javed and Krishan Narayan for their help with data collection. This research was not funded by any funding bodies. Electronic Supplemental Materials The electronic supplementary material is available with the online version of the article at http://dx.doi.org/10.1027/ 1618-3169/a000347 ESM 1. Data file (doc). Raw data for recall performance for Experiments 1 and 2. ESM 2. Data file (doc). Raw data of n-back performance for Experiments 1 and 2.

References Anderson, M. C. (2005). The role of inhibitory control in forgetting unwanted memories: A consideration of three methods. In C. MacLeod & B. Uttl (Eds.), Dynamic cognitive processes (pp. 159–189). Tokyo, Japan: Springer. Anderson, M. C., Bjork, R. A., & Bjork, E. L. (1994). Retrieval can cause forgetting: Retrieval dynamics in long term memory. Journal of Experimental Psychology, 20, 1063–1087. Anderson, M. C., & Green, C. (2001). Suppressing unwanted memories by executive control. Nature, 410, 366–369. doi: 10.1038/35066572 Anderson, M. C., & Huddleston, E. (2012). Towards a cognitive and neurobiological model of motivated forgetting. In R. F. Belli (Ed.), True and false recovered memories: Toward a reconciliation of the debate (Vol. 58: Nebraska Symposium on Motivation, pp. 53–120). New York, NY: Springer. Anderson, M. C., & Huddleston, E. (2012). Reassessing critiques of the independent-probe method for studying inhibition. Journal of Experimental Psychology: Learning, Memory and Cognition, 38, 1408–1418. Anderson, M. C., Ochsner, K. N., Kuhl, B., Cooper, J., Robertson, E., Gabrieli, S. W., . . . Gabrieli, J. D. E. (2004). Neural systems underlying the suppression of unwanted memories. Science, 303, 232–235. doi: 10.1126/science.1089504 Anderson, M. C., & Spellman, B. A. (1995). On the status of inhibitory mechanisms in cognition: Memory retrieval as a model case. Psychological Review, 102, 68–100. Anderson, M. C., & Weaver, C. (2009). Inhibitory control over action and memory. In L. R. Squire (Ed.), The new encyclopaedia of neuroscience (pp. 153–163). Oxford, UK: Elsevier Ltd. doi: 10.1016/b978-008045046-9.00421-6 Aslan, A., Zellner, M., & Bauml, K.-H. T. (2010). Working memory capacity predicts listwise directed forgetting in adults and children. Memory, 18, 442–450. doi: 10.1080/ 09658211003742698 Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck Depression Inventory – II. San Antonio, TX: Psychological Corporation. Benoit, R. G., & Anderson, M. C. (2012). Opposing mechanisms support the voluntary forgetting of unwanted memories. Neuron, 76, 450–460. doi: 10.1016/j.neuron.2012.07.025 Benoit, R. G., Hulbert, J. C., Huddleston, E., & Anderson, M. C. (2015). Adaptive top-down suppression of hippocampal activity and the purging of intrusive memories from consciousness.

Ó 2017 Hogrefe Publishing

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

Journal of Cognitive Neuroscience, 27, 96–111. doi: 10.1162/ jocn_a_00696 Bergström, Z. M., de Fockert, J., & Richardson-Klavehn, A. (2009a). Event-related potential evidence that automatic recollection can be voluntarily avoided. Journal of Cognitive Neuroscience, 21, 1280–1301. doi: 10.1162/jocn.2009.21075 Bergström, Z. M., de Fockert, J., & Richardson-Klavehn, A. (2009b). ERP and behavioural evidence for direct suppression of unwanted memories. NeuroImage, 48, 726–737. doi: 10.1016/j.neuroimage.2009.06.051 Bjork, E. L., Bjork, R. A., & MacLeod, M. D. (2006). Types and consequences of forgetting: Intended and unintended. In L.-G. Nilsson & N. Ohta (Eds.), Memory and society: Psychological perspectives (pp. 134–158). Hove and New York: Psychology Press. Bulevich, J. B., Roediger, H., Balota, D. A., & Butler, A. C. (2006). Failures to find suppression of episodic memories in the think/ no-think paradigm. Memory & Cognition, 34, 1569–1577. Camp, G., Pecher, D., Schmidt, H. G., & Zeelenberg, R. (2009). Are independent-probes truly independent? Journal of Experimental Psychology: Learning, Memory, and Cognition, 35, 934–942. doi: 10.1037/a0015536 De Fockert, J. W. (2013). Beyond perceptual load and dilution: A review of the role of working memory in selective attention. Frontiers in Psychology, 4, 287. doi: 10.3389/fpsyg. 2013.00287 Del Prete, F., Hanczakowski, M., Bajo, M. T., & Mazzoni, G. (2015). Inhibitory effects of thought substitution in the think/no-think task: Evidence from independent cues. Memory, 23, 507–517. doi: 10.1080/09658211.2014.907429 Depue, B. E., Burgess, G. C., Willcutt, E. G., Ruzic, L., & Banich, M. T. (2010). Inhibitory control of memory retrieval and motor processing associated with the right lateral prefrontal cortex: Evidence from deficits in individuals with ADHD. Neuropsychologia, 48, 3909–3917. doi: 10.1016/j.neuropsychologia. 2010.09.013 Depue, B. E., Curran, T., & Banich, M. T. (2007). Prefrontal regions orchestrate suppression of emotional memories via a twophase process. Science, 317, 215–219. doi: 10.1126/science. 1139560 Depue, B. E., Orr, J. M., Smolker, H. R., Naaz, F., & Banich, M. T. (2015). Organization of right prefrontal networks that coordinate inhibitory regulation of cognitive, emotional, and motor processes. Cerebral Cortex, 1–13. doi: 10.1093/cercor/bhu324 Gagnepain, P., Henson, R. N., & Anderson, M. C. (2014). Suppressing unwanted memories reduces their unconscious influence via targeted cortical inhibition. Proceedings of the National Academy of the United States of America, 111, E1310–E1319. doi: 10.1073/pnas.1311468111 Garavan, H., Ross, T. J., Murphy, K., Roche, R. A. P., & Stein, E. A. (2002). Dissociable executive functions in the dynamic control of behavior: Inhibition, error detection, and correction. NeuroImage, 17, 1820–1829. Glucksberg, S., & King, J. L. (1967). Motivated forgetting mediated by implicit verbal chaining: A laboratory analog of repression. Science, 152, 517–519. Hanslmayr, S., Leipold, P., & Bauml, K.-H. (2010). Anticipation boosts forgetting of voluntarily suppressed memories. Memory, 18, 252–257. doi: 10.1080/09658210903476548 James, W. (1950). The principles of psychology: In two volumes, V. 2. New York, NY: Dover Publications. Kahneman, D. (1973). Attention and effort. NJ: Prentice-Hall. Kane, M. J., & Engle, R. W. (2003). Working-Memory capacity and the control of attention: The contributions of goal neglect, response competition, and task set to Stroop interference. Journal of Experimental Psychology: General, 132, 47–70.

Ó 2017 Hogrefe Publishing

Kopp, B., Rist, F., & Mattler, U. (1996). N200 in the flanker task as a neurobehavioral tool for investigating executive control. Psychophysiology, 33, 282–294. Lavie, N., Hirst, A., de Fockert, J. W., & Viding, E. (2004). Load theory of selective attention and cognitive control. Journal of Experimental Psychology: General, 133, 339–354. doi: 10.1037/ 0096-3445.133.3.339 Levy, B. J., & Anderson, M. C. (2008). Individual differences in suppressing unwanted memories: The executive deficit hypothesis. Acta Psychologica, 127, 623–635. doi: 10.1016/j.actpsy. 2007.12.004 Levy, B. J., & Anderson, M. C. (2012). Purging of memories from conscious awareness tracked in the human brain. Journal of Neuroscience, 32, 16785–16794. doi: 10.1523/JNEUROSCI. 2640-12.2012 MacLeod, M. D. (2002). Retrieval-induced forgetting in eyewitness memory: Forgetting as a consequence of remembering. Applied Cognitive Psychology, 16, 135–149. MacLeod, M. D., & Macrae, C. N. (2001). Gone today but here tomorrow: The transient nature of retrieval-induced forgetting. Psychological Science, 12, 148–152. Mecklinger, A., Parra, M., & Waldhauser, G. T. (2009). ERP correlates of intentional forgetting. Brain Research, 1255, 132–147. doi: 10.1016/j.brainres.2008.11.073 Menon, V., Adleman, N. E., White, C. D., Glover, G. H., & Reiss, A. L. (2001). Error-related brain activation during a Go/No Go response inhibition task. Human Brain Mapping, 12, 131–143. Navon, D., & Gopher, D. (1979). On the economy of the humanprocessing system. Psychological Review, 86, 214–255. Noreen, S., Bierman, R. N., & MacLeod, M. D. (2014). Forgiving you is hard, but forgetting seems easy: Can forgiveness facilitate forgetting? Psychological Science, 25, 1295–1302. doi: 10.1177/0956797614531602 Noreen, S., & MacLeod, M. D. (2013). It’s all in the detail: Intentional forgetting of autobiographical memories using the autobiographical think/no-think task. Journal of Experimental Psychology: Learning, Memory and Cognition, 39, 375–393. doi: 10.1037/a0028888 Noreen, S., & MacLeod, M. D. (2014). To think or not to think, that is the question: Individual differences in suppression and rebound effects in autobiographical memory. Acta Psychologica, 145, 84–97. doi: 10.1016/j.actpsy.2013.10.011 Noreen, S., & MacLeod, M. D. (2015). What do we really know about cognitive inhibition? Task demands and inhibitory effects across a range of memory and behavioural tasks. PLos One, 10, e0134951. doi: 10.1371/journal.pone.0134951 Ortega, A., Gomez-Ariza, C. J., Roman, P., & Bajo, M. T. (2012). Memory inhibition, aging and the executive deficit hypothesis. Journal of Experimental Psychology: Learning, Memory and Cognition, 38, 178–186. doi: 10.1037/a0024510 Paz-Alonso, P. M., Ghetti, S., Matlen, B. J., Anderson, M. C., & Bunge, S. A. (2009). Memory suppression is an active process that improves over childhood. Frontiers in Human Neuroscience, 3, 24. doi: 10.3389/neuro.09.024.2009 Racsmány, M., Conway, M. A., Keresztes, A., & Krajcsi, A. (2012). Inhibition and interference in the think no-think task. Memory and Cognition, 40, 168–176. doi: 10.3758/s13421-011-0144-6 Redick, T. S., Heitz, R. P., & Engle, R. W. (2007). Working memory capacity and inhibition: Cognitive and social consequences. In D. S. Gorfein & C. M. MacLeod (Eds.), Inhibition and Cognition (chapter 7) (pp. 125–142). Washington, DC: American Psychological Association. Roman, P., Soriano, M. P., Gomez-Ariza, C. J., & Bajo, M. T. (2009). Retrieval-induced forgetting and executive control. Psychological Science, 20, 1053–1058. doi: 10.1111/j.14679280.2009.02415.x

Experimental Psychology (2017), 64(1), 14–26

S. Noreen & J. W. de Fockert, The Role of Cognitive Load in Intentional Forgetting

Rosenzweig, S., & Mason, G. (1934). An experimental study of memory in relation to the theory of repression. British Journal of Psychology, 24, 247–265. doi: 10.1111/j.2044-8295.1934. tb00701.x Shaw, J. S., Bjork, R. A., & Handal, A. (1995). Retrieval-induced forgetting in an eyewitness-memory paradigm. Psychonomic Bulletin and Review, 2, 249–253. doi: 10.3758/BF03210965 Wang, Y., Cao, Z., Zhu, Z., Cai, H., & Wu, Y. (2015). Cueindependent forgetting by intentional suppression – Evidence for inhibition as the mechanism of intentional forgetting. Cognition, 143, 31–35. doi: 10.1016/j.cognition.2015.05.025

Saima Noreen Faculty of Life and Health Sciences De Montfort University The Gateway Leicester LE1 9BH United Kingdom saima.noreen@dmu.ac.uk

Received March 7, 2016 Revision received October 25, 2016 Accepted November 1, 2016 Published online February 20, 2017

Experimental Psychology (2017), 64(1), 14–26

Ó 2017 Hogrefe Publishing

Short Research Article

Competent and Warm? How Mismatching Appearance and Accent Inﬂuence First Impressions Karolina Hansen,1 Tamara Rakić,2 and Melanie C. Steffens3 1

Faculty of Psychology, University of Warsaw, Poland

Department of Psychology, Lancaster University, UK

Faculty of Psychology, University of Koblenz-Landau, Germany

Abstract: Most research on ethnicity has focused on visual cues. However, accents are strong social cues that can match or contradict visual cues. We examined understudied reactions to people whose one cue suggests one ethnicity, whereas the other cue contradicts it. In an experiment conducted in Germany, job candidates spoke with an accent either congruent or incongruent with their (German or Turkish) appearance. Based on ethnolinguistic identity theory, we predicted that accents would be strong cues for categorization and evaluation. Based on expectancy violations theory we expected that incongruent targets would be evaluated more extremely than congruent targets. Both predictions were confirmed: accents strongly influenced perceptions and Turkish-looking German-accented targets were perceived as most competent of all targets (and additionally most warm). The findings show that bringing together visual and auditory information yields a more complete picture of the processes underlying impression formation. Keywords: nonnative speech, stereotypes, ethnolinguistic identity, expectancy violations, impression formation, person perception

In today’s world of migration, people one meets have different cultural backgrounds (Davis, D’Odorico, Laio, & Ridolfi, 2013). National and ethnic distinctions in use for centuries are becoming outdated and inaccurate. As societies become more multicultural, we increasingly encounter people of mixed ethnicity, whose appearance and accent may violate expectations (King-O’Riain, Small, Mahtani, Song, & Spickard, 2014). In Germany, for instance, people may expect that a Turkish-looking person speaks German with a Turkish accent, and they may be surprised to hear native-like German (Hansen, Steffens, Rakić, & Wiese, 2016). Up to now, social psychological research has largely overlooked the existence of such individuals and how impressions of them are formed. In the current research, we investigate how people evaluate others based on their appearance and accent, when one of these cues indicates that the person is an outgroup member (e.g., looks Turkish) and the other that the person is an ingroup member (e.g., speaks with a standard German accent). When people encounter others, several cues indicating their ethnicity can be congruent or incongruent with each other. In the following, we focus on physical appearance (Dion, Berscheid, & Walster, 1972) and voice information (Zuckerman & Driver, 1989), specifically accent, as two powerful cues indicating social category memberships. Ó 2017 Hogrefe Publishing

Language and manner of speaking are at the core of ethnolinguistic identity theory (ELIT; Giles, Bourhis, & Taylor, 1977; Giles & Johnson, 1981, 1987). Based on social identity theory (Tajfel, Billig, Bundy, & Flament, 1971; Tajfel & Turner, 1979), ELIT focuses on the importance of language and accent in identity formation and maintenance. ELIT postulates that language is the most important marker of ethnic identity and others’ impressions are often based on accents. Researchers have shown that people who speak with a nonstandard accent are perceived as less intelligent and of lower social status (Fuertes, Gottdiener, Martin, Gilbert, & Giles, 2012; Giles & Powesland, 1975), but can also be seen as more loyal and sociable (Fuertes et al., 2012; Giles, 1971). Accent-based discrimination is an unrecognized potential threat often overlooked in research and in real life (Crandall, Eshleman, & O’Brien, 2002; Hansen, Rakić, & Steffens, 2014; Ng, 2007). In the US 21% of the population speaks a language other than English at home and among them, 42% speak English less than very well (US Census Bureau, 2011). In Germany about 9% of the population speaks a language other than German at home and 63% of them speak German less than very well (Haug, 2008). Thus, native speakers may expect that a foreign-looking person speaks with a foreign accent (Cheryan & Monin, 2005; Hirschfeld & Gelman, 1997). Experimental Psychology (2017), 64(1), 27–36 DOI: 10.1027/1618-3169/a000348

Appearance and accent of a person can both be indicators of this person’s ethnicity. Therefore, people with mismatching appearance and accent could be difficult to categorize, others could be surprised when encountering them, and they could be evaluated differently than nonsurprising people. There is very little research on reactions to people who embody conflicting cues about their categorization, such as mismatching appearance and accent. To the best of our knowledge, only a few studies directly contrasted the role of appearance and accent in person perception. Two early studies did not aim at contrasting different types of cues, but found stronger effects of speech style than of racial labels on the perception of targets (Jussim, Coleman, & Lerch, 1987; McKirnan, Smith, & Hamayan, 1983). Later studies explicitly aimed at contrasting appearance and accent and showed that accent is a stronger cue than appearance for ethnic categorization in adults (Rakić, Steffens, & Mummendey, 2011), ingroup favoritism in children (Kinzler, Dupoux, & Spelke, 2007; Kinzler, Shutts, Dejesus, & Spelke, 2009), and beliefs about general knowledge of foreign-accented speakers (Rödin & Özcan, 2011). The effects were observed across cultures (US, France, Germany, Sweden) and with different dependent variables. Both Rakić and colleagues (2011) and Pietraszewski and Schwartz (2014b) independently ran similar who-said-what experiments in Germany and the US, to reveal that accent is crucial in social categorization. Going beyond mere categorization, it is interesting how appearance-accent (mis)matches influence evaluations. A possible mechanism at work here could be expectancy violations. Expectancy violations theory postulates that violations of expectations produce more extreme outcomes than situations that match those expectations (e.g., Burgoon & Burgoon, 2001; Roese & Sherman, 2007). For example, Blacks with strong academic qualifications were evaluated as more competent than Whites with similar credentials, which represented positive violations of expectations based on the stereotype that Blacks are less academically oriented (Jackson, Sullivan, & Hodge, 1993). Similarly, women with top credentials were evaluated more favorably as leaders than similarly qualified men because they violated stereotypical gender expectations (Rosette & Tost, 2010). Conversely, Whites who spoke nonstandard English were viewed more negatively than Blacks who did, representing negative expectancy violations (Jussim et al., 1987).

The Current Research The present research examined how appearance and accent, suggesting the same or different ethnicities, influence how people are evaluated. We let our participants evaluate others on the two fundamental stereotype Experimental Psychology (2017), 64(1), 27–36

K. Hansen et al., Accent, Appearance, and First Impressions

dimensions competence and warmth (Abele & Wojciszke, 2007; Fiske, Cuddy, Glick, & Xu, 2002). Because Turks are the largest immigrant group in Germany, we chose Germans and Turks as targets (Federal Ministry of the Interior, 2007). In Germany, as in the US and in many other countries, Turks (and Muslims more broadly) are stereotypically perceived as low on competence and warmth (Asbrock, 2010; Froehlich, Martiny, Deaux, & Mok, 2016). In contrast, the majority ingroup tends to self-stereotype as high on both dimensions (Fiske, Cuddy, & Glick, 2007; Fiske et al., 2002). In the case of Germans, in some studies they perceive themselves as competent and warm (Asbrock, 2010; Eckes, 2002), in some as competent and moderately warm (Froehlich et al., 2016), and still in others as competent, but not warm (e.g., Cuddy et al., 2009). In a computer-based experiment, we studied the influence of auditory and visual cues to ethnicity on the perceived competence and warmth of job candidates. We expected incongruent targets to violate participants’ expectations. Therefore, we also included a categorization task and tested whether incongruent targets were unexpected and thus categorized more slowly than congruent targets. We used photographs of male targets and recordings of speech in congruent or incongruent combinations. Male targets were used because stereotypes of nationalities apply more to men than women (Eagly & Kite, 1987) and for Germans the prototypical Turk is a man (e.g., Klingst & Drieschner, 2005). Hypotheses 1–2 establish the basis for testing our main (evaluation) hypotheses. Based on ELIT (e.g., Giles & Johnson, 1987), we expected that accent would be a strong cue for social categorization. Conceptually replicating previous studies (Pietraszewski & Schwartz, 2014b; Rakić et al., 2011), but now in the German-Turkish context, targets should be categorized more according to their accent than appearance (Hypothesis 1a). We were especially interested in incongruent targets and we expected that Turkish-looking targets speaking standard German would be generally categorized as German (Hypothesis 1b) and German- looking targets with a Turkish accent would be categorized as non-German (Hypothesis 1c). Research on expectancy violations shows that counterstereotypical people evoke more effortful cognitive processing than stereotypical people (Bettencourt, Dill, Greathouse, & Charlton, 1997; Roese & Sherman, 2007). When people meet a counter-stereotypical person, the discrepancy leads to recategorization until an appropriate relevant category or subcategory is found (Fiske & Neuberg, 1990; Hutter & Crisp, 2006; Kunda & Thagard, 1996). Thus, we hypothesized that incongruent targets should be more difficult to categorize, which would be reflected by longer categorization reaction times (RTs, Hypothesis 2). Based on ELIT, we predicted that accents would strongly influence evaluations (Hypothesis 3). Based on extensive Ó 2017 Hogrefe Publishing

K. Hansen et al., Accent, Appearance, and First Impressions

Table 1. Pretest ratings of photographs of faces and recordings of voices Faces

Voices

M (SD)German

M (SD)Turkish

M (SD)German

M (SD)Turkish

Attractiveness

3.18 (1.21)

2.82 (1.04)

1.86

.07

3.44 (1.36)

3.21 (1.38)

0.96

.34

Pleasantness

4.47 (0.89)

4.14 (1.05)

1.27

.21

4.61 (1.14)

4.52 (0.89)

0.39

.70

Typically German

5.33 (1.29)

1.62 (0.70)

15.95

< .001

4.80 (1.64)

1.49 (0.82)

7.63

< .001

Typically Turkish

1.34 (0.47)

3.66 (1.71)

6.01

< .001

1.61 (0.92)

3.20 (1.52)

5.70

< .001

–

1.63 (0.84)

4.85 (1.14)

13.22

< .001

90%

93%

Accent strength Categorization

–

100%

93%

–

research showing that nonstandard speakers are evaluated as less competent than standard speakers (Fuertes et al., 2012) and on the fact that Turks are perceived in Germany as incompetent (Asbrock, 2010; Froehlich et al., 2016), we expected Turkish-accented speakers to be evaluated as less competent than standard German speakers (Hypothesis 4). As findings regarding warmth of nonstandard speakers (Fuertes et al., 2012) as well as perceived warmth of Germans and Turks in Germany are mixed (e.g., Froehlich et al., 2016), we did not formulate specific predictions for this dimension. Our main hypothesis was that incongruent targets would be evaluated differently than congruent targets. Based on the expectancy violation theory (e.g., Burgoon & Burgoon, 2001; Roese & Sherman, 2007), we expected that incongruent targets would be evaluated more extremely than congruent targets in the direction of the valence of the violation. Again, as Turks are consistently perceived in Germany as incompetent and Germans as competent, but perceptions of their warmth differ between studies (Froehlich et al., 2016), we formulated these hypotheses for the competence dimension, but only explored the warmth dimension. Specifically, we hypothesized that when participants see a Turkish-looking person speaking standard German, their negative expectations would be positively violated and they would evaluate the target as very competent (Hypothesis 5a). Conversely, we expected that German-looking targets speaking with a Turkish accent would negatively violate participants’ expectations, and therefore be evaluated as incompetent (Hypothesis 5b).

Method Pretests and Selection of Stimulus Materials We used portrait photographs of faces from an online database (Minear & Park, 2004) and added several of 1

our own photographs of Turkish men. All men were young, with a neutral facial expression, without glasses, and with a neutral modern haircut. Pictures were converted into black and white. Short voice samples of young German and Turkish native speakers were recorded. All speakers said the same neutral everyday phrase, “Good morning. Nice to meet you” (in German: “Guten Tag. Es freut mich, dass wir uns kennen lernen”), ensuring that any influence of the content of the statement was excluded and that accented sentences were easy to understand. Speakers were briefly trained, speech rate was held constant, and voice samples were approximately 3 s long. To avoid the “what is beautiful is good” phenomenon (Dion et al., 1972; Zuckerman & Driver, 1989) and ensure that the stimuli were perceived as typical for their respective groups, all stimuli were pretested for attractiveness, pleasantness, ethnic typicality, and voices also for accent strength (Ryan, Carranza, & Moffie, 1977). Pretest participants (N = 29, 13 men, Mage = 22.73, SD = 3.42) were from the same population as participants in the main experiment, but participated only in the pretest. The pretest consisted of a block of faces and a block of voices. After each face or voice was presented in random order, participants answered how (1) attractive, (2) pleasant, (3) typically German, and (4) typically Turkish targets appeared or sounded (1 = not at all to 7 = very much). Voices were also evaluated regarding accent strength (1 = no accent at all to 7 = very strong accent). From the pretested photographs of faces, we selected four moderately attractive and pleasant German and four moderate Turkish-looking faces; all of them were typical for their respective groups (Table 1).1 Similarly, from the pretested voices, we selected four plus four moderately attractive and pleasant, but typical voices (Table 1). In the main experiment, we wanted to cross accents with appearance cues and study categorization and evaluation of mixed people. To be sure that accent and appearance had

One could worry that Turkish-looking faces were descriptively less attractive than German-looking faces. However, in the later evaluations Turkish-looking targets were perceived as most competent when they spoke standard German, but as least competent when they spoke with a Turkish accent. Thus, the descriptive difference in facial attractiveness of targets cannot account for the findings.

Ó 2017 Hogrefe Publishing

Experimental Psychology (2017), 64(1), 27–36

the same categorization baselines on their own, not yet being combined, we ran another pretest (N = 18, 4 men, Mage = 26.06, SD = 6.31). We used the same categorization task that we later used in the main study, but for the pretest, we presented faces and voices separately (in two blocks with randomized block and stimulus order). The results showed that: German faces were in 100% of cases categorized as German, Turkish faces in 93% as Turkish, German voices in 90% as German, and Turkish voices in 93% as Turkish (Table 1). A 2 2 chi-square test showed no differences between these percentages, w2 = 0.26, p = .61. Experimental Design The experiment had a 2 (Appearance: German vs. Turkish) 2 (Accent: standard German vs. German with a Turkish accent) within-subject design. Thus, there were four target types: German appearance/German accent (GG, congruent), Turkish appearance/Turkish accent (TT, congruent), German appearance/Turkish accent (GT, incongruent), and Turkish appearance/German accent (TG, incongruent). Stimulus composition was counterbalanced: any given voice (e.g., speaking standard German) was matched with a congruent picture (German-looking person) in one version of the experiment and with an incongruent picture (Turkish-looking person) in the second version. For generalization and control reasons, there were initially also two between-participants factors: context and face-voice sequence. As evaluations of others could depend on the context (e.g., Abele & Wojciszke, 2007; Vonk, 1999), we tested for the generalizability of our findings in the contexts of a roommate search and a job interview. The sequence of presentation was counterbalanced between participants: half of them first saw the face of a target and then immediately heard the voice, and half heard the voice and then immediately saw the face. Participants We stopped collecting data after achieving at least 50 participants per between-subjects condition (context and sequence). Participants were 226 undergraduate students of various faculties of a German university. After excluding the data of 11 participants who were not native German speakers, the final sample consisted of 215 participants (72 men, Mage = 22.33, SD = 3.24). They were compensated with either €1 and a chocolate bar or with partial course credit. 2

K. Hansen et al., Accent, Appearance, and First Impressions

Procedure and Measures After being welcomed by an experimenter unaware of the study’s hypotheses, participants were seated in front of a computer screen and signed informed consent. The experiment consisted of an evaluation and a categorization block, with the same targets in each. First, participants were asked to imagine that either they were helping in a recruitment process at their workplace or that they had a free room for rent in their apartment (later analyses showed no differences between these two contexts). All participants first saw two “filler” congruent German targets for training purposes and to set a common base. Then, the main targets were presented in an individual random order. Targets’ faces and voices were presented with 1 s in between. For all targets, participants were asked to look at a face and listen to a voice and answer on a separate screen how competent (competent, competitive, independent, α = .93) and warm (likable, warm, good-natured, α = .91) the person appeared (1 = not at all to 7 = very much; Asbrock, 2010; Fiske et al., 2002). After this, participants saw and heard the same targets (in a different sequence) again and were asked to answer the question “Is this person German?” with yes and no as quickly as possible; reaction times (RTs) were collected. For categorization, we added female targets and questions about the gender of the target as filler items in order to prevent mental preparation to responding always to the same question, avoiding falsely short reaction times.2 Finally, participants answered demographic questions, provided their email address for debriefing, were given their reward, thanked, and dismissed.

Results Preliminary Analyses Two preliminary 2 (Appearance: German vs. Turkish) 2 (Accent: German vs. Turkish) 2 (Context: job interview vs. students’ apartment) mixed analyses of variance (ANOVAs) yielded no effects involving context on competence or warmth evaluations (all Fs < 1). Similar analyses including presentation sequence (appearance-accent vs. accent-appearance) yielded no main effects of sequence (Fs < 1.94, ps = .17), and only one out of six possible interactions on the warmth dimension.3 Therefore, data were collapsed across these factors.

A few supplementary questions (manipulation check: accent strength, cooperativeness, trustworthiness, suggested salary/room rent) yielded similar results but will not be reported for space concerns. Motivation to respond without prejudice was assessed at the end and did not moderate the findings. An interaction of appearance, accent, and sequence on warmth evaluations, F(1, 197) = 18.93, p < .001, ηp2 = .09, boiled down to the following finding: German-looking Turkish-accented targets were perceived as warmer when their German appearance was presented first (M = 4.91, SD = 1.58) than when their Turkish accent was first (M = 4.56, SD = 1.63), F(1, 197) = 4.76, p = .03, ηp2 = .02, but Turkish-looking Germanaccented targets were evaluated similarly in both presentation sequences, F < 1. This suggests that the sequence played only a minor and selective (or even random) role for evaluations.

Experimental Psychology (2017), 64(1), 27–36

Ó 2017 Hogrefe Publishing

K. Hansen et al., Accent, Appearance, and First Impressions

Table 2. Logistic regression results for accent and appearance predicting categorization of targets as German or non-German B

B (SD)

Intercept

2.64

0.20

Accent

4.54

0.24

Appearance

2.00

0.23

Accent Appearance

0.98

0.33

95% CI [ 3.03,

173.66

< .001

[4.06, 5.02]

349.35

< .001

[1.55, 2.44]

77.25

< .001

8.65

.003

[ 1.64,

2.24]

Wald

0.33]

Figure 1. Percent of targets categorized as Germans or non-Germans (left) and mean reaction times of categorization by target type (right). Error bars represent standard errors of the mean.

Social Categorization Prerequisites for analyzing the evaluations were that the data replicate the strong influence of accent on social categorization and that incongruent stimuli are expectancy-violating and thus people take longer to categorize them. As can be seen in Figure 1, targets were categorized more according to their accent than appearance, which was tested by means of a binomial logistic regression for repeated measures using the generalized estimating equations method (Zeger & Liang, 1986; see Table 2). As the Wald statistic shows, accent was a significant and strong predictor of categorization. The influence of appearance was much weaker and there was an interaction effect of appearance and accent. Follow-up analyses showed that German-looking targets were more often categorized as Germans than Turkish-looking targets, and this effect was stronger for German-accented speakers, McNemar’s w2 = 50.21, p < .001, than for Turkish-accented speakers, McNemar’s w2 = 10.62, p = .001, which could be due to a floor effect for Turkish-accented speakers. The results confirmed the Hypothesis 1a that accent would play a stronger role for categorization than appearance. Hypotheses 1b and 1c were also confirmed as Turkishlooking German-accented targets were mostly (65%) categorized as German and German-looking Turkish-accented targets as non-German (87%). Reaction times We excluded responses that were ±3 standard deviations from the mean. We computed a 2 (Accent: German vs. Ó 2017 Hogrefe Publishing

Turkish) 2 (Congruence of targets: congruent vs. incongruent) repeated-measures ANOVA. The analysis showed that accent did not influence RTs, F < 1, but congruence did, F(1, 197) = 7.61, p = .006, ηp2 = .04 (Figure 1). Incongruent targets were categorized more slowly (M = 1,347.28 ms, SD = 539.57 ms) than congruent targets (M = 1,250.58 ms, SD = 432.98 ms), corroborating Hypothesis 2 that incongruent targets are more difficult to categorize (interaction: F < 1). Having confirmed that incongruent stimuli were expectancy-violating, we analyzed the effects of appearance and accent on evaluations.

Competence Impressions A 2 (Appearance: German vs. Turkish) 2 (Accent: German vs. Turkish) repeated-measures ANOVA showed that targets speaking standard German were evaluated as more competent (M = 4.83, SD = 0.75) than Turkishaccented targets (M = 4.20, SD = 0.84), F(1, 197) = 85.74, p < .001, ηp2 = .30 (Figure 2, Hypothesis 4). Competence evaluations also depended on appearance, but to a smaller extent, F(1, 197) = 6.21, p = .01, ηp2 = .03 (Hypothesis 3). More importantly, evaluations depended on specific combinations of accent and appearance, as reflected by an interaction effect, F(1, 197) = 20.63, p < .001, ηp2 = .10. Analyses of simple main effects showed that among German-accented targets, Turkish-looking (i.e., incongruent) targets were perceived as more competent than German-looking (i.e., congruent) targets, F(1, 197) = 21.30, p < .001, ηp2 = .10 (Hypothesis 5a). Turkish-accented Experimental Psychology (2017), 64(1), 27–36

Figure 2. Mean evaluations of competence and warmth by target type.

targets were evaluated as similarly competent whether they were German- or Turkish-looking, F(1, 197) = 2.47, p = .12, ηp2 = .01 (Hypothesis 5b).

Warmth Impressions An ANOVA on the warmth dimension showed that neither accent itself, F < 1, nor appearance itself, F(1, 197) = 3.68, p = .06, ηp2 = .02, influenced warmth evaluations in a significant way. Only the combination of appearance and accent influenced warmth perceptions, interaction: F(1, 197) = 38.52, p < .001, ηp2 = .16. As depicted in Figure 2, incongruent targets were evaluated as warmer than congruent targets. More precisely, among German-accented targets, those who also looked German were perceived as less warm than those who looked Turkish, F(1, 197) = 31.38, p < .001, ηp2 = .14; among Turkish-accented targets, those who also looked Turkish were perceived as less warm than those who looked German, F(1, 197) = 13.52, p < .001, ηp2 = .06. These results show that incongruent targets were perceived as warmer than both congruent German and congruent Turkish targets.

Additional Analyses In sum, Turkish-looking German-accented targets were evaluated as both most competent and, together with other incongruent targets, as most warm. German-looking Turkish-accented targets were along with TurkishTurkish targets evaluated as least competent. Overall,

Experimental Psychology (2017), 64(1), 27–36

K. Hansen et al., Accent, Appearance, and First Impressions

competence and warmth evaluations were correlated, r = .66, p < .001. As targets were outgroup members for women, but gender-ingroup members for men, one may wonder whether effects differed between female and male participants. Exploratory analyses showed that the pattern of results for both genders was similar, but the differences between targets were larger for men. Especially the advantage of the Turkish-looking German-accented target over the German-German target was larger for men (both for competence and warmth). Considering the higher percentage (67%) of women in the sample, we conclude that the observed effects would be larger in a more balanced sample. Studies show that it is crucial how a stereotypeincongruent person is categorized (e.g., Bless, Schwarz, Bodenhausen, & Thiel, 2001). To check potential influence of categorization on evaluation, we compared evaluations of the Turkish-looking German-accented targets between participants who categorized them as German (35%) or non-German (65%). Results showed no significant differences (ts < |1.7|, ps > .10 for competence, ts < |1| for warmth), indicating that categorization did not affect evaluations.

Discussion When people encounter others, they often see and hear them. Their appearance and speech, as well as the combination of those two, can influence how people evaluate each other. Although such cross-modal effects are frequent in real life, they are relatively underrepresented in psychology (see also Freeman & Ambady, 2011; Zuckerman, Miyake, & Hodgins, 1991). The present research provides an original contribution to understanding the influence of visual and auditory cues on impression information. Targets were seen in photographs and heard in short voice recordings. They appeared Turkish or German and spoke standard German or German with a Turkish accent. Participants evaluated targets’ competence and warmth, and categorized them as Germans or nonGermans. In a pretest, appearance and accent presented separately were similarly used to infer ethnicity. When pitted against each other, accent was more diagnostic for social categorization and evaluation. Such a strong role of accent is in line with ethnolinguistic identity theory (Giles & Johnson, 1987) and results of research conducted in the US (Kinzler et al., 2009; Pietraszewski & Schwartz, 2014b), Germany (Rakić et al., 2011), and Sweden (Rödin & Özcan, 2011). Nevertheless, it is an open question whether this would replicate everywhere or would depend on the diagnosticity of accents and appearance in a specific cultural context Ó 2017 Hogrefe Publishing

K. Hansen et al., Accent, Appearance, and First Impressions

(see Pietraszewski & Schwartz, 2014a). Future crosscultural research or experimental manipulations of diagnosticity could shed more light on this issue. Our results also showed that standard German speakers were overall evaluated as more competent than Turkishaccented speakers. However, the evaluation of targets depended on the combination of their appearance and accent. As expectancy violations theory predicted (Burgoon & Burgoon, 2001), effects of appearance-accent mismatch went beyond “the sum of the elements” and Turkishlooking German-accented targets were perceived as most competent. German-looking Turkish-accented targets were, together with congruent Turkish targets, evaluated as the least competent. Thus, our expectancy violations hypotheses were confirmed. As earlier findings about the perceived warmth of Turks in Germany (e.g., Froehlich et al., 2016) and of foreign-accented speakers (Fuertes et al., 2012) were inconclusive, we did not formulate hypotheses for warmth. Nonetheless, results on this dimension were very interesting: the two types of incongruent targets were evaluated as warmer than the two types of congruent targets. Congruent Turkish targets were perceived as relatively cold (see also Asbrock, 2010; Eckes, 2002), but targets who had only one Turkish feature were evaluated as warmer. Possibly, perceived threat changes evaluations (Cottrell & Neuberg, 2005): A young Turkish-looking Turkish-accented man might be too threatening to be perceived as nice and friendly, but when he possesses only one Turkish trait, stereotypes about warm Turks might be activated and expressed. An alternative explanation pertains both to the warmth and the competence findings: according to expectancy violations theory or to the “black sheep” effect (Biernat, Vescio, & Billings, 1999), German-looking Turkishaccented targets should be evaluated as least competent, but they were evaluated as similarly (in)competent as the congruent Turkish targets. This result suggests that other cognitive processes could also contribute to the observed effects. It could be that (appearance or accent) cues change their meaning in the context of other cues (e.g., Anderson, 1971; Kunda & Thagard, 1996). In a study where participants indicated how they interpreted surprising combinations of appearance and accents, German-looking faces sometimes changed the perception of Turkish accents: Some participants reinterpreted the targets as Northern or Eastern Europeans (Hansen, 2013). This shows how surprising combinations of accent and appearance can strongly change people’s perceptions (Kunda & Thagard, 1996; Remedios, Chasteen, Rule, & Plaks, 2011). Turkish-looking but German-accented targets were evaluated as both warmest and most competent. The other incongruent targets, German-looking but speaking with a Ó 2017 Hogrefe Publishing

Turkish accent, were perceived as low in competence, but high in warmth. While the latter result could be interpreted as compensatory stereotyping (Yzerbyt, Provost, & Corneille, 2005), the earlier one could not. More research is needed to better understand the obtained findings on the warmth dimension and generally, the inconsistent findings for warmth evaluations in research on accents (Fuertes et al., 2012). The positive evaluation of Turkish-looking targets who spoke standard German was in line with expectancy violations: Participants were positively surprised by these targets (which was reflected in longer categorization latencies) and evaluated them extremely well (Burgoon, 2009; Roese & Sherman, 2007). Previous research has shown, for example, that Blacks with strong academic qualifications were evaluated as more competent than Whites with similar credentials (Jackson et al., 1993). Along with other results interpreted as expectancy violations, these results can be also seen as an effect of lower linguistic standards that Germans might have for foreign-looking people. People may be evaluated in comparison to the average of their group (Biernat & Manis, 1994). Stereotypeincongruent targets can be contrasted from the group norm and described in such terms, for example: “For a Turk he speaks German very well” (Collins, Biernat, & Eidelman, 2009). Similar contrastive judgment patterns can also occur when expectations are violated. The present experiment offers no direct evidence of expectancy violation. However, in cases in which the same result can be based on different cognitive processes, measuring its neural correlates can be a useful tool for constraining explanations of such behavioral data (Bartholow, 2010). Research related to the present study, combining accent and appearance ethnicity cues, has shown that incongruent targets evoke brain reactions that can be interpreted as expectancy violations (Hansen et al., 2016). The results of studies like the present one may depend on the cultural context where they are conducted, for example, in a traditionally monocultural or multicultural country. The results could also depend on the characteristics and beliefs about a specific ethnic group. We chose Turks as targets as they are the biggest and the most prototypical immigrant group in Germany (Federal Ministry of the Interior, 2007). We are not aware of any data directly showing relationships between Turkish appearance, accent strength, and other variables. However, existing data show that about 7% of the German population speak German less than very well (Haug, 2008) and children of Turkish origin often have problems at school because of their low German language competencies (e.g., Becker, 2010). Thus, indirectly it can be inferred that Turkish appearance and Turkish accent are significantly related with each other in Germany, and also that Turkish accent could be related Experimental Psychology (2017), 64(1), 27–36

to low perceived acculturation. We cannot generalize our findings, for example, to Asians, who are often perceived to be particularly competent and hardworking, and people could hold different assimilation expectations regarding them (Asbrock, 2010; Fiske et al., 2002). Similarly, we had few and pretested stimuli per condition, which made the experiment well controlled and its interpretations cleaner. The chosen stimuli were judged as typical for their groups, but we cannot know how representative they were of the Turkish- and German-origin populations in Germany. Nevertheless, we think that even if a specific cultural context or stimulus choice may have influenced the results, the mechanism is still interesting: If people expect from a Turkish-/Moroccan-/Indian- or German-/French-/American-looking person to speak with a specific accent, but the person speaks with a different one, this can be surprising, new qualities can emerge from such atypical combinations of features, and they can strongly influence evaluations. Our results suggest that Turks in Germany would benefit from learning German at an early age, as foreign-looking people who speak standard German evoke an especially positive impression. We think that these are reasonable conclusions, but we would also like to draw attention to the other side of the coin. A widespread approach to communication problems between native and nonnative speakers is to reduce the accent of the nonnative speaker (e.g., Carlson & McHenry, 2006; Shah, 2012). This focuses attention only on one person’s responsibility, and eradicating accent in speech is very difficult or even impossible to achieve (Gluszek & Dovidio, 2010). We believe that to diminish language-based discrimination, it is important to address the role of native speakers’ consciousness, for example by using perspective-taking interventions (Hansen et al., 2014; Subtirelu & Lindemann, 2014).

Conclusions An important implication of the present research is that researchers should pay more attention to the interactions of appearance, accent, and other cues in impression formation. Reactions to people with features suggesting different ethnicities have been little studied, but with increasing global mobility they are timely and important. With our research, we hope to pave the way for future research on the social perception of people whose social categorization is ambiguous. Acknowledgments The current research was supported by the ProExzellenz program of the state of Thuringia, Polish National Science

Experimental Psychology (2017), 64(1), 27–36

K. Hansen et al., Accent, Appearance, and First Impressions

Centre (NCN Grant Fuga DEC-2013/08/S/HS6/00573), and Foundation for Polish Science scholarship (FNP START 030.2015-W) awarded to Karolina Hansen, as well as by a grant of the German Research Foundation (DFG STE 938/ 10-2; FOR 1097) to Melanie C. Steffens and Tamara Rakić. We thank Claudia Niedlich for her help in data collection, Wiktor Soral for his help with converting the dataset, as well as Aysan Ashoee, Jana Meyer, the editor, and two anonymous reviewers for their comments on a previous version of this manuscript. Electronic Supplementary Material The electronic supplementary material is available with the online version of the article at http://dx.doi.org/10.1027/ 1618-3169/a000348 ESM 1. Data file (xls). Raw data.

References Abele, A. E., & Wojciszke, B. (2007). Agency and communion from the perspective of self versus others. Journal of Personality and Social Psychology, 93, 751–763. doi: 10.1037/0022-3514. 93.5.751 Anderson, N. H. (1971). Integration theory and attitude change. Psychological Review, 78, 171–206. doi: 10.1037/h0030834 Asbrock, F. (2010). Stereotypes of social groups in Germany in terms of warmth and competence. Social Psychology, 41, 76–81. doi: 10.1027/1864-9335/a000011 Bartholow, B. (2010). Event-related brain potentials and social cognition: On using physiological information to constrain social cognitive theories. Social Cognition, 28, 723–747. doi: 10.1521/soco.2010.28.6.723 Becker, B. (2010). Wer profitiert mehr vom Kindergarten? [Who profits more from preschool attendance?]. Kölner Zeitschrift für Soziologie und Sozialpsychologie, 62, 139–163. doi: 10.1007/ s11577-010-0090-5 Bettencourt, B. A., Dill, K. E., Greathouse, S. A., & Charlton, K. (1997). Evaluations of ingroup and outgroup members: The role of category-based expectancy violation. Journal of Experimental Social Psychology, 33, 244–275. doi: 10.1006/ jesp.1996.1323 Biernat, M., & Manis, M. (1994). Shifting standards and stereotype-based judgments. Journal of Personality and Social Psychology, 66, 5–20. doi: 10.1037/0022-3514.66.1.5 Biernat, M., Vescio, T. K., & Billings, L. S. (1999). Black sheep and expectancy violation: Integrating two models of social judgment. European Journal of Social Psychology, 29, 523–542. doi: 10.1002/(SICI)1099-0992(199906)29:4<523::AIDEJSP944>3.0.CO;2-J Bless, H., Schwarz, N., Bodenhausen, G. V., & Thiel, L. (2001). Personalized versus generalized benefits of stereotype disconfirmation: Trade-offs in the evaluation of atypical exemplars and their social groups. Journal of Experimental Social Psychology, 37, 386–397. doi: 10.1006/jesp.2000.1459 Burgoon, J. K. (2009). Expectancy violations theory. In E. Griffin (Ed.), A first look at communication theory (pp. 84–97). New York, NY: McGraw-Hill.

Ó 2017 Hogrefe Publishing

K. Hansen et al., Accent, Appearance, and First Impressions

Burgoon, J. K., & Burgoon, M. (2001). Expectancy theories. In W. P. Robinson & H. Giles (Eds.), The new handbook of language and social psychology (2nd ed., pp. 79–102). Sussex, UK: Wiley. Carlson, H. K., & McHenry, M. A. (2006). Effect of accent and dialect on employability. Journal of Employment Counseling, 43, 70–83. doi: 10.1002/j.2161-1920.2006.tb00008.x Cheryan, S., & Monin, B. (2005). Where are you really from? Asian Americans and identity denial. Journal of Personality and Social Psychology, 89, 717–730. doi: 10.1037/0022-3514.89.5.717 Collins, E. C., Biernat, M., & Eidelman, S. (2009). Stereotypes in the communication and translation of person impressions. Journal of Experimental Social Psychology, 45, 368–374. doi: 10.1016/j.jesp.2008.09.005 Cottrell, C. A., & Neuberg, S. L. (2005). Different emotional reactions to different groups: A sociofunctional threat-based approach to “prejudice”. Journal of Personality and Social Psychology, 88, 770–789. doi: 10.1037/0022-3514.88.5.770 Crandall, C. S., Eshleman, A., & O’Brien, L. T. (2002). Social norms and the expression and suppression of prejudice: The struggle for internalization. Journal of Personality and Social Psychology, 82, 359–378. doi: 10.1037/0022-3514.82.3.359 Cuddy, A. J., Fiske, S. T., Kwan, V. S., Glick, P., Demoulin, S., Leyens, J. P., . . . Sleebos, E. (2009). Stereotype content model across cultures: Towards universal similarities and some differences. The British Journal of Social Psychology, 48, 1–33. doi: 10.1348/014466608X314935 Davis, K. F., D’Odorico, P., Laio, F., & Ridolfi, L. (2013). Global spatio-temporal patterns in human migration: A complex network perspective. PLoS One, 8, e53723. doi: 10.1371/ journal.pone.0053723 Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of Personality and Social Psychology, 24, 285–290. doi: 10.1037/h0033731 Eagly, A. H., & Kite, M. E. (1987). Are stereotypes of nationalities applied to both women and men? Journal of Personality and Social Psychology, 53, 451–462. doi: 10.1037/00223514.53.3.451 Eckes, T. (2002). Paternalistic and envious gender stereotypes: Testing predictions from the stereotype content model. Sex Roles, 47, 99–114. doi: 10.1023/A:1021020920715 Federal Ministry of the Interior. (2007). Deutsche und Ausländer mit Migrationshintergrund [Germans and foreigners with an immigrant background]. Retrieved from http://www.en.bmi. bund.de/ Fiske, S. T., Cuddy, A. J. C., & Glick, P. (2007). Universal dimensions of social cognition: Warmth and competence. Trends in Cognitive Sciences, 11, 77–83. doi: 10.1016/j.tics. 2006.11.005 Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82, 878–902. doi: 10.1037/0022-3514.82.6.878 Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from category-based to individuating processes: Influences of information and motivation on attention and interpretation. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1–74). New York, NY: Academic Press. Freeman, J. B., & Ambady, N. (2011). A dynamic interactive theory of personal construal. Psychological Review, 118, 247–279. doi: 10.1037/a0022327 Froehlich, L., Martiny, S. E., Deaux, K., & Mok, S. Y. (2016). “It’s their responsibility, not ours”: Stereotypes about competence and causal attributions for immigrants’ academic

Ó 2017 Hogrefe Publishing

underperformance. Social Psychology, 47, 74–87. doi: 10.1027/ 1864-9335/a000260 Fuertes, J. N., Gottdiener, W. H., Martin, H., Gilbert, T. C., & Giles, H. (2012). A meta-analysis of the effects of speakers’ accents on interpersonal evaluations. European Journal of Social Psychology, 42, 120–133. doi: 10.1002/ejsp.862 Giles, H. (1971). Patterns of evaluation to R.P., South Welsh and Somerset accented speech. The British Journal of Social and Clinical Psychology, 10, 280–281. doi: 10.1111/j.2044-8260. 1971.tb00748.x Giles, H., Bourhis, R. Y., & Taylor, D. M. (1977). Towards a theory of language in ethnic group relations. In H. Giles (Ed.), Language, ethnicity, and intergroup relations (pp. 307–348). London, UK: Academic Press. Giles, H., & Johnson, P. (1981). The role of language in ethnic group relations. In J. C. Turner & H. Giles (Eds.), Intergroup behavior (pp. 199–243). Oxford, UK: Blackwell. Giles, H., & Johnson, P. (1987). Ethnolinguistic identity theory: A social psychological approach to language maintenance. International Journal of the Sociology of Language, 68, 69–99. doi: 10.1515/ijsl.1987.68.69 Giles, H., & Powesland, P. F. (1975). Speech style and social evaluation. Oxford, UK: Academic Press. Gluszek, A., & Dovidio, J. F. (2010). The way they speak: A social psychological perspective on the stigma of nonnative accents in communication. Personality and Social Psychology Review, 14, 214–237. doi: 10.1177/1088868309359288 Hansen, K. (2013). Interactions of accent and appearance in social categorization, impression formation, and economic decisions. (PhD Thesis). Jena, Germany: Friedrich Schiller University. Retrieved from http://www.db-thueringen.de/servlets/ DerivateServlet/Derivate-27328/Diss/HansenKarolina_Dissertation. pdf Hansen, K., Rakić, T., & Steffens, M. C. (2014). When actions speak louder than words: Preventing discrimination of nonstandard speakers. Journal of Language and Social Psychology, 33, 66–75. doi: 10.1177/0261927x13499761 Hansen, K., Steffens, M. C., Rakić, T., & Wiese, H. (2016). When appearance does not match accent: Neural correlates of ethnicity-related expectancy violations. Social Cognitive and Affective Neuroscience. Advance online article. doi: 10.1093/ scan/nsw148. Haug, S. (2008). Sprachliche Integration von Migranten in Deutschland [Language integration of immigrants in Germany]. Federal Government for Migration and Refugees. Retrieved from http://www.bamf. de/SharedDocs/Anlagen/DE/Publikationen/WorkingPapers/wp14sprachliche-integration.pdf?__blob=publicationFile Hirschfeld, L. A., & Gelman, S. A. (1997). What young children think about the relationship between language variation and social difference. Cognitive Development, 12, 213–238. doi: 10.1016/S0885-2014(97)90014-9 Hutter, R. C., & Crisp, R. (2006). Implications of cognitive busyness for the perception of category conjunctions. The Journal of Social Psychology, 146, 253–256. doi: 10.3200/ SOCP.146.2.253-256 Jackson, L. A., Sullivan, L. A., & Hodge, C. N. (1993). Stereotype effects of attributions, predictions, and evaluations: No two social judgments are quite alike. Journal of Personality and Social Psychology, 65, 69–84. doi: 10.1037/0022-3514. 65.1.69 Jussim, L., Coleman, L. M., & Lerch, L. (1987). The nature of stereotypes: A comparison and integration of three theories. Journal of Personality and Social Psychology, 52, 536–546. doi: 10.1037/0022-3514.52.3.536 King-O’Riain, R. C., Small, S., Mahtani, M., Song, M., & Spickard, P. (2014). Global mixed race. New York, NY: NYU Press.

Experimental Psychology (2017), 64(1), 27–36

Kinzler, K. D., Dupoux, E., & Spelke, E. S. (2007). The native language of social cognition. Proceedings of the National Academy of Sciences of the United States of America, 104, 12577–12580. doi: 10.1073/pnas.0705345104 Kinzler, K. D., Shutts, K., Dejesus, J., & Spelke, E. S. (2009). Accent trumps race in guiding children’s social preferences. Social Cognition, 27, 623–634. doi: 10.1521/soco.2009.27.4.623 Klingst, M., & Drieschner, F. (2005). Raus aus dem Ghetto [Get out of the the ghetto]. Die Zeit, 46. Retrieved from http://www.zeit. de/2005/46/dt__Integration Kunda, Z., & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A parallel-constraintsatisfaction theory. Psychological Review, 103, 284–308. doi: 10.1037/0033-295X.103.2.284 McKirnan, D. J., Smith, C. E., & Hamayan, E. V. (1983). A sociolinguistic approach to the belief-similarity model of racial attitudes. Journal of Experimental Social Psychology, 19, 434–447. doi: 10.1016/0022-1031(83)90021-5 Minear, M., & Park, D. C. (2004). A lifespan database of adult facial stimuli. Behavior Research Methods, Instruments & Computers, 36, 630–633. doi: 10.3758/BF03195583 Ng, S. H. (2007). Language-based discrimination: Blatant and subtle forms. Journal of Language and Social Psychology, 26, 106–122. doi: 10.1177/0261927X07300074 Pietraszewski, D., & Schwartz, A. (2014a). Evidence that accent is a dedicated dimension of social categorization, not a byproduct of coalitional categorization. Evolution and Human Behavior, 35, 51–57. doi: 10.1016/j.evolhumbehav.2013.09.005 Pietraszewski, D., & Schwartz, A. (2014b). Evidence that accent is a dimension of social categorization, not a byproduct of perceptual salience, familiarity, or ease-of-processing. Evolution and Human Behavior, 35, 43–50. doi: 10.1016/ j.evolhumbehav.2013.09.006 Rakić, T., Steffens, M. C., & Mummendey, A. (2011). Blinded by the accent! The minor role of looks in ethnic categorization. Journal of Personality and Social Psychology, 100, 16–29. doi: 10.1037/ a0021522 Remedios, J. D., Chasteen, A. L., Rule, N. O., & Plaks, J. E. (2011). Impressions at the intersection of ambiguous and obvious social categories: Does gay + Black = likable? Journal of Experimental Social Psychology, 47, 1312–1315. doi: 10.1016/ j.jesp.2011.05.015 Rödin, M., & Özcan, G. (2011). Is it how you look or speak that matters? An experimental study exploring the mechanisms of ethnic discrimination (Research Paper No. 2011:12). Retrieved from http://econpapers.repec.org/RePEc:hhs: sunrpe:2011_0012 Roese, N. J., & Sherman, J. W. (2007). Expectancy. In A. W. Kruglanski & E. T. Higgins (Eds.), Social psychology: Handbook of basic principles (2nd ed., pp. 91–115). New York, NY: Guilford Press. Rosette, A. S., & Tost, L. P. (2010). Agentic women and communal leadership: How role prescriptions confer advantage to top women leaders. Journal of Applied Psychology, 95, 221–235. doi: 10.1037/a0018204 Ryan, E. B., Carranza, M. A., & Moffie, R. W. (1977). Reactions toward varying degrees of accentedness in the speech of

Experimental Psychology (2017), 64(1), 27–36

K. Hansen et al., Accent, Appearance, and First Impressions

Spanish-English bilinguals. Language and Speech, 20, 267–273. Shah, A. P. (2012). Accent modification for speakers with Indian accents: Speech characteristics and an assessment framework. Perspectives on Communication Disorders and Sciences in Culturally and Linguistically Diverse Populations, 19, 12–20. doi: 10.1044/cds19.1.12 Subtirelu, N. C., & Lindemann, S. (2014). Teaching first language speakers to communicate across linguistic difference: Addressing attitudes, comprehension, and strategies. Applied Linguistics, 20, 1–20. doi: 10.1093/applin/amu068 Tajfel, H., Billig, M., Bundy, R. P., & Flament, C. (1971). Social categorization and intergroup behaviour. European Journal of Social Psychology, 2, 149–178. doi: 10.1002/ejsp.2420010202 Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33–47). Monterey, CA: Brooks-Cole. US Census Bureau. (2011). Language use in the United States: 2011. Retrieved from http://www.census.gov/prod/2013pubs/ acs-22.pdf Vonk, R. (1999). Effects of other-profitability and self-profitability on evaluative judgements of behaviours. European Journal of Social Psychology, 29, 833–842. doi: 10.1002/(SICI)1099-0992 (199911)29:7<833::AID-EJSP950>3.0.CO;2-Z Yzerbyt, V., Provost, V. R., & Corneille, O. (2005). Not competent but warm. . . really? Compensatory stereotypes in the Frenchspeaking world. Group Processes & Intergroup Relations, 8, 291–308. doi: 10.1177/1368430205053944 Zeger, S. L., & Liang, K.-Y. (1986). Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 42, 121–130. doi: 10.2307/2531248 Zuckerman, M., & Driver, R. E. (1989). What sounds beautiful is good: The vocal attractiveness stereotype. Journal of Nonverbal Behavior, 13, 67–82. doi: 10.1007/BF00990791 Zuckerman, M., Miyake, K., & Hodgins, H. S. (1991). Crosschannel effects of vocal and physical attractiveness and their implications for interpersonal perception. Journal of Personality and Social Psychology, 60, 545–554. doi: 10.1037/ 0022-3514.60.4.545 Received March 4, 2016 Revision received October 31, 2016 Accepted November 2, 2016 Published online February 20, 2017

Karolina Hansen University of Warsaw Faculty of Psychology Stawki 5/7 00-183 Warszawa Poland karolina.hansen@psych.uw.edu.pl

Ó 2017 Hogrefe Publishing

Short Research Article

Automatic Retrieval of Newly Instructed Cue-Task Associations Seen in Task-Conflict Effects in the First Trial after Cue-Task Instructions Nachshon Meiran and Maayan Pereg Department of Psychology and Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel

Abstract: Novel stimulus-response associations are retrieved automatically even without prior practice. Is this true for novel cue-task associations? The experiment involved miniblocks comprising three phases and task switching. In the INSTRUCTION phase, two new stimuli (or familiar cues) were arbitrarily assigned as cues for up-down/right-left tasks performed on placeholder locations. In the UNIVALENT phase, there was no task cue since placeholder’s location afforded one task but the placeholders were the stimuli that we assigned as task cues for the following BIVALENT phase (involving target locations affording both tasks). Thus, participants held the novel cue-task associations in memory while executing the UNIVALENT phase. Results show poorer performance in the first univalent trial when the placeholder was associated with the opposite task (incompatible) than when it was compatible, an effect that was numerically larger with newly instructed cues than with familiar cues. These results indicate automatic retrieval of newly instructed cue-task associations. Keywords: intention-based reflexivity, NEXT paradigm, procedural working-memory

Almost every reaction-time experiment begins with instructions and healthy human adults typically follow the instructions without any apparent difficulty. Nonetheless, how symbolic/verbal instructions translate into motor action with such immediacy and efficiency remains a mystery. One hypothesis is that instruction information is translated into associations between a stimulus (or a stimulus category, e.g., “living”) and an action concept (e.g., “press the left key”) and these associations are held in workingmemory (WM), defined as a system for holding novel bindings between familiar elements (Baddeley, 2000; Oberauer, 2009). Once the relevant stimulus is presented, the action concept is retrieved, and the action itself is then retrieved via preexisting associations (“left” associated with the left-sided movement, see Meiran, Cole, & Braver, 2012). Some elements of the aforementioned hypothesis have already gained support. First, since instructions for a new task are by definition (of “new task”) not yet stored in long-term memory, they must be stored in WM. Accordingly, the involvement of WM in storing stimulusresponse links has already been acknowledged, especially Ó 2017 Hogrefe Publishing

through the notion of “procedural WM” (Oberauer, 2009; Oberauer, Souza, Druey, & Gade, 2013). A related distinction is between nonarbitrary stimulus-response mapping (such as mapping the digits “1,” “2,” “3,” and “4” to the leftmost, middle-left, middle-right, and rightmost keys) and arbitrary mapping (in which “9,” “3,” “8,” and “1” are mapped to the same keys). In the former case, a known (hence, long-term memory-based) rule relates stimuli to responses. In the latter case, no such rule exists, and the links must be stored in WM. Thus, WM-based stimulusresponse links are especially evident when the mapping is arbitrary. This premise has gained support from an individual differences study showing a high correlation between a WM latent variable and a variable describing the unique variance in choice reaction-time tasks that is due to stimulus-response rule arbitrariness (Wilhelm & Oberauer, 2006). Additionally, WM load manipulations selectively influence choice response times (RT) involving arbitrary rules, a fact that has been attributed to the rate of rule retrieval from WM (Shahar, Teodorescu, Usher, Pereg, & Meiran, 2014). Experimental Psychology (2017), 64(1), 37–48 DOI: 10.1027/1618-3169/a000349

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

Finally, the automatic response retrieval seen when the stimulus is presented was shown in various paradigms (Cohen-Kdoshay & Meiran, 2007; De Houwer, Beckers, Vandorpe, & Custers, 2005; Liefooghe, Wenke, & De Houwer, 2012; Wenke, Gaschler, & Nattkemper, 2007) producing a phenomenon termed “intention-based reflexivity” (IBR; Meiran et al., 2012). In these works, WM involvement is evident via the abolishment of IBR by WM load (CohenKdoshay & Meiran, 2007; Meiran & Cohen-Kdoshay, 2012). The automatic retrieval of actions based on instructions alone seems to offer a very promising answer to the question concerning how verbal/symbolic/semantic information translates immediately into action, and in the present work, we wanted to examine the generality of this principle. Specifically, most of the aforementioned works dealt with action retrieval (stimulus-response rules). Nonetheless, instructions are not restricted to (direct) action retrieval. For example, in cued task switching experiments (Meiran, 2014, for review) the instructions sometimes associate stimuli (“task cues”) with tasks in a broader sense. For example, “X” and “Y” might indicate a color and a shape task, respectively. In this example, the task cue (e.g., “X”) is not linked to any particular key, but to a decision type (e.g., respond according to the target’s color). In the present work, we thus asked the following: Research Question 1 (RQ1): would such newly instructed stimulus-task associations be automatically retrieved as do stimulus-response associations? To address this question, we designed a new version of the NEXT paradigm (Meiran, Pereg, Kessler, Cole, & Braver, 2015), which was previously used to study stimulusresponse associations. The original NEXT paradigm consisted of miniblocks of trials, each miniblock consisting of three phases. In the INSTRUCTION phase, two stimuli that were not presented in the experiment beforehand (e.g., two letters, two symbols) were arbitrarily mapped to the right and left keys. In the NEXT phase, participants were presented with stimuli colored in RED indicating that they should just advance the screen. Critically, the screen advancement responses were carried out by pressing the right key (or the left key for the other half of the participants). Thus, screen advancement responses could be either compatible (the stimulus is associated with the right key and the right key is used to advance the screen) or incompatible (the stimulus is associated with the left key, in this example) with the newly instructed, and not yet executed task. After a variable number of NEXT trials, a very brief GO phase began in which participants received the stimuli in GREEN and executed the instructed task. The most important finding is that compatible NEXT responses were quicker than Experimental Psychology (2017), 64(1), 37–48

incompatible NEXT responses, indicating a NEXT compatibility effect. This effect shows that seeing the stimulus leads to reflexive, unintentional retrieval of the response that has been linked to it in the instructions. The original NEXT paradigm consists of familiar responses (always the right and the left key press), and novel (hence, unfamiliar) stimulus-to-response associations and is thus suitable to study the automatic retrieval of novel stimulus-response associations. In the new paradigm used here, we created a cue-task analog in which we assessed the automatic retrieval of a cue-task association. To this end, we employed the task-conflict paradigm (Braverman, Berger, & Meiran, 2014; Braverman & Meiran, 2010, 2014), and specifically, its version used by Braverman and Meiran (2010). In Braverman and Meiran’s (2010) study, participants made right-left and up-down judgments performed on placeholder locations, and the authors took advantage of the fact that the identity of the placeholder was irrelevant for the task at hand. The critical manipulation took place with univalent targets in which placeholder’s locations afforded only one task (e.g., a stimulus in which the placeholder occupies the middle upward location, and thus one that could only be judged as UP, see Figure 1). Univalent stimuli do not require a task cue because they clearly indicate which task to execute. The manipulation in that study involved the (irrelevant) identity of the placeholder. Sometimes, this placeholder was compatible with the task (e.g., an up-down double-headed arrow occupying the upper location) and at other times, it was incompatible with it (e.g., a right-left double-headed arrow occupying the upper location). The authors also encouraged participants to process these double-headed arrows as task cues. This was achieved by including bivalent stimuli in which placeholder’s location afforded both tasks. Specifically, these locations were upper-left, upper-right, lower-left, or lowerright, that could be judged both along the right-left axis and the up-down axis. Since bivalent stimuli do not indicate which task to execute, they were presented together with double-headed arrows that served as task cues, informing participants which task to execute. For example, a stimulus in which the placeholder was placed in the upper-left location could be presented together with a right-left arrow, indicating that the task is right-left. The most important finding was a task-conflict effect, indicating poorer performance in incompatible univalent trials as compared with compatible univalent trials. This task-conflict effect was interpreted as evidence for the automatic retrieval of the task identity associated with the double-headed arrow, an identity that conflicted with the task identity dictated by the univalent stimulus. For example, if a right-left doubleheaded arrow occupied the upper location, the (irrelevant) arrow retrieved the right-left task, but the (relevant) Ó 2017 Hogrefe Publishing

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

(A)

(B)

Figure 1. Trial progression in a familiar (A) and an example novel miniblock (B).

location retrieved the up-down task, and this created a conflict between two task-identity representations that consequently led to poor performance. In the adapted NEXT paradigm used in the present paper, we also looked for task-conflict effects, but this time instead of using transparent and overlearned cues (doubleheaded arrows) as placeholders, we used newly instructed cues as placeholders. Like in the original NEXT paradigm, Ă&#x201C; 2017 Hogrefe Publishing

the modified paradigm consisted of miniblocks, each involving three phases. In the first, INSTRUCTION phase, we introduced two new stimuli that would serve as task cues for the two tasks in the last two trials of the miniblock (the BIVALENT phase, see below). The INSTRUCTION phase was followed by two phases involving task switching between up-down and right-left tasks. In the UNIVALENT phase (analogous to the original NEXT phase), there was no Experimental Psychology (2017), 64(1), 37â&#x20AC;&#x201C;48

need for a task cue because the targets included placeholders whose location afforded only one task. In this phase, the newly instructed cues served as placeholders that could be compatible or incompatible with the relevant task as in Braverman and Meiran’s (2010) experiments. This phase ended at an unexpected point in time, when the BIVALENT phase began (analogous to the original GO phase). In the BIVALENT phase, the newly instructed task cues had to be used as cues for the first time given that there was no other way to tell which task to execute because the placeholder’s location afforded both tasks. This phase involved only two trials, a fact which forced participants to maintain high readiness to process the task cues including during the UNIVALENT phase that preceded the BIVALENT phase. We predicted that, if the newly instructed task cues lead to automatic retrieval of task rules, better performance is expected for compatible than for incompatible univalent targets. We also compared the size of this task-conflict effect observed with newly instructed task cues to that observed with double-headed arrows, which are familiar and nonarbitrary cues. For these familiar cues, we predicted a replication of Braverman and Meiran’s “task-conflict effect,” although this was a relatively minor issue in the present work. Only when this paper was almost ready for submission, we became aware of the fact that Liefooghe (personal communication, April 11th, 2016) had just completed an analogous study. In his study, participants switched between font-related tasks, and the stimuli were nonwords (whose identity was irrelevant in this critical phase). These nonwords were newly instructed task cues and were used as cues immediately after the critical phase. Across two experiments, Liefooghe found reliable task-conflict effects from newly instructed task cues. He, however, did not examine how the effect changed across trials and did not compare the task-conflict effect to the effect found with familiar task cues as we have done. Focusing on how the effect changes with trial progression in the UNIVALENT phase is theoretically important, since only in the very first trial, a task-conflict effect provides relatively unequivocal evidence for intention-based reflexivity (Cohen-Kdoshay & Meiran, 2009). The rationale is based on a distinction between skill-based automaticity (e.g., Logan, 1988) reflecting retrieval from long-term memory; and intention-based reflexivity, reflecting retrieval of novel information that has not yet been placed in long-term memory. Accordingly, automaticity effects that are seen only in advanced trials and not in the first trial that follows the instructions could reflect some form of fast learning (e.g., Oberauer et al., 2013) in which the new information has been stored in long-term memory. For these reasons, only the first-trial effect provides a relatively unequivocal evidence for intention-based reflexivity. In other words, Experimental Psychology (2017), 64(1), 37–48

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

Liefooghe’s result could potentially arise from the advanced trials of the run and not from the first trials after the instructions. As such, they leave slight doubt regarding whether his task-conflict effect truly reflects intention-based reflexivity. Consequently, our analytic strategy was to test for taskconflict effects in the first univalent trial in miniblocks with novel cues using a focused contrast. This analysis provides an answer to our core question concerning whether newly instructed stimulus-task associations are automatically retrieved? We have already used this analytic strategy beforehand, for the reasons specified above (Meiran, Pereg, Kessler, Cole, & Braver, 2014) and have started using it routinely. In the next phase of the analysis, we examine the same effect in miniblocks with the double-headed arrows that represent familiar task cues, in order to answer the question whether this automatic retrieval of cue-task associations is stronger when the cues are familiar and nonarbitrary? Finally, we tested whether the task-conflict effect is also seen in the more advanced trials of the UNIVALENT phase by using exploratory Analysis of Variance (ANOVA).

Experiment 1a Method Participants Forty-one undergraduate students from the Ben-Gurion University of the Negev (35 women, mean age = 23.44, SD = 1.33) participated in the experiment. The compensation was either monetary (40 NIS, $11 US) or course credit. All the participants reported having normal or corrected-to-normal vision, and not having diagnosed attention deficits or learning disabilities. Sample size was determined using G-Power (Faul, Erdfelder, Lang, & Buchner, 2007) to provide Power > .80 for a two-sided t-test detecting dz = .40 (η2 = .061). Response keys were counterbalanced between participants, such that half of the participants responded with the keys “9” and “1” in the numerical keypad (where they occupy the up-right and lower-left location, respectively, and thus indicate up/right and down/left, respectively), and half responded with the keys “7” and “3” (up-left and down-right, respectively).

Materials and Procedure The procedure consisted of a series of short miniblocks each with either novel or familiar task cues. In each miniblock, the participants applied the instructed task cues only twice, in two bivalent trials at the end of the miniblock. Prior to the bivalent trials, participants were presented with Ó 2017 Hogrefe Publishing

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

0–5 univalent trials (sampled from a close-to exponential distribution, to make the point of transition from the UNIVALENT phase to the BIVALENT phase unpredictable). In miniblocks with familiar cues, the task cues were right-left or up-down double-headed arrows. In miniblocks with novel-cues, two arbitrary task cues were mapped to the tasks. These cues were the same stimuli used by Meiran et al. (2015) and were English and Hebrew letters, digits, symbols, and shapes that were randomly selected on each miniblock under the constraint that both belonged to the same broad category (digits, Hebrew/English letters, symbols, shapes). Since the task was rather complex, participants first practiced the tasks on bivalent stimuli, first the horizontal and vertical tasks, separately (36 trials each), and then in task switching (90 trials). Afterwards, participants practiced with univalent trials (10 trials of each task, separately, and 40 in a task-switching context). Finally, they practiced the new paradigm in four miniblocks, one with the familiar cues and three with novel cues. The experiment included 165 miniblocks (108 with novel cues and 57 with familiar cues, randomly ordered). The smaller number of familiar-cue miniblocks reflects the fact that this issue is of lesser interest in the present work. The INSTRUCTION phase ended when participants pressed the <enter> key, but not sooner than after 3 s has elapsed. In the UNIVALENT phase, no task cue was used, and in the BIVALENT phase, a cue that was presented for 500 ms (during the experimental phase and 300 ms in the practice phase) and then immediately masked by white noise scramble during the target presentation. Masking was used to force participants to maintain high readiness to quickly encode the task cues. Given their smaller number, miniblocks with familiar cues included three univalent trials (to ensure a sufficiently high number of trials per condition), except for one miniblock with zero univalent trial and one with five univalent trials (that were introduced at the beginning of the experiment to create an expectancy and are not further analyzed). To encourage participants to encode the cue-task association and then maintain it in a state of high readiness, we took the following steps. First, 75% of the bivalent trials were incompatible, and thus ones in which task cues must have been encoded. (Incompatible bivalent trials produce different responses according to the two task rules, and thus, produce errors when participants execute the wrong task because of failing to use the task cue). Second, we used a short cue-target interval and masked the cue. Third, we used an individually-tailored response deadline during the BIVALENT phase of the experimental part, which was set to be two times longer than the mean RT observed when participants reacted to the bivalent trials during practice. The reason for doubling RT is that the experimental part Ó 2017 Hogrefe Publishing

included novel cues that were expected to require longer to process. In the UNIVALENT phase, we set a response deadline of 2,000 ms. If participants did not respond until the deadline, the target disappeared and the trial was considered an error. Data Analysis Trials with an error on were omitted from all RT analyses as were correct trials with RT shorter than 100 ms (anticipation errors) or longer than 3,000 ms (outliers).

Results Univalent Trials First Novel Trial Our first question was whether the task-conflict effect can emerge immediately following novel stimuli instructions (Figure 2). In the proportion of errors (PE), there was a significant, albeit small 1% task-conflict effect (PE), F(1, 40) = 5.35, p = .02, MSE = 0.0004, η2p = .12. In response times (RT), the effect (4 ms) failed to reach significance, F(1, 40) = 0.91, p = .35, MSE = 334.94, η2p = .02. Novel Versus Familiar Rules Our second question concerned the difference between novel and familiar stimuli. In the first familiar trial, the task-conflict effect did not reach significance in either RT (6 ms), F(1, 40) = 0.71, p = .40, MSE = 1,136.70, η2p = .02, or PE (0.6%), F(1, 40) = 1.25, p = .27, MSE = 0.0002, η2p = .03. Thus, there was a significant task-conflict effect with novel cue-task associations and a nonsignificant effect with familiar associations. However, this does not imply that the difference between the effects is significant. To address this question, we used ANOVA with the within-subjects independent variables: Compatibility (compatible-incompatible) and Block-Type (novel-familiar). Since there was no taskconflict effect in RT for either novel or familiar rules, we only tested PE. The results showed a significant main effect for Trial-Type, F(1, 40) = 10.72, p < .01, MSE = 0.0005, η2p = .21, and Compatibility, F(1, 40) = 6.01, p = .02, MSE = 0.0003, η2p = .13, but the interaction (testing the difference between the two effects) did not reach significance, F(1, 40) = 1.61, p = .21, MSE = 0.0003, η2p = .04. Thus, despite their differential trends, we cannot conclude that the effect seen with novel cue-task associations is larger than that seen with familiar cue-task associations (Figure 2). Advanced Trials Finally, we tested whether the task-conflict effect emerged in the advanced Trials 2 and 3, and whether there was a difference between the novel and familiar task cues. To test this, we performed an ANOVA with three within-subjects Experimental Psychology (2017), 64(1), 37–48

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

Figure 2. First univalent trials – Experiment 1a. Interaction between Block-Type and Compatibility. Error bars represent within-subject confidence intervals (Jarmasz & Hollands, 2009).

460 440

RT (ms)

420 400 380 360 340 320 300 novel compatible

familiar incompatible

0.04 0.02 0 novel compatible

familiar incompatible

variables: Block-Type (novel-familiar), Compatibility (compatible-incompatible), and Univalent-Trial (2–3). In RT, a significant effect for Block-Type was found, F(1, 40) = 34.56, p < .001, MSE = 1,826.83, η2p = .46, indicating higher RT in the novel condition. The three-way interaction reached significance as well, F(1, 40) = 8.70, p < .01, MSE = 542.33, η2p = .18, showing a positive task-conflict effect only in the third familiar trial, F(1, 40) = 2.19, p = .15, MSE = 477.78, η2p = .05, 7 ms, and a reversed effect in the third novel trial, F(1, 40) = 6.08, p = .02, MSE = 525.72, η2p = .13, 12 ms. This reversed effect was however accompanied by an opposite trend seen in PE, suggesting that it might reflect speed-accuracy tradeoff rather that processing-efficiency differences (Figure 3). None of the other interactions including Compatibility reached significance. In PE, none of the main effects reached significance, as well as none of the interactions including Compatibility. Thus, there was no evidence for consistent task-conflict effect with novel cues in the advanced trials of the UNIVALENT phase. Bivalent Trials The only reason for including bivalent trials was to force participants to maintain the newly instructed cue-task Experimental Psychology (2017), 64(1), 37–48

associations in a highly accessible state. Although we did not have any interesting predictions regarding this phase, we report analyses of PE and RT for completeness sake (Figure 4). Results were analyzed in two-way ANOVAs with the within-subjects independent variables: Block-Type (novel-familiar) and Bivalent Trial (1–2). In RT, both main effects reached significance, F(1, 40) = 128.16, p < .001, MSE = 1,286.81, η2p = .76 (Block-Type); F(1, 40) = 120.60, p < .001, MSE = 962.55, η2p = .75 (Bivalent Trial), showing that RT were higher for novel rules, and that the first trial was slower than the second trial (i.e., the bivalent-trial effect). The interaction was nonsignificant, F(1, 40) = 3.75, p = .06, MSE = 275.34, η2p = .08, (68 vs. 58 ms for the novel and the familiar conditions, respectively). Similar results were found for PE, F(1, 40) = 9.51, p < .01, MSE = 0.002, η2p = .19 (Block-Type); F(1, 40) = 35.40, p < .001, MSE = 0.0008, η2p = .47 (Bivalent Trial); F(1, 40) = 0.36, p = .55, MSE = 0.0008, η2p < .01, (Interaction, 2.0% vs. 2.5% Bivalent Trial effect for novel and familiar, respectively). The Block-Type effect is quite trivial, and it would have been surprising not to find it. The Bivalent Trial effect closely resembles a similar effect seen in the NEXT paradigm (where it was called “Go-Trial Effect”). Ó 2017 Hogrefe Publishing

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

460

Figure 3. Advanced univalent trials – Experiment 1a. Interaction between Block-Type, Univalent Trial, and Compatibility. Error bars represent within-subject confidence intervals (Jarmasz & Hollands, 2009).

440

RT (ms)

420 400 380 360 340 320 300 Trial 2

Trial 3

Trial 2

novel

familiar

compatible

Trial 3

incompatible

0.04 0.02 0 Trial 2

Trial 3

Trial 2

novel compatible

familiar incompatible

Experiment 1b This experiment tested the novel-cue results of Experiment 1a, with an attempt to increase the effect by further increasing participants’ motivation to succeed in the bivalent phase.

Method Participants Thirty-four undergraduate students from Ben-Gurion University of the Negev (21 women, mean age = 24.59, SD = 2.29), similar in attributes to participants from the previous experiment, took part in the experiment for monetary compensation (40–50 NIS, 10–12 USD, depending on their performance, as will be elaborated on in the following section). The sample size was determined using G-Power (Faul et al., 2007) to detect a two-sided t-test dz = .5 with power > .80. Materials and Procedure The procedure was very similar to that in Experiment 1a, with a few minor changes. First, we focused only on novel rules, such that familiar instructions were not included in

Ó 2017 Hogrefe Publishing

Trial 3

the experimental part (after the practice phase). Second, we offered participants extra payment for excellent bivalent performance immediately at the end of the experiment. Participants were told that if they will be highly accurate and fast, they will receive extra 10 NIS. As in Experiment 1a, feedback regarding bivalent performance was given on every miniblock. Extra payment was given to participants whose RT/accuracy score (yet, with accuracy > .90) was above the 75th percentile of the participants from Experiment 1a.

Results Univalent Trials First Trial A pattern similar to that of Experiment 1a emerged in both RT and PE. In RT, the task-conflict effect (10 ms) failed to reach significance, F(1, 33) = 1.54, p = .22, MSE = 1,142.81, η2p = .04, but in PE, there was a significant, 5.5% effect, F(1, 33) = 8.09, p < .01, MSE = 0.006, η2p = .20 (Figure 5). Advanced Trials As in Experiment 1a, Trials 2 and 3 were tested in a twoway ANOVA with the within-subjects independent variables

Experimental Psychology (2017), 64(1), 37–48

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

Figure 4. Bivalent trials – Experiment 1a. Interaction between Block-Type and Bivalent Trial. Error bars represent within-subject confidence intervals (Jarmasz & Hollands, 2009).

550 500

RT (ms)

450 400 350 300 250 novel

familiar Trial 1

Trial 2

0.15 0.1 0.05 0 novel

familiar Trial 1

Trial 2

Trial (2–3) and Compatibility (compatible-incompatible). The interaction did not reach significance in RT, F(1, 33) = 0.45, p = .50, MSE = 902.38, η2p = .01, with a positive (9 ms and 1 ms) effect in Trials 2 and 3 (respectively), which did not reach significance. However, in PE, the effect was positive in Trial 2 (3.2%) and negative in Trial 3 ( 2.5%) and the interaction reached significance, F(1, 33) = 4.36, p = .04, MSE = 0.006, η2p = .12 (Figure 5). Bivalent Trials The results replicate Experiment 1a, with a Bivalent Trial effect in both RT, F(1, 33) = 76.27, p < .001, MSE = 1,595.99, η2p = .70, and PE, F(1, 33) = 19.14, p < .001, MSE = 0.001, η2p = .37] (Figure 6). Combining the Experiments Given that the results in the two experiments were similar, we decided to combine the two experiments (including only univalent trials, just the novel trials of Experiment 1a) in order to increase statistical power. In addition, since there was no evidence for speed-accuracy tradeoff we used LISAS (a linear integrated speed-accuracy score; Vandierendonck, 2016) as an integrated speed-accuracy measure.

Experimental Psychology (2017), 64(1), 37–48

Results The results were subjected to a three-way ANOVA with the between-subjects independent variable Experiment (1a, 1b) and the within-subjects independent variables Trial (1–3) and Compatibility (compatible-incompatible). All three main effects reached significance: The Experiment main effect shows generally worse univalent performance in Experiment 1b, F(1, 73) = 14.00, p < .001, MSE = 81,382.06, η2p = .16; the Compatibility main effect indicates a task-conflict effect of 19 LISAS units, F(1, 73) = 4.12, p = .04, MSE = 9,334.84, η2p = .05; and the main effect for Trial showed that the worse performance was in the first trial, as expected, F(2, 146) = 19.09, p < .001, MSE = 4,797.61, η2p = .21. Importantly, the three-way interaction did not reach significance, F(2, 146) = 2.54, p = .08, MSE = 5,506.97, η2p = .03, although there was a trend toward a larger task-conflict effect in the first two trials of Experiment 1b (Figure 7A). Finally, the two-way interaction between Trial and Compatibility reached significance, F(2, 146) = 7.44, p < .001, MSE = 5,506.97, η2p = .09, showing the largest effect in Trial 1, F(1, 73) = 12.77, p < .001, MSE = 6,476.49,

Ó 2017 Hogrefe Publishing

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

Figure 5. Univalent trials – Experiment 1b. Interaction between Compatibility and Trial. Error bars represent within-subject confidence intervals (Jarmasz & Hollands, 2009).

460 440

RT (ms)

420 400 380 360 340 320 300 Trial 1

Trial 2 compatible

Trial 3

incompatible

0.15 0.1 0.05 0 Trial 1

Trial 2 compatible

incompatible

η2p = .15, a smaller effect in Trial 2, F(1, 73) = 4.07, p = .047, MSE = 6,357.92, η2p = .05, and a nonsignificantly-reversed effect in Trial 3, F(1, 73) = 1.57, p = .21, MSE = 7,514.38, η2p = .02 (Figure 7B).

550

RT (ms)

500 450 400 350

Bayesian Analyses

300

The LISAS first-trial scores were subjected to a repeatedmeasures Bayesian ANOVA using the freely available JASP software (Love et al., 2015). This ANOVA included the between-subjects independent variable Experiment and the within-subjects independent variable Compatibility. The main effect for Compatibility had BF10 = 18.44. This result indicates that, given equal prior probability for H1 and H0, the posterior probability based on the current results indicates that H1 is 18.44 times more probable than H0. Additionally, the model with the highest Bayes Factor was the model with both main effects and an interaction (BF10 = 620.75), which is considered very strong evidence for H1; whereas the unique contribution of the interaction

250 Trial 1

Trial 2

Trial 1

Trial 2

0.15

Trial 3

0.1 0.05 0

Figure 6. Bivalent trial effect – Experiment 1b. Error bars represent within-subject confidence intervals (Jarmasz & Hollands, 2009).

Ó 2017 Hogrefe Publishing

Experimental Psychology (2017), 64(1), 37–48

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

Figure 7. LISAS – combined speed and accuracy scores. (A) The interaction between Experiment, Trial, and Compatibility. (B) The interaction between Trial and Compatibility pooled across experiments. Error bars represent within-subject confidence intervals (Jarmasz & Hollands, 2009).

(A) 650 600

LISAS

550 500 450 400 350 300 Trial 1

Trial 2

Trial 3

Trial 1

Experiment 1a compatible

Trial 2

Trial 3

Experiment 1b incompatible

(B) 650 600

LISAS

550 500 450 400 350 300 Trial 1

Trial 2 compatible

incompatible

between Experiment and Compatibility over a model with just the main effects (BF10 = 323.78) was relatively small (BF10 = 1.91). Finally, the two main effect model was clearly better than an Experiment-only model (BF10 = 18.50), and the relative contribution of Compatibility from this perspective had BF10 = 17.50. Thus, no matter how we compute it, there is strong evidence for a task-conflict effect in the first trial.

Discussion The present work asked whether newly instructed cue-task associations are retrieved automatically. To answer this question we combined elements from the NEXT paradigm (Meiran et al., 2015) and the task-conflict paradigm

Experimental Psychology (2017), 64(1), 37–48

Trial 3

(Braverman & Meiran, 2010) and asked whether newly instructed task cues would produce task-conflict effects before they have ever been used as cues beforehand? We also examined whether this task-conflict effect already shows in the first univalent trials which follows the instructions and whether it remains visible in subsequent univalent trials. Finally, we compared the task-conflict effects to those seen using familiar (and repeatedly used) task cues. Across the two experiments, we obtained strong decisive support for a task-conflict effect in the LISAS scores of the first novel trial, which is when the new task cues were never responded to beforehand. We can additionally say that the effect found with novel cues in the first UNIVALENT trial was not smaller and possibly larger than the effect seen in more advanced trials (with novel cues) or with familiar cues.

Ó 2017 Hogrefe Publishing

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

The lack of clear task-conflict effect with familiar cues seems like a failure replicating the original effect reported by Braverman and Meiran (2010). We, however believe this mismatch is probably due to the many procedural differences between the experiments, since the task-conflict effect seems quite robust and has been repeatedly found (see also Braverman et al., 2014; Braverman & Meiran, 2014).1 In conclusion, the present experiments show evidence that newly instructed task cues lead to automatic retrieval of task rules. This is seen in poorer performance in univalent trials (in which no task cue is needed) when seeing a cue related to the alternative task as compared to seeing a cue related to the required task.

Acknowledgments This research was supported by a research grant from the USA-Israel Binational Science Foundation #2011246 awarded to Nachshon Meiran and Todd S. Braver. Electronic Supplemental Materials The electronic supplementary material is available with the online version of the article at http://dx.doi.org/10.1027/ 1618-3169/a000349 ESM 1. Data file (csv). Raw data. ESM 2. Data file (xls). TCE data for analyses.

References Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4, 417–423. doi: 10.1016/S1364-6613(00)01538-2 Braverman, A., Berger, A., & Meiran, N. (2014). The hierarchy of task decision and response selection: A task-switching event related potentials study. Brain and Cognition, 88, 35–42. doi: 10.1016/j.bandc.2014.04.006 Braverman, A., & Meiran, N. (2010). Task conflict effect in task switching. Psychological Research, 74, 568–578. doi: 10.1007/ s00426-010-0279-2 Braverman, A., & Meiran, N. (2014). Conflict control in task conflict and response conflict. Psychological Research, 88, 35–42. doi: 10.1007/s00426-014-0565-5 Cohen-Kdoshay, O., & Meiran, N. (2007). The representation of instructions in working memory leads to autonomous response activation: Evidence from the first trials in the flanker paradigm. The Quarterly Journal of Experimental Psychology, 60, 1140–1154. doi: 10.1080/17470210600896674 Cohen-Kdoshay, O., & Meiran, N. (2009). The representation of instructions operates like a prepared reflex: Flanker compatibility effects found in first trial following S-R instructions.

Experimental Psychology, 56, 128–133. doi: 10.1027/16183169.56.2.128 De Houwer, J., Beckers, T., Vandorpe, S., & Custers, R. (2005). Further evidence for the role of mode-independent short-term associations in spatial Simon effects. Perception & Psychophysics, 67, 659–666. doi: 10.3758/ BF03193522 Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191. doi: 10.3758/BF03193146 Jarmasz, J., & Hollands, J. G. (2009). Confidence intervals in repeated measures designs: The number of observations principle. Canadian Journal of Experimental Psychology, 63, 124–138. doi: 10.1037/a0014164 Liefooghe, B., Wenke, D., & De Houwer, J. (2012). Instructionbased task-rule congruency effects. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38, 1325–1335. doi: 10.1037/a0028148 Logan, G. D. (1988). Toward an instance theory of automatization. Psychological Review, 95, 492–527. doi: 10.1037/0033-295X. 95.4.492 Love, J., Selker, R., Marsman, M., Jamil, T., Dropmann, D., Verhagen, A. J., & Wagenmakers, E. J. (2015). JASP (Version 0.7) [Computer software]. Meiran, N. (2014). The task cuing paradigm: A user’s guide. In J. A. Grange & G. Houghton (Eds.), Task switching and cognitive control (pp. 45–73). New York, NY: Oxford University Press. Meiran, N., & Cohen-Kdoshay, O. (2012). Working memory load but not multitasking eliminates the prepared reflex: Further evidence from the adapted flanker paradigm. Acta Psychologica, 139, 309–313. doi: 10.1016/j.actpsy.2011.12.008 Meiran, N., Cole, M. W., & Braver, T. S. (2012). When planning results in loss of control: Intention-based reflexivity and working-memory. Frontiers in Human Neuroscience, 6, 104. doi: 10.3389/fnhum.2012.00104 Meiran, N., Pereg, M., Kessler, Y., Cole, M. W., & Braver, T. S. (2014). Reflexive activation of newly instructed stimulusresponse rules: Evidence from lateralized readiness potentials in no-go trials. Cognitive, Affective, & Behavioral Neuroscience, 15, 365–373. doi: 10.3758/s13415-014-0321-8 Meiran, N., Pereg, M., Kessler, Y., Cole, M. W., & Braver, T. S. (2015). The power of instructions: Proactive configuration of stimulus-response translation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 41, 768–786. doi: 10.1037/xlm0000063 Oberauer, K. (2009). Design for a working memory. Psychology of Learning and Motivation, 51, 45–100. Oberauer, K., Souza, A. S., Druey, M. D., & Gade, M. (2013). Analogous mechanisms of selection and updating in declarative and procedural working memory: Experiments and a computational model. Cognitive Psychology, 66, 157–211. doi: 10.1016/j.cogpsych.2012.11.001 Shahar, N., Teodorescu, A. R., Usher, M., Pereg, M., & Meiran, N. (2014). Selective influence of working memory load on exceptionally slow reaction times. Journal of Experimental Psychology: General, 143, 1837–1860. doi: 10.1037/a0037190 Vandierendonck, A. (2016). A comparison of methods to combine speed and accuracy measures of performance: A rejoinder on the binning procedure. Behavior Research Methods. Advance online publication. doi: 10.3758/s13428-016-0721-5

In a visit to Konstanz by Meiran that took place a few years ago, Grzyb and Hübner reported having replicated this effect in their laboratory too.

Ó 2017 Hogrefe Publishing

Experimental Psychology (2017), 64(1), 37–48

Wenke, D., Gaschler, R., & Nattkemper, D. (2007). Instructioninduced feature binding. Psychological Research, 71, 92–106. doi: 10.1007/s00426-005-0038-y Wilhelm, O., & Oberauer, K. (2006). Why are reasoning ability and working memory capacity related to mental speed? An investigation of stimulus-response compatibility in choice reaction time tasks. European Journal of Cognitive Psychology, 18, 18–50. doi: 10.1080/09541440500215921

Experimental Psychology (2017), 64(1), 37–48

N. Meiran & M. Pereg, Newly Instructed Cue-Task Associations

Received April 17, 2016 Revision received September 14, 2016 Accepted November 8, 2016 Published online February 20, 2017 Nachshon Meiran Department of Psychology and Zlotowski Center for Neuroscience Ben-Gurion University of the Negev 84105 Beer-Sheva Israel nmeiran@bgu.ac.il

Ó 2017 Hogrefe Publishing

line n o e e e fr e issu l p m sa

Journal of Individual Differences Editor-in-Chief André Beauducel Rheinische Friedrich-Wilhelms-Universität, Bonn, Germany Associate Editors Philip J. Corr, UK Sam Gosling, USA Jürgen Hennig, Germany Philipp Y. Herzberg, Germany Aljoscha Neubauer, Austria Thomas Rammsayer, Switzerland Karl-Heinz Renner, Germany

ISSN-Print 1614-0001 ISSN-Online 2151-2299 ISSN-L 1614-0001 4 issues per annum (= 1 volume)

Subscription rates (2017) Libraries / Institutions US $324.00 / € 249.00 Individuals US $159.00 / € 114.00 Postage / Handling US $16.00 / € 12.00

www.hogrefe.com

About the Journal Researchers, teachers, and students interested in all areas of individual differences (e.g., gender, temperament, personality, intelligence) and their assessment in human and animal research will find the Journal of Individual Differences useful. It publishes manuscripts dealing with individual differences in behavior, emotion, cognition, and their developmental aspects. This includes human as well as animal research. The Journal of Individual Differences is conceptualized to bring together researchers working in different areas ranging from, for example, molecular genetics to theories of complex behavior. Moreover, it places emphasis on papers dealing with special methodological and conceptual issues in basic science as well as in their applied fields (assessment of personality and intelligence).

Willibald Ruch, Switzerland Stefan Schmukle, Germany Astrid Schütz, Germany Andrzej Sekowski, Poland Jutta Stahl, Germany Martin Voracek, Austria

Manuscript Submissions All manuscripts should be submitted online at www.editorialmanager.com/jindivdiff, where full instructions to authors are also available. Electronic Full Text The full text of the journal – current and past issues (from 1999 onward) – is available online at econtent.hogrefe.com/loi/jid (included in subscription price). A free sample issue is also available here. Abstracting Services The journal is abstracted / indexed in Current Contents / Social and Behavioral Sciences (CC / S&BS), Social Sciences Citation Index (SSCI), Scopus, PsycINFO, and PSYNDEX. Impact Factor (Journal Citation Reports®, Thomson Reuters): 2015 = 1.135

Fredrike Bannink

Handbook of Positive Supervision for Supervisors, Facilitators, and Peer Groups 2015, xii + 204 pp. US $49.00 / € 34.95 ISBN 978-0-88937-465-2 Also available as eBook

Positive supervision focuses on what actually works instead of on problems and on supervisees’ strengths rather than on their deficits. The task of supervisors using this approach is – unlike the more traditional problemsolving – to create solutions with their supervisees and to teach them to apply the same approach when working with their own clients.

Essential reading for all supervisors, this book introduces a new form of supervision, based on positive psychology and solution-focused brief therapy, that is shorter, more positive and hopeful, and more cost-effective than traditional methods.

Fredrike Bannink

Handbook of Solution-Focused Conflict Management 2010, xii + 180 pp., hardcover US $49.00 / € 34.95 ISBN 978-0-88937-384-6 Also available as eBook

Benjamin Franklin once said: “Every problem is an opportunity in disguise.” In the new and highly successful approach of solution-focused conflict management described here, the focus is on discovering these opportunities to find the “win-win” scenario. The key lies in asking eliciting questions about goals, exceptions, and competencies and in motivating clients to change. Clients’ perspectives are considered primary, and they are empowered to formulate their own hopes for the future and to devise ways to make them happen. Focusing on the preferred future facilitates change in the desired direction.

www.hogrefe.com

This book is essential reading for all those who need to manage conflicts. It provides a detailed description of the highly successful solution-focused model, its theoretical background, and practical applications in conflict management practice: divorce, workplace, family, neighbors, personal injury cases, victim-offender conflicts. Kenneth Cloke, one of the most inspiring mediators in the world, wrote the Foreword and Epilogue.

Short Research Article

Individual Differences in the Flexibility of Peripersonal Space Samuel B. Hunley,1 Arwen M. Marker,2 and Stella F. Lourenco1 1

Department of Psychology, Emory University, Atlanta, GA, USA

Clinical Child Psychology Program, University of Kansas, KS, USA

Abstract: The current study investigated individual differences in the flexibility of peripersonal space (i.e., representational space near the body), specifically in relation to trait claustrophobic fear (i.e., fear of suffocating or being physically restricted). Participants completed a line bisection task with either a laser pointer (Laser condition), allowing for a baseline measure of the size of one’s peripersonal space, or a stick (Stick condition), which produces expansion of one’s peripersonal space. Our results revealed that individuals high in claustrophobic fear had larger peripersonal spaces than those lower in claustrophobic fear, replicating previous research. We also found that, whereas individuals low in claustrophobic fear demonstrated the expected expansion of peripersonal space in the Stick condition, individuals high in claustrophobic fear showed less expansion, suggesting decreased flexibility. We discuss these findings in relation to the defensive function of peripersonal space and reduced attentional flexibility associated with trait anxieties. Keywords: peripersonal space, spatial attention, claustrophobic fear, perception, line bisection

Although the physical environment contains no separation between the space near the body and the space farther away, there is now ample evidence that the brains of primates impose such a boundary. Specifically, there are dissociable representations for the space close to the body, known as “peripersonal” space, and the space far from the body, known as “extrapersonal” space. It has been suggested that this dissociation is due to the special status afforded to peripersonal space, which supports an organism’s interactions with objects in the environment and which is where an organism is most susceptible to threats (e.g., de Vignemont & Iannetti, 2015; Graziano & Cooke, 2006; Previc, 1998; Rizzolatti, Scandolara, Matelli, & Gentilucci, 1981). Neural codes for peripersonal space have been found in brain regions such as the macaque periarcuate cortex (Rizzolatti, Matelli, & Pavesi, 1983; Rizzolatti et al., 1981), as well as the human intraparietal sulcus (IPS) and supramarginal gyrus (SMG; Brozzoli, Gentile, Petkova, & Ehrsson, 2011; Holt et al., 2014; Makin, Holmes, & Zohary, 2007), confirming neural specialization for the space near the body. Other evidence suggests that a critical feature of peripersonal space is its flexibility (Ackroyd, Riddoch, Humphreys, Nightingale, & Townsend, 2002; Berti & Frassinetti, 2000; Canzoneri, Ubaldi, Rastelli, Finisguerra, Bassolino, & Serino, 2013; Longo & Lourenco, 2006; Noel, Pfeiffer, Blanke, & Serino, 2015; Serino, Canzoneri, Marzolla, Di Pellegrino, & Magosso, 2015; Taffou & Viaud-Delmon, 2014). It has been found Ó 2017 Hogrefe Publishing

that the boundary associated with peripersonal space is flexible, adapting to changes in motor capabilities, social input, and threat level. In the current study, we examined the extent of this flexibility in relation to trait claustrophobic fear. In one of the first experimental demonstrations of the flexibility of peripersonal space, Iriki, Tanaka, and Iwamura (1996) demonstrated peripersonal space expansion in monkeys after tool use, using a single unit recording paradigm. Specifically, neurons that initially fired only for objects near the hand, subsequently fired for objects placed near the end of a tool that had been used to complete the task of reaching a target. Analogous effects have been reported in human beings using a line bisection task (e.g., Ackroyd et al., 2002; Berti & Frassinetti, 2000; Longo & Lourenco, 2006). In this task, neurologically-healthy human participants bisect lines slightly to the left of center when a line is presented relatively close to the body, an effect known as pseudoneglect (e.g., Jewell & McCourt, 2000). Critically, the bias shifts rightward as the distance of the presented lines increases from the body, such that, when a line is presented far from the body, many participants will display a rightward bias (Kinsbourne, 1987; Longo, Trippier, Vagnoni, & Lourenco, 2015). The relative rate of the rightward shift can be used to characterize the extent, or “size,” of an individual’s peripersonal space. A more gradual rightward shift in bias with increasing distance suggests a larger peripersonal space, whereas Experimental Psychology (2017), 64(1), 49–55 DOI: 10.1027/1618-3169/a000350

a more rapid rightward shift in bias with increasing distance suggests a smaller peripersonal space (Farnè & Làdavas, 2000; Longo & Lourenco, 2007). This task has been used to demonstrate both expansion of peripersonal space with tool use (e.g., sticks; Ackroyd et al., 2002; Berti & Frassinetti, 2000; Longo & Lourenco, 2006) and contraction of peripersonal space when one’s ability to act is impeded such as when bisecting lines while wearing wrist weights (Lourenco & Longo, 2009). Though there is debate about the mechanisms that support changes in peripersonal space, one possibility is that the so-called body schema is altered (Maravita & Iriki, 2004). For instance, a tool that extends one’s reach is incorporated into one’s representation of the body (see Holmes, 2012, for alternate explanation). Regardless of the exact mechanism, shifts in peripersonal space have been replicated across a variety of populations, including amputees who show expansion of peripersonal space when wearing their prosthetic limb as compared to when it is removed (Canzoneri, Marzolla, Amoresano, Verni, & Serino, 2013) and blind individuals who have expanded peripersonal spaces when using their cane compared to when they are without it (Serino, Bassolino, Farnè, & Làdavas, 2007). Other research suggests that the flexibility of peripersonal space is not limited to direct manipulations of the body’s ability to act. For instance, illusions that produce feelings of ownership of another person’s body or body parts, such as the full body illusion or enfacement illusion (i.e., illusory possession of another person’s face), lead to expansion of peripersonal space so as to include the newly “possessed” body or body parts (Maister, Cardini, Zamariola, Serino, & Tsakiris, 2015), even if the other body belongs to a virtual avatar (Noel et al., 2015). Additionally, researchers examining peripersonal space using a measure of attentional bias found that participants had larger peripersonal spaces in response to a threatening auditory stimulus (i.e., a growling dog) when compared to a nonthreatening auditory stimulus (i.e., a sheep bleating; Taffou & Viaud-Delmon, 2014). Likewise, expanding peripersonal space through active tool use leads participants to respond to potential threats sooner (Rossetti, Romano, Bolognini, & Maravita, 2015). Taken together, the extant findings suggest flexible peripersonal space representations that accommodate to changing action-based demands, sensorimotor feedback, and environmental threats. Nevertheless, an open question is whether there are meaningful individual differences in the flexibility of peripersonal space. In the current paper, we investigated this possibility by examining peripersonal space in relation to claustrophobic fear. Previous research suggests that individuals high in claustrophobic fear exhibit distortions both in their representations of peripersonal space Experimental Psychology (2017), 64(1), 49–55

S. B. Hunley et al., Peripersonal Space Flexibility

(Lourenco, Longo, & Pathman, 2011) and their perceptions of egocentric distance (Hunley, Park, Longo, & Lourenco, 2016) compared to individuals low in claustrophobic fear. Given these distortions, here we tested whether a relation exists between claustrophobic fear and the flexibility of peripersonal space. Participants were given a line bisection task in which they bisected physical lines of different lengths from different viewing distances using either a laser pointer or a stick of appropriate length. When participants employ a laser pointer to respond on a line bisection task, it allows for an assessment of the default size of peripersonal space. When participants employ a stick to bisect lines, it serves to extend the range of one’s reach, and thus allows for an assessment of the extent of expansion of peripersonal space (e.g., Ackroyd et al., 2002; Longo & Lourenco, 2006). In both cases, the rightward shift in bias with increasing distance is used to capture the size of peripersonal space, such that a more gradual rightward shift indicates a larger peripersonal space (e.g., Longo et al., 2015). Consequently, by examining participants’ performance on a line bisection task under these different conditions, we were in a position to examine individual differences in flexibility as they related to claustrophobic fear.

Method Participants A total of 70 undergraduate students (49 females) between 18 and 22 years of age (M = 19.43 years) participated for course credit or payment. One participant failed to complete the study in its entirety due to an unrelated illness and was thus excluded from statistical analyses. Sample size was decided a priori based on previous work examining peripersonal space expansion (e.g., Ackroyd et al., 2002; Berti & Frassinetti, 2000; Longo & Lourenco, 2006) and individual differences in peripersonal space (see Lourenco et al., 2011). Most participants were right-handed (M = 76.83, range: 72.73 to 100), as measured by the Edinburgh Handedness Inventory (EHI; Oldfield, 1971). All had normal or corrected-to-normal vision. Procedures were approved by the local Ethics Committee.

Materials and Procedure Prior to testing, participants completed the EHI. The line bisection task took place in a large, square room (wall length: 3.8 m; height: 2.9 m) where participants were tested individually. They were randomly assigned to either the Laser or the Stick condition in which they were instructed to indicate the center of individually presented lines while Ó 2017 Hogrefe Publishing

S. B. Hunley et al., Peripersonal Space Flexibility

standing at one of four distances from the lines (30, 60, 90, or 120 cm) using either a laser pointer (Laser condition; n = 34) or a stick (Stick condition; n = 35). A betweensubjects design was utilized to reduce the likelihood of cross-over effects between Laser and Stick conditions. All lines were presented in a horizontal orientation (height: 1 mm) and centered on legal-sized sheets of paper (21.6 35.6 cm). We used three different line lengths (10, 20, and 30 cm). The sheets of paper were attached to a wall 56 cm above the ground. In the Laser condition, the laser pointer was attached to the head of a tripod that was kept at a constant height of 115 cm. Following existing research (e.g., Lourenco & Longo, 2009), the tripod was placed adjacent to the right side of each participant so that the tip of the laser pointer was even with the front of his or her body at each of the distances. In the Stick condition, participants held each stick in a comfortable grasp, with their arms tucked close to their body, not outstretched. There were four sticks, one corresponding to each of the distances, and each tapered to a point at one end to allow participants to bisect lines precisely. We did not include an active training component with the sticks, given that previous research with the line bisection task has not found this manipulation necessary to produce expansion of peripersonal space (e.g., Ackroyd et al., 2002; Berti & Frassinetti, 2000; Longo & Lourenco, 2006). In both conditions, participants indicated when they were satisfied with each of their responses and an experimenter (who had been out of view of the participant and was blind to the experimental hypotheses) marked each response on the sheet of paper. Participants were given unlimited time to respond, but they were encouraged to respond quickly. Each condition consisted of two fully crossed sets of trials (3 line lengths 4 distances 2 = 24 trials), with each participant receiving three blocks (72 trials total). The procedure used for the line bisection task has been shown to have good test-retest reliability (r = .82; Longo & Lourenco, 2007). Following completion of the line bisection task, participants were given the Claustrophobia Questionnaire (CLQ; Radomsky, Rachman, Thordarson, McIsaac, & Teachman, 2001), a self-report measure with 26 items (2 subscales: suffocation [SS] and restriction [RS]). Each item corresponded to a specific situation (SS: e.g., “using an oxygen mask”; RS: e.g., “in a crowded train stopped between stations”). Participants rated each item in terms of how anxious they would feel in the specific situation. Items were rated on a scale of 0–4, with 0 indicating “not at all anxious” and 4 indicating “extremely anxious.” The CLQ has high internal consistency (Cronbach’s α = .95) and excellent test-retest reliability (r = .89; Radomsky et al., 2001). Ó 2017 Hogrefe Publishing

Results Bisection responses were measured off-line by two coders who never disagreed by more than 0.25 mm. We estimated the size of peripersonal space using least-squares linear regression to determine the rate at which bias shifted rightward with increasing distance, as in previous studies (e.g., Longo & Lourenco, 2006; Lourenco et al., 2011). For each participant, we regressed rightward bias (% of line length) on distance to compute the slope of the best-fitting line (see Electronic Supplementary Material 1 for raw data). Participants in both conditions demonstrated a rightward shift in bias over distance, as exhibited by the slope of the regression line in each condition (Laser condition: M = 1.30% line length/m, SD = 1.02, t[33] = 7.42, p < .001, d = 2.58; Stick condition: M = 0.64% line length/m, SD = 0.95, t[34] = 4.02, p < .001, d = 1.38). Critically, however, participants in the Laser condition had steeper slopes than participants in the Stick condition (Mdiff = 0.66, SEdiff = 0.24, 95% CI [0.19, 1.13]), t(67) = 2.77, p = .007, d = 0.67 (see Figure 1). This finding is consistent with smaller peripersonal spaces in the Laser than Stick condition, and it replicates previous research which suggests that wielding a stick serves to expand peripersonal space (e.g., Ackroyd et al., 2002; Berti & Frassinetti, 2000; Longo & Lourenco, 2006). To investigate the relationship between claustrophobic fear and performance on the line bisection task, we conducted a linear regression analysis, with condition (Laser, Stick) and CLQ (total) scores (see Table 1) as factors, and the slopes of participants’ performance on the line bisection task as the dependent variable. This analysis revealed a significant main effect of condition (β = 1.11, 95% CI [ 1.66, 0.56]), t(68) = 4.01, p < .001, d = 1.16, as well as a significant interaction between condition and CLQ scores (β = 1.33, 95% CI [0.48, 2.17]), t(68) = 3.12, p = .003, d = 0.77. Thus, participants’ line bisection performance differed across condition as a function of their level of claustrophobic fear. There was no main effect of CLQ scores (β = 0.28, 95% CI [ 0.60, 0.04]), t(68) = 1.72, p = .090. To follow up on the significant interaction revealed in the regression analysis above, we further analyzed each condition. We first examined performance in the Laser condition using a linear regression model of participants’ slopes on the line bisection task with CLQ (total) scores as a predictor variable. This analysis revealed that CLQ (total) scores did not significantly predict participants’ slopes on the line bisection task (β = 0.27, 95% CI [ 0.62, 0.07]), t(33) = 1.60, p = .12. However, separate linear regression models for each of the CLQ subscales revealed a significant negative relationship between participants’ scores on the SS subscale and their performance on Experimental Psychology (2017), 64(1), 49–55

S. B. Hunley et al., Peripersonal Space Flexibility

Figure 1. Mean bias at each distance in Laser and Stick conditions, as well as the regression lines for each condition (solid line for the Laser condition and dashed line for the Stick condition). Positive numbers reflect rightward bias. The bars are ±SEM.

Table 1. Mean scores on the CLQ for participants in the Laser and Stick conditions (SD in parentheses). Scores in each condition were consistent with normative data (e.g., Lourenco et al., 2011; Radomsky et al., 2001) Scores Laser Condition

Stick Condition

36.94 (15.18)

35.03 (16.27)

14.06 (7.38)

13.23 (7.64)

22.88 (9.07)

21.80 (9.46)

CLQ (total)

the line bisection task (β = 0.38, 95% CI [ 0.71, 0.04]), t(33) = 2.31, p = .028, d = 0.82 (see Figure 2), as reported in previous research (Lourenco et al., 2011). There was no relationship between the RS subscale and line bisection performance (β = 0.15, 95% CI [ 0.51, 0.21]), t(33) = 0.85, p = .40, d = 0.30. Thus, in the Laser condition, higher claustrophobic fear, as measured by the SS subscale, was associated with more gradual line bisection slopes, indicating larger peripersonal space representations than for those lower in claustrophobic fear. However, a concern with this finding is that it was restricted to the SS subscale. In the original study of Lourenco and colleagues (2011), the relation between claustrophobic fear and line bisection performance was observed for both subscales on the CLQ. Thus, to ensure the reliability of the effect reported here, we calculated the 95% CI of the β values reported by Lourenco and colleagues (2011) and compared them to those reported here. The β’s of the current study fell within those reported by Lourenco and colleagues (2011) for the the SS subscale (95% CI [ 0.79, 0.15]) and the CLQ (total) scores (95% CI [ 0.81, 0.19]), though not the RS subscale (95% CI [ 0.78, 0.16]). Thus, although Experimental Psychology (2017), 64(1), 49–55

Figure 2. Scatterplot (with best-fitting regression line [r = .38]) relating participants’ performance in the Laser condition (regression slopes) and CLQ (SS) scores. No outliers were identified.

our findings are generally within the range of the previously reported effects, confirming the relation between claustrophobic fear and peripersonal space, the effects may be inconsistent for the RS subscale specifically. We next examined performance in the Stick condition in relation to claustrophobic fear. A linear regression model revealed a significant positive relationship between CLQ scores and participants’ slopes on the line bisection task, total CLQ: (β = 0.46, 95% CI [0.15, 0.78]), t(34) = 2.98, p = .005, d = 1.04 (see Figure 3); SS: (β = 0.43, 95% CI [0.11, 0.75]), t(34) = 2.71, p = .011, d = 0.94; RS: (β = 0.45, 95% CI [0.13, 0.76]), t(34) = 2.87, p = .007, d = 1.00. Notably, this effect was in the opposite direction as that reported in the Laser condition. Higher claustrophobic fear was associated with steeper line bisection slopes, indicating that participants higher in claustrophobic fear had smaller peripersonal spaces in this condition. Because the line bisection slopes in this condition are an indication of the degree of expansion of peripersonal space when using a stick, the relation between bisection performance and CLQ scores suggests that participants who were higher in claustrophobic fear experienced less expansion relative to participants lower in claustrophobic fear. However, an alternative interpretation of this effect is that individuals who were higher in claustrophobic fear already had maximally expanded peripersonal spaces. That is, given that higher claustrophobic fear is associated with larger peripersonal space, as suggested by the relationship between CLQ scores and bisection performance when using a laser pointer (see Figure 3; see also Lourenco et al., 2011), perhaps peripersonal space could not expand further, and consequently, the stick manipulation had no effect. We would argue that this is unlikely because a comparison of participants who were high and low in claustrophobic fear based on a median split of CLQ total scores (low [< 36], high [ 36]) revealed that participants high in Ó 2017 Hogrefe Publishing

S. B. Hunley et al., Peripersonal Space Flexibility

Figure 3. Scatterplot (with best-fitting regression line [r = .48]) relating participants’ performance in the Stick condition (regression slopes) and CLQ (total) scores. No outliers were identified.

claustrophobic fear in the Laser condition had significantly steeper rightward shifts in bias over distance (Mslope = 1.23, SDslope = 1.02) compared to participants low in claustrophobic fear in the Stick condition (Mslope = 0.29, SDslope = 0.81), t(36) = 3.16, p = .003, d = 1.03. In other words, participants low in claustrophobic fear in the Stick condition had larger peripersonal spaces than participants high in claustrophobic fear in the Laser condition, suggesting ample room for expansion of peripersonal space, as measured by this task, among the higher claustrophobic individuals.

Discussion The current study provides evidence of individual differences in the flexibility of peripersonal space. Although previous research has demonstrated that peripersonal space accommodates to situational factors (e.g., Longo & Lourenco, 2006; Noel et al., 2015; Taffou & Viaud-Delmon, 2014), our findings suggest that the flexibility associated with an expansion of peripersonal space to such factors may not be experienced equally across individuals. Specifically, in the current study, we found that individuals with higher trait claustrophobic fear did not experience the expected expansion of peripersonal space when using a stick to bisect lines at farther distances. This finding builds on work showing that individuals who are higher in claustrophobic fear have larger default peripersonal spaces (Lourenco et al., 2011) by demonstrating that higher claustrophobic fear may also be associated with less flexible representations of peripersonal space. Why might individuals who are higher in claustrophobic fear demonstrate decreased flexibility of peripersonal space? Although we cannot provide a definitive answer to this question, we put forward two possible explanations Ó 2017 Hogrefe Publishing

for further empirical investigation. One explanation is based on the recent claim that peripersonal space is subserved by two distinct systems, one for defending the body and the other for non-defensive (visuomotor) actions (de Vignemont & Iannetti, 2015). Though others have argued that peripersonal space serves multiple functions (Cooke & Graziano, 2003; Graziano & Cooke, 2006; Hall, 1966; Hediger, 1955; Lourenco et al., 2011; Sambo, Forster, Williams, & Iannetti, 2012; Sambo, Liang, Cruccu, & Iannetti, 2012), de Vignemont and Iannetti (2015) recently claimed that these different functions are supported by distinct systems of peripersonal space, with each obeying its own principles (see also, Sambo & Ianetti, 2013). Following from this proposal, we would suggest that although peripersonal space for non-defensive behaviors must dynamically respond to changing action capabilities in order to guide action effectively, this may not be the case for defensive peripersonal space. Indeed, it may be more adaptive to have a representation of defensive peripersonal space that is relatively fixed, along with a relatively larger defensive than non-defensive peripersonal space (Lourenco et al., 2011), in order to keep potential threats at a safe distance from the body. The decreased flexibility we report here could be due to an overactivation of defensive peripersonal space, thereby accounting for both relatively larger and less flexible peripersonal space representations even under nonthreatening circumstances among individuals high in claustrophobic fear. Future studies are needed to provide direct evidence for the dissociation between defensive and non-defensive systems of peripersonal space. One potential avenue would be to test both the size and flexibility of peripersonal space under conditions designed to elicit either defensive or non-defensive reactions. Another possible, though not mutually exclusive, explanation of the decreased flexibility of peripersonal space is rooted in a general attentional mechanism. A growing body of evidence suggests that high trait anxiety is associated with decreased attentional flexibility (Derryberry & Reed, 2002; Eysenck, Derakshan, Santos, & Calvo, 2007; Pacheco-Unguetti, Acosta, Callejas, & Lupiáñez, 2010) and disrupted activation in fronto-parietal networks associated with attention (e.g., Bishop, 2009; Sylvester et al., 2012). Importantly, regions that fall within these networks have also been associated with peripersonal space representations (Brozzoli et al., 2011; Holt et al., 2014; Serino, Canzoneri, & Avenanti, 2011). Thus, another account of the current findings is that the decreased flexibility in peripersonal space among individuals higher in claustrophobic fear may reflect reduced attentional flexibility associated with high trait anxiety. On this account, the decreased flexibility is not specific to defensive or non-defensive functions of peripersonal space, but rather, Experimental Psychology (2017), 64(1), 49–55

applies broadly across the two types of representations and to other attentional contexts. Future studies would do well to address questions related to the role of general attentional mechanisms in maintaining peripersonal space representations. Electronic Supplementary Material The electronic supplementary material is available with the online version of the article at http://dx.doi.org/10.1027/ 1618-3169/a000350 ESM 1. Data file (csv). Raw data collected in the study.

References Ackroyd, K., Riddoch, M. J., Humphreys, G. W., Nightingale, S., & Townsend, S. (2002). Widening the sphere of influence: Using a tool to extend extrapersonal visual space in a patient with severe neglect. Neurocase, 8, 1–12. doi: 10.1093/neucas/8.1.1 Berti, A., & Frassinetti, F. (2000). When far becomes near: Remapping of space by tool use. Journal of Cognitive Neuroscience, 12, 415–420. doi: 10.1162/089892900562237 Bishop, S. J. (2009). Trait anxiety and impoverished prefrontal control of attention. Nature Neuroscience, 12, 92–98. doi: 10.1038/nn.2242 Brozzoli, C., Gentile, G., Petkova, V. I., & Ehrsson, H. H. (2011). fMRI adaptation reveals a cortical mechanism for the coding of space near the hand. The Journal of Neuroscience, 31, 9023–9031. doi: 10.1523/JNEUROSCI.1172-11.2011 Canzoneri, E., Marzolla, M., Amoresano, A., Verni, G., & Serino, A. (2013). Amputation and prosthesis implantation shape body and peripersonal space representations. Scientific Reports, 3, 1–8. doi: 10.1038/srep02844 Canzoneri, E., Ubaldi, S., Rastelli, V., Finisguerra, A., Bassolino, M., & Serino, A. (2013). Tool-use reshapes the boundaries of body and peripersonal space representations. Experimental Brain Research, 228, 25–42. doi: 10.1007/ s00221-013-3532-2 Cooke, D. F., & Graziano, M. S. (2003). Defensive movements evoked by air puff in monkeys. Journal of Neurophysiology, 90, 3317–3329. doi: 10.1152/jn.00513.2003 de Vignemont, F., & Iannetti, G. (2015). How many peripersonal spaces? Neuropsychologia, 70, 327–334. doi: 10.1016/ j.neuropsychologia.2014.11.018 Derryberry, D., & Reed, M. A. (2002). Anxiety-related attentional biases and their regulation by attentional control. Journal of Abnormal Psychology, 111, 225. doi: 10.1037/0021-843X. 111.2.225 Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007). Anxiety and cognitive performance: Attentional control theory. Emotion, 7, 336–353. doi: 10.1037/1528-3542.7.2.336 Farnè, A., & Làdavas, E. (2000). Dynamic size-change of hand peripersonal space following tool use. Neuroreport, 11, 1645–1649. doi: 10.1097/00001756-200006050-00010 Graziano, M. S. A., & Cooke, D. F. (2006). Parieto-frontal interactions, personal space, and defensive behavior. Neuropsychologia, 44, 2621–2635. doi: 10.1016/ j.neuropsychologia.2005.09.011 Hall, E. T. (1966). The Hidden Dimension. New York, NY: Doubleday.

Experimental Psychology (2017), 64(1), 49–55

S. B. Hunley et al., Peripersonal Space Flexibility

Hediger, H. (1955). Studies of the psychology and behavior of captive animals in zoos and circuses. Oxford, England: Criterion Books. Holmes, N. P. (2012). Does tool use extend peripersonal space? A review and re-analysis. Experimental Brain Research, 218, 273–282. doi: 10.1007/s00221-012-3042-7 Holt, D. J., Cassidy, B. S., Yue, X., Rauch, S. L., Boeke, E. A., Nasr, S., . . . Coombs, G. (2014). Neural correlates of personal space intrusion. The Journal of Neuroscience, 34, 4123–4134. doi: 10.1523/JNEUROSCI.0686-13.2014 Hunley, S. B., Park, E., Longo, M. R., & Lourenco, S. F. (2016). Claustrophobic fear is associated with compression of visual space. Manuscript in preparation. Iriki, A., Tanaka, M., & Iwamura, Y. (1996). Coding of modified body schema during tool use by macaque postcentral neurones. Neuroreport, 7, 2325–2330. Jewell, G., & McCourt, M. E. (2000). Pseudoneglect: A review and meta-analysis of performance factors in line bisection tasks. Neuropsychologia, 38, 93–110. doi: 10.1016/S0028-3932(99) 00045-7 Kinsbourne, M. (1987). Mechanisms of unilateral neglect. Advances in psychology, 45, 69–86. doi: 10.1016/S0166-4115 (08)61709-4 Longo, M. R., & Lourenco, S. F. (2006). On the nature of near space: Effects of tool use and the transition to far space. Neuropsychologia, 44, 977–981. doi: 10.1016/ j.neuropsychologia.2005.09.003 Longo, M. R., & Lourenco, S. F. (2009). Space perception and body morphology: extent of near space scales with arm length. Experimental Brain Research, 177, 285-290. doi: 10.1007/ s00221-007-0855-x. Longo, M. R., Trippier, S., Vagnoni, E., & Lourenco, S. F. (2015). Right hemisphere control of visuospatial attention in near space. Neuropsychologia, 70, 350–357. doi: 10.1016/ j.neuropsychologia.2014.10.035 Lourenco, S. F., & Longo, M. R. (2009). The plasticity of near space: Evidence for contraction. Cognition, 112, 451–456. doi: 10.1016/j.cognition.2009.05.011 Lourenco, S. F., Longo, M. R., & Pathman, T. (2011). Near space and its relation to claustrophobic fear. Cognition, 119, 448–453. doi: 10.1016/j.cognition.2011.02.009 Maister, L., Cardini, F., Zamariola, G., Serino, A., & Tsakiris, M. (2015). Your place or mine: Shared sensory experiences elicit a remapping of peripersonal space. Neuropsychologia, 70, 455–461. doi: 10.1016/j.neuropsychologia.2014.10.027 Makin, T. R., Holmes, N. P., & Zohary, E. (2007). Is that near my hand? Multisensory representation of peripersonal space in human intraparietal sulcus. The Journal of Neuroscience, 27, 731–740. doi: 10.1523/JNEUROSCI.3653-06.2007 Maravita, A., & Iriki, A. (2004). Tools for the body (schema). Trends in Cognitive Sciences, 8, 79–86. doi: 10.1016/j.tics. 2003.12.008 Noel, J.-P., Pfeiffer, C., Blanke, O., & Serino, A. (2015). Peripersonal space as the space of the bodily self. Cognition, 144, 49–57. doi: 10.1016/j.cognition.2015.07.012 Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9, 97–113. doi: 10.1016/0028-3932(71)90067-4 Pacheco-Unguetti, A. P., Acosta, A., Callejas, A., & Lupiáñez, J. (2010). Attention and anxiety different attentional functioning under state and trait anxiety. Psychological Science, 21, 298–304. doi: 10.1177/0956797609359624 Previc, F. H. (1998). The neuropsychology of 3-D space. Psychological Bulletin, 124, 123–164. doi: 10.1037/0033-2909. 124.2.123

Ó 2017 Hogrefe Publishing

S. B. Hunley et al., Peripersonal Space Flexibility

Radomsky, A. S., Rachman, S., Thordarson, D. S., McIsaac, H. K., & Teachman, B. A. (2001). The claustrophobia questionnaire. Journal of Anxiety Disorders, 15, 287–297. doi: 10.1016/S08876185(01)00064-0 Rizzolatti, G., Matelli, M., & Pavesi, G. (1983). Deficits in attention and movement following the removal of postarcuate (area 6) and prearcuate (area 8) cortex in macaque monkeys. Brain, 106, 655–673. doi: 10.1093/brain/106.3.655 Rizzolatti, G., Scandolara, C., Matelli, M., & Gentilucci, M. (1981). Afferent properties of periarcuate neurons in macaque monkeys: II. Visual responses. Behavioural Brain Research, 2, 147–163. doi: 10.1016/0166-4328(81)90053-X Rossetti, A., Romano, D., Bolognini, N., & Maravita, A. (2015). Dynamic expansion of alert responses to incoming painful stimuli following tool use. Neuropsychologia, 70, 486–494. doi: 10.1016/j.neuropsychologia.2015.01.019 Sambo, C., Forster, B., Williams, S., & Iannetti, G. (2012). To blink or not to blink: Fine cognitive tuning of the defensive peripersonal space. The Journal of Neuroscience, 32, 12921–12927. doi: 10.1523/JNEUROSCI.0607-12.2012 Sambo, C. F., & Iannetti, G. D. (2013). Better safe than sorry? The safety margin surrounding the body is increased by anxiety. The Journal of Neuroscience, 33, 14225–14230. doi: 10.1523/ JNEUROSCI.0706-13.2013 Sambo, C. F., Liang, M., Cruccu, G., & Iannetti, G. D. (2012). Defensive peripersonal space: The blink reflex evoked by hand stimulation is increased when the hand is near the face. Journal of Neurophysiology, 107, 880–889. doi: 10.1152/ jn.00731.2011 Serino, A., Bassolino, M., Farnè, A., & Làdavas, E. (2007). Extended multisensory space in blind cane users. Psychological Science, 18, 642–648. doi: 10.1111/j.1467-9280.2007.01952.x

Ó 2017 Hogrefe Publishing

Serino, A., Canzoneri, E., & Avenanti, A. (2011). Fronto-parietal areas necessary for a multisensory representation of peripersonal space in humans: An rTMS study. Journal of Cognitive Neuroscience, 23, 2956–2967. doi: 10.1162/jocn_a_00006 Serino, A., Canzoneri, E., Marzolla, M., Di Pellegrino, G., & Magosso, E. (2015). Extending peripersonal space representation without tool-use: Evidence from a combined behavioralcomputational approach. Frontiers in Behavioral Neuroscience, 9, 1–14. doi: 10.3389/fnbeh.2015.00004 Sylvester, C., Corbetta, M., Raichle, M., Rodebaugh, T., Schlaggar, B., Sheline, Y., . . . Lenze, E. (2012). Functional network dysfunction in anxiety and anxiety disorders. Trends in Neurosciences, 35, 527–535. doi: 10.1016/j.tins.2012.04.012 Taffou, M., & Viaud-Delmon, I. (2014). Cynophobic fear adaptively extends peri-personal space. Frontiers in Psychiatry, 5, 1–7. doi: 10.3389/fpsyt.2014.00122 Received September 14, 2016 Revision received November 11, 2016 Accepted November 21, 2016 Published online February 20, 2017

Samuel Hunley Department of Psychology Emory University 36 Eagle Rd., Suite 150 Atlanta, GA 30322 USA shunley@emory.edu

Experimental Psychology (2017), 64(1), 49–55

Short Research Article

The Role of Embodiment and Individual Empathy Levels in Gesture Comprehension Karine Jospe,1,2 Agnes Flöel,3 and Michal Lavidor1,2 1

Department of Psychology, Bar Ilan University, Ramat Gan, Israel

The Gonda Brain Research Center, Bar Ilan University, Israel Department of Neurology Charité – Universitätsmedizin Berlin, Germany

Abstract: Research suggests that the action-observation network is involved in both emotional-embodiment (empathy) and actionembodiment (imitation) mechanisms. Here we tested whether empathy modulates action-embodiment, hypothesizing that restricting imitation abilities will impair performance in a hand gesture comprehension task. Moreover, we hypothesized that empathy levels will modulate the imitation restriction effect. One hundred twenty participants with a range of empathy scores performed gesture comprehension under restricted and unrestricted hand conditions. Empathetic participants performed better under the unrestricted compared to the restricted condition, and compared to the low empathy participants. Remarkably however, the latter showed the exactly opposite pattern and performed better under the restricted condition. This pattern was not found in a facial expression recognition task. The selective interaction of embodiment restriction and empathy suggests that empathy modulates the way people employ embodiment in gesture comprehension. We discuss the potential of embodiment-induced therapy to improve empathetic abilities in individuals with low empathy. Keywords: embodiment, empathy, hand gestures, mimicry, mimicry-restriction

To live in a socially complex world we need the ability to understand the intentions and actions of others and to recognize their emotional state. One of the things we tend to use is an ability termed “empathy.” Empathy is considered to have two aspects to it: a cognitive aspect also labeled “perspective-taking” and its role is to cognitively understand the other better (Hawk, Fischer, & Van Kleef, 2011) and an affective aspect, which manifests by experiencing feelings or taking the proper action that suits the emotional understanding achieved in the cognitive aspect (Shamay-Tsoory, 2011). These emotional experiences are aroused by automatic mimicry of the nonverbal expression of the other’s feelings (Knafo, Zahn-Waxler, Van Hulle, Robinson, & Rhee, 2008). A connection between empathy and mimicry can be found, for example, in the research conducted by Richter and Kunzmann (2011), where they report that watching happy, angry, or frightened faces activated corresponding muscles in the observer’s face. This connection was also found by Decety, Echols, and Correll (2010). They showed that regions related to first-hand experience of pain, such as the anterior cingulate cortex (ACC), the insula, and other parts of the pain matrix, responded to both felt and

Experimental Psychology (2017), 64(1), 56–64 DOI: 10.1027/1618-3169/a000351

observed pain. This indicates that empathizing with other people’s pain is associated with brain activity that is similar to the activity that occurs when they feel pain themselves. This phenomenon of automatic mimicry is also regarded as grounded cognitive understanding or “embodiment” (Barsalou, 2008). Murata, Saito, Schug, Ogawa, and Kameda (2016) found that mimicry increased when participants were explicitly asked to infer the emotional state of a facial expression. This suggests that mimicry is essential to the comprehension process of observed emotions. Another line of support to the automatic mimicry process was made by Hess and Fischer (2013), who found that mimicry was related to the understanding and involved in regulating one’s relation with the other person, rather than being the synchronization of meaningless individual muscle actions. An individual difference in cognitive understanding can be found within empathy scores. Aziz-Zadeh, Sheng, and Gheytanchi (2010) found that individual differences in empathy scores were correlated with cognitive performance of prosody comprehension. In addition, Chartrand and Bargh (1999) found that individuals who spontaneously imitate another person’s behavior are more empathetic

Ó 2017 Hogrefe Publishing

K. Jospe et al., Embodiment Modulation and Empathy Levels

compared to individuals who do not. It was also found by Williams, Nicolson, Clephan, Grauw, and Perrett (2013) that participants with high empathy scores made fewer errors when asked to imitate facial expressions. Mimicry and imitation are not only involved in empathic understanding, they may also be crucial to understanding perceived action, like hand gestures (Parzuchowski, Szymkow, Baryla, & Wojciszke, 2014). Fadiga, Fogassi, Pavesi, and Rizzolatti (1995) found that the motor evoked potential (MEP) of participant’s hands increased while watching someone else grasp objects. Moreover, increased MEP was present in those muscles that would have been moved by the participant had he made the movement himself. Furthermore, even the mere observation of an outcome to a hand-related action was sufficient to activate the observer’s motor cortex (Heimann, Umilta, & Gallese, 2013). These findings were specific to goal-related actions and not to meaningless hand movements, suggesting their involvement in comprehension and not synchronization of meaningless muscle activation. More support to this claim can be found in the work of Alibali and Hostetter (2010). They introduced the Gesture Simulated Action (GSA) framework, aiming to explain how representational gestures might arise from an embodied cognitive system. The GSA was originally introduced to explain the use of gestures by speakers and the role of these gestures in expressing the intended information. Taking into consideration the role of the gestures performed by the speakers and the contribution of gestures in comprehending complex materials, Alibali and Hostetter (2010) suggest that a similar mechanism is involved in the comprehension of gestures. To further understand the mechanism of mimicry and its involvement in comprehension, several previous studies aimed to affect this mechanism. For example, Chiavarino, Bugiani, Grandi, and Colle (2013) found that automatic imitation can be modulated by bottom-up features of the observed action, while Strack, Martin, and Stepper (1988) showed that modulating the ability to imitate (thus modulating bottom-up features of the observer) affected the perception of situations. In their experiment, participants were asked to hold a pen with their lips (to prevent smiling) while evaluating cartoons as less funny than a control group holding the pen in their hands. Furthermore, participants who were asked to hold the pen with their teeth, without the lips touching (creating a forced smile), rated the cartoons as funniest, compared to both groups. Rychlowska et al. (2014) found that blocking mimicry resulted in having true and false smiles judged as equally genuine. These studies imply that restricting the ability to imitate a smile changes the way we comprehend the respective situations even though the imitation is not visibly executed. Ó 2017 Hogrefe Publishing

The involvements of mimicry in understanding hand gestures and in empathy have been vastly researched separately, as we discussed above. We also reviewed research that found individual differences in gesture mimicry. In the current study, we therefore wanted to explore whether we can find individual differences in hand gesture comprehension that are based on individual differences in empathy. Next, we wanted to assess whether this comprehension can be modulated by restricting imitation, or in other words, interfering with the action-observation process. Specifically, we wanted to show that restricting imitation of one action (e.g., restricting the hand movement) will affect only the understanding of a related action (e.g., hand gesture comprehension task), while not affecting understanding of unrelated actions (e.g., facial expressions recognition task). This prediction follows Thomas, Sink and Haggard’s findings (2013), where perceptual enhancement was reported only when the action observed and the stimulated body part were congruent, and the model’s action did not modulate non-corresponding sites on the observer’s body. Furthermore, since the literature showed different mimicry action and accuracy for individuals with varying empathy scores (Williams et al., 2013) we expect to see this difference under restriction as well. Meaning, if the mimicry itself is different between participants with varying empathy levels, the effect of the restriction should be different too. We expect it to be more profound for individuals with high empathy levels, since it seems they use this mechanism of mimicry more than less empathetic participants. For this research, we adapted the paradigm from Strack et al.’s study from 1988, and assessed whether restriction of hand movements would alter the comprehension of gestures while not altering the comprehension of a nongestural control task (facial expressions). Moreover, participants with a range of empathy levels were tested, in order to examine whether baseline comprehension, as well as modulation by hand movement restriction, would vary between participants with different empathy levels. Based on the literature reviewed above, we hypothesize that: Hypothesis 1 (H1): Empathy levels will be positively correlated with action comprehension; Hypothesis 2 (H2): Restricting the participants’ hands will impair their performance in a gesture comprehension task, but will not impair their performance in a facial expression recognition task; Hypothesis 3 (H3): The effect of restricting the participants’ hands will be modulated by individual empathy levels. Experimental Psychology (2017), 64(1), 56–64

K. Jospe et al., Embodiment Modulation and Empathy Levels

Method Participants One hundred twenty participants (85 females, 35 males) took part in the study (mean age = 25.53, range: 18–50, SD = 5.42), with an average empathy score of 48 as measured by the empathy questionnaire (“Empathy Quotient,” Baron-Cohen & Wheelwright, 2004). All participants were healthy, right handed (as measured by the Edinburgh Handedness Inventory, Oldfield, 1971), with normal/corrected to normal vision, and native speakers of Hebrew, and provided informed consent to participate in this study. Figure 1. An example of a hand gesture stimulus (screen shot from video).

Stimuli The experiment was constructed of two blocks, with semantic judgment of hand gesture clips in one and facial expression recognition in the other block. In the gestures block, participants were shown 40 right-hand gestures video clips, each 1-second-long, which were created and used in a previous study in our laboratory (CohenMaximov, Avirame, Flöel, & Lavidor, 2015). Each video was followed by a written description of the performed gesture. Half of the video clips were followed by a correct description and half by an incorrect description. Participants were asked to judge whether the description following the video is correct or not. A screenshot example of one of the video clips is presented in Figure 1. In this video clip, the congruent description was /LO SHOME’A/ (“Can’t hear”) and the incongruent description was /NOSE’A LE’AT/ (“driving slowly”). This video clip is provided in the Electronic Supplementary Material 1. In the facial expression recognition block, participants were shown 89 pictures of facial expression taken from Ekman and Friesen (1978). Each picture was followed by a correct or incorrect written description of the facial expression. Participants were asked to judge whether the description matches the facial expression or not. An example of these stimuli is presented in Figure 2. Examples of all six facial expressions are provided in the Electronic Supplementary Material 2.

Procedure Participants completed the Empathy Questionnaire (EQ; see Baron-Cohen & Wheelwright, 2004) which resulted in an empathy score (EQ score) on a scale of 0–80. Actual scores ranged in between 23 and 69 with mean and median of 48.28. Participants were divided to one of the two hand restrictions conditions (restricted or unrestricted hand) in a Experimental Psychology (2017), 64(1), 56–64

Figure 2. An example of a happy facial expression.

pseudorandom order, providing that the average EQ score in each group remained around 48. Participants were tested individually in a quiet room in front of a computer monitor that presented both the gesture video clips and the facial expression tasks under two conditions: restricted and unrestricted dominant hand. Half (n = 60) of the participants performed the tasks under a restricted hand condition; they were asked to sit on their right hand. The other half (n = 60) performed the tasks without restriction. All participants were asked to respond with the index and middle finger of their left hand. Ó 2017 Hogrefe Publishing

K. Jospe et al., Embodiment Modulation and Empathy Levels

Figure 3. Scatter-plots and linear correlations between EQ scores and RT (in ms) in the facial expression recognition task under hand restriction conditions.

Each participant completed, in random order, a gesture comprehension and facial expression recognition tasks. In both blocks stimuli were balanced, so half of the items were congruent and half incongruent, and the blocks order was counterbalanced between participants.

Results We excluded correct responses that exceeded the average reaction time (RT) in more than 3 SD. We calculated a mean RT of all six facial expressions, thus creating one variable for the facial expression recognition task. Raw data can be found in the Electronic Supplementary Material 3, ESM 3. To test our first hypothesis, that empathy levels will be positively correlated with action comprehension, we correlated RT and EQ scores in each task under each condition. For the facial expression recognition task, we found a negative correlation under both the restricted and unrestricted conditions (r = .294, n = 60, p = .022; r = .309, n = 60, p = .016, respectively). The higher the EQ Score, the faster the RT (see Figure 3). For the gesture comprehension task, we found a negative correlation for the unrestricted condition (r = .408, n = 60, p = .001): the higher the EQ Score, the faster the RT. But most interestingly, this correlation was reversed under the restricted condition (r = .300, n = 60, p = .020). Under this condition we found that, the lower the EQ, the faster the RT (see Figure 4). To assess the effect of the experimental condition and empathy level on the RT of both gesture comprehension and facial expression recognition tasks, two two-step hierarchical linear regressions were employed: one for the gesture comprehension task and one for the facial Ó 2017 Hogrefe Publishing

expression recognition task. The hierarchical approach was chosen to enable us to test the model containing our main effects and to analyze the effects of the interaction between them. The results from the regression can be found in Table 1. For the gesture comprehension task, at the first step, the EQ scores and condition were entered to test for main effects. At the second step, the interaction variable of EQ Score Hand Restriction condition was entered. The model for the first step was not significant, R2 = .005, p = .766. Showing that neither EQ score nor conditions are effective predictors by themselves, but adding the interaction variable to the model in Step 2 did produce a significant model, F(3, 116) = 5.822, p = .001, R2 = .131, Δr2 = .126. Adding the interaction variable to the model rendered all predictors significant. This indicates that the effect of the hand restriction condition on performance was modulated by the participants’ EQ scores. The same regressions were employed for the facial expression recognition task as the predicted variable; at the first step, the EQ scores and condition were entered to test for main effects. At the second step, the interaction variable of EQ Score Hand Restriction condition was entered. Here we found that the first model was significant, F(2, 117) = 6.106, p = .003, R2 = .095. And the second model was also significant, but the interaction variable did not add to the prediction power of the regression F(3, 116) = 4.110, p = .008, R2 = .096, Δr2 = .002. Only the EQ score was a significant predictor in both the first and second models. In our second hypothesis we predicted that hand restriction will impair performance in a gesture comprehension task, but will not impair performance in a facial expression recognition task. The results from these regression analyses affirm the second part of the hypothesis, but only partially affirm the first part. Specifically, facial expression recognition was not affected by hand restriction condition, but Experimental Psychology (2017), 64(1), 56–64

K. Jospe et al., Embodiment Modulation and Empathy Levels

Figure 4. Scatter-plots and linear correlations between EQ scores and RT (in ms) in the gesture comprehension task under hand restriction conditions.

Table 1. Summary of the two two-step hierarchical analysis for predicting performance in the gesture comprehension and facial expression recognition task with hand restriction manipulation and EQ scores as predictors (N = 120) R2

variable

Gesture comprehension Step 1

.005

EQ score

0.062

.787

Hand restriction manipulation

0.025

.502

Step 2

.131

EQ score

0.425

.001**

Hand restriction manipulation

1.751

.000**

EQ Score Hand condition

1.839

.000**

EQ score

0.300

.001**

Hand restriction condition

0.072

.406

0.206

.653

Facial expression recognition Step 1

Step 2

.095

.096

EQ Score Hand restriction Notes. EQ = Empathy Quotient, **p < .01.

hand gesture comprehension was affected by the hand restriction condition. However, the direction of the effect that was predicted to be negative was not necessarily negative, but rather inordinate and modulated by the empathy level. To better understand the participants’ performance in each condition, and the reversed correlation between the two conditions in the gesture comprehension task we divided the participants in each condition into two groups 1

by their empathy scores, creating a high empathy group (participants of an EQ of 49 or higher) and a low empathy group (participants of an EQ of 48 and lower).1 Mean RT of correct responses in the gesture comprehension and facial expression recognition tasks as a function of hand condition and empathy group are presented in Table 2. After dividing the groups, we performed a two-way MANOVA, using correct RT of the gesture comprehension and the facial expression recognition tasks as the dependent variables, and using the experimental condition and EQ groups as factors. A main effect for empathy group was found significant (F(2, 115) = 10.894, p = .000, η2p = .159). A two-way ANOVA revealed a significant main effect for empathy group only for the facial expression recognition task (F(1, 116) = 16.602, p = .000, η2p = .125); high empathy participants were significantly faster in their RT than low empathy participants (mean = 796.40, SE = 20; mean = 914.34, SE = 20; respectively). Crucially, the general MANOVA also revealed a significant multivariate interaction between the experimental condition of hand restriction and EQ groups (F(2, 115) = 7.730, p = .001, η2p = .119). The following two-way ANOVA revealed a significant interaction only in the gesture comprehension task (F(1, 116) = 10.175, p = .002, η2p = .081), and not in the facial expression recognition task (F(1, 116) = 0.003, p = .955, η2p = .000). As shown in Figure 5, post hoc analyses with Bonferroni adjustment revealed that under the unrestricted condition, participants in the high empathy group were significantly faster than the participants in the low empathy group (F(1, 116) = 6.668, p = .011, η2p = .054). Also, we found

MANCOVA using correct RT of both tasks as dependent variables, experimental condition as a factor, and EQ scores as a covariant indicated a significant interaction between the condition and the EQ score only in the gesture comprehension task, but not in the facial recognition task (F(1, 116) = 16.958, p = .000, η2 = 128; F(1, 116) = 0.203, p = .653, η2 = .002, respectively). This suggests that the differences of the RT in the gesture comprehension task as a factor of the condition vary as a function of the EQ scores. In contrast, the RT differences in the facial recognition task did not reflect condition and empathy interaction.

Experimental Psychology (2017), 64(1), 56–64

Ó 2017 Hogrefe Publishing

K. Jospe et al., Embodiment Modulation and Empathy Levels

Table 2. Mean RTs and SE for gesture comprehension and facial expression recognition according to hand restriction condition and EQ group Experimental condition Unrestricted

Restricted

Total

EQ group

Gesture comprehension (SE), in ms

Facial-expression recognition (SE), in ms

High

773 (32)

808 (29)

Low

891 (32)

825 (29)

Total

832 (23)

867 (20)

High

886 (32)

785 (29)

Low

798 (32)

904 (29)

Total

842 (23)

844 (20)

High

830 (23)

796 (20)

Low

844 (23)

914 (20)

Notes. EQ = Empathy Quotient; RT = reaction time; SE = standard error (in parentheses); N = number of subjects.

Figure 5. Mean RTs and SE for gesture comprehension and facial expression recognition according to hand restriction manipulation and EQ group. *p < .05.

that participants in the high empathy group were significantly slower under the restricted condition than the unrestricted condition (F(1, 116) = 6.077, p = .015, η2p = .050). Last, consistently with the correlations, we found that participants in the low empathy group were significantly faster under the restricted condition than the unrestricted condition (F(1, 116) = 4.186, p = .043, η2p = .035). This last finding, confirmed both in the regression model and the ANOVA, is a novel finding that was not reported in the literature before. Accuracy levels in the gesture comprehension task were very high under both conditions, with mean accuracy = 97% and 94% for the unrestricted and restricted condition, respectively. This probably reflects a ceiling effect, similar to the findings in a previous study that used the same materials (Cohen-Maximov et al., 2015). In the facial expression recognition task, accuracy was also very high (91%–95%). No main effects or interactions were found for accuracy. No main effect or interactions were found for gender, which was tested as a background variable. Ó 2017 Hogrefe Publishing

Discussion In the current study, we tested the effect of mimicry restriction on hand gesture comprehension and facial expression recognition in participants with a range of empathy levels. The main finding of our experiment was that while under natural conditions (where hand gestures are not restricted), people with higher empathy perform better the hand gesture comprehension task, compared to people with lower empathy scores. When hand gestures were restricted, high empathy individuals deteriorated in their hand gesture comprehension. We explicitly predicted this finding based on previous studies using smile restriction and emotion recognition (Strack et al., 1988). However, the surprising and novel finding of the current study is that the performance of the low empathy group was significantly better under the restricted condition compared to the unrestricted condition. This means that people with lower empathy actually benefit from a hand restriction while judging hand gestures. Experimental Psychology (2017), 64(1), 56–64

These findings suggest an individual difference in hand gesture comprehension that relies on both empathy levels and mimicry. When the ability to mimic is restricted, the ability to comprehend hand gesture breaks down if the empathy levels are high, but improves when empathy levels are low. We argue that the ability to comprehend hand gestures relies on mimicry, which is utilized differently in people with different empathy levels. In our second finding, we were also able to show that restricting hand movement affected only hand gesture comprehension and did not affect other social perceptions such as facial expression recognition (affirming our second hypothesis). What mechanisms may underlie these findings? As for the significant difference between high and low empathy participants under the natural condition, this finding was not surprising. It is rather intuitive that people with high empathy will comprehend other people better than people with low empathy scores. Such interpretation is supported by Chartrand and Bargh (1999), who found that individuals who spontaneously imitate another person’s behavior are more empathetic compared to individuals who do not. Further support comes from Kaplan’s and Iacoboni’s (2006) findings, reporting that the emotional empathy of participants was correlated with the intensity of the activity in the premotor areas (areas associated with the Action-Observation Network (AON)) while observing people carrying out an action with different intentions. Also, Gazzola, Aziz-Zadeh, and Keysers (2006) found correlations between the activity in brain areas involved in the AON and participants score on the empathy scale. We stipulate that under natural conditions, people with high empathy rely on the ability to mimic, and the freedom to do so seems to give them an advantage in comprehending others, similar in principle to the findings of empathy and prosody comprehension (Aziz-Zadeh et al., 2010). As for the deteriorating effect of mimicry restriction, this finding was not surprising either, since similar results were found by Strack et al. (1988) and by Rychlowska et al. (2014) for facial expressions. This finding thus supports our second hypothesis. The unexpected finding was an improvement in performance of the low empathy participants under the restricted condition. Unlike Strack et al. (1988), restriction of the hand did not exert a general detrimental effect on performance in the gesture task, but rather affected performance dependent on baseline level of empathy: hand restriction significantly assisted the gesture comprehension of participants with low empathy scores but impaired the gesture comprehension in the high empathy group, without affecting the performance in a facial expression recognition task. This affirms our third

Experimental Psychology (2017), 64(1), 56–64

K. Jospe et al., Embodiment Modulation and Empathy Levels

hypothesis of a different effect of hand restriction on participants with different empathy levels. But yet, while predicting that hand restriction will affect participants with varying empathy levels differently, we expected to find the effect for both groups in the same direction, only with different intensities. We suggest that improving the performance of participants with low empathy implies that sitting on the hand helped their embodiment process and facilitated their understanding of others. We argue that the explanations of the embodiment process found in the literature today are suitable for the strategy high empathy participants employ; in other words, under natural conditions, the understanding of observed gestures relies on the grounded cognition and imitation (as suggested by Barsalou, 2008). When interfering with this process by restricting the hand movement, grounded imitation is less available and performance level is reduced (as was found with facial muscles and emotion recognition by Strack et al., 1988). In contrast, for participants with low empathy, the embodiment process and its utilization are fundamentally different (see findings mentioned above by Aziz-Zadeh et al., 2010; Gazzola et al., 2006; Kaplan & Iacoboni, 2006). Here we demonstrated that low empathy participants were benefiting from hand restriction; and in fact were able to reach the performance level that the empathetic participants showed under natural conditions (without any restriction). These findings suggest that individuals with low empathy may have difficulties relying on the grounded imitation. Their performance in understanding hand gestures was improved, however, when asked to sit on their hands. Here, we hypothesize that people with low empathy poorly utilized the embodiment process under normal (unrestricted) conditions. Asking them to sit on their dominant hand while performing the task, might have focused their attention to the hand and its cortical representation. This enhanced attention may have enabled or facilitated the embodiment mechanism, which resulted in an outcome of better (observed) gesture comprehension. At the same time, participants with high empathy scores utilize the embodiment process in an optimal way, and interfering with this process (in the hand restriction condition) only reduced its affectivity. Further studies are required to shed light on the exact mechanisms underlying improved performance in these individuals, using measurement of motor cortex activity during imitation restriction, using EEG or magnetic resonance-based brain imaging. Importantly though, these findings open the exciting possibility that in individuals with low empathy scores, known to have deficits in social cognition (Baron-Cohen & Wheelwright, 2004), behavioral interventions like embodiment training may help to

Ó 2017 Hogrefe Publishing

K. Jospe et al., Embodiment Modulation and Empathy Levels

improve empathetic abilities. By focusing attention to different body parts and activating the relevant area in the motor cortex, individuals with low empathy may be able to improve their understanding of the people around them, with profound implications for social interaction. The current study thus makes a unique contribution to embodiment theories. While the current perception is that embodiment is an automatic process, here we show that its utilization is subject to individual differences, with empathetic individuals capable of a more efficient utilization. The good news is that the embodiment module is flexible, and can be activated more efficiently following simple manipulations of motor activation or attention to body parts. Acknowledgments This study was supported by the German-Israel Foundation Grant No. I-1299-105.4/2015. Electronic Supplementary Materials The electronic supplementary material is available with the online version of the article at http://dx.doi.org/10.1027/ 1618-3169/a000351 ESM 1. Video film (.wmv). Hand Gesture Video example. ESM 2. Pictures (.jpg). Facial Expression examples. ESM 3. Data (.sav). Raw data collected in the experiment.

References Alibali, M. W., & Hostetter, A. B. (2010). Mimicry and simulation in gesture comprehension. Behavioral and Brain Sciences, 33, 433–434. doi: 10.1017/S0140525X10001445 Aziz-Zadeh, L., Sheng, T., & Gheytanchi, A. (2010). Common premotor regions for the perception and production of prosody and correlations with empathy and prosodic ability. PLoS One, 5, 1–8. doi: 10.1371/journal.pone.0008759 Baron-Cohen, S., & Wheelwright, S. (2004). The empathy quotient: an investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences. Journal of Autism and Developmental Disorders, 34, 163–175. doi: 10.1023/B:JADD.0000022607.19833.00 Barsalou, L. W. (2008). Grounded cognition. Annual Reviews Psychology, 59, 617–645. doi: 10.1146/annurev.psych.59. 103006.093639 Chartrand, T. L., & Bargh, J. A. (1999). The Camaleon effect: The perception-behavior link and social interaction. Journal of Personality and Social Psychology, 76, 893–910. doi: 10.1037/ 0022-3514.76.6.893 Chiavarino, C., Bugiani, S., Grandi, E., & Colle, L. (2013). Is automatic imitation based on goal coding or movement coding? Experimental Psychology, 60, 213–225. doi: 10.1027/16183169/a000190

Ó 2017 Hogrefe Publishing

Cohen-Maximov, T., Avirame, K., Flöel, A., & Lavidor, M. (2015). Modulation of gestural-verbal semantic integration by tDCS. Brain Stimulation, 8, 493–498. doi: 10.1016/j.brs.2014.12.001 Decety, J., Echols, S., & Correll, J. (2010). The blame game: the effect of responsibility and social stigma on empathy for pain. Journal of Cognitive Neuroscience, 22, 985–997. doi: 10.1162/ jocn.2009.21266 Ekman, P., & Friesen, W. V. (1978). Facial action coding system: A technique for the measurement of facial movement. Palo Alto, CA: Consulting Psychologists Press. Fadiga, L., Fogassi, L., Pavesi, G., & Rizzolatti, G. (1995). Motor facilitation during action observation: A magnetic stimulation study. Journal of Neurophysiology, 73, 2608–2611. Gazzola, V., Aziz-Zadeh, L., & Keysers, C. (2006). Empathy and somatotopic auditory mirror system in humans. Current Biology, 16, 1824–1829. doi: 10.1016/j.cub.2006.07.072 Hawk, S. T., Fischer, A. H., & Van Kleef, G. A. (2011). Taking your place or matching your face: Two paths to empathic embarrassment. Emotion, 11, 502–513. doi: 10.1037/a0022762 Heimann, K., Umilta, M. A., & Gallese, V. (2013). How the motorcortex distinguishes among letters, unknown symbols and scribbles. A high density EEG study. Neuropsychologia, 51, 2833–2840. doi: 10.1016/j.neuropsychologia.2013.07.014 Hess, U., & Fischer, A. (2013). Emotional mimicry as social regulation. Personality and Social Psychology Review, 17, 142–157. doi: 10.1177/1088868312472607 Kaplan, J. T., & Iacoboni, M. (2006). Getting a grip on other minds: Mirror neurons, intention understanding, and cognitive empathy. Social Neuroscience, 1, 175–183. doi: 10.1080/ 17470910600985605 Knafo, A., Zahn-Waxler, C., Van Hulle, C., Robinson, J. L., & Rhee, S. H. (2008). The developmental origins of a disposition toward empathy: Genetic and environmental contributions. Emotion, 8, 737–752. doi: 10.1037/a0014179 Murata, A., Saito, H., Schug, J., Ogawa, K., & Kameda, T. (2016). Spontaneous facial mimicry is enhanced by the goal of inferring emotional states: Evidence for moderation of “automatic” mimicry by higher cognitive processes. PLoS One, 11(4). doi: 10.1371/journal.pone.0153128 Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9, 97–113. Parzuchowski, M., Szymkow, A., Baryla, W., & Wojciszke, B. (2014). From the heart: Hand over heart as an embodiment of honesty. Cognitive Processing, 15, 237–244. doi: 10.1007/s10339-0140606-4 Richter, D., & Kunzmann, U. (2011). Age differences in three facets of empathy: Performance-based evidence. Psychology and Aging, 26, 60–70. doi: 10.1037/a0021138 Rychlowska, M., Canadas, E., Wood, A., Krumhuber, E. G., Fischer, A., & Niedenthal, P. M. (2014). Blocking mimicry makes true and false smiles look the same. PLoS One, 9(3). doi: 10.1371/ journal.pone.0090876 Shamay-Tsoory, S. G. (2011). The neural bases for empathy. The Neuroscientist, 17, 18–24. doi: 10.1371/journal.pone. 0090876 Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facilitating conditions of the human smile: A non-obtrusive test of the facial feedback hypothesis. Journal of Personality and Social Psychology, 54, 768–777. doi: 10.1037/00223514.54.5.768 Thomas, R., Sink, J., & Haggard, P. (2013). Sensory effects of action observation: Evidence for perceptual enhancement driven by sensory rather than motor simulation. Experimental Psychology, 60, 335–346. doi: 10.1027/16183169/a000203

Experimental Psychology (2017), 64(1), 56–64

Williams, J. H. G., Nicolson, A. T. A., Clephan, K. J., Grauw, H.d., & Perrett, D. I. (2013). A novel method testing the ability to imitate composite emotional expressions reveals an association with empathy. PLoS One, 8, e61941. doi: 10.1371/journal.pone. 0061941 Received February 7, 2016 Revision received December 6, 2016 Accepted December 6, 2016 Published online February 20, 2017

Experimental Psychology (2017), 64(1), 56â&#x20AC;&#x201C;64

K. Jospe et al., Embodiment Modulation and Empathy Levels

Karine Jospe Department of Psychology The Gonda Multidisciplinary Brain Research Center Bar Ilan University Ramat Gan, 5290002 Israel KarineJospe@gmail.com

Ă&#x201C; 2017 Hogrefe Publishing

Instructions to Authors Experimental Psychology publishes innovative, original, high-quality experimental research. The scope of the journal is defined by experimental methodology and thus papers based on experiments from all areas of psychology are welcome. To name just a few fields and domains of research, Experimental Psychology considers manuscripts reporting experimental work on human learning, memory, perception, action, language, thinking, problem-solving, judgment and decision making, social cognition, and neuropsychological aspects of these topics. Apart from the use of experimental methodology, a primary criterion for publication is that research papers make a substantial contribution to theoretical research questions. For experimental papers that have a mainly applied focus, Experimental Psychology is not the appropriate outlet. A major goal of Experimental Psychology is to provide a particularly fast outlet for such research. Authors usually receive an editorial decision within 6 weeks of manuscript submission. Experimental Psychology publishes the following types of article: Research Articles, Short Research Articles, Theoretical Articles, and Registered Reports. Replication studies should be submitted as a Registered Report. Manuscript Submission. All manuscripts should in the first instance be submitted electronically at http://www.editorial manager.com/exppsy. Detailed instructions to authors are provided at http://www.hogrefe.com/j/exppsy Copyright Agreement. By submitting an article, the author confirms and guarantees on behalf of him-/herself and any coauthors that he or she holds all copyright in and titles to the submitted contribution, including any figures, photographs, line drawings, plans, maps, sketches, tables, raw data, and other electronic supplementary material, and that the article and its contents does not infringe in any way on the rights of third parties. ESM and raw data files will be published online as received from the author(s) without any conversion, testing, or reformatting. They will not be checked for typographical errors or functionality. The author indemnifies and holds harmless the publisher from any third party claims. The author agrees, upon acceptance of the article for publication, to transfer to the publisher the exclusive right to reproduce and distribute the article and its contents, both physically and in nonphysical, electronic, and other form, in the journal to which it has been submitted and in other independent publications, with no limits on the number of copies or on the form or the extent of the distribution. These rights are transferred for the duration of copyright as defined by international law.

Ó 2017 Hogrefe Publishing

Furthermore, the author transfers to the publisher the following exclusive rights to the article and its contents: 1. The rights to produce advance copies, reprints, or offprints of the article, in full or in part, to undertake or allow translations into other languages, to distribute other forms or modified versions of the article, and to produce and distribute summaries or abstracts. 2. The rights to microfilm and microfiche editions or similar, to the use of the article and its contents in videotext, teletext, and similar systems, to recordings or reproduction using other media, digital or analogue, including electronic, magnetic, and optical media, and in multimedia form, as well as for public broadcasting in radio, television, or other forms of broadcast. 3. The rights to store the article and its content in machinereadable or electronic form on all media (such as computer disks, compact disks, magnetic tape), to store the article and its contents in online databases belonging to the publisher or third parties for viewing or downloading by third parties, and to present or reproduce the article or its contents on visual display screens, monitors, and similar devices, either directly or via data transmission. 4. The rights to reproduce and distribute the article and its contents by all other means, including photomechanical and similar processes (such as photocopying or facsimile), and as part of so-called document delivery services. 5. The right to transfer any or all rights mentioned in this agreement, as well as rights retained by the relevant copyright clearing centers, including royalty rights to third parties. Online Rights for Journal Articles Hogrefe will send the corresponding author of each accepted paper free of charge an e-offprint (PDF) of the published version of the paper when it is first released online. This e-offprint is provided exclusively for the author’s personal use, including for sharing with coauthors. Other uses of the e-offprint/ published version of record, including but not limited to the following, are not permitted except with the express written permission of the publisher: posting the e-offprint/published version of record to a personal or institutional website or to an institutional or disciplinary repository; changing or modifying the digital file; reproducing, distributing, or licensing the article in whole or in part for commercial use. If you wish to post the article to your personal or institutional website or to archive it in an institutional or disciplinary repository, please use either a pre-print or a post-print of your manuscript in accordance with the publication release for your article and the document ‘‘Guidelines on sharing and use of articles in Hogrefe journals’’ on the journal’s web page at www.hogrefe. com/j/exppsy. December 2016

Experimental Psychology (2017), 64(1)

How to meet people’s increasing need to develop and manage their own lives and careers “A must read for all career development professionals and students alike. The Handbook of the Life Design is a new and essential resource for those working to improve career services in line with today’s challenges and conditions.” Sara Santilli, PhD, writing in Career Convergence, February 2015

Laura Nota / Jérôme Rossier (Editors)

Handbook of Life Design

From Practice to Theory and from Theory to Practice 2015, vi + 298 pp., hardcover US $54.00 / € 38.95 ISBN 978-0-88937-447-8 Also available as eBook Our lives and careers are becoming ever more unpredictable. The “lifedesign paradigm” described in detail in this ground-breaking handbook helps counselors and others meet people’s increasing need to develop and manage their own lives and careers. Life-design interventions, suited to a wide variety of cultural settings, help

www.hogrefe.com

individuals become actors in their own lives and careers by activating, stimulating, and developing their personal resources. This handbook first addresses life-design theory, then shows how to apply life designing to different age groups and with more at-risk people, and looks at how to train life-design counselors.

Methodology

line n o e e e fr e issu l p m sa

European Journal of Research Methods for the Behavioral and Social Sciences Methodology is the official organ of the European Association of Methodology (EAM)

ISSN-Print 1614-1881 ISSN-Online 1614-2241 ISSN-L 1614-1881 4 online issues and 1 print compendium per annum (= 1 volume)

Subscription rates (2017) Libraries / Institutions US $324.00 / € 249.00 Individuals US $159.00 / € 114.00 Postage / Handling US $8.00 / € 6.00

www.hogrefe.com

Editors Peter Lugtig Utrecht University, The Netherlands

Managing Editors Charlotte Rietbergen, Utrecht, The Netherlands

Jose-Luis Padilla University of Granada, Spain

Luis-Manuel Lozano, Granada, Spain

About the Journal Methodology is the official organ of the European Association of Methodology (EAM), a union of methodologists working in different areas of the social and behavioral sciences (e.g., psychology, sociology, economics, educational and political sciences). The journal provides a platform for interdisciplinary exchange of methodological research and applications in the different fields, including new methodological approaches, review articles, software information, and instructional papers that can be used in teaching. Three main disciplines are covered: data analysis, research methodology, and psychometrics. The articles published in the journal are not only accessible to methodologists but also to more applied researchers in the various disciplines.

Manuscript Submissions All manuscripts should be submitted to one of the two editors, preferably by email: • Peter Lugtig MethodologyJournal.fsw@uu.nl •J ose-Luis Padilla methodologyjournal@ugr.es Electronic Full Text The full text of the journal – current and past issues (from 2005 onward) – is available online at econtent.hogrefe.com/loi/med (included in subscription price). A free sample issue is also available here. Abstracting Services The journal is abstracted / indexed in Social Sciences Citation Index (SSCI) and Current / Social & Behavioral Sciences Contents (CC / S&BS) (since 2009), PsycINFO, PSYNDEX, ERIH, and Scopus. Impact Factor (Journal Citation Reports®, Thomson Reuters): 2015 = 1.935

How to assess the social atmosphere in forensic hospitals and identify ways of improving it “All clinicians and researchers who want to help make forensic treatment environments safe and effective should buy this book.” Mary McMurran, PhD, Professor of Personality Disorder Research, Institute of Mental Health, University of Nottingham, UK

Norbert Schalast / Matthew Tonkin (Editors)

The Essen Climate Evaluation Schema – EssenCES A Manual and More

2016, x + 108 pp. US $49.00 / € 34.95 ISBN 978-0-88937-481-2 Also available as eBook The Essen Climate Evaluation Schema (EssenCES) described here is a short, well-validated questionnaire that measures three essential facets of an institution’s social atmosphere. An overview of the EssenCES is followed by detailed advice on how to administer

www.hogrefe.com

and score it and how to interpret findings, as well as reference norms from various countries and types of institutions. The EssenCES “manual and more” is thus a highly useful tool for researchers, clinicians, and service managers working in forensic settings.