About

Mahzarin R. Banaji is the Richard Clarke Cabot Professor of Social Ethics at Harvard University and has served as the Harvard College Professor from 2014 to 2019. She studies thinking and feeling as they unfold in social context, with a focus on mental systems that operate in implicit or unconscious mode. Her research explores social attitudes and beliefs in adults and children, especially those rooted in group membership, and examines the implications of these processes for individual responsibility and social justice in democratic societies. Her current research interests include the origins of social cognition and applications of implicit cognition to improve individual decisions and organizational policies. Prior to her tenure at Harvard, Banaji taught at Yale from 1986 to 2002, where she was the Ruben Post Halleck Professor of Psychology. Since 2002, she has held the position of Richard Clarke Cabot Professor of Social Ethics in the Department of Psychology at Harvard, and has also served as the first Carol K. Pforzheimer Professor at the Radcliffe Institute for Advanced Study, as well as the George A. and Helen Dunham Cowan Chair in Human Dynamics at the Santa Fe Institute. She has been recognized as a fellow of the Society for Experimental Psychologists, the Society for Experimental Social Psychology, and the American Academy of Arts and Sciences. She was named Herbert A. Simon Fellow of the American Academy of Political and Social Science and received the William James Fellow for her lifetime contributions to the science of psychology. Banaji is also the author of the book 'Blindspot: Hidden Biases of Good People,' published in 2013 with Anthony Greenwald. Her research interests encompass implicit social cognition, social development, attitudes and preferences, beliefs and stereotypes, intergroup relations, judgment and decision making, person perception, and research methods.

Research topics

• Psychology
• Sociology
• Social psychology
• Developmental psychology
• Computer Science
• Artificial Intelligence
• Political Science
• Social Science
• Gender studies
• Psychiatry
• Criminology
• Linguistics

Selected publications

• Reducing Implicit Racial Preferences: I. A Comparative Investigation of 17 Interventions

  OSF Preprints (OSF Preprints) · 2026-03-31

  other

  Many methods for reducing implicit prejudice have been identified, but little is known about their relative effectiveness. We held a research contest to experimentally compare interventions for reducing the expression of implicit racial prejudice. Teams submitted seventeen interventions that were tested an average of 3.70 times each in four studies (total N = 17,021), with rules for revising interventions between studies. Eight of seventeen interventions were effective at reducing implicit preferences for Whites compared to Blacks, particularly ones that provided experience with counterstereotypical exemplars, used evaluative conditioning methods, and provided strategies to override biases. The other nine interventions were ineffective, particularly ones that engaged participants with others’ perspectives, asked participants to consider egalitarian values, or induced a positive emotion. The most potent interventions were ones that invoked high self-involvement or linked Black people with positivity and White people with negativity. No intervention consistently reduced explicit racial preferences. Furthermore, intervention effectiveness only weakly extended to implicit preferences for Asians and Hispanics.

  Publisher

• Patterns of Implicit and Explicit Attitudes V. Increase in bias from 2021-2024

  2026-01-14

  articleOpen accessSenior author

  Between 2007-2020, implicit and explicit intergroup attitudes declined in bias steadily and were forecasted to continue toward attitude neutrality. New data from 2.5 million U.S. respondents (2021-2024) reveal that these encouraging trends have stalled or reversed. The largest increases in bias emerged for sexuality, transgender, race and skin-tone bias; 10-108% increases on explicit and 6-13% increases on implicit measures. Age, disability, and body weight bias also increased, but at slower rates. Exploratory breakpoint analyses showed that implicit attitudes were leading indicators of change, reversing trend earlier than explicit reports. Reversals were widespread across demographic groups for most topics, though strongest among conservatives for sexuality and transgender biases. Surprisingly, younger respondents (who had previously shown the largest decreases in bias) now showed greater increases in bias. Even after robust bias reduction spanning over 14 years, the new observed bias increases since 2021 highlight how minds get reshaped by sweeping sociocultural change.

  PublisherDOI

• Who is American? A comprehensive analysis of the American = White/Foreign = Asian stereotype (2007–2023)

  Scientific Reports · 2025-01-27 · 1 citations

  articleOpen accessSenior author

  Against the backdrop of increasing ethnic diversity in the U.S., we replicate, extend, and challenge previous examinations of the American = White/Foreign = Asian stereotype in the largest sample to date (N = 666,623 respondents) over 17 years (2007-2023). Six key findings emerged. First, a robust American = White association emerged on implicit (Cohen's d = 0.50) and explicit (Cohen's d = 0.51) measures. Second, the strength of this effect varied by respondents' race/ethnicity with implicit stereotypes strongest among White respondents (Cohen's d = 0.86) and absent among East Asian respondents (Cohen's d = 0.02). Third, the strength of implicit stereotypes was modulated by age, religion, and ideology-older, Christian, and conservative respondents displayed stronger implicit American = White associations-but not gender or education. Fourth, respondents living in U.S. metropolitan areas with greater Asian representation or a history of voting for Democratic candidates exhibited weaker implicit American = White associations. Fifth, over the past 17 years, implicit and explicit American = White associations decreased by 41% and 47%, respectively, and 14/14 demographic subgroups changed towards neutrality. Finally, we observed suggestive evidence that implicit stereotype trends towards neutrality were temporarily disrupted during the COVID-19 pandemic for White Americans but not Asian Americans.

  PublisherOA PDFDOI

• Kernels of Selfhood: GPT-4o shows humanlike patterns of cognitive consistency moderated by free choice

  ArXiv.org · 2025-01-27

  preprintOpen accessSenior author

  Large Language Models (LLMs) show emergent patterns that mimic human cognition. We explore whether they also mirror other, less deliberative human psychological processes. Drawing upon classical theories of cognitive consistency, two preregistered studies tested whether GPT-4o changed its attitudes toward Vladimir Putin in the direction of a positive or negative essay it wrote about the Russian leader. Indeed, GPT displayed patterns of attitude change mimicking cognitive consistency effects in humans. Even more remarkably, the degree of change increased sharply when the LLM was offered an illusion of choice about which essay (positive or negative) to write. This result suggests that GPT-4o manifests a functional analog of humanlike selfhood, although how faithfully the chatbot's behavior reflects the mechanisms of human attitude change remains to be understood.

  PublisherOA PDFDOI

• Kernels of Selfhood: GPT-4o shows humanlike patterns of cognitive consistency moderated by free choice

  2025-02-03 · 1 citations

  preprintOpen accessSenior author

  Large Language Models (LLMs) show emergent patterns that mimic human cognition. We explore whether they also mirror other, less deliberative human psychological processes. Drawing upon classical theories of cognitive consistency, two preregistered studies tested whether GPT-4o changed its attitudes toward Vladimir Putin in the direction of a positive or negative essay it wrote about the Russian leader. Indeed, GPT displayed patterns of attitude change mimicking cognitive consistency effects in humans. Even more remarkably, the degree of change increased sharply when the LLM was offered an illusion of choice about which essay (positive or negative) to write. This result suggests that GPT-4o manifests a functional analog of humanlike selfhood, although how faithfully the chatbot’s behavior reflects the mechanisms of human attitude change remains to be understood.

  PublisherDOI

• Kernels of selfhood: GPT-4o shows humanlike patterns of cognitive dissonance moderated by free choice

  Proceedings of the National Academy of Sciences · 2025-05-14 · 10 citations

  articleOpen accessSenior authorCorresponding

  Large language models (LLMs) show emergent patterns that mimic human cognition. We explore whether they also mirror other, less deliberative human psychological processes. Drawing upon classical theories of cognitive consistency, two preregistered studies tested whether GPT-4o changed its attitudes toward Vladimir Putin in the direction of a positive or negative essay it wrote about the Russian leader. Indeed, GPT displayed patterns of attitude change mimicking cognitive dissonance effects in humans. Even more remarkably, the degree of change increased sharply when the LLM was offered an illusion of choice about which essay (positive or negative) to write, suggesting that GPT-4o manifests a functional analog of humanlike selfhood. The exact mechanisms by which the model mimics human attitude change and self-referential processing remain to be understood.

  PublisherDOI

• Reply to Cummins et al.: GPT reveals cognitive dissonance that is both irrational and alarmingly humanlike

  Proceedings of the National Academy of Sciences · 2025-08-27 · 1 citations

  articleOpen accessSenior authorCorresponding

  We (Lehr et al., LSHVB) reported that GPT-4o a) shifts its attitudes after writing pro-or anti-Putin essays, and b) shifts attitudes more so after ostensibly choosing (versus being assigned to) which essay to write ( 1 ).Cummins et al. (CEH, 2 ) call these results "context window effects," disputing that they reflect cognitive dissonance.We clearly stated that GPT-4o mimics humanlike cognitive dissonance, and that these results do not indicate large language model (LLM) sentience (p.4).We therefore reject CEH's attribution of anthropomorphization. Here, we clarify our argument and offer new data that invalidate CEH's evidentiary claim-and thank them for the opportunity.CEH's Mischaracterization.We agree with CEH that context windows (CWs) "are a fact, not a rival hypothesis."In fact, they are a core feature of LLM architecture and also of human cognition (cf.assimilation, priming, recency effects).Although CEH claim we rejected it, our data actually supported the CW hypothesis: GPT's attitudes shifted dramatically in both choice and no-choice conditions.The irrationality we highlighted lies in GPT's inability to respect its own priors.GPT has millions of words about Putin in its training data; yet a tiny 600-word essay shifted its evaluation sharply-a monumental failure of Bayesian updating that should be concerning in any rational agent.CEH's Evidence/Our Counterevidence.We showed that GPT-4o shifts attitudes after merely writing an essay.CEH report that GPT-4o shifts attitudes after merely reading one.We welcome this conceptual replication of our experiment, but reading versus writing is not theoretically relevant: We know that dissonance effects arise from both encountering and producing information in humans too (3).What is theoretically critical is that GPT's behavior differed under choice versus no-choice.CEH ignore this key result and do not include the crucial choice/no-choice variation.In a new study, we correct this omission.GPT read essays (CEH's preferred paradigm), under one of three conditions: choice-granted, nochoice, and choice-disregarded (Table 1).Why do we expect the results hypothesized in the "Predictions: Context + Dissonance" cell of Table 1 ?Previous research teaches that when humans freely choose X, they

  PublisherOA PDFDOI

• The Development of Social Group Cognition in Infancy and Childhood

  Oxford University Press eBooks · 2024-08-21 · 1 citations

  book-chapterSenior author

  Abstract Although they are far from biological or social maturity, infants and children show surprising early-emerging capacities in social group cognition. This chapter reviews research on when and how infants and children categorize, evaluate, stereotype, and behave differently toward social groups defined by gender, race, age, and language. Research across these groups reveals three thematic conclusions. First, an early-emerging preference for the familiar (e.g., looking at faces most prevalent in infants’ environments), beyond similarity or in-group status. Second, generally similar trajectories across group targets (e.g., looking preferences at three to six months, evaluative associations formed around nine to twelve months) suggesting domain-general cognitive developments may scaffold infant social group cognition. Third, an additional internalization of culturally dominant beliefs and norms of fairness in early to middle childhood (e.g. the emergence of socially desirable, fair responding in middle childhood). Understanding social group cognition is advanced by understanding its origins in early in life.

  PublisherDOI

• Implicit Attitudes Evoked by a Singular American Slur: Five Experiments on <i>N***er</i> and <i>N***a</i> in Samples of Black and White Americans

  Social Cognition · 2024-06-01 · 4 citations

  articleOpen accessSenior author

  Five studies examined implicit (IAT) attitudes toward the slurs n***er and n***a among Black and White Americans (total N = 3,226). Both groups showed strong implicit negativity toward n***er/a combined relative to socially acceptable contrast terms such as Black or African American. Controlling for baseline Black-White race attitudes, Black Americans who engaged in conscious reappropriation exhibited similar implicit negativity toward n***er/a as White Americans. When the rhotic and non-rhotic forms were directly contrasted, n***er was more implicitly negative than n***a, with Black Americans distinguishing the two more strongly than did White Americans. However, even Black American reappropriators showed implicit negativity toward n***a relative to Black. In sum, both n***er and n***a evoke automatic negative meaning in a broad sample of Americans today. At the same time, the relatively more positive meaning of n***a over n***er demonstrates the power of reappropriation to wrest control of word meaning.

  PublisherOA PDFDOI

• The Romance of Research Methods

  Cambridge University Press eBooks · 2024-12-11

  book-chapter1st authorCorresponding
  PublisherDOI

Recent grants

• NIH Grant R01MH057672

  NIH · $1.2M · 2005

Frequent coauthors

• Brian A. Nosek

  Center for Open Science

  132 shared

• Anthony G. Greenwald

  116 shared

• Benedek Kurdi

  73 shared

• Elizabeth A. Phelps

  New York University

  42 shared

• Jason P. Mitchell

  Harvard University Press

  39 shared

• William A. Cunningham

  31 shared

• John C. Gore

  Vanderbilt University

  28 shared

• Tessa Elizabeth Sadie Charlesworth

  Northwestern University

  27 shared

Labs

• Group and Intergroup RelationsPI

Education

• Ph.D., Psychology

  Harvard University

  1986

• B.A., Psychology

  University of Delhi

  1981

Awards & honors

• Fellow of the Society for Experimental Psychologists
• Fellow of the Society for Experimental Social Psychology
• Fellow of the American Academy of Arts and Sciences
• Herbert A. Simon Fellow of the American Academy of Political…
• William James Fellow for a lifetime of significant intellect…