Memory Bias and Social Networks

Michèle Belot, Department of Economics, Cornell University & Marina Schröder, Institute of Economic Policy, Leibniz University, Hannover

Imagine you attend a social event and someone approaches you with a smile on their face. They clearly know who you are and are about to greet you. You, on the other hand, can’t really recall who they are. We have all been there. We often navigate such awkward social situations by attempting to pretend we remember. Because failing to remember someone feels like an insult.

Systematic bias on race and gender

Memory has always played an important role in building and maintaining social networks. And social networks are important in life: they can be the key to a great job, or to hear about relevant opportunities in general. If we can’t recall people, we can’t pass them information that is relevant to them. Systematic biases along attributes of race or gender in remembering those we encounter could lead to discrimination.

In a recent paper we published in the Journal of Behavioral and Experimental Economics, we studied how good people are at remembering others, and how attributes such as gender and race mediate our ability to recall relevant information about a particular person. 

Experiments in field and lab

We started by collecting data for a field experiment in a context that we know well: academic conferences. After all, academia is a context in which we know that social networks matter, and in which women and ethnic minorities have fared less well than others. One month after attending a (plenary only) conference in San Diego (2012) or Edinburgh (2014), we surveyed conference participants and asked them to match pictures of presenters with titles of papers presented. We gave them a choice between four possible answers. We then replicated the study in a controlled computer-setting, using the subject pool of the experimental lab at the University of Cologne. In this lab experiment we randomly matched pictures of non-academics with titles of scientific papers, 

In both cases, we found that women and ethnic minorities are more accurately remembered in settings where there are few of them. In such cases, gender or ethnicity are a distinctive attribute that enhances recall. But as their fraction increases, these  individuals are then more likely to be confused with others who share the same attributes. Women are more likely to be confused with other women, ethnic minorities are more likely to be confused with other ethnic minorities. White men on other hand, are much less likely to be confused with other white men. 

Included more, but remembered less?

These results are consistent with a theory of “categorization”, such that we lump together certain groups of people in our memory. As a consequence, we find it hard to recall who a specific person is, and that appears to be particularly the case when this person is a woman or an ethnic minority. . In settings where there are few people with minority attributes, recall is enhanced. In settings where there are more of them, this leads to confusion. Bias in recall seems an important but understudied topic.

Given the importance of social networks in life and the importance of remembering plays in network formation, we conjecture that such biases may have real implications for career paths and broader labour market outcomes. Behavioural strategies will need to be developed if improvements in the visibility of women and ethnic minorities in conferences, on boards, etc., has the unintended consequence of their identities becoming “blended” in our memories.

Full Article: Belot, M., & Schröder, M. (2023). Remember me? The role of gender and racial attributes in memory. Journal of Behavioral and Experimental Economics, 102008.