Category-Based Learning About Deviant Outgroup Members Hinders Performance in Trust Decision Making

Social and non-social categorisation in investment decisions and learning

Categorical processes allow us to make sense of the environment effortlessly by grouping stimuli sharing relevant features. Although these processes occur in both social and non-social contexts, motivational, affective, and epistemic factors specific to the social world may motivate individuation over categorisation of social compared with non-social stimuli. In one experiment, we tested this hypothesis by analysing the reliance on categorical versus individuating information when making investment decisions about social and non-social targets. In an adaptation of the iterative trust game, participants from three experimental groups had to predict the economic outcomes associated with either humans (i.e., social stimuli), artificial races (i.e., social-like stimuli), or artworks (i.e., non-social stimuli) to earn economic rewards. We observed that investment decisions with humans were initially biased by categorical information in the form of gender stereotypes, but later improved through an individuating learning approach. In contrast, decisions made with non-social stimuli were initially unbiased by categorical information, but the category-outcomes associations learned through repeated interactions were quickly used to categorise new targets. These results are discussed along with motivational and perceptual mechanisms involved in investment decisions and learning about social and non-social agents.

Social overshadowing: Revisiting cue-competition in social interactions

In a large variety of contexts, it is essential to use the available information to extract patterns and behave accordingly. When it comes to social interactions for instance, the information gathered about interaction partners across multiple encounters (e.g., trustworthiness) is crucial in guiding one's own behavior (e.g., approach the trustworthy and avoid the untrustworthy), a process akin to trial-by-trial learning. Building on associative learning and social cognition literatures, the present research adopts a domain-general approach to learning and explores whether the principles underlying associative learning also govern learning in social contexts. In particular, we examined whether overshadowing, a well-established cue-competition phenomenon, impacts learning of the cooperative behaviors of unfamiliar interaction partners. Across three experiments using an adaptation of the iterated Trust Game, we consistently observed a 'social overshadowing' effect, that is, a better learning about the cooperative tendencies of partners presented alone compared to those presented in a pair. This robust effect was not modulated by gender stereotypes or beliefs about the internal communication dynamics within a pair of partners. Drawing on these results, we argue that examining domain-general learning processes in social contexts is a useful approach to understanding human social cognition.

Ethnic discrimination unlearned: experience in the repeated Trust Game reduces trust bias

Introduction

Discrimination toward ethnic minorities is a persistent societal problem. One reason behind this is a bias in trust: people tend to trust their ingroup and comparatively distrust outgroups.

Methods

In this study, we investigated whether and how people change their explicit trust bias with respect to ethnicity based on behavioral interactions with in- and outgroup members in a modified Trust Game.

Results

Subjects' initial explicit trust bias disappeared after the game. The change was largest for ingroup members who behaved unfairly, and the reduction of trust bias generalized to a small sample of new in- and outgroup members. Reinforcement learning models showed subjects' learning was best explained by a model with only one learning rate, indicating that subjects learned from trial outcomes and partner types equally during investment.

Discussion

We conclude that subjects can reduce bias through simple learning, in particular by learning that ingroup members can behave unfairly.

What is cued by faces in the face-based context-specific proportion congruent manipulation?

In a typical context-specific proportion congruent manipulation, participants are presented with Stroop stimuli in one of two contexts. In one context, stimuli are mostly incongruent. In the other context, stimuli are mostly congruent. Despite task-wide instructions to ignore the word and to name the color in which the word appears, the size of the congruency effect varies as a function of context. Specifically, the size of the congruency effect is reduced for the mostly incongruent relative to the mostly congruent context. The purpose of the current series of experiments is to explore the mechanisms underlying this context-specific proportion congruent (CSPC) effect when faces are used as the context. Existing manipulations have reported a face-based CSPC effect, however the results of these studies are confounded with contingency learning biases leaving an open question as to what processes faces serve to cue. In the four experiments reported here both inducer (contingency-biased) and diagnostic (contingency-unbiased) stimuli were included in a face-based context level manipulation. Across four experiments, a face-based CSPC effect is observed for inducer but not diagnostic stimuli, suggesting that this effect is driven by participants learning context + stimulus associations.

Older and Younger Adults Perform Similarly in an Iterated Trust Game

In social contexts, aging is typically associated with a greater reliance on heuristics, such as categorical information and stereotypes. The present research examines younger and older adults’ use of individuating and age-based categorical information when gauging whether or not to trust unfamiliar targets. In an adaptation of the iterated Trust Game, participants had to predict the cooperative tendencies of their partners to earn economic rewards in first encounters – in a context in which they knew nothing about their partners, and across repeated interactions – in a context in which they could learn the individual cooperative tendency of each partner. In line with previous research, we expected all participants to rely on stereotypes in first encounters, and progressively learn to disregard stereotypes to focus on individuating behavioral cues across repeated interactions. Moreover, we expected older participants to rely more on social categories than younger participants. Our results indicate that overall, both the elderly and the young adopted an individuating approach to predict the cooperative behaviors of their partners across trials. However, older adults more consistently relied on gender (but not age) stereotypes to make cooperation decisions at zero acquaintance. The impact of context, motivation, and relevance of categorical information in impression formation is discussed.

How Adolescents and Adults Learn About Changes in the Trustworthiness of Others Through Dynamic Interaction

Whether to trust or distrust another individual is a complex interpersonal challenge, especially when such individuals behave inconsistently. It is still unclear as to how individuals learn and adapt to fluctuations in the trustworthiness of others and how this process changes from adolescence to adulthood. To address these issues, we implemented repeated rounds of a trust game within the context of a complicated and changeable interpersonal environment. Specifically, adolescents and adults played the role of trustors who had to decide whether to invest money in two anonymous partners carrying the risk of no reciprocation. Unbeknownst to participants, these two partners had different trustworthiness profiles: one partner initially yielded a higher initial return rate (70%) while the other initially yielded a lower initial return rate (30%). Crucially, over repeated rounds, these two partners gradually changed their responses to the point where, finally, return rates were both neutral (50%). Results indicated that all participants showed less updating in the negative direction in response to good-to-neutral partners while more updating in the positive direction in response to the bad-to-neutral partner. Compared to adults, this behavioral disparity in responses to good-to-neutral and bad-to-neutral partners was less pronounced in adolescents. Based on the computational modeling approach, the potential mechanisms underlying their behavioral patterns were revealed: the higher learning rate promoted flexible adaptions in participants to untrustworthy trustees as they changed to neutral. The less pronounced distinction between good-to-neutral and bad-to-neutral partners in adolescents was related to their lower learning rate. Overall, our study extends the understanding of trust behavior to a fluctuating social context and highlights the role of social learning in social emotion and interaction.

The face-specific proportion congruency effect: social stimuli as contextual cues

Previous research shows that larger interference is observed in contexts associated with a high proportion of congruent trials than in those associated with a low proportion of congruent trials. Given that one of the most relevant contexts for human beings is social context, researchers have recently explored the possibility that social stimuli could also work as contextual cues for the allocation of attentional control. In fact, it has been shown that individuals use social categories (i.e., men and women) as cues to allocate attentional control. In this work, we go further by showing that individual faces (instead of the social categories they belong to) associated with a high proportion of congruent trials can also lead to larger interference effects compared to individual faces predicting a relatively low proportion of congruent trials. Furthermore, we show that faces associated with a high proportion of congruent trials are more positively evaluated than faces associated with a high proportion of incongruent trials. These results demonstrate that unique human faces are potential contextual cues than can be employed to apply cognitive control when performing an automatic task.