Influence of observation on other-regarding attitude: a computational exploration

Study aims. Observing others' choices allows people to learn useful information about the social context and can potentially change their preferences. We investigate how other-regarding attitudes shift after participants are asked to predict the interpersonal choices made by others. We hypothesize that acquaintance with social choices of a group should push individual attitude towards them. We also test how two cognitive models of preferences are able to accommodate the shift in social attitude. The Preference Temperature model assumes that preferences are stable but are implemented with noise. The Preference Uncertainty model states that preferences are expressed without noise but change continuously gravitating towards a particular state. Methodology. 98 participants chose repeatedly in a task in which they could reduce their own earnings in order to increase or reduce earnings of an unknown other. Between the first and second parts of the task, participants made a series of predictions about the choices made in the same task by a group of previous participants. This observed group consisted of individuals who expressed an attitude similar to that of the participant but extreme (e.g., very altruistic or very spiteful). We estimated the two models of participants' attitudes before and after the manipulation using Markov Chain Monte Carlo estimation. Findings. Participants' attitudes shifted in the direction of the attitude expressed by the observed group. Notably, this effect was driven by altruistic participants, whereas spiteful participants appeared not to be consistently affected by observation. The size of the shift was the same irrespective of how distant participants' attitudes were from the observed group's and of how consistent participants were in their choices. In accordance to previous findings, when comparing the cognitive models, we found that the Preference Uncertainty model accounted for the shift in attitude better than the Preference Temperature model. Conclusions. Attitude towards other individuals appears to be influenced by learning about others' behavior. However, this effect is different for altruistic and spiteful people. This difference could imply that these two types of participants draw upon qualitatively different cognitive mechanisms. The Preference Uncertainty model, which explains our data better, yields testable predictions about brain activation: value representation areas, such as the ventromedial prefrontal cortex and ventral striatum, should display a variability in signal in absence of information about other's choices, whereas any influence of observation should produce clearly detectable changes in activation.