Abstract
Interacting with other individuals to negotiate how we want to distribute resources (e.g. money) is an important part of our social life. Considering that not all of our requests are granted, the outcomes of such social interactions are, by their nature probabilistic, and therefore risky. While perception of risk has been widely studied in non-social contexts, cognitive/computational models of risk processing in human social interactions have not been established to a comparable degree. Here, we address this knowledge gap by modelling the proposer behaviour in the Ultimatum Game, which we used as an experimental probe of human interpersonal negotiations, where one side (i.e. the proposer) offers terms that the other side (i.e. the responder) may accept/reject. In two social decision-making experiments, our participants (acting as proposers) chose between making offers covering a wide range of monetary splits, while others accepted or rejected them probabilistically, in relation to their valuation of social reward distributions. We show that in interpersonal negotiations, people make use of their knowledge about their opponents’ valuation processes while dynamically adjusting their risk-preferences to make optimal expected value calculations. Our results suggest that, this dynamic adjustment is influenced by a number of factors such as one’s own degree of prosociality, opponents’ estimated prosociality, and the uncertainty of these estimates. Thus, we provide the first evidence to suggest that dynamic risk taking in human interpersonal negotiations follows similar computational principles to non-social value-based decision-making but it is shaped by the mental models that we construct about others.
Significance Humans are capable of developing sophisticated strategies before entering into a negotiation. This study describes a cognitive/computational mechanism underlying how people evaluate the risk inherent in such uncertain social situations where their terms may be accepted or rejected depending on how the others value them. Surprisingly, despite its everyday and political importance, evaluation of risk in human interpersonal negotiations has not previously been studied using a cognitive/computational framework. In an ecologically valid experimental design, we show that people have a dynamic risk-preference in such social interactions, which is influenced by the mental models they construct about their opponents.