Deciding to simulate: Cognitive mechanisms of predicting the decisions of others

People often rely on simulating the other person’s state of mind when trying to understand and predict their behaviour. Likewise, converging evidence suggests similar cognitive and neural processes are engaged when making decisions for ourselves and when predicting the decisions of others, pointing towards mental simulation as the underlying mechanism of prediction. In this study, we tested this hypothesis directly, leveraging the fact that, under the simulation hypothesis, any predictions made with a biased decision-making process should also be biased. We first induced a bias in participant’s risk perception by adapting them to contexts with high or low risks, and then asked them to predict decisions of a risk-seeking and risk-averse agent in a mediumrisk context. Our findings provide partial support for the simulation hypothesis; the adaptation manipulation resulted in biased predicted decisions, suggesting that these predictions were made with participants’ own (biased) decision-making system. Surprisingly, this behavioural effect only emerged during predictions for a risk-averse, but not risk-seeking agent. Modeling predictions with the drift-diffusion model suggested that the prediction bias mapped onto a group difference in the risk tendency parameter when predicting for the risk-averse, but not risk-seeking agent. Furthermore, participants’ risk tendency when deciding for themselves correlated with predicted behaviour of the risk-averse, but not risk-seeking agent, suggesting the perceived similarity of the other agent as a mediating factor. Altogether, our findings support the view that people engage in simulating others’ decisions with their own mind, but also suggest that the perceived similarity with the other person may determine their prediction strategy.