Abstract

We are constantly flooded with information (e.g., via the Web, advertising, colleagues, and friends) that helps us form predictions about the future and make decisions. Classic theories in psychology, economics and machine learning all implicitly assume that beliefs are adjusted in a similar manner in response to desirable and undesirable information. In this talk I will provide evidence challenging these assumptions. I will show that while people incorporate good news into their existing beliefs in a normative manner, they display an aversion to incorporating bad news, discounting the strength of the information. This asymmetry in how we incorporate information can have important societal implications including the generation of financial market bubbles and ill preparedness in the face of natural disasters. However, we find that the asymmetry is not set in stone, but rather fluctuates in response to threats in the environment, in a way that may be adaptive. Using a combination of fMRI, TMS and DTI we characterise the neural system supporting this asymmetry and show that changes in this circuit are related to individual differences in behaviour and mental health, with more balanced coding of information observed in depression and middle age.