Collective behaviour of cognitive agents (renewal)

Human interactions are substantially more complex and intensive than our nearest primate relations, as well as other vertebrate species. Nonetheless, as in animal groups human sociality is an adaptive response and can be understood as the product of natural selection. However, more than most animal species, humans now live in a physical and social environment far removed from that in which these social responses evolved, and one that continues to change rapidly. Novel transport and communications technologies exposed individuals to an ever-increasing number of potential interactions throughout modern history, and individuals can obtain social benefits, status and belonging in a number of loosely connected communities, such as their workplace, friendship groups and online forums. The rapid pace of change means that our instinctive social responses may now be unsuitable for the types of groups we inhabit, and for the quantity and quality of social information that we receive.

The potential pitfalls of human collective behaviour are well established, ranging from crowd panics to stock market crashes. Many scholars have described taxonomies of cognitive biases in human decision-making as explanations for apparent human irrationality. However, others have argued that such irrationality is not simply the result of cognitive limitations but stems from the divergence between the environment humans evolved in and that they now inhabit. A limitation of mapping human biases is that such a map serves to document how decisions are made at the time the map is made, rather than to predict how further changes will influence decision-making. To understand how ongoing changes in the social world will be reflected in human decision-making, we must identify what individuals are trying to accomplish when they make decisions, and what implicit knowledge and beliefs they draw on in trying to achieve those goals.

This fellowship extension will develop a first-principles theory of human...