Information selection efficiency in networks: A neurocognitive-founded agent-based model

Abstract

Heterogeneity in beliefs in financial markets constitutes a major obstacle to coordination. Beliefs being heavily driven by information, efficient coordination and selection of information in financial markets are interconnected matters. The dialectic between different mechanisms at play, behavioral, cognitive, and interactive, has received little investigation. However, it appears fundamental in opening the black box of agents’ decision-making in the context of information selection. We introduce an agent-based model of information signal selection. In line with the Agent Zero neurocognitive-founded model (Epstein in Agent zero: toward neurocognitive foundations for generative social science. Princeton University Press 2014), agents are endowed with deliberative, affective, and social modules, through which they acquire disposition to adopt a signal. Agents receive signal values and estimate their variance with memory, i.e., ability to remember past signal values; they emotionally react to shocks and transmit individual dispositions through social networks. Information selection efficiency and opinion volatility are studied activating different combinations of cognitive modules and considering various network topologies. Human cognition is significantly more complex. Yet, this simple model generates a rich set of behaviors, and complex general mechanisms of information selection emerge. Simulation results obtained with novel model exploration methodologies outline that the impact of network structure and of memory abilities on market coordination, information selection efficiency and opinion stability appear to vary in magnitude and in sign depending on the combinations of cognitive modules considered. The behavior of the aggregate cognition framework appears to be different from the aggregate behavior of its modules taken separately, and interactions between cognitive modules appear crucial to explain information selection at the micro-level.