How incentive motivation reduces cognitive noise and improves decision-making under uncertainty – MONODEC

Making decisions under uncertainty, from perceptual judgments to value-based choices, requires combining multiple pieces of ambiguous or conflicting information. Recent research has identified imprecise decision computations as an important contributor to the variability of perceptual and value-based decisions made under uncertainty. However, whether and how this ‘cognitive noise’ can be regulated to improve decision-making remains unknown. This project uses incentive motivation to test the idea of a resource-rational control of cognitive noise. We hypothesize that incentive motivation improves decision-making by reducing cognitive noise through more precise decision computations in the human brain. We propose to test this original hypothesis by developing a set of behavioral tasks and a computational modeling framework that quantifies the selective effect of incentive motivation on cognitive noise. We will measure cognitive noise at the behavioral level by fitting our computational model to human decisions, and at the neural level by measuring the precision with which we can decode model-predicted decision variables from dynamic brain activity patterns recorded in magnetoencephalography (MEG). Using dual-task variants of our core behavioral task, we will test whether incentive motivation reduces cognitive noise by allocating limited resources to concurrent tasks following resource-rational principles. We will assess the generalizability of the obtained findings by comparing incentive-based effects between perceptual and symbolic variants of our core behavioral task. By investigating how incentive motivation improves decision-making from behavior and brain activity using a single computational framework, this project provides a unique window into the mechanisms of human decision-making and their contextual regulation.