Many decisions under uncertainty entail dynamic belief updating: multiple pieces of evidence informing about the state of the environment are accumulated to choose an appropriate action. Traditionally, this process has been conceptualised as a linear and perfect (i.e. without loss) integration of sensory information along purely feedforward sensory-motor pathways. Yet, natural environments can undergo hidden changes, which requires an adaptive, non-linear accumulation of decision evidence that strikes a tradeoff between stability and flexibility. How this adaptive computation is implemented in the brain has remained unknown.

I will present an approach that my laboratory has developed to identify evidence accumulation signatures in human behaviour and neural population activity (measured with magnetoencephalography, MEG), across a large number of well-defined cortical areas. Applying this approach to data recorded during visual evidence accumulation tasks with change-points, we find that behaviour and neural activity in frontal and parietal regions involved in action selection exhibit hallmarks signatures of adaptive evidence accumulation. The same signatures of adaptive behaviour and neural activity emerge naturally from simulations of a detailed model of a recurrent cortical microcircuit. The MEG data further show that decision dynamics in downstream action-selective areas are mirrored by a selective modulation of the state of early visual cortex. This state modulation (i) is expressed in the alpha frequency-band, (ii) tracks the evolving belief state encoded in downstream areas, (iii) adapts to the environmental volatility, and (iv) is amplified by pupil-linked arousal elicited by inferred change points. Our findings link normative decision computations to recurrent cortical circuit dynamics and highlight the adaptive nature of decision-related feedback processing in the brain.