Dynamic interactions between sensory and decision-making circuits shape temporal evidence weighting

Perceptual decisions hinge on the brain's ability to merge sensory data over time, yet the brain's method for adjusting this integration varies with task requirements. The exact origin of this flexibility in sensory evidence weighting between sensory and decision circuits is unknown. To decipher these mechanisms, researchers merged behavioral observations, neural data, and computer simulations to explore the core processes involved.

In a motion-discrimination task, macaques were asked to discern motion patterns under varying sensory evidence statistics, favoring integration at different times. Behavioral measurements showed that the influence of evidence ebbed and flowed over time, linked to changes in motion perception accuracy and decision-making signals in visual area MT.

A recurrent network model was crafted to simulate sensory-decision interactions, featuring bidirectional connections and a fluctuating modulatory signal akin to urgency. The model mirrored the behavioral and neural patterns observed, adjusting the timing and progression of evidence weighting without greatly affecting raw sensory signals.

The MT's decision-making signals can be split into two parts: one originating from sensory input and the other from within the decision-making process. These results suggest that the brain's capacity for flexible sensory integration arises from the interplay between sensory transmission, internal decision circuits, and signals modulating urgency, offering a framework for how neural systems can adapt sensory evidence processing to meet evolving task needs.