Back

Human primary auditory cortex and insula encode perceptual decisions, not stimulus features

Des Lauriers, C.; Pruvost-Robieux, E.; Zanello, M.; Filipescu, C.; Moiraghi, A.; Landre, E.; Pallud, J.; Defrance, G.; Sharshar, T.; Aucouturier, J.-J.; Gavaret, M.; Llorens, A.

2026-07-03 neuroscience
10.64898/2026.07.03.736338 bioRxiv
Show abstract

The traditional view of perceptual decision-making assumes a largely feedforward cortical hierarchy, in which sensory regions encode stimulus features that are progressively integrated with top-down signals in higher order associative areas to guide decisions. However, recent work has cast doubt on whether stimulus- and decision-related signals actually follow such a predicted spatiotemporal organization, especially in naturalistic situations where sensory cues are subtle and decisions more strongly driven by internal strategies. Leveraging the unique spatiotemporal precision of human intracerebral recordings, we map here how stimulus and decision variables are represented along the auditory cortical hierarchy as patients engage in a social voice decision task with realistic, low-salience cues. Contrary to feedforward predictions, we found no clear spatial or temporal gradient separating stimulus- and decision-related effects; rather, decision effects emerged early during stimulus exposure and, strikingly, flowed back all the way to the most primary regions of the superior temporal gyrus. In addition, the direction of these early and primary decision signals closely reflected the variability in patients' specific decision criteria. Taken together, these results strongly challenge the traditional feedforward model, supporting a view in which the activity of early auditory regions does not reflect subtle stimulus categories but is instead dynamically configured by task-dependent priors and response strategies.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.