The project investigated how top-down inputs from the secondary somatosensory cortex (S2) influence sensory processing and perceptual sensitivity in primary somatosensory cortex (S1). To address this, we combined anatomical circuit mapping, large-scale functional imaging, a tactile detection task for head-fixed mice, and chemogenetic manipulations of long-range cortical inputs.
We first mapped the organization of S2-to-S1 projections using complementary retrograde and anterograde tracing approaches. Distinct populations of S2 neurons were found to project selectively to specific S1 subregions corresponding to different body representations, demonstrating target specificity within this top-down pathway. Anterograde labeling further revealed that S2 axons preferentially arborize in layer 1 of S1, contacting the distal dendrites of S1 pyramidal neurons.
Animals were trained to report whisker stimulation by licking a water spout to obtain water reward in a whisker-detection task. Stimulus intensity was progressively reduced to reach the perceptual threshold, defined as the intensity at which the stimulus was detected in approximately 50% of trials. Under these conditions, chemogenetic silencing of S2 axon terminals in S1 led to a significant increase in detection threshold and a marked impairment in task performance, establishing a causal role for S2 inputs in modulating tactile sensitivity.
To characterize the functional dynamics of these projections, we performed wide-field calcium imaging of S2 axons expressing a genetically encoded calcium sensor within S1. Axonal activity was registered to individual somatotopic maps using anatomical landmarks derived from layer-4 neurons expressing a red fluorescent protein, enabling alignment across animals. We then compared S2 input activity on trials in which stimuli of identical intensity were either detected (hits) or not detected (misses).
Analysis of trial-to-trial variability revealed that perceptual outcome was strongly predicted by prestimulus S2 input activity. Hit and miss trials diverged in their axonal activity patterns well before stimulus onset, suggesting that baseline cortical state biases subsequent perception. Spatially resolved analyses showed that successful detection was preceded by a selective increase in S2 input activity within the task-relevant S1 representation, accompanied by suppression in other S1 regions. These prestimulus spatial patterns reliably predicted behavioral outcomes, as confirmed by receiver-operating-characteristic and decoding analyses.
To determine whether these patterns generalize across sensory contexts, we compared S2 input dynamics during detection of stimuli applied to different body parts. Prestimulus activity patterns were conserved across different whiskers (C2 and B1) representations but differed during hindpaw detection, reflecting a redistribution of S2 input activity toward the behaviorally relevant S1 region. Despite these differences, prestimulus activity remained highly predictive of perceptual outcome across tasks.
Finally, we examined whether baseline S2 input activity is passively fluctuating or actively tuned by task demands. Using a block-based design with alternating whisker- and hindpaw-detection contexts, we found that prestimulus S2 input activity shifted systematically at block transitions.
Overall, the project established that S2 inputs to S1 are anatomically specific, dynamically organized, and causally involved in shaping perceptual sensitivity. A major achievement of this work is the identification of task- and body-part–specific baseline input patterns that predict perceptual outcome, providing direct evidence that top-down cortical pathways actively configure activity of primary sensory cortex to support adaptive behavior.
