Nicotine causes highly localized and tuned changes in thalamocortical gain that cascade strongly and selectively through the cortical circuit.

In layer 4 of the primary sensory cortices, sensory data meets contextual information carried by modulatory systems, establishing a potentially important point of control. Accordingly, modulatory receptor expression at this point often differs from other nearby layers and cortical areas. Altering the processing state at even a small number of layer 4 thalamocortical synapses should be impactful on downstream processing, as this is the point at which most information about the outside world enters cortex. It is, however, often assumed-based on diffuse innervation of cortex by subcortical nuclei, and dense receptor expression in the association cortices-that modulatory signals are shared across large swaths of tissue, with the primary site of action being in 'higher' levels of cortex. Hypothesizing that modifying the feedforward input to cortex might be an important capability of the cholinergic system, we recorded across the depth of cortex while delivering nicotine to a small proportion of thalamocortical synapses in macaque primary visual cortex, leveraging selective receptor expression to achieve spatial control. We observed bidirectional response changes throughout all layers, including layers in which nicotine is known to have no effect when applied locally. The pattern of these changes carried the signature of having passed through a normalizing circuit, rather than signatures of drug spread: the magnitude and tuning dependence of the response changes were well-accounted for by a normalization model with a tuned, multiplicative gain field. The circuit apparently did not compensate these changes; perceived contrast in a behavioral task was biased with the same normalization-predicted tuning dependence.