Patterns of neuronal activity underlying behavioral decisions

National Institute Of Mental Health

Investigators

Linked publications & trials

Abstract

Cortical neurons fire in complex patterns of activity during behavior and cognition. In sensory regions of the cortex, animals' interactions with the sensory world evoke neural activity in the cortex. But not all patterns of cortical activity can be elicited by sensory stimuli. We study how non-sensory, artificially induced activity patterns can be used by animals to make behavioral responses. These studies shed light on the limits of cortical function, and how circuit properties like neural connections constrain the set of activity patterns the cerebral cortex can process. Trained animals (Emx1-Cre) report non-sensory optogenetic stimuli delivered to excitatory cells (Flex-ChrimsonR) in primary visual cortex. We find that animals improve their detection performance for such stimuli over the course of days, exhibiting learning in both sensitivity (more accurate responses to stimuli) and speed (faster responses to stimuli). Animals show a mean decrease in reaction time per session for constant stimulus intensity (p<0.01, Wilcoxon rank sum test against median of 0). Over many days, subjects improve their detection sensitivity by several orders of magnitude in stimulus intensity. Recent findings show that cortical responses to the optogenetic input also increase, suggesting a remodeling of recurrent connectivity that can increase amplification of the input as animals learn to use that input. The implications are broad: that cortical networks use recurrent connectivity to amplify certain patterns of input, and that this amplification can developed, via learning, in the adult. We have also obtained results on gain changes in excitatory neurons that affect cortical computation and are dependent on the cortical recurrent network in mouse V1. Transforming patterns of inputs into output firing is a central aspect of neural computation, but how large-scale networks of many recurrently-connected cells impact these transformations remains challenging to study in vivo. One proposed mechanism is recurrent cortical networks can amplify desired inputs or filter out extraneous inputs, a computation that can benefit perceptual behavior. Here, we investigate whether mouse visual cortex (V1) amplifies or attenuates responses to visual input via an adjustment of the input-output relationship of its component neurons. Using holographic stimulation methods, we precisely target individual cells based on their visually-evoked activity, and directly provide a fixed optogenetic input. To test if the local network might change these neurons input-output functions, we stimulated V1 pyramidal cells (n=342) both during and without visual stimulation to measure changes in optogenetically-evoked output. In visually-driven cells, we find no evidence for changes in input-output relationships. However, visually-suppressed cells show significantly weaker outputs to fixed input when the visual stimulus is present compared to without. These results indicate V1 neurons are pulled into a sublinear input-output regime when suppressed by the network, but largely remain in a linear regime when driven. In support of the stabilized supralinear network model of cortical function, this suggests cortical circuitry may use supralinear transfer functions a property of cells embedded in a noisy network to filter selected input patterns by attenuating those unrelated to the visual stimulus, not by amplifying responses. Together with other work in the lab, these findings contribute to our understanding of how cortical networks compute that is, how the connectivity within and between brain areas transforms inputs to process information, a function that is at the core of how brains work.