Visually-evoked choice behavior driven by distinct population computations with non-sensory neurons in visual cortical areas

Abstract

During visual detection tasks, subjects sometimes fail to respond to identical visual stimuli even when the stimuli are registered on their retinas. It is widely assumed that variability in detection performance is attributed to the fidelity of the visual responses in visual cortical areas, which could be modulated by fluctuations of subjective internal states such as vigilance, attention, and reward experiences. However, it is not clear what neural ensembles represent such different internal states. Here, we utilized a behavioral task that differentiated distinct perceptual states to identical stimuli, and analyzed neuronal responses simultaneously recorded from both primary visual cortex (V1) and posterior parietal cortex (PPC) during the task. We found that population activity differed across choice types with the major contribution of non-sensory neurons, rather than visually-responsive neurons, in V1 as well as PPC. The distinct population-level activity in V1, but not PPC, was restricted within the stimulus presentation epoch, which was distinguished from pre-stimulus background activity and was supported by near-zero noise correlation. These results indicate a major contribution of non-sensory neurons in V1 for population-level computation that enables behavioral responses from visual information.2 instance, in sensory detection task, human or animal subjects are instructed or well-trained to 3 reliably report the presence and absence of sensory stimuli to obtain rewards. When the sensory 4 evidence is near threshold for the decision criterion, the subjects' reports vary across trials 5 despite the best efforts of subjects to get rewards. Interestingly, even if they report the absence 6 of stimuli, it is sometimes possible that they could correctly guess the contents of the stimuli 7 above chance level if they are forced to answer 1-8 . Revealing the neural mechanisms underlying 8 such trial-by-trial variability of perceptual reports is crucial to understand how the brain exploits 9 sensory information for optimal decision making. 11The trial-by-trial variance of the responses to identical stimuli is believed to reflect noises in 12 conversion of sensory information into motor outputs 9 . It has been demonstrated that variability 13 of firing rates of sensory neurons is responsible for the trial-by-trial variability of choices 10-14 .14 However, the accumulating evidence suggests that perceptual decision is also significantly 15 affected by latent subjective states reflecting task engagement 15,16 . For instance, it is known that 16 behavioral response variability is correlated with mind wondering in humans 17 and fluctuations 17 of physiological and behavioral states in animals [18][19][20][21][22] . These drifts of subjective states could be 18 partially attributed to fluctuation of cortical states 22-28 , in which synchronization and 19 desynchronization of many neurons in particular areas of cortex could affect efficiency of the 20 population coding 29,30 . In addition, the task engagement is known t...

An intra-oral flavor discrimination task in freely moving mice

“…The model fit these predictors. To quantify the impact of each task parameter, we calculated the difference between the explained variances (R 2 ) of the full and partial models 45 . We used 5-fold cross-validation by leaving a 20% subset of trials for prediction to avoid overfitting.…”