Learning Enhances Sensory and Multiple Non-sensory Representations in Primary Visual Cortex

Abstract

SummaryWe determined how learning modifies neural representations in primary visual cortex (V1) during acquisition of a visually guided behavioral task. We imaged the activity of the same layer 2/3 neuronal populations as mice learned to discriminate two visual patterns while running through a virtual corridor, where one pattern was rewarded. Improvements in behavioral performance were closely associated with increasingly distinguishable population-level representations of task-relevant stimuli, as a result of stabilization of existing and recruitment of new neurons selective for these stimuli. These effects correlated with the appearance of multiple task-dependent signals during learning: those that increased neuronal selectivity across the population when expert animals engaged in the task, and those reflecting anticipation or behavioral choices specifically in neuronal subsets preferring the rewarded stimulus. Therefore, learning engages diverse mechanisms that modify sensory and non-sensory representations in V1 to adjust its processing to task requirements and the behavioral relevance of visual stimuli.

Cortical states control visual spatial perception

“…In our visual spatial task, monocular detection was more difficult than binocular detection, and this may have accentuated the effects of cortical state (Chen et al, 2008;McGinley et al, 2015;Spitzer et al, 1988). Our study isolated the effects of cortical state on perception in stationary conditions, minimizing complicated interactions between arousal, visual motion, and locomotion (Niell and Stryker, 2010;Poort et al, 2015;Saleem et al, 2013;Vaiceliunaite et al, 2013;Vinck et al, 2015). Understanding the effects of cortical states on perception across modalities, during a variety of behavioral contexts, remains an important topic for future study.…”

Hunger-Dependent Enhancement of Food Cue Responses in Mouse Postrhinal Cortex and Lateral Amygdala

“…The responses of primary sensory cortical neurons are known to be fairly stable across days in untrained (Mank et al, 2008) or trained (Chen et al, 2015; Poort et al, 2015) mice, while the excitability of hippocampal neurons varies substantially across days (Ziv et al, 2013). However, little is known about across-day stability of representations in intervening regions such as POR, which is both a higher visual cortical area and an entry point into the hippocampal memory formation (Burwell and Amaral, 1998a, b).…”

“…It is possible that these effects could arise from inputs from early sensory cortical neurons, which can reflect aspects of stimulus value and reward (Fritz et al, 2003; Komiyama et al, 2010). For example, studies of V1 have shown evidence of response changes with learning (Makino and Komiyama, 2015; Poort et al., 2015) and the presence of reward timing (Shuler and Bear, 2006). We observed that while trial history had a weak influence on a subset of V1 neurons, effects in POR were more pronounced.…”

Random synaptic feedback weights support error backpropagation for deep learning

“…It therefore seems likely (perhaps inevitable) that this top-down modulation of neuronal activity will impact the learning that goes on in the lower area neuron's synapses. Indeed, recent work has demonstrated that learning in sensorimotor tasks alters representations in earlier cortical areas51. For higher layers to deliver error information that could enable lower layers to make useful changes to their synaptic weights, neurons in the lower layers should, at least in part, be able to differentiate a top-down error signal from activity originating in the forward pathway.…”