Continuous multiplexed population representations of task context in the mouse primary visual cortex

Abstract

Primary visual cortex (V1) neurons integrate motor and multisensory information with visual inputs during sensory processing. However, whether V1 neurons also integrate and encode higher-order cognitive variables is less understood. We trained mice to perform a context-dependent cross-modal decision task where the interpretation of identical audio-visual stimuli depends on task context. We performed silicon probe population recordings of neuronal activity in V1 during task performance and showed that task context (whether the animal should base its decision on visual or auditory stimuli) can be decoded during both intertrial intervals and stimulus presentations. Context and visual stimuli were represented in overlapping populations but were orthogonal in the population activity space. Context representation was not static but displayed distinctive dynamics upon stimulus onset and offset. Thus, activity patterns in V1 independently represent visual stimuli and cognitive variables relevant to task execution.

Not so spontaneous: Multi-dimensional representations of behaviors and context in sensory areas

“…This dissimilarity between spontaneous and sensory-driven patterns increases across zebrafish development (Avitan et al, 2021). Other internally generated signals, such as task context and behavioral choice, are also represented by neural directions orthogonal to sensory-driven patterns (Goris et al, 2017;Hajnal et al, 2021;Zhao et al, 2020). Together, these studies suggest that sensory areas process information on a background of high-dimensional SA that can represent many variables.…”

“…(E) The discriminant direction in neural population space which best distinguishes between visual stimuli was found to be orthogonal to the discriminant direction which best distinguishes the context of the task (visual or auditory) in a context-dependent behavioral task in mice (Hajnal et al, 2021).…”

“…Therefore, the separation of behavioral and stimulus neural subspaces enables high-fidelity decoding from neural populations, regardless of the animal's actions (Goris et al, 2017;Montijn et al, 2016;Stringer et al, 2021; Figure 2C). During decision-making tasks, several dimensions of internally generated activity in sensory areas correspond to representations of choice and behavioral context (Hajnal et al, 2021;Kim and Shadlen, 1999;Lee et al, 2020;Musall et al, 2019;Reinert et al, 2021;Salkoff et al, 2020;Steinmetz et al, 2019). This internally generated activity is also represented in different neural subspaces from sensorydriven activity.…”

“…This internally generated activity is also represented in different neural subspaces from sensorydriven activity. For instance, in the mouse visual cortex, visual information and task context (auditory versus visual) are optimally decoded from the neural population using orthogonal vectors (Hajnal et al, 2021;Figure 2E). In macaque middle temporal visual area (MT), visual information and choice information are represented in orthogonal subspaces (Zhao et al, 2020).…”

Estimating How Sounds Modulate Orientation Representation in the Primary Visual Cortex Using Shallow Neural Networks

“…Note that we had already shown that arousal and locomotion could not account for those results ( McClure and Polack, 2019 ), and we therefore did not consider those parameters further in this study. Moreover, several studies have recently demonstrated that the modulation of the V1 neuronal activity by behavioral parameters such as locomotion and arousal are orthogonal to orientation encoding ( Hajnal et al, 2021 ; Stringer et al, 2019 ).…”