Distributed context-dependent choice information in mouse posterior cortex.

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Choice information is distributed across brain networks, intermingled with sensory, motor, and cognitive variables.
• The posterior cortex's role in decision-making is unclear due to variable choice signals and their area specificity.

Purpose of the Study:

• To investigate the nature and computational significance of choice signals in the mouse posterior cortex.
• To determine if choice signals represent an independent decision variable within posterior cortical networks.

Main Methods:

• Examined mesoscale neural activity in the mouse posterior cortex during a visual task.
• Utilized low-dimensional embedding to analyze choice signals.
• Trained a recurrent neural network with animal choices to model decision variables.

Main Results:

• Choice signals formed a decision variable in a low-dimensional space, primarily in the ventral visual stream.
• This decision variable subspace was largely separate from sensory and motor representations.
• Modulations by task difficulty and attention state were observed, alongside agreement with a recurrent neural network model.

Conclusions:

• The posterior cortex contains an independent, multi-area decision variable.
• This variable is modulated by task features and cognitive demands, suggesting a role in contextual inference.
• Findings contribute to understanding decision-making in dynamic animal-environment interactions.