The orbitofrontal cortex updates beliefs for state inference.

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Computational Neuroscience

Background:

• The orbitofrontal cortex (OFC) is known to be involved in learning and inferring latent states.
• However, the specific computations the OFC performs for state inference remain unclear.

Purpose of the Study:

• To investigate the precise computation performed by the rat OFC during state inference.
• To identify neural signatures associated with state inference in the OFC.

Main Methods:

• Rats were trained on a temporal wagering task with hidden reward states.
• OFC inactivations were performed to assess the role of the OFC in belief updating.
• Electrophysiology and novel population analysis methods were used to analyze OFC neural dynamics.

Main Results:

• Rats utilized state inference to guide their decisions in the temporal wagering task.
• OFC inactivation impaired the rats' ability to update beliefs about hidden states.
• Latent neural factors reflecting inferred states were identified in OFC recordings from rats performing inference, but not from those using alternative strategies.
• Neural firing rates and population factors exhibited abrupt changes correlated with state transitions.

Conclusions:

• The rat OFC actively updates beliefs about hidden states, a process critical for decision-making.
• OFC neural dynamics contain decipherable information about inferred states, providing neural signatures of this cognitive process.
• These findings elucidate a specific computation performed by the OFC and its neural basis in state inference.