Recurrent pattern completion drives the neocortical representation of sensory inference

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Perception

Background:

• Perceptual inference, the brain's ability to interpret incomplete sensory data using prior expectations, is fundamental to perception.
• The precise neural mechanisms underlying sensory inference remain largely unknown.

Purpose of the Study:

• To elucidate the neural codes and circuits responsible for sensory inference in the brain.
• To investigate how the primary visual cortex (V1) processes illusory contours (ICs).

Main Methods:

• Utilized illusory contours (ICs) as a model system.
• Employed multi-Neuropixels recordings and mesoscale two-photon (2p) calcium imaging in mice.
• Applied 2p holographic optogenetics for precise neural circuit manipulation.

Main Results:

• Discovered a distinct neuronal population in V1 that responds to illusory bars, not just component stimuli.
• Demonstrated that activating these 'IC-encoders' via optogenetics can evoke the perception of ICs without visual input.
• Showed these neurons are critical for integrating top-down expectations with sensory data.

Conclusions:

• Identified specialized neurons in V1 that encode sensory inference through pattern completion.
• Suggests these neurons facilitate perception by reinforcing expected patterns via top-down signaling.
• Highlights the role of pattern completion circuits in lower cortical areas for perceptual inference.