Behavior-relevant top-down cross-modal predictions in mouse neocortex.

Area of Science:

• Neuroscience
• Cognitive Science

Background:

• Animals predict their environment and actions for adaptation.
• The predictive coding hypothesis suggests the brain generates predictions and compares them to sensory input.
• Mechanisms for reconciling conflicting predictions and sensory information are not fully understood.

Purpose of the Study:

• To investigate how the brain resolves conflicts between top-down predictions and bottom-up sensory information.
• To explore the role of hierarchical predictive coding in behavior.

Main Methods:

• Simultaneous imaging of neuronal populations in the mouse somatosensory barrel cortex and posterior parietal cortex.
• Utilizing an auditory-cued texture discrimination task with mismatched stimuli.
• Analyzing information flow dominance (top-down vs. bottom-up).

Main Results:

• Mismatched auditory cues and textures created conflicts between predictions and sensory input.
• Dominant top-down information flow altered texture representations, leading to choices based on predictions.
• Dominant bottom-up information flow resulted in correct texture representations and behavioral choices.

Conclusions:

• Hierarchical predictive coding operates in the mouse neocortex.
• The balance between top-down and bottom-up processing determines behavioral outcomes when predictions conflict with sensory input.