Cortical Transformation of Stimulus Space in Order to Linearize a Linearly Inseparable Task

Area of Science:

• Cognitive neuroscience
• Neuroimaging
• Computational neuroscience

Background:

• The human brain excels at learning complex categorization tasks with linearly inseparable boundaries.
• The neural mechanisms underlying this computational ability remain largely unknown.

Purpose of the Study:

• To investigate how the brain achieves computational feats in challenging categorization tasks.
• To explore neural representational changes associated with learning linearly inseparable categories.

Main Methods:

• Participants were trained on a morph shape space animal categorization task with a linearly inseparable prototype structure.
• Functional magnetic resonance imaging (fMRI) was used to scan participants before and after four days of behavioral training.

Main Results:

• Widespread representational changes were observed across the brain post-training.
• Neural patterns for the categories became more linearly separable after training.
• These task-dependent neural changes occurred in frontal and parietal areas, not ventral temporal cortices.

Conclusions:

• The brain flexibly transforms neural representational space to solve computationally challenging tasks.
• Neural changes reflect attentional and decisional reweighting rather than altered object recognition templates.
• Learning complex categories involves dynamic reorganization of neural representations.