A Model of Representational Spaces in Human Cortex.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Current models of human cortical function focus on coarse-scale features, neglecting fine-scale activity patterns.
• There's a need for models that account for population responses encoding information in detailed neural activity.

Purpose of the Study:

• To present a linear model of shared representational spaces in the human cortex.
• To capture fine-scale distinctions in population responses using common response-tuning basis functions.
• To model cortical patterns with individual-specific topographic basis functions.

Main Methods:

• Developed a novel algorithm, searchlight hyperalignment, to derive a common model space for the entire cortex.
• Utilized complex, dynamic stimuli to sample visual, auditory, and social percepts broadly.
• Employed between-subject multivariate pattern classification and intersubject correlation of representational geometry.

Main Results:

• Successfully aligned neural representations across brains in occipital, temporal, parietal, and prefrontal cortices.
• Demonstrated that structural principles for shared neural representations are consistent across diverse information domains.
• Provided a model accounting for individual variability in coarse-scale topographies (e.g., retinotopy, category selectivity).

Conclusions:

• The developed model accounts for fine-scale neural activity patterns multiplexed with coarse-scale topographies.
• These fine-scale patterns carry finer distinctions in information encoding.
• Structural principles for shared neural representations are fundamental across the human cortex.