A humanness dimension to visual object coding in the brain

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Human Object Recognition

Background:

• Previous neuroimaging research on object recognition focused on limited categories (faces, bodies, scenes, vehicles).
• Broader studies explored dichotomies like animate/inanimate and continuous features like biological similarity.
• Existing research often neglects objects that challenge strict animate-inanimate categorization.

Purpose of the Study:

• To investigate the brain's representation of objects, including those that blur the animate-inanimate distinction.
• To compare the explanatory power of different object coding models, including animacy, agency, experience, and human similarity.
• To identify the temporal dynamics of object representation in the human brain.

Main Methods:

• Developed a novel stimulus set featuring standard objects and ambiguous items (e.g., robots, toy animals).
• Employed Magnetoencephalography (MEG) time-series decoding to analyze brain activity.
• Tested various computational models (shape, animacy, agency, experience, human similarity) against neural data.

Main Results:

• Early brain responses (0-200 ms) were primarily predicted by stimulus shape and retinotopic models.
• Later neural representations were better explained by higher-order models of agency and experience.
• A model based on human similarity provided the strongest account of brain representations after initial perceptual processing.

Conclusions:

• Object recognition in the human brain involves a temporal progression from low-level features to higher-level conceptual properties.
• Agency and experience are crucial dimensions for coding object representations beyond simple animacy.
• Human similarity emerges as a key factor in the brain's object representation, suggesting a 'human-centric' coding dimension.