Dynamic Representation of Multidimensional Object Properties in the Human Brain

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Visual Perception

Background:

• Human visual system rapidly identifies objects using complex representations.
• Understanding how object properties are neurally encoded over time is crucial.

Purpose of the Study:

• To investigate the temporal dynamics of multidimensional object representations in the human brain.
• To determine how behavior-derived object dimensions are reflected in neural signals over time.

Main Methods:

• Utilized magnetoencephalography (MEG) to record neural signals from four participants viewing thousands of object images.
• Developed an analysis approach integrating behavioral similarity judgments with time-resolved MEG data.
• Modeled time-resolved MEG signals to capture unfolding object representations.

Main Results:

• All behavior-derived object dimensions were reflected in neural signals.
• Object property processing exhibited two temporal patterns: early peaks (~125 ms) and late peaks (~300 ms).
• Early processing was stimulus-specific and stable across participants; later processing was more participant-specific and variable.

Conclusions:

• Early neural responses primarily reflect visual object dimensions, while later responses reflect conceptual dimensions.
• Conceptual object representations show greater inter-individual variability than visual representations.
• These findings provide a comprehensive model of how object properties unfold in the brain, underpinning rich object vision.