Measuring internal representations from behavioral and brain data.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Psychology

Background:

• Internal knowledge representations are crucial for cognition, guiding sensory data processing.
• A key hypothesis suggests the brain uses internal models to simplify high-dimensional sensory input for decision-making.
• Understanding these internal representations is vital for explaining perception and higher cognitive functions.

Purpose of the Study:

• To directly visualize the content of subjective internal knowledge representations.
• To investigate how the brain utilizes internal models to interpret ambiguous sensory information.
• To determine the neural timing and location of internal knowledge influencing perception.

Main Methods:

• Employed a pure top-down, knowledge-based task where observers detected illusory faces in white noise.
• Utilized reverse correlation on behavioral data to reconstruct observers' internal representations.
• Applied reverse correlation to electroencephalogram (EEG) data to map neural activity related to internal knowledge.

Main Results:

• Successfully reconstructed subjective internal representations from behavioral and brain data.
• Demonstrated that internal representations guide the interpretation of white noise as a face.
• Showed that internal knowledge content drives neural activity sequentially in frontal and then occipitotemporal cortex.

Conclusions:

• Internal knowledge representations can be experimentally reconstructed and visualized.
• The brain actively uses internal models to interpret sensory input, not just passively receive it.
• Neural activity patterns reflect the dynamic interplay between internal knowledge and sensory processing.