Two representations of a high-dimensional perceptual space

Area of Science:

• Neuroscience
• Cognitive Science
• Psychophysics

Background:

• Perceptual spaces are internal representations of sensory information crucial for decision-making.
• While color perception is well-studied, the neural basis of generic perceptual spaces, especially high-dimensional ones, remains unclear.
• Understanding these spaces is key to deciphering how the brain transforms sensory signals into actions.

Purpose of the Study:

• To investigate how the brain represents complex, high-dimensional perceptual spaces.
• To explore the underlying combinatorial strategies used in these representations.
• To compare different methods of measuring perceptual distances within a visual texture domain.

Main Methods:

• Utilized psychophysical measurements in human observers (N=4).
• Determined perceptual distances using a segmentation threshold task.
• Assessed perceptual distances via a suprathreshold border salience comparison task.
• Focused on a 10-dimensional domain of visual textures.

Main Results:

• Quantitative and qualitative differences were observed between the two measurement tasks.
• Segmentation thresholds suggested a coordinate-based representation with Euclidean geometry.
• Border salience comparisons indicated a curved space and activity patterns across broadly tuned elements.
• Observer uncertainty from border salience did not align with segmentation thresholds.

Conclusions:

• The brain employs at least two distinct representational strategies for complex perceptual spaces.
• These strategies differ in their underlying geometry and combinatorial mechanisms.
• Findings suggest flexibility in how the brain constructs and utilizes perceptual spaces for processing sensory information.