Attraction and repulsion in perception and working memory as complementary outcomes of learning

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Visual Perception

Background:

• Visual perception is inherently imperfect and influenced by environmental structure.
• Learned experiences create biases in perception and memory, shifting representations towards or away from reference points.
• These biases are acquired over various timescales, reflecting an interaction between the observer and their environment.

Purpose of the Study:

• To propose a unifying principle explaining diverse bias phenomena in perception and memory.
• To demonstrate how the brain optimizes noisy sensory data for behavioral efficiency under resource constraints.
• To link these biases to established frameworks like predictive and adaptive coding.

Main Methods:

• Theoretical modeling based on predictive and adaptive coding principles.
• Analysis of how experience stabilizes representations through stimulus integration.
• Examination of how adaptation maintains stimulus discriminability.

Main Results:

• Biases in perception and memory arise from a shared adaptive principle optimizing sensory representations.
• This principle explains how the brain integrates information for stability while using adaptation for discrimination.
• Predictive and adaptive coding frameworks naturally generate attraction and repulsion biases within tasks.

Conclusions:

• Perceptual and memory biases are emergent properties of a hierarchical statistical learning system.
• This system dynamically balances stability and flexibility in processing sensory information.
• Understanding these biases offers insight into the brain's adaptive strategies for perception and memory.