Multi-stable perception balances stability and sensitivity.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Visual Perception

Background:

• Multi-stable perception involves spontaneous reversals in phenomenal appearance during continuous viewing.
• Perceptual dynamics are influenced by stabilizing and destabilizing factors, previously modeled computationally.

Purpose of the Study:

• To characterize the statistical dynamics of multi-stable perception in individual observers.
• To identify a consistent operating regime for multi-stable perception.
• To determine if this regime represents a functional optimization balancing stability and sensitivity.

Main Methods:

• Statistical analysis of dominance times (mean, variance) and history dependence (correlation, time-constant).
• Computational modeling incorporating competition, adaptation, and noise.
• Identification of model parameter volumes reproducing empirical data and comparison with theoretical 'sweet spots'.

Main Results:

• A consistent "operating regime" for multi-stable perception was identified in 21 of 24 datasets, with low probability (<0.02).
• This regime was found to be marginally stable, near oscillatory instability.
• The empirical operating regime largely coincided with the theoretical "sweet spot" balancing stability and sensitivity.

Conclusions:

• Multi-stable perception operates in a functionally optimized regime, balancing stability and sensitivity.
• This optimization suggests a mechanism for effective visual inference.
• Perceptual dynamics in multi-stable viewing are not arbitrary but serve a functional purpose.