Outlier responses reflect sensitivity to statistical structure in the human brain.

Area of Science:

• Cognitive Neuroscience
• Neurophysiology
• Sensory Processing

Background:

• The brain extracts environmental regularities for prediction.
• Previous research focused on repetitive sequences or stimulus-outcome associations.
• Implicit encoding of non-sequential stochastic regularities remains less understood.

Purpose of the Study:

• To investigate whether humans implicitly encode non-sequential stochastic regularities.
• To detect violations of these statistical regularities.
• To identify neurophysiological markers associated with this process.

Main Methods:

• Novel experimental design using pure-tone sounds with frequencies from a Gaussian distribution.
• Behavioral and magnetoencephalographic (MEG) metrics were collected.
• Source reconstruction was used to identify brain regions involved.

Main Results:

• Sounds from the distribution's tail evoked larger responses than central sounds, similar to mismatch negativity (MMN).
• Responses to identical outliers were larger with narrower distributions, indicating sensitivity to uncertainty.
• Statistical-context-sensitive responses originated in a temporo-parietal network.

Conclusions:

• Humans implicitly encode uncertainty from random sensory distributions.
• An early neurophysiological marker for encoding complex statistical structure was identified.
• This sensitivity supports perceptual inference in noisy environments and decision-making under uncertainty.