Can the Brain Build Probability Distributions?

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Psychology

Background:

• Human cognition efficiently processes vast data by representing the environment using probability distributions.
• Probabilistic accounts suggest latent variable representation, but neural underpinnings of global pattern recognition remain under-explored.

Purpose of the Study:

• To investigate the neural mechanisms enabling the human brain to represent complex, global features of probability distributions.
• To develop and apply a novel analysis of electroencephalography (EEG) data for single-trial examination of neural responses to probabilistic stimuli.

Main Methods:

• Participants listened to sequences of tones forming either normal or bimodal probability distributions.
• A novel single-trial EEG analysis was employed to identify neural responses indexing item likelihood within the context of presented distributions.

Main Results:

• A distinct neural response was identified that correlates with the likelihood of individual tones based on preceding stimuli.
• This neural response accurately reflected the global structure of the presented auditory probability distributions (normal vs. bimodal).

Conclusions:

• The adult human brain possesses the capacity to construct representations of complex probability distributions.
• The novel EEG analysis provides a valuable tool for further research into the neural basis of probabilistic cognition and perception.