Modelling sensory attenuation as Bayesian causal inference across two datasets.

Area of Science:

• Cognitive Neuroscience
• Computational Psychiatry
• Sensory Processing

Background:

• Distinguishing self-generated from external sensory information is vital for environmental interaction.
• Sensory consequences of self-movements typically elicit attenuated neural and behavioral responses compared to external stimuli.
• Sensory attenuation (SA) is proposed to occur when an internal cause for sensory information is inferred.

Purpose of the Study:

• To propose and validate a computational model of sensory attenuation (SA) based on Bayesian Causal Inference (BCI).
• To investigate the role of inferred internal causes in sensory attenuation.
• To model empirical patterns of SA across different experimental paradigms.

Main Methods:

• Developed a sequential Bayesian Causal Inference (BCI) model.
• Utilized a hierarchical Markov Model (HMM) and variational message passing for simulations.
• Optimized participant-specific model parameters for two experiments involving tactile and delay detection tasks.

Main Results:

• The BCI model successfully captured empirical patterns of sensory attenuation in both experiments.
• Participant-specific model parameters showed good agreement between data and model predictions.
• The model accurately predicted tactile detections in Experiment 1 and delay detections in Experiment 2.

Conclusions:

• Bayesian Causal Inference (BCI) provides a robust framework for modeling human sensory attenuation.
• Computational models of SA can unify findings across sensory modalities and experimental paradigms.
• This approach may enhance understanding of sensory processing deficits in conditions like schizophrenia.