Multiple Concurrent Predictions Inform Prediction Error in the Human Auditory Pathway.

Area of Science:

• Neuroscience
• Auditory Perception
• Computational Neuroscience

Background:

• Predictive coding theory posits that the brain generates predictions to interpret sensory input, with prediction errors signaling discrepancies.
• The role of prediction errors when conflicting predictions arise from different sensory processing stages remains unclear.

Purpose of the Study:

• To investigate how the auditory system encodes prediction errors when faced with contradictory statistical and rule-based predictions.
• To test whether neural responses reflect prediction errors relative to statistical, rule-based, or a combination of both prediction types.

Main Methods:

• Two fMRI experiments were conducted with human participants listening to auditory stimuli (pure tones and frequency-modulated sweeps).
• Repetition was used to induce stimulus statistics-informed predictions, while task instructions provided rule-informed predictions.
• Neural responses in auditory pathway regions were analyzed to determine the encoding of prediction errors.

Main Results:

• Neural populations in the inferior colliculus, medial geniculate body, and auditory cortices encoded prediction errors based on a combination of both statistical and rule-informed predictions.
• This indicates that prediction error signaling is not solely based on one type of prediction.

Conclusions:

• Auditory processing integrates multiple, potentially conflicting, sources of prediction.
• The findings challenge the notion of linear transmission of predictions in the sensory hierarchy, suggesting a non-linear integration mechanism.
• This non-linear integration may be essential for processing complex auditory information, such as speech.