A normative model of peripersonal space encoding as performing impact prediction

Area of Science:

• Neuroscience
• Computational Neuroscience
• Cognitive Science

Background:

• Accurate prediction of body-environment contact is crucial for survival.
• Peripersonal space (PPS) is hypothesized to involve multisensory neurons encoding near-body stimuli.
• A normative model linking impact prediction and PPS encoding was previously lacking.

Purpose of the Study:

• To develop a normative model based on Bayesian Decision Theory for impact prediction.
• To demonstrate how this model recapitulates known characteristics of peripersonal space (PPS).

Main Methods:

• Leveraged Bayesian Decision Theory to formulate a computational model.
• Analyzed model outputs to identify characteristics aligning with PPS encoding.

Main Results:

• The model predicts a graded boundary for PPS.
• PPS size dynamically enlarges with increasing stimulus speed.
• Contact prediction is stronger for looming than receding stimuli, but present for receding stimuli under uncertainty.
• Model predictions scale with the perceived value of environmental objects.
• The model accounts for variations in PPS size across different body parts.

Conclusions:

• Peripersonal space (PPS) likely reflects the neural computation of impact prediction.
• The developed model provides a normative framework for understanding PPS.
• The study generates testable predictions for future empirical research on PPS and impact prediction.