Prediction-based attention computing: a proof of concept study.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Computational Psychology

Background:

• Extended reality (XR) and data modeling offer new avenues for adaptive simulations.
• Measuring and responding to individual neuropsychological states in real-time remains a challenge.
• Optimal metrics for real-time data capture and user experience enhancement in adaptive XR are under investigation.

Purpose of the Study:

• To examine prediction-based attention computing (PbAC), a novel adaptive XR approach.
• To determine if PbAC can adaptively capture users' internal state predictions.
• To assess PbAC's ability to modulate associated neuropsychological responses.

Main Methods:

• An XR-based racquetball paradigm was employed.
• In-situ eye tracking assessed participants' prior beliefs and prediction errors.
• Sensorimotor behaviors were compared across PbAC, probability-controlled, and matched-order conditions.

Main Results:

• Sensorimotor responses were significantly influenced by the expectedness of XR stimuli.
• PbAC conditions demonstrated clear prediction-related biases.
• High prediction error trials in PbAC elicited marked surprisal responses, comparable to control conditions.

Conclusions:

• The findings provide proof of concept for prediction-based attention computing (PbAC).
• PbAC shows potential for tailoring simulations to individual cognitive and sensorimotor states.
• This research supports the further development of PbAC in future XR innovations.