Empirical Movement Models for Brain Computer Interfaces.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Standardized performance metrics for brain-computer interfaces (BCIs) with continuous position control are lacking.
• Fitts's Law, commonly used for aimed movements, has been applied to BCIs but exhibits limitations.
• Existing applications of Fitts's Law to BCIs show fragile predictive performance and unsupported information transfer rate estimations.

Purpose of the Study:

• To adapt and validate an alternative model to Fitts's Law for BCI performance evaluation.
• To address the limitations of Fitts's Law in the context of BCI tasks.
• To provide a more robust and accurate method for assessing BCI performance.

Main Methods:

• Reviewed existing studies applying Fitts's Law to BCI tasks.
• Adapted and validated the Shannon-Welford model for BCI performance assessment.
• Compared the predictive power of the Shannon-Welford model against Fitts's Law.

Main Results:

• The Shannon-Welford model demonstrated superior predictive performance compared to Fitts's Law.
• The Shannon-Welford model showed robust predictions even when target distance and width disproportionately affected task difficulty.
• Model parameters from the Shannon-Welford model offer a new way to quantify the impact of control-display gain on speed-accuracy trade-offs in brain control.

Conclusions:

• The Shannon-Welford model is a more suitable metric than Fitts's Law for evaluating continuous control BCIs.
• The Shannon-Welford model provides a robust framework for understanding speed-accuracy trade-offs in brain control.
• This research contributes a validated, alternative performance metric for advancing BCI technology.