BCI Control of Heuristic Search Algorithms.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Brain-Computer Interfaces (BCI) offer novel human supervision for complex Artificial Intelligence (AI) systems.
• Heuristic search is a fundamental AI mechanism used in various applications.
• Prefrontal Cortex (PFC) asymmetry correlates with motivational states and is controllable via Neurofeedback (NF).

Purpose of the Study:

• To develop a BCI mechanism for controlling heuristic search in AI systems.
• To investigate harnessing users' mental disposition to influence heuristic search performance.
• To enable faster AI solutions through user-driven adjustments.

Main Methods:

• Utilized functional Near-Infrared Spectroscopy (fNIRS)-based BCI to capture PFC asymmetry.
• Employed weighted variants of the A* algorithm (WA*) for adjustable solution speed-complexity.
• Mapped PFC asymmetry values to the dynamic weighting parameter of the WA* algorithm.
• Conducted experiments using 8-puzzle and path planning tasks.

Main Results:

• Subjects successfully modulated PFC asymmetry through NF during heuristic search tasks.
• User-modulated PFC asymmetry led to faster computation of solutions.
• Demonstrated user ability to influence heuristic search progression via BCI control.
• Achieved faster solutions by reducing the search space in WA*.

Conclusions:

• Established a novel mechanism for human intervention in AI heuristic search.
• Showcased the potential of fNIRS-based BCI for real-time AI control.
• Opened avenues for hybrid cognitive systems integrating human and AI capabilities.
• Highlighted the link between affective BCI signals and AI system performance.