Using brain imaging to track problem solving in a complex state space.

Area of Science:

• Cognitive Science
• Neuroscience
• Computational Psychology

Background:

• Tracking mental states during complex problem-solving is challenging due to vast state spaces.
• Integrating diverse data streams offers a more comprehensive understanding of cognitive trajectories.

Purpose of the Study:

• To develop and validate a method for tracking participants' paths through complex cognitive state spaces.
• To combine behavioral, imaging, and computational models for enhanced mental state classification.

Main Methods:

• Utilized a Hidden Markov Model (HMM) incorporating an Imperfect Memory Model for transition probabilities.
• Integrated behavioral data (event timing) and neuroimaging data (multivoxel pattern analysis).
• Applied HMM to a complex memory game with 625 states and numerous possible paths.

Main Results:

• Achieved over 80% accuracy in identifying participants' most probable paths through the state space.
• Demonstrated the efficacy of combining behavioral, imaging, and model-based data for classification.
• Successfully tracked cognitive trajectories in a high-dimensional problem-solving environment.

Conclusions:

• The combined data and HMM approach provides a robust methodology for tracking mental states.
• This technique offers a generalizable framework for analyzing individual problem-solving episodes.
• Supports the use of computational modeling with multimodal data in cognitive neuroscience research.