The fluid events model: Predicting continuous task action change.

Area of Science:

• Cognitive Psychology
• Behavioral Science
• Human-Computer Interaction

Background:

• Human behavior in interactive events is influenced by both environmental stimuli and prior experiences.
• Predicting action-switching in real-time, dynamic situations remains a challenge in behavioral science.

Purpose of the Study:

• To introduce and validate the fluid events model, a novel behavioral model for predicting action-switching in ongoing interactive events.
• To determine the relative contributions of event structure and experience-based factors in predicting behavioral shifts.

Main Methods:

• Assessed the fluid events model using data from diverse tasks: figure-drawing, decision-making, and probability matching.
• Analyzed trial-by-trial action switching, incorporating event-structure factors (structure changes, action suitability, time on task) and experience-based factors (recent/frequent shifts, performance dips, propensity to switch).

Main Results:

• The fluid events model achieved 70% accuracy in predicting trial-by-trial action switching, a 34% improvement over chance.
• Demonstrated high correlations (mean r = 0.98) between predicted and actual overall switch rates.
• Experience-based factors were found to be more significant predictors of action switching than event-structure factors in the tested tasks.

Conclusions:

• The fluid events model offers a robust framework for predicting behavioral adaptation in dynamic environments.
• Prior experience significantly shapes immediate behavioral choices, often more than the immediate situational context.
• The model's success highlights the importance of integrating historical data into predictive behavioral models.