[4K1-GS-1] Fundamental AI, theory: algorithm

[4K1-GS-1-02] A model of mind wandering based fluctuations of thinking and autonomic nervous system

〇Jumpei Nishikawa¹, Kazuma Nagashima¹, Ryo Yoneda¹, Junya Morita¹, Tetsuya Terada² (1. Shizuoka University, 2. Mazda Motor Corporation)

Keywords:Mind wandering, Homeostasis, Perceptual motor task, ACT-R, Cognitive model

To maintain performance on perceptual-motor tasks, it is necessary to regulate the arousal level of participant appropriately. However, the arousal level inside the participant does not directly appear outside. In order to predict and regulate this, we integrate models built in previous studies using the cognitive architecture ACT-R to represent the emergence of mind wandering during perceptual-motor tasks. Our model probabilistically switches between a task-focused mode and a mind-wandering mode during the task. The strength of memories (activation) associated with each mode is involved in this switching. Based on the assumptions that the stochastic fluctuations on activity corresponds to changes in the autonomic nervous system, we constructed a model in which the noise of memory activation decreases with the time, and activated by stimulus events. We confirmed that preliminary simulations using this model represented the qualitative characteristics of data from behavioral experiments obtained in perceptual-motor tasks with similar settings.