Presentation Information
[A-8-04]Predicting Masticatory Duration from Temporal Scaling Properties of Chewing Sounds: Hierarchical Bayesian Modeling of Prediction-Error Convergence in Older Dentate and Edentulous Groups
〇Shunji Sugimoto1, Toshikazu Miura2, Takahiro Okawa2, Shizuka Tarukawa2, Manaka Koga2, Asako Suzuki2, Satoshi Horihata2, Masayasu Ito2, Yasuhiko Kawai2 (1. Toyohashi Univ. of Tech., 2. Nihon Univ.)
Keywords:
chewing sound,temporal scaling,fractal,prediction-error minimization,hierarchical Bayesian model
Chewing is a sensorimotor integration process in which the food bolus progressively converges toward a swallowable state through oral sensory feedback. In this study, we investigated the predictive relationship between the temporal scaling exponent αPSD of chewing sounds and masticatory duration in a dentate group (D) and a complete denture wearer group (CD) using a hierarchical Bayesian model. Based on a prediction-error minimization framework, we formulated a model in which αPSD is linked to the convergence rate of prediction error and constructed a linear predictor that separates variance into between-subject and within-subject components. Results showed that at the between-subject level, larger individual mean αPSD was associated with longer masticatory duration, and this tendency was more pronounced in the CD group. In contrast, the relationship between within-subject (trial-to-trial) variation in αPSD and masticatory duration remained unclear. By separating between- and within-subject variance, we demonstrated that the association between αPSD and masticatory duration is primarily driven by inter-individual differences. The steeper slope observed in the CD group suggests that reduced oral sensory feedback amplifies the association between the temporal correlation structure of chewing sounds and the timing of swallowing decisions.