Regular updates of internal sensory predictions drive sensorimotor adaptation in optimization-based models of speech production

This paper presents a version of an optimization-based model of speech production that reproduces key acoustic and articulatory features of sensorimotor adaptation to altered sensory feedback. In the presented approach, the mechanism of sensorimotor adaptation is solely driven by independently motivated regular updates, based on the sensory feedback perceived by the speaker, of two of the speakers' internal models used for computing (near)-optimal articulation. These internal models predict acoustic and somatosensory consequences of articulatory states, respectively. The paper presents simulations that replicate real adaptation experiments. These show successful reproduction of known aspects of sensorimotor adaptation, including gradual and incomplete adaptation when auditory feedback is suddenly altered (F1-shifted), adaptation to non-uniform auditory perturbations, and generalization of adaptation behavior to untrained vowels. They also show that the rate and magnitude of adaptation behavior depend on a small number of parameters in the proposed model. Variation in the values of these parameters explains inter-speaker differences in terms of adaptation behavior, including sensory preference.