A state-space model for finger tapping with applications to cognitive inference

Sensory-motor functions have been repeatedly linked to both cognitive and physical functions. One common test of sensory-motor performance frequently used for neuropsychological evaluation is the Halstead-Reitan finger tapping test (FTT). While this test has been normed and used extensively, the underlying sensory, motor and cognitive processes mediating tapping behavior during the test are not well understood. As a first step towards investigating the behavioral aspects manifested by these processes, we describe a state-space model for finger tapping during the FTT. This state-space model exploits quasiperiodicity to decompose tapping into a set of time-varying states corresponding to the instantaneous amplitude of the finger oscillation, the instantaneous frequency (or speed) of tapping, and a phase that keeps track of the current finger position during the cycle. We evaluate the model by showing a good fit between estimated and actual measurements, and outline an experiment that will relate features from the model to cognitive function.