Dynamic decoding of ongoing perception

Abstract

Decoding of perceptual and mental states using multivariate analysis methods has received much recent attention. It relies on selective responses to experimental conditions in single trials, aggregated across voxels. In this study, we show that decoding is also possible when the state of interest changes continuously over time. It is shown that both orientation and rotation direction of a continuously rotating grating can be decoded with high accuracy using linear dynamical systems and hidden Markov models. These findings extend the decoding results for static gratings and are of importance in the decoding of ongoing changes in mental state.