Testing generalization from static to dynamic faces using magnetoencephalography (MEG)

Poster Session E - Monday, April 15, 2024, 2:30 – 4:30 pm EDT, Sheraton Hall ABC

Sebastian Montesinos¹ (sebastian.montesinos@nih.gov), Shruti Japee¹, Lina Teichmann¹, Chris Baker¹; ¹National Institute of Mental Health

Studies of visual perception frequently make use of rapidly presented static images, despite the fact that our natural visual input is dynamic. The onset of these stimuli provokes reliable responses across the visual cortex. Here, we seek to investigate the extent to which the representations evoked during the presentation of static stimuli generalize to dynamic movies that involve those same stimuli. To probe this question, the current study uses magnetoencephalography (MEG) to compare the time course of signals while participants view static frames and dynamic movies of human faces and objects. The static faces vary in terms of head orientations while the dynamic movies show these faces moving naturally from one orientation to another, such that the static orientations always occur sometime during the movie. Time-resolved MEG decoding methods allow us to compare brain MEG signal patterns between the different face orientations. Preliminary results indicate that head orientation can be reliably decoded during static viewing after 100ms. We are now investigating whether models trained on the patterns during static viewing will generalize to the dynamic movies, such that we can predict the time-course of the movies based on these models. The results of this study will help elucidate the connection between static and dynamic representations in the brain.

Topic Area: PERCEPTION & ACTION: Vision