Quals research talk: A Dynamic Neural Field Model of Sequence Perception

November 6, 2017
Halligan 127
Speaker: Andrew Valenti, Tufts University
Host: Matthias Scheutz


We show how temporal and spatial information can be represented as stable patterns in a dynamical system. We hypothesize that human sensory input is encoded as patterns. These patterns encode spatiotemporal characteristics of the sensory input independent of whether the input is visual, auditory, or somatosensory. We describe a biologically plausible model in which category perception arises as temporal patterns are generated incrementally from the input sequence. These patterns are then used to identify a set of words whose prefix is the corresponding sequence, consistent with the Cohort theory of spoken word recognition.