Home | People | Papers | Projects | Methods | Links | Graduate Programs | Subjects
Evidence suggests that a relatively early representation in the visual processing of orthography (before lexical access or the representation of phonology) is specific to our writing system and does not reflect fundamental visual properties such as shape: Different stimuli that have little or no visual similarity (e.g., "A" and "a") are represented with similar codes. Does the visual system really use such a seemingly non-visual representation for reading and if so, how does it arise?
Our second line of research has looked for neural evidence for the existence of such abstract letter identities and word forms and has investigated how such a representation could arise. First, we will describe a neuroimaging study that found evidence for just such an abstract representation of words in left inferior occipitotemporal cortex. Then we will describe a neural network model and behavioral study that support a correlation-based hypothesis of how such an abstract representation could arise.