The Computational Exploration of Visual Word Recognition in a Split Model
Neural Computation
A Neural Network Model of Lateralization during Letter Identification
Journal of Cognitive Neuroscience
A Computational Model of Lateralization and Asymmetries in Cortical Maps
Neural Computation
Towards Novel Neuroscience-Inspired Computing
Emergent Neural Computational Architectures Based on Neuroscience - Towards Neuroscience-Inspired Computing
Hi-index | 0.00 |
This chapter concerns the influence of the bihemispheric structure of the brain on processing. The extent to which the hemispheres operate on different aspects of information, and the nature of integrating information between the hemispheres are vexed topics open to artificial neural network modelling. We report a series of studies of split-architecture neural networks performing visual word recognition tasks when the nature of the stimuli vary. When humans read words, the exterior letters of words have greater saliency than the interior letters. This "exterior letters effect" (ELE) is an emergent effect of our split model when processing asymmetrical (word-like) stimuli. However, we show that the ELE does not emerge if the stimuli are symmetrical, or are mixed (symmetrical and asymmetrical). The influence of split processing on task performance is inextricably linked to the nature of the stimuli, suggesting that the task determines the nature of the separable processing in the two hemispheres of the brain.