Visual crowding and category specific deficits for pictorial stimuli: A neural network model

This paper describes a series of modular neural network simulations of visu al object processing. In a departure from much previous work in this domain , the model described here comprises both supervised and unsupervised modul es and processes real pictorial representations of items from different obj ect categories. The unsupervised module carries out bottom-up encoding of v isual stimuli, thereby developing a "perceptual" representation of each pre sented picture. The supervised component then classifies each perceptual re presentation according to a target semantic category. Model performance was assessed (1) during learning, (2) under generalisation to novel instances, and (3) after lesion damage at different stages of processing. Strong cate gory effects were observed throughout the different experiments, with livin g things and musical instruments eliciting greater recognition failures rel ative to other categories. This pattern derives from within-category simila rity effects at the level of perceptual representation and our data support the view that visual crowding can be a potentially important factor in the emergence of some category-specific impairments. The data also accord with the cascade model of object recognition, since increased competition betwe en perceptual representations resulted in category-specific impairments eve n when the locus of damage was within the semantic component of the model. Some strengths and limitations of this modelling approach are discussed and the results are evaluated against some other accounts of category-specific recognition failure.