A connectionist model of category learning by individuals with high-functioning autism spectrum disorder

Abstract

Individuals with autism spectrum disorder (ASD) show atypical patterns of learning and generalization. We explored the possible impacts of autism-related neural abnormalities on perceptual category learning using a neural network model of visual cortical processing. When applied to experiments in which children or adults were trained to classify complex two-dimensional images, the model can account for atypical patterns of perceptual generalization. This is only possible, however, when individual differences in learning are taken into account. In particular, analyses performed with a self-organizing map suggested that individuals with high-functioning ASD show two distinct generalization patterns: one that is comparable to typical patterns, and a second in which there is almost no generalization. The model leads to novel predictions about how individuals will generalize when trained with simplified input sets and can explain why some researchers have failed to detect learning or generalization deficits in prior studies of category learning by individuals with autism. On the basis of these simulations, we propose that deficits in basic neural plasticity mechanisms may be sufficient to account for the atypical patterns of perceptual category learning and generalization associated with autism, but they do not account for why only a subset of individuals with autism would show such deficits. If variations in performance across subgroups reflect heterogeneous neural abnormalities, then future behavioral and neuroimaging studies of individuals with ASD will need to account for such disparities.

Learning, plasticity, and atypical generalization in children with autism

“…As previously shown (Dovgopoly & Mercado, 2013), A Type 1 children showed generalization patterns similar to those previously found for TD children after training with distorted prototypes (Church et al, 2010). However, they showed relatively poorer generalization after training with the PrototypeOnly task (Fig.…”

“…Previous research showed that this percentage of endorsements of random shapes predicts the distinctive A Type 1 and A Type 2 generalization patterns seen in children with HFASD (Dovgopoly & Mercado, 2013). Also, the TD children in our previous research (Church et al, 2010) meet this A Type 1 criterion.…”

“…Numerous qualitative and computational models have been developed to account for specific behavioral and neural abnormalities associated with ASD (for reviews, see Björne & Balkenius, 2005;Dovgopoly & Mercado, 2013). Most of these models make only general predictions about the behavioral or perceptual consequences of these abnormalities.…”

“…Simulations with this NN model led to the novel and counterintuitive prediction that the subgroup of individuals with HFASD who are atypical learners in a standard task will perform better when trained with highly simplified input sets consisting of a single prototypical exemplar rather than multiple variable examples of the category (Dovgopoly & Mercado, 2013). This occurs because repeated experiences with an "ideal" stimulus, combined with degraded plasticity, leads to an internal representation that readily generalizes to distortions of the prototype, but not random patterns.…”

“…Recent simulations using a neural network (NN) model of visual processing (Henderson & McClelland, 2011) to investigate the possible effects of neural abnormalities on visual category learning suggest that atypical generalization may reflect disrupted cortical reorganization resulting in slower perceptual learning in some children with HFASD (Dovgopoly & Mercado, 2013). These simulations replicated atypical generalization patterns (Church et al, 2010) by taking into account two subgroups of participants with HFASD: typical learners (A Type 1) who show no differences in performance from typically developing children and atypical learners (A Type 2).…”