Deep dyslexia: A case study of connectionist neuropsychology

Plaut, David C.; Shallice, Tim

doi:10.1080/02643299308253469

Cited by 756 publications

(680 citation statements)

References 107 publications

Supporting

Mentioning

630

Contrasting

Unclassified

Order By: Relevance

“…Interestingly, 59% of these incorrect responses were the exact semantics of an alternative word, and these ''word'' errors were more semantically and/or visually similar to the correct word than would be expected by chance. This co-occurrence of visual and semantic similarity in error responses is analogous to aspects of the error pattern produced by deep dyslexic patients (also see Hinton & Shallice, 1991;Plaut & Shallice, 1993a).…”

Section: Relearning After Damage In Connectionist Networkmentioning

confidence: 62%

“…The implications of the nature of recovery in damaged networks for patient rehabilitation would be far better established in a simulation that corresponded more closely to the tasks carried out by the patients. To this end, the current work investigates the effects of recovery in networks (Hinton & Shallice, 1991;Plaut & Shallice, 1993a) that, when damaged, replicate the qualitative error pattern exhibited by patients with impaired reading via meaning.…”

Section: Hinton and Plaut (1987)mentioning

confidence: 99%

“…Much of this work has been focused on deficits in word reading-the acquired dyslexias (Mozer & Behrmann, 1990;Patterson, Seidenberg, & McClelland, 1990;Plaut & Shallice, 1993a)-but there has also been considerable work in other domains, including spelling (Shallice, Glasspool, & Houghton, 1995;Olson & Caramazza, 1994), speech production (Harley & MacAndrew, 1992;Martin, Dell, & Schwartz, 1994), face recognition (Burton, Young, Bruce, Johnston, & Ellis, 1991;Farah, O'Reilly, & Vecera, 1993), visual object naming (Gordon, 1982;Plaut & Shallice, 1993b), spatial attention (Cohen, Romero, Servan-Schreiber, & Farah, 1994;Humphreys, Freeman, & Müller, 1992), learning and memory McClelland, McNaughton, & O'Reilly, 1995), semantic memory (Farah & McClelland, 1991;Horn, Ruppin, Usher, & Hermann, 1993), and control of action and responding (Bapi & Levine, 1990;Cohen & Servan-Schreiber, 1992;Levine & Prueitt, 1989). Although still in its infancy, the relative success of this work suggests that connectionist modeling may provide an appropriate formalism within which to explore how disorders of brain function give rise to disorders of cognition.…”

Section: Introductionmentioning

confidence: 99%

“…Following this, previous preliminary work on the effects of relearning after damage in connectionist networks is described. Most of the simulations described in this paper are based on retraining a network that has been used previously to model impaired reading via meaning in a form of acquired dyslexia known as deep dyslexia (Hinton & Shallice, 1991;Plaut & Shallice, 1993a). The implemented network can be viewed as constituting part of a larger but unimplemented framework for lexical processing (Seidenberg & McClelland, 1989).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Relearning after Damage in Connectionist Networks: Toward a Theory of Rehabilitation

1996

Self Cite

View full text Add to dashboard Cite

Connectionist modeling offers a useful computational framework for exploring the nature of normal and impaired cognitive processes. The current work extends the relevance of connectionist modeling in neuropsychology to address issues in cognitive rehabilitation: the degree and speed of recovery through retraining, the extent to which improvement on treated items generalizes to untreated items, and how treated items are selected to maximize this generalization. A network previously used to model impairments in mapping orthography to semantics is retrained after damage. The degree of relearning and generalization varies considerably for different lesion locations, and has interesting implications for understanding the nature and variability of recovery in patients. In a second simulation, retraining on words whose semantics are atypical of their category yields more generalization than retraining on more typical words, suggesting a counterintuitive strategy for selecting items in patient therapy to maximize recovery. In a final simulation, changes in the pattern of errors produced by the network over the course of recovery is used to constrain explanations of the nature of recovery of analogous braindamaged patients. Taken together, the findings demonstrate that the nature of relearning in damaged connectionist networks can make important contributions to a theory of rehabilitation in patients.

show abstract

Section: Relearning After Damage In Connectionist Networkmentioning

confidence: 62%

Section: Hinton and Plaut (1987)mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Relearning after Damage in Connectionist Networks: Toward a Theory of Rehabilitation

1996

Self Cite

View full text Add to dashboard Cite

show abstract

“…Semantic memory impairments are then simulated by lesioning the model to establish the nature of the interaction between neuropathology and cognitive deficit (e.g. Hinton and Shallice, 1991;Plaut and Shallice, 1993). A compelling example of this sort of approach is the connectionist model of Farah and McClelland (1991): patterns of category-specific deficits led Warrington and McCarthy (1987) to suggest that an animate/inanimate distinction could be understood in terms of a differential dependence on functional and structural (perceptual) features for recognition.…”

Section: Category Specificity and Connectionismmentioning

confidence: 99%

Functional integration and inference in the brain

Friston

2002

Progress in Neurobiology

273

201

View full text Add to dashboard Cite

Self-supervised models of how the brain represents and categorises the causes of its sensory input can be divided into two classes: those that minimise the mutual information (i.e. redundancy) among evoked responses and those that minimise the prediction error. Although these models have similar goals, the way they are attained, and the functional architectures employed, can be fundamentally different. This review describes the two classes of models and their implications for the functional anatomy of sensory cortical hierarchies in the brain. We then consider how empirical evidence can be used to disambiguate between architectures that are sufficient for perceptual learning and synthesis.Most models of representational learning require prior assumptions about the distribution of sensory causes. Using the notion of empirical Bayes, we show that these assumptions are not necessary and that priors can be learned in a hierarchical context. Furthermore, we try to show that learning can be implemented in a biologically plausible way. The main point made in this review is that backward connections, mediating internal or generative models of how sensory inputs are caused, are essential if the process generating inputs cannot be inverted. Because these processes are dynamical in nature, sensory inputs correspond to a non-invertible nonlinear convolution of causes. This enforces an explicit parameterisation of generative models (i.e. backward connections) to enable approximate recognition and suggests that feedforward architectures, on their own, are not sufficient. Moreover, nonlinearities in generative models, that induce a dependence on backward connections, require these connections to be modulatory; so that estimated causes in higher cortical levels can interact to predict responses in lower levels. This is important in relation to functional asymmetries in forward and backward connections that have been demonstrated empirically.To ascertain whether backward influences are expressed functionally requires measurements of functional integration among brain systems. This review summarises approaches to integration in terms of effective connectivity and proceeds to address the question posed by the theoretical considerations above. In short, it will be shown that functional neuroimaging can be used to test for interactions between bottom-up and top-down inputs to an area. The conclusion of these studies points toward the prevalence of top-down influences and the plausibility of generative models of sensory brain function.

show abstract