Lesioning an attractor network: investigations of acquired dyslexia.

  title={Lesioning an attractor network: investigations of acquired dyslexia.},
  author={Geoffrey E. Hinton and Tim Shallice},
  journal={Psychological review},
  volume={98 1},
A recurrent connectionist network was trained to output semantic feature vectors when presented with letter strings. When damaged, the network exhibited characteristics that resembled several of the phenomena found in deep dyslexia and semantic-access dyslexia. Damaged networks sometimes settled to the semantic vectors for semantically similar but visually dissimilar words. With severe damage, a forced-choice decision between categories was possible even when the choice of the particular… 

Figures and Tables from this paper

Deep dyslexia: A case study of connectionist neuropsychology

Abstract Deep dyslexia is an acquired reading disorder marked by the Occurrence of semantic errors (e.g. reading RIVER as “ocean”). In addition, patients exhibit a number of other symptoms, including

Perseverative and Semantic Influences on Visual Object Naming Errors in Optic Aphasia: A Connectionist Account

A connectionist simulation of visual object naming based on a small set of general connectionist principles strongly suggests that these principles provide important insights into the nature of semantic processing of visual information and its breakdown following brain damage.

Word Reading in Damaged Connectionist Networks: Computational and Neuropsychological Implications

A complementary motivation for studying the effects of damage in networks is to extend the understanding of the nature of computation in the networks themselves, which is not just with the development of a network that accomplishes a task, but with understanding how the network accomplishes the task.


A recurrent connectionist network that pronounces both concrete and abstract words via their semantics, defined so that abstract words have fewer semantic features reproduces the main effects of abstractness on deep dyslexics.

Dyslexic and Category-Specific Aphasic Impairments in a Self-Organizing Feature Map Model of the Lexicon

DISLEX is an artificial neural network model of the mental lexicon. It was built to test computationally whether the lexicon could consist of separate feature maps for the different lexical

A connectionist approach to naming disorders of Japanese in dyslexic patients

A Japanese triangle model computed phonology directly from orthography for both Kanji and Kana strings, with additional input to phonology from a component representing putative word semantics, successfully simulated certain effects seen in the performance of Japanese skilled readers.

Phonology, reading acquisition, and dyslexia: insights from connectionist models.

Compared with simple feedforward networks, representing phonological knowledge in an attractor network yielded improved learning and generalization and provides a computationally explicit account of many aspects of reading acquisition using connectionist principles.

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

The data support the view that visual crowding can be a potentially important factor in the emergence of some category-specific impairments and accord with the cascade model of object recognition, since increased competition between perceptual representations resulted in category- specific impairments even when the locus of damage was within the semantic component of the model.

Connectionist Perspectives on Category-Specific Deficits

Theories of semantic memory tend between two poles. At one extreme is a view often associated with the connectionist enterprise: that the semantic system is a unitary, homogeneous mass, without

Representation , Learning , Generalization and Damage in Neural Network Models of Reading Aloud

A new class of neural network models of reading aloud based on Sejnowski & Rosenberg’s NETtalk, which are not restricted to mono-syllabic words, require no complicated inputoutput representations and require no preprocessing to align the letters and phonemes in the training data.



Semantic access dyslexia.

The orthodox interpretation of the dyslexia without dysgraphia in terms of a disconnection syndrome is shown to be inappropriate, at least for the present case, where the deficit appears to be within the semantic domain.

Where Do Semantic Errors Come From?

Selective impairment of semantics in lexical processing

Abstract We report the performance of a neurologically impaired patient, KE, whose frequent errors in reading, writing, naming, and comprehension were nearly always semantically related to the target

What is right with “deep” dyslexic patients?

On the underlying causes of semantic paralexias in a patient with deep dyslexia

Category specific semantic impairments.

A quantitative investigation of the visual identification and auditory comprehension deficits of 4 patients who had made a partial recovery from herpes simplex encephalitis finds category specificity in the organization of meaning systems that are also modality specific semantic systems.

Word Recognition in a Phonemic Dyslexic Patient

It is argued that these findings support a dual encoding model of word recognition, the present case illustrating the impairment of the phonemic route, a direct graphemic-semantic route being relatively spared.

Evidence for preserved reading in 'pure alexia'.

Four patients who developed pure alexia after infarctions of the left cerebral hemisphere regained the ability to explicitly identify briefly presented words; these subjects were more accurate with nouns than functors and words of high as compared with low imageability.

Modality specific word comprehension deficits in deep dyslexia.

It is argued that deep dyslexia can result from a modality-specific deficit in attaining the meaning of words together with a disorder of the phonological reading route.