Category-specific impairments? No. A critique of Sartori et al.

In this paper we examine the claim by Sartori, Miozzo, and Job (this issue) that their patient Michelangelo exhibits a category-specific deficit in naming living things. We conclude that their data are flawed in many respects and offer no reasonable basis for the conclusions they wish to draw. In addition their further claim that the data reported by Stewart, Parkin, and Hunkin (1992) are artefactual is not supported in any substantial way by examination of the findings.     

Category-specific naming impairments? Yes

Recently, Stewart, Parkin, and Hunkin (1992) have questioned previously reported cases of selective damage in processing items from categories of animate objects, arguing that there has been a lack of adequate control for visual familiarity, visual complexity, and name frequency of the stimuli employed. When re-testing Michelangelo (see Sartori & Job, 1988), one of the patients cited by Stewart et al. (1992), with a set of materials matched on all three factors, the asymmetry in naming animal and artefact items still remains. An analogous pattern is obtained when--in addition to such factors--the visual similarity within the sub-sets of animals and artefacts is taken into account. These results constitute empirical evidence for category-specific impairments and cannot be interpreted as being due to isolated or conjoint effects of visual familiarity, visual complexity, or name frequency.     

Naming impairments following recovery from herpes simplex encephalitis: Category-specific?

An apparently clear case of category-specific naming impairment selectively affecting animals was detected in a patient who had recovered from herpes simplex encephalitis. However, subsequent investigation demonstrated that these category-specific effects could be eliminated by controlling simultaneously for three factors in picture naming: word frequency, concept familiarity, and visual complexity. The results emphasize the importance of controlling for all factors pertinent to picture naming when attempting to demonstrate category specificity in picture naming. Further testing indicated that deficits were also apparent when naming to definition was required, and some impairment in the ability to answer questions about objects and living things was also noted. Theoretical implications of these data are considered.     

Sources of error in picture naming under time pressure

We used a deadline procedure to investigate how time pressure may influence the processes involved in picture naming. The deadline exaggerated errors found under naming without deadline. There were also category differences in performance between living and nonliving things and, in particular, for animals versus fruit and vegetables. The majority of errors were visuallyand semantically related to the target (e. celery-asparagus), and there was a greater proportion of these errors made to living things. Importantly, there were also more visual-semantic errors to animals than to fruit and vegetables. In addition, there were a smaller number of pure semantic errors (e.g., nut-bolt), which were made predominantly to nonliving things. The different kinds of error were correlated with different variables. Overall, visual-semantic errors were associated with visual complexity and visual similarity, whereas pure semantic errors were associated with imageability and age of acquisition. However, for animals, visual-semantic errors were associated with visual complexity, whereas for fruit and vegetables they were associated with visual similarity. We discuss these findings in terms of theories of category-specific semantic impairment and models of picture naming.     

A computational model of semantic memory impairment: modality specificity and emergent category specificity.

It is demonstrated how a modality-specific semantic memory system can account for category-specific impairments after brain damage. In Experiment 1, the hypothesis that visual and functional knowledge play different roles in the representation of living things and nonliving things is tested and confirmed. A parallel distributed processing model of semantic memory in which knowledge is subdivided by modality into visual and functional components is described. In Experiment 2, the model is lesioned, and it is confirmed that damage to visual semantics primarily impairs knowledge of living things, and damage to functional semantics primarily impairs knowledge of nonliving things. In Experiment 3, it is demonstrated that the model accounts naturally for a finding that had appeared problematic for a modality-specific architecture, namely, impaired retrieval of functional knowledge about living things. Finally, in Experiment 4, it is shown how the model can account for a recent observation of impaired knowledge of living things only when knowledge is probed verbally.     

Hierarchies, similarity, and interactivity in object recognition: "category-specific" neuropsychological deficits.

Category-specific impairments of object recognition and naming are among the most intriguing disorders in neuropsychology, affecting the retrieval of knowledge about either living or nonliving things. They can give us insight into the nature of our representations of objects: Have we evolved different neural systems for recognizing different categories of object? What kinds of knowledge are important for recognizing particular objects? How does visual similarity within a category influence object recognition and representation? What is the nature of our semantic knowledge about different objects? We review the evidence on category-specific impairments, arguing that deficits even for one class of object (e.g., living things) cannot be accounted for in terms of a single information processing disorder across all patients; problems arise at contrasting loci in different patients. The same apparent pattern of impairment can be produced by damage to different loci. According to a new processing framework for object recognition and naming, the hierarchical interactive theory (HIT), we have a hierarchy of highly interactive stored representations. HIT explains the variety of patients in terms of (1) lesions at different levels of processing and (2) different forms of stored knowledge used both for particular tasks and for particular categories of object.     

Sometimes a noun is just a noun: comments on Bird, Howard, and Franklin (2000)

Bird, Howard, and Franklin (2000) have proposed a semantic-conceptual explanation of grammatical category-specific deficits that attributes impairments in noun and verb processing to two distinct mechanisms. According to their account, apparent deficits in verb production are not category specific, but rather result from the lower imageability of verbs compared to concrete nouns. Noun deficits are said to result from differences in the distribution of semantic feature types such that damage to sensory features results in disproportionate impairments in naming nouns, especially animate nouns, compared to verbs. However, this hypothesis, which we call the "extended sensory/functional theory" (ESFT), fails on several counts. First, the assumption that representations of living things are more heavily freighted with sensory features than are those of nonliving objects does not have any reliable empirical basis. Second, the ESFT incorrectly predicts associations between deficits in processing sensory features and living things or functional features and nonliving things. Finally, there are numerous cases of patients with grammatical category-specific deficits that do not seem to be consistent with damage at the semantic level. All of this suggests that the ESFT is not a useful model for considering grammatical (or semantic) category-specific deficits.     

Category-specific naming and modality-specific imagery

Some attempts to explain category-specific disorders have stressed how different modality knowledge bases (i.e., visual knowledge vs motoric/functional knowledge) may underlie the distinction between living and nonliving things. This study examined 60 normal subjects for the relationship between picture naming in four subcategories (animals, fruit/vegetables, praxic and nonpraxic objects) and imagery vividness in seven modalities. Participants made more nonliving than living errors; and females made more nonliving errors than males. There was a significant correlation between naming of animals and fruits/vegetables and visual imagery vividness; however, this association was also significant for praxic and nonpraxic object naming. There was no evidence of associations between kinesthetic imagery and praxic object naming or gustatory/olfactory imagery and fruit/vegetable naming. These findings accord with the notion of a general association between visual imagery and picture naming, but provide no support for more specific links between modality-specific imagery vividness and naming in different categories.     

The influence of surface and edge-based visual similarity on object recognition

The role of 'visual similarity' has been emphasised in object recognition and in particular, for category-specific agnosias. [Laws and Gale, 2002] recently described a measure of pixel-level visual overlap for line drawings (Euclidean Overlap: EO[line]) that distinguished living and nonliving things and predicted normal naming errors and latencies ( [Laws et al., 2002]). Nevertheless, it is important to extend such analyses to stimuli other than line drawings. We therefore developed the same measure for greyscale versions of the same stimuli (EO[grey]), i.e., that contain shading and texture information. EO[grey], however, failed to differentiate living from nonliving things and failed to correlate with naming latencies to the greyscale images. By contrast, EO[line] did correlate with the naming latencies. This suggests that similarity of edge information is more influential than similarity of surface characteristics for naming and for categorically separating living and nonliving things (be they line drawings or greyscale images).     

Analyzing the factors underlying the structure and computation of the meaning of chipmunk,cherry, chisel,cheese, and cello (and many other such concrete nouns)

Seven trends regarding the categories that tend to be impaired/preserved in category-specific semantic deficits were identified. The authors hypothesized that these trends arise despite the multiple sources of variation in patient testing because numerous factors that structure semantic memory probabilistically converge to make some categories of knowledge more susceptible to damage than others. Analysis of semantic feature norms and corpus data for 541 concepts revealed that differences in the distribution of knowledge types across categories are sufficient to explain 6 of the trends and are necessary to explain loss of knowledge about nonliving things. Feature informativeness, concept confusability, visual complexity, familiarity, and name frequency contributed to this patterning and provide insight into why knowledge about living things is most often impaired.     

Category-specific naming errors in normal subjects: the influence of evolution and experience

The importance of "artifactual" variables (such as conceptual familiarity) have been highlighted in current accounts of category-specific disorders for living things (e.g., Funnell & Sheridan, 1992). The difficulties experienced by patients are essentially viewed as an exaggeration of normal processes and the implication is that normal subjects should also have greater difficulty naming living items (because they have lower conceptual familiarity than nonliving things). The current study examined normal subjects' ability to name pictures of artifact-matched sets of living and nonliving things in a naming-to-deadline paradigm. Contrary to the prediction, normal subjects made more nonliving naming errors. Furthermore, female subjects made more nonliving-thing errors than male subjects. These findings could not be reduced to differences in either category-based or gender-based familiarity ratings. Rather, it is proposed that an elaborated domain-specific evolutionary model parsimoniously explains both the greater incidence of living thing deficits in patients and the better performance of normal subjects with living things.     

Changes to the object recognition system in patients with dementia of the Alzheimer's type

Do DAT patients show category-specific deficits in object identification, and do they arise from semantic or visual damage? Participants decided whether line drawings of living and nonliving objects matched names at superordinate, basic, or subordinate levels. Patients were most impaired with superordinate decisions. Controls had most difficulty with subordinate decisions. No category-specific deficit was found with patients. Impaired superordinate decisions by the patients support semantic damage. If category-specific deficits arise from damaged semantics, they should have been found. Since they were not, and since patients performed subordinate decisions the best, a visual basis to category specificity is supported. Finally, a living advantage was found with normal observers which cannot be spurious due to differences in concept familiarity since living and nonliving objects were matched for this variable.     

The visual basis of category effects in object identification: evidence from the visual hemifield paradigm

The basis for the category specific living things advantage in object recognition (i.e., faster and more accurate identification of living compared to nonliving things) was investigated in two experiments. It was hypothesised that the global shape of living things on average provides more information about their basic level identity than the global shape of nonliving things. In two experiments subjects performed name-picture or picture-name verification tasks, in which blurred or clear images of living and nonliving things were presented in either the right or the left visual hemifield. With blurred images, recognition performance was worst for nonliving things presented to the right visual field/left hemisphere, indicating that the lack of visual detail in the stimulus combined with a left hemisphere bias toward processing high frequency visual elements proved detrimental for processing nonliving stimuli in this condition. In addition, an overall living things advantage was observed in both experiments. This advantage was considerably larger with blurred images than with clear. These results are compatible with the global shape hypothesis and converge with evidence using other paradigms.     

Shape Integration for Visual Object Recognition and Its Implication in Category-Specific Visual Agnosia

A series of experiments was conducted on a patient (ELM) with bilateral inferior temporal lobe damage and category-specific visual agnosia in order to specify the nature of his functional impairment. In Experiment 1, ELM performed a task of picture/word matching that used line drawings of fruits and vegetables as stimuli. The pattern of confusions exhibited by the patient suggested a failure in processing the full range of shape features necessary for the unique specification of the target relative to other structurally related items. This hypothesis of a shape integration impairment was tested and verified by subsequent visual recognition experiments (Experiments 2-4), which used synthetic stimuli with shapes precisely defined on the dimensions of elongation, curvature, and tapering. Furthermore, it was determined (Experiment 5) that the integration deficit is specific to the retrieval of shape knowledge from memory and does not affect the encoding of the properties of visual stimuli. It is argued that these findings have critical implications for cognitive theories of visual object recognition and for an interpretation of the visual function of the inferior temporal cortex. Finally, it was shown that the patient's deficit for structural knowledge integration is modulated by the semantic properties of the objects (Experiment 6), thereby demonstrating the applicability of the present findings to an explanation of category-specific visual agnosia.     

Semantic processing of living and nonliving concepts across the cerebral hemispheres

Studies of patients with category-specific semantic deficits suggest that the right and left cerebral hemispheres may be differently involved in the processing of living and nonliving domains concepts. In this study, we investigate whether there are hemisphere differences in the semantic processing of these domains in healthy volunteers. Based on the neuropsychological findings, we predicted a disadvantage for nonliving compared to living concepts in the right hemisphere. Our prediction was supported, in that semantic decisions to nonliving concepts were significantly slower and more error-prone when presented to the right hemisphere. In contrast there were no hemisphere differences for living concepts. These findings are consistent with either differential representation or processing of concepts across right and left hemispheres. However, we also found a disadvantage for nonliving things compared to living things in the left hemisphere, which is not consistent with a simple representation account. We discuss these findings in terms of qualitatively different semantic processing in right and left hemispheres within the framework of a distributed model of conceptual representation.     

Distinctiveness and correlation in conceptual structure: behavioral and computational studies

Patients with category-specific deficits have motivated a range of hypotheses about the structure of the conceptual system. One class of models claims that apparent category dissociations emerge from the internal structure of concepts rather than fractionation of the system into separate substores. This account claims that distinctive properties of concepts in the living domain are vulnerable because of their weak correlation with other features. Given the assumption that mutual activation among correlated properties produces faster activation in the normal system, the authors predicted a disadvantage for the distinctive features of living things for unimpaired adults. Results of a speeded feature verification study supported this prediction, as did a computational simulation in which networks mapped from orthography to semantics.     

An attenuation of the 'normal' category effect in patients with Alzheimer's disease: a review and bootstrap analysis

There is a consensus that Alzheimer's disease (AD) impairs semantic information, with one of the first markers being anomia i.e. an impaired ability to name items. Doubts remain, however, about whether this naming impairment differentially affects items from the living and nonliving knowledge domains. Most studies have reported an impairment for naming living things (e.g. animals or plants), a minority have found an impairment for nonliving things (e.g. tools or vehicles), and some have found no category-specific effect. A survey of the literature reveals that this lack of agreement may reflect a failure to control for intrinsic variables (such as familiarity) and the problems associated with ceiling effects in the control data. Investigating picture naming in 32 AD patients and 34 elderly controls, we used bootstrap techniques to deal with the abnormal distributions in both groups. Our analyses revealed the previously reported impairment for naming living things in AD patients and that this persisted even when intrinsic variables were covaried; however, covarying control performance eliminated the significant category effect. Indeed, the within-group comparison of living and nonliving naming revealed a larger effect size for controls than patients. We conclude that the category effect in Alzheimer's disease is no larger than is expected in the healthy brain and may even represent a small diminution of the normal profile.     

Names,concepts, features and the living/nonliving things dissociation

The present paper evaluates different hypotheses for explaining the living/nonliving things dissociation phenomenon in terms of feature type, considering the role of this dimension in the organization of conceptual semantic representations and in the activation of name representations. For this purpose we used Sloman and associates' (Memory and Cognition 27(3) (1999) 526; Cognitive Science 22(2) (1998) 189) name centrality and conceptual centrality tasks and asked subjects to judge functional and perceptual/visual features of living and nonliving items. Conceptual centrality results are more in accordance with a "single feature-domain connection hypothesis" where visual features are more important than functional features for the representation of living things and no feature type advantage is found for nonliving things. Name centrality results show that functional features are more important than sensory/visual features overall, a result that is not predicted by any of the hypotheses considered. The fact that the two judgments diverge emphasizes their importance for evaluating the role of feature type in the living/nonliving dissociation. Implications for explaining this phenomenon are also discussed.     

Category-specificity can emerge from bottom-up visual characteristics: evidence from a modular neural network

The role of bottom-up visual processes in category-specific object recognition has been largely unexplored. We examined the role of low-level visual characteristics in category specific recognition using a modular neural network comprising both unsupervised and supervised components. One hundred standardised pictures from ten different categories (five living and five nonliving, including body parts and musical instruments) were presented to a Kohonen self-organising map (SOM) which re-represents the visual stimuli by clustering them within a smaller number of dimensions. The SOM representations were then used to train an attractor network to learn the superordinate category of each item. The ease with which the model acquired the category mappings was investigated with respect to emerging category effects. We found that the superordinates could be separated by very low-level visual factors (as extracted by the SOM). The model also accounted for the well documented atypicality of body parts and musical instrument superordinates. The model has clear relevance to human object recognition since the model was quicker to learn more typical category exemplars and finally the model also accounted for more than 20% of the naming variance in a sample of 57 brain injured subjects. We conclude that purely bottom-up visual characteristics can explain some important features of category-specific phenomena.     

Category-specific effects on the identification of non-manipulable objects

Theories of category-specific effects on visual object identification predict easier identification of non-living than living objects. The Sensory-Functional theory credits greater representational weighting of the visual properties of living objects independent of greater weighting of the functional properties of non-living objects. It predicts a lost or reversed non-living advantage for non-manipulable objects. Normal participants matched pictures of non-manipulable objects with words describing three levels of identity while visual object similarity, and concept familiarity were controlled. Consistent with the Sensory-Functional theory, living objects were matched faster than non-living objects. Concept familiarity facilitated subordinate matches. Visual similarity hampered subordinate matches and facilitated basic matches.