Category-Specific Semantic Deficits

The task of explaining the organization of conceptual knowledge is both promising and difficult. It is promising because if conceptual knowledge is organized, then in principle it should be possible to form theories which would account for the nature of conceptual organization in normal brains and brains which present for cognitive deficits of one kind or another. We should think that this organization would obtain modularly; that is, we should think that conceptual knowledge should be organized according to functionally individuated cognitive mechanisms. There is reason to be wary of this promise, however. The most relevant body of possible evidence for this task spells difficulty for cognitive theorists. Subjects who present for selective deficits of semantic processing of one kind or another are notoriously difficult to explain. Cases of category-specific semantic deficits, in which individuals behave as if unable to name and/or recognize items of a particular semantic category, presently resist satisfactory explanation and therefore prevent the success of any theory of semantic architecture. The attempt to explain subjects who present with apparent categorical deficits has resulted in the formation of three competing hypotheses: the Sensory/Functional theory, the Organized-Unitary-Content hypothesis, and the Domain-Specific hypothesis. While each affords a measure of support, there are too numerous counter-examples for each to rectify one apparent victor. The number of dissociation's for cases of category-specific semantic deficits may be due in part to methodological concerns and possible inconsistencies. These will be discussed in latter sections. Regardless, what is clear is that each theory posits mutually contrary predictions as to the course and cause of these cases.

Section I will treat the history and brief development of category-specific semantic deficits and the corresponding theories of semantic organization in the brain. Section II will focus on an analysis of the Sensory Functional Theory, the first projected and supported hypothesis of category-deficits. Section III will focus on domain-specific theory. Finally, while unitary-content-hypothesis represents one of the more recent efforts to explain category-specific semantic deficits, the complexity and implications inherent in both the Sensory-Functional account and the Domain-Specific hypothesis disallow the current discussion to treat this third hypothesis if this current discussion is to at all remain committed to a standard of through analysis. In all cases, we shall recourse to the experiments said to support each, and criticize methodology and interpretation where reasonable. In many instances, initial support has lead to retrospective denial of the same evidence-class. We shall treat these instances where appropriate. Suffice it to say that it is clear that all these theories obtain some positive instances, some negative instances, and many, many undetermined instances. With that in mind, it should be noted from the start that any discussion of the ramifications for cognition are speculative at best, this in addition to the lack of standard interpretation for many of the cases, has resulted in some confusion with respect to the data and the basic attributes of each theory. When and if possible, these discrepancies will be ironed out.

I: History and Development of Category-Specific Semantic Deficits

Warrington and colleagues (Warrington & McCarthy, 1983, Warrington and Shalice, 1984) provide the first empirical investigation into the possible causes of category-specific semantic deficits. Warrington and Shalice (1984) report four patients who presented for disproportionate impairment in both the production and comprehension of the names of living things. All four patients were recovering from herpes simplex encephalitis. The quantitative investigation took form in visual identification and auditory comprehension tasks. Results indicated selective impairment and selective preservation of certain categories of visual stimuli in all four cases. (Warrington and Shalice, 1984) Sometimes the selective impairment and selective preservation was almost binary; for instance, Warrington and Shalice report that one of their cases, J.B.R., was able to recognize a mere 2 of 48 living things, but accurately described and named 45 of 48 nonliving things. It seemed peculiar that he could define briefcase as "small case used by students to carry papers," but responded with "don't know," when asked to define the word parrot. On the other hand, they also report the case of S.B.Y., who correctly recognized 36 of 48 nonliving things, yet could not recognize any living things. The other two cases, K.B. and I.N.G. also presented poor performance for a picture naming task in animals with a relative sparing of nonliving things. All this suited the inference that certain kinds of localized brain damage could result in category-specific semantic deficits.

Subsequent findings confirmed the dissociation of deficits in the category of living versus relative sparing of the category of nonliving things (Basso, Capitani & Laiacona, 1988; Damasio, Grabowski, Tranel, Hichwa, & Damasio, 1996; De Renzi & Lucchelli, 1994; Farrah, Hammond, Mehta, & Radcliff, 1989). Oppositely, numerous reports also indicated the reverse dissociation: subjects presented for impaired performance in naming nonliving things with intact ability in naming living things (Hillis & Caramazza, 1991; Warrington & McCarthy, 1983). Given that the category 'living things' can be damaged independently of the category 'nonliving things', the plausible inference to suggest is that the semantic system is organized according to categorical schemas. However, though some suggested this hypothesis initially (for instance, Warrington, 1981), it was subsequently discarded; the interpretation of the data did not fit the notion that the organization of the semantic system reflected definite categorical boundaries.

Warrington, who found reason to doubt the initial (Warrington, 1981) hypothesis that category-specific semantic deficits may result from damage to the categorical organization of the semantic system, proposed the Sensory-Functional hypothesis. Experimental findings did not seem to fit with the interpretation that the seemingly categorical nature of these deficits reflected the categorical organization of the semantic system. The paradigm cases of J.B.R. and S.B.Y. presented for deficits in not only animal and plant categories, but also in foods. Warrington and colleagues concluded that there is no tenable way to subsume the food category within the animal and plant categories, given that only animal and plant categories are further part of the category of living things. Moreover, further tests on J.B.R. indicated he showed impairment in defining musical instruments, stones, and metals. All of these are inanimate things, thus there seemed to be even more evidence that was contrary to the suggestion that semantic knowledge was organized along definable categorical boundaries. This violation of the binary opposition between living and nonliving things served as the primary motivation for the positing of the Sensory-Functional hypothesis.

II: Sensory-Functional hypothesis

The basic tenets of this theory have been supported and reiterated by numerous cases (Warrington & McCarthy, 1987; Warrington & Shalice, 1984) and was later described again from a computational perspective (Farah & McClelland, 1991). In lieu of the suggestion that there is a relative independence between the kinds of category-specific semantic deficits presented (i.e. since the totality of case-studies thus far indicated instances of subjects experiencing naming difficulties for some living things, say animals and plants, and some nonliving things, say tools and musical instruments, with a relative sparing in some living things, say fruits, then the nature of the condition must not reflect a semantic system which is organized along strict categorical lines), Warrington and colleagues thought instead that the nature of a given semantic deficit could be attributed to damage in a specific modality or type of knowledge. Along these lines they proposed the visual semantic subsystem and the functional semantic subsystem. As these are essential to the theory, some remarks about them are necessary.

With respect to the paradigm cases of J.B.R. and S.B.Y., Warrington and Shallice (1984) purported that functional and visual (i.e. sensory) attributes are differentially important for both the proper naming of living and nonliving things, as well as for other categories, like food. Thus, if a subject presents with difficulty naming artifacts, then the Sensory-Functional hypothesis posits that there has been damage to the functional semantic subsystem. They assumed that the proper identification of an object of the category artifacts, say a hammer, was contingent on sufficient retrieval of information about what the object does, that is, the functional role the artifact plays. What seems essential to the identification of a hammer is that we use a hammer to, for instance, pound nails into wood. If someone lacked information about the utility of a hammer, he/she should not be able to name it, if there is any kind of association between the name of a tool and one's memory of the use of that tool. With respect to the latter subsystem, Warrington and Shalice (1984) claimed that damage to the visual semantic subsystem results in disproportionate impairment of living things (categories of animals and fruits/vegetables) and foods, because a necessary condition for the proper identification of animals, fruits/vegetables, and foods is that the subject can accurately represent knowledge about visual properties of those entities. Thus, anyone who presents for a categorical deficit in naming living things and foods (like patient S.B.Y.) was thought to have damage to their sensory semantic subsystem; that is, they could not retrieve information about the visual characteristics and properties that are necessary for the proper identification of those objects.

Another important aspect of sensory-functional theory is its prediction that, in most cases, subjects will present with multiple category-specific semantic deficits. This is because it is likely that a given type or modality of knowledge is somehow involved in the correct identification of items of several categories. This seems to explain much of the early case studies of category deficits. In the case of J.B.R., who presented with deficits in the categories of animals, plants, and foods, it may be the case that damage to the visual semantic subsystem, as purported by the theory, results in an inability to identify any items which rely on a fully functional visual semantic subsystem. Of course, one of the more difficult questions to work out is how much a given modality is involved in the correct processing of items of a given category, but at the least the theory seems consistent with the earlier cases on a basic level.

If the apparent categorical-nature of category-specific semantic deficits can be explained away by reference to damage in non-categorical and modality-specific subsystems of the semantic system, then the semantic system should not be organized according to definite categorical boundaries. However, there are problems with this position, most of which have to do with methodology and interpretation of the relevant data and case studies, while another portion has to do with theoretical implications that many have found reason to seriously doubt. We will conclude this section with mention of these flaws.

To start first with the methodological considerations: Warrington and Shalice (1984) failed to account for frequency and familiarity in the naming task in three out of four of their cases. This has lead some, including Funnell and Sheridan (1992), Gaffan and Heywood (1993), and Stewart, Parkin, and Hunkin (1992), to question the validity of Warrington and Shalice's experiments. They all essentially ask whether the cases reflect the fact that the tested items were generally less familiar and more visually complex than items in the nonliving condition. Evidently, given that visual complexity, frequency, and familiarity are all properties of the stimuli which researchers use to test their hypotheses, it is not unreasonable to question findings which left uncontrolled these factors.

The interpretation of the case study of subject H.O. (Stewart et al., 1992) reveals the possibly significant impact these stimulus factors can impart on the outcome of something as simple as a picture naming task. Stewart and colleagues report that H.O., who was recovering from herpes simplex encephalitis, presented with a nonrandom category effect. His performance in the Snodgrass and Vanderwart picture naming task (1980) indicated that he had a category-deficit for living things (65%) versus a comparatively spared ability for the category of nonliving things (86%). However, when the lists were corrected for frequency, familiarity, and visual complexity, the once-significant difference between the two categories failed to obtain (42% for living things, 36% for nonliving things). More importantly, when Stewart and colleagues apparently redefined the stimuli so that sensory (and not functional attributes) were more essential in the determination of the meaning of category members, there was no marked difference between categories (57% for animals, vegetables, fruit, large buildings, and 58% for tools and kitchen utensils). This seems to oppose the basic tenets of the Sensory-Functional hypothesis, because it predicts that sensory and functional attributes are differentially import for the proper identification of items in a particular category. These findings and more suggest some methodological issues that should not be so easily dropped. However, methodological inconsistencies are not the only source of criticism.

Sensory-Functional hypothesis opposes the idea that the semantic system is organized categorically. Rather than allow that the apparent categorical nature of these cases reflects damage to the boundaries of semantic categories, Sensory-Functional hypothesis assumes, contrarily, that these cases are a result of damage to either sensory or functional systems which relay information about either the sensory properties of an object (i.e. visual properties) or the functional properties of an object. Moreover, this hypothesis assumes that both are differentially important for the identification of items in a particular category. Another way to understand this dichotomy is the following: in any subject who presents with disproportionate category-specific semantic deficits, the nature of the disproportion can be explained by reference to damage to either (a) the modules which represent visual properties (visual semantic subsystem) or to (b) the modules which represent functional properties (functional semantic subsystem). The theoretical problems with this result from the purported dichotomy of function v. sensory information. It is not the case that in any given instance of category-specific deficits the error in the relevant cognitive system is either the functional semantic system or the sensory semantic system, for this presupposes both of the following: (a) that sensory and functional information are the most essential to any category-specific deficit and (b) that the correct identification of a given item of a given category is contingent on functional information but not sensory information, or sensory information but not functional information.

Reports of patients who present with category-specific deficits in animals but not foods (Hart & Gordon, 1992) are inconsistent with the purported dichotomy of functional information versus sensory information in the Sensory-Functional hypothesis. If a subject presents with semantic deficits in the naming of animals but not foods, then the alleged proposal of the Sensory-Functional hypothesis, namely that identification of items in both the animal category and the foods category are contingent on visual information, seems implausible, because the presentation of semantic deficits in naming animals with a spared ability in naming foods suggests that there is more to the problem than merely the correct usage of visual information; if it was the case that only visual information represented via the visual semantic subsystem played the necessary role, then both animal and food items should be problematic for someone with damage to the visual semantic subsystem. Hart and Gordon's case is inconsistent with this basic claim of the Sensory-Functional hypothesis.

Of final consideration, although it is more an explanatory gripe and not something which necessarily invalidates the theory, the Sensory-Functional theory fails to indicate why there are far more numerous cases of individuals with category-specific deficits in the category of living things than individuals with category-specific deficits in the category of nonliving things. These considerations and more reveal the basic motivations for the proposal of the Domain-Specific hypothesis (Caramzza & Shelton, 1998), which attempts to correct some of these manifest shortcomings of the Sensory-Functional hypothesis.

III: Domain-Specific hypothesis

This section will focus on the domain-specific hypothesis; first, a general construal of the hypothesis is in order, followed by a brief description of its chief proponent's conclusions regarding the case of E.W., which they (Caramazza & Shelton, 1998) purport falsifies the competing Sensory-Functional hypothesis. After that, some discussion of both the ramifications for cognition this theory postulates, as well as the possible weak points that come to attention under certain conditions.

The appeal to evolution in the Domain-Specific hypothesis (Caramazza and Shelton, 1998) instantiates the prevalent motivation of cognitive scientists to form theories which establish some link to biology and the theory of evolution. The central position of the Domain-Specific hypothesis is that evolutionary restrictions have resulted in neural circuits dedicated to the processing of distinct categories of objects. These kinds of processes may be perceptually or conceptually realized. Thus, with respect to category-specific semantic deficits, Domain-Specific hypothesis reveals that a subject's inability to properly name items in the category of animate things, such as animals or vegetables, can be explained with recourse to evolutionary fitness and survivability. In fact the qualification is more restrictive than this: the Domain-Specific hypothesis provides an independent criterion for determining what constitutes a conceptual category in the semantic system. This independent criterion, instantiated in the appeal to evolution, posits that the constitution of a conceptual category in the semantic system is restricted to only those categories "for which rapid and efficient identification could have had survival and reproductive advantages" (Caramazza & Mahon, 2003). Contrary to the Sensory-Functional account, this theory restricts what scientists can properly call a 'category-specific semantic deficit.' This results in a precision of method that the former theory does not prescribe. From the viewpoint of experimental psychology, this restriction is warranted and desirable.

Caramazza & Shelton (1998) invoke their theory around the implications of the case study of E.W. E.W. is a brain-damaged subject who has a disproportionate category-specific semantic deficit for the category of animals; she is impaired equally for both visual and functional attributes of this category, and so it is not something which the Sensory-Functional theory can coherently account for. The results of E.W.'s testing can be summed up as follows:

(a) E.W.'s category-specific semantic deficit is restricted to the category of animate objects; that is, her performance on other living things, such as fruits and vegetables, is within the normal range; the same is true of her performance on any food item.

(b) E.W.'s category-specific semantic deficit manifests in both visual and auditory comprehension tasks for the category of animals; furthermore, this deficit is not caused by way of some general perceptual processing error because this deficit cannot be accounted for by visual complexity or auditory complexity alone; her performance in complex visual processing tasks is within the normal range.

(c) Her category-specific deficit corresponds to language comprehension of the relevant objects; in attribute processing tasks her performance was significantly decreased with respect to statements about animals, yet she performed normally for statements about other living things and for items of the 'artifacts' category.

(d) Given that E.W. exhibits problems for both visual and auditory comprehension of objects of the animal category, as well as information pertaining to attributes of animals, it is clear that her deficit is not restricted to a modality of knowledge. "...she consistently performed equally poorly with visual and functional/associative statements of animate objects and equally well, and within normal limits, for aoll attributes of inanimate objects." (Caramazza & Shelton, 1998)

The fact that E.W.'s deficit is witnessed in language comprehension tasks for members of the animal categories implies that her problem does not reside merely in the damage to some modality of knowledge; furthermore, E.W.'s damaged system does not reflect visual and functional/associative knowledge in distinct subsystems. Furthermore, if E.W.'s case serves as a counterexample to the Sensory-Functional hypothesis, it is probably that the organization of semantic knowledge is not necessitated by some basic organizing principle that instantiates specific sensory or functional information specific to a given entity of a semantic category. In E.W.'s case, Caramazza and Shelton have shown the limitations and incoherence of the Sensory-Functional theory; this is evident in that these implications invite the general conclusion that the semantic system is amodal in its organization. The Sensory-Functional hypothesis obviously posits the opposite: namely, that the semantic system is organized according to specific modalities.

The appeal to evolutionary theory as a kind of backdrop to form the Domain-Specific hypothesis is more pragmatic than it is anything else. Caramazza and Shelton prescribe the appeal to evolutionary theory as a way to boost the explanatory justification of their theory, and though the appeal to adaptation and survivability does not at all run counter to their central assumption that category-specific deficits can reflect a categorically-organized semantic system, is it not the case that such an appeal is exactly apparent with respect to explaining E.W.'s case. It is interesting that, after using E.W. as a case to refute the Sensory-Functional hypothesis, Caramazza and Shelton propose a theory that attempts to link an explanation of their case with basic pillars of evolutionary theory that do not seem wholly relevant. Nevertheless, it is desirable that we appreciate how it is they came to form these sorts of judgments.

The link with evolution begins when they posit that "it is not implausible to assume that evolutionary pressures led to specific adaptations for recognizing and responding to animal and plant life." (Caramazza & Shelton, 20) True; it is not implausible, but that only means that to assert the possibility of such is not contradictory; the question as to upon what ground does such an assertion stand on remains to be asked. They then reason that, for instance, the ability to properly identify animals could be taken as process which would produce an immediately high fitness value due to the fact that successful completion of such a task would allow one to evade potential predators. For similar reasons, they propose that the ability to identify food sources, such as fruit and vegetables, manifestly contributes to survivability. Through repeated use of the neural mechanisms which allow one to accurately identify what an animal is or what constitutes an edible piece of fruit, Caramazza and Shelton further assume that "the neural adaptations would involve circuits that include the limbic system... [If such assumptions were correct] we would have the independent motivation we have been seeking in support of the assumption that conceptual knowledge is organized categorically in the brain." (Caramazza & Shelton, 20) Essentially, evolutionary adaptations would serve the purpose of providing the frameworks for which the relevant neural structures would be built from. What follows from these neural structures is the categorical presentation of the conceptual, perceptual and linguistic knowledge humans possess and can manipulate.

This recourse to evolutionary perspectives and the categorical nature of human knowledge (in all modalities) invites the inference that theorists can restrict what should count as adaptive semantic categories. Adaptive semantic categories are categories for which it is plausible to posit (at least the possibility) that the successful recognition of their members would have fitness value, that is, would increase survivability. Furthermore, the only true cases of category-specific deficits are those which somehow involve the categories of animals, plants/vegetables, and any other category which it is plausible to assert that the recognition of such would increase the overall fitness value. A further implication from all of this is the conclusion that those cases in which individuals presented with category-specific deficits which were not the categories of animals or plants/vegetables (or anything else that may count as an adaptive semantic category) were not 'truly' category-specific semantic deficits, at least according to the Domain-Specific hypothesis. In this instance, Caramazza and Shelton would propose that an explanation for these types of cases (cases which subjects presented with category-specific deficits that may not count as an adaptive semantic category) can be found in the Sensory-Functional theory, such as cases in which the individual has trouble naming items of the musical instruments category, or cases in which an individual has trouble naming a specific kind of object but can depict specific visual features of that object, et cetera.

The assumption that evolutionary adaptations provide the necessary conditions for the possibility of categorical organization of knowledge in the brain implies that the frequency of subjects who present with category-specific deficits for living things should be greater than the frequency of subjects who present with category-specific deficits of non-living things. Caramazza and Shelton explicate that the relevant specialized neural mechanisms are "more likely to be highly localized and consequently more prone to selective damage." (Caramazza & Shelton, 20) This claim is highly supported in the literature; most reports of category-specific deficits involve the category of living things.

That the literature indicates a high frequency of reported cases of individuals who present with category-specific deficits in living things compared to non-living things is something that Caramazza and Shelton take seriously. Their theory postulates that this distinction arose from evolutionary constraints, but there is more to it than that. According to Domain-Specific hypothesis, the acquisition of the animate-inanimate distinction involves domain-specific mechanisms which are fundamentally composed of a semantic system ordered according to categories. At the most basic level of this categorical hierarchy is the binary opposition of the animate category versus the inanimate category. To sum up, then: (a) the case of E.W. shows the inadequacy of the Sensory-Functional hypothesis because E.W.'s case could not be explained in terms of the inadequate processing of sensory or functional attributes seeing as how E.W. was equally bad at both; (b) in light of this refutation, Caramazza and Shelton propose the Domain-Specific hypothesis, which posits that a 'true' case of a category-specific semantic deficit reflects the categorical organization of the semantic system, and that the nature of this categorical organization can be explained in terms of the evolutionary constrains which necessitated the formation of neural mechanisms which allowed for the possibility of distinguishing between animate and inanimate things; finally (c) the presuppositions of the Domain-Specific hypothesis restrict what may be called 'true' instances of category-specific semantic deficits; the only true instances of this condition are those which one may reasonably assert as adaptive semantic categories-that is-they must somehow have produced an immediate impact on fitness.

While Caramazza and Shelton, it seems, have adequately shown that the Functional-Sensory hypothesis is at least theoretically uninviting in certain respects, the parallel distributed processing model of Farah and McClelland (1991) proves an obstacle because of its purported support for the Sensory-Functional hypothesis.

Farah and McClelland's PDP model showed that damage to the visual semantic subsystem of a semantic system organized according to specific modalities could result in typical responses of a subject who presents with category-specific deficits. This effort was essentially an attempt to reorient the Sensory-Functional theory to suit the perspective of computational psychology. Their assumptions in the formation of this model are twofold: first, visual and functional semantic properties are represented in semi-independent networks; second, the ratio of visual to functional properties for members of the category of living things is significantly larger than for members of the category of nonliving things. Their model exhibits semantic knowledge as subdivided into visual and functional components. Farah and McClelland then show that, after visual components of several networks are damaged, disproportionate category-specific deficits for living things may present. They then conclude that this pattern of performance serves as a basic pillar for the notion that category-specific deficits can be obtained without recourse to the notion of categorical organization of the semantic system.

On the surface this may seem to spell trouble for Caramazza and Shelton's Domain-Specific theory. However, Caramazza and Shelton explicitly treat this potential obstacle,

"In some respects the results of the implemented model merely reflect their assumption that the meaning of living things is more dependent on visual properties than is the case for nonliving things...damage to the visual semantic network could hardly result in anything but more severe difficulty for living than nonliving things. Put differently, the highly disproportionate ratios of visual to functional properties for the categories of living and nonliving things can be seen as roughly approximating a categorical distinction between the two semantic domains." (Caramazza & Shelton, 18)

Their contention that, in some sense, the implemented model reflects the basic assumption that living things are named according to visual properties first and foremost is reasonable. However, the subsequent reduction of Farah and McClelland's PDP model as merely showing an approximated version of a categorical distinction between the two semantic domains of living versus nonliving things is doubtful and probably unwarranted. Farah and McClelland's motive is show that the onset of category-specific semantic deficits is supervenient on the visual and semantic subsystem. Essentially, their model instantiates the notion that the onset of disproportionate category-specific semantic deficits are supervenient on, caused by, and implemented as a result of damage to the specific way in which a particular type of knowledge is processed. Thus, Caramazza and Shelton's claim that Farah and McClelland's model can be reinterpreted and reduced to the mere approximation of categorical distinctions between two semantic domains completely alters the essential motive of this PDP model, and partakes in an element of equivocation, because the meaning of the supervenience exhibited by the PDP cannot be transmitted to Caramazza's and Shelton's thesis of domain-specificity, seeing as how the notion of 'domain-specificity' presupposes a fundamental role in the organization of semantic knowledge. Seeing as how this level of fundamental import is not present in Farah and McClelland's PDP model, which purportedly supports the Functional-Sensory hypothesis, a transmission of this kind should not be easily forgotten on the part of Caramazza and Shelton.

Of final consideration is the assumption of Caramazza and Shelton that it was necessary for the sake of explanatory power that, in the explanation of categorical-deficits in terms of the categorical organization of the semantic system in the brain, one must selective independent grounds for such an explanation, and that these grounds could take the form of links with evolutionary theory. It is not immediately clear that this move is at all necessary: clearly cognitive theorists should not endeavor to propose theories which are inherently contradictory with respect to biological theories, like evolution, but that is not to suggest that it is necessary that cognitive scientists recourse to the theories of biology for the acclaimed independent justification of a given explanation they purport. It is true that in many respects the sciences are fragmented, or at least there is the sense that, under certain conditions, theories of quantum mechanics just fail to seem relevant to the Krebb's cycle of biology. Is that really a problem though? Should quantum theorists attempt to reconcile their theories with recourse to the Krebb's cycle? It would be nice if they did so, perhaps for biologists, but it is not at all necessary; what is necessary is that the sciences do not contradict one another, but again, that is not to say that they must explicitly supervene on one another, they need only implicitly do so.

With all of this in mind, it is clear that the Sensory-Functional theory can explain things in a relatively simpler manner than the Domain-Specific theory. Moreover, it is inherently the case that Sensory-Functional theory is a more readily testable hypothesis compared to the Domain-Specific theory; this is undoubtedly the case because the latter seeks explanatory justification with respect to the theory of evolution, and thus it becomes clear that the notion of an 'adaptive semantic category' is not something which can be tested in the traditional sense. As Caramazza and Shelton make explicit, the entire reasoning behind the notion of an 'adaptive semantic category' is one which only requires that the assertion 'category x is adaptive' be reasonable, not empirically verifiable!

While the Sensory-Functional theory has manifestly more instances of counter-examples, it is not clear that theorists should hastily accept its competitor, Domain-Specific theory, because this theory is only theoretically pleasing in its recourse to biological principles, it is not something which one could easily verify in the strictest sense. These theories are remarkable in that they attempt to explain such a confusing condition, but there is no clear indication that either one of them has the whole story about category-specific semantic deficits right.

References

Barsalou, L., & Simmons, K. (2003) The Similarity-In-Topography Principle: Reconciling Theories of Conceptual Deficits, Cognitive Neuropsychology, 20: 3, 4, 5, 6, 451-486

Caramazza, A., & Bradford M. (2003) The Organization of Conceptual Knowledge: the Evidence from Category-Specific Semantic Deficits, Trends in Cognitive Science, 7:8, 354-361

Caramazza A., & Shelton, J. (1998) Domain-Specific Knowledge Systems in the Brain: The Animate-Inanimate Distinction, Journal of Cognitive Neuroscience, 10:1, 1-34

Farah, M.J., & McClelland, J.L. (1991) A Computational Model of Semantic Memory Impairment: Modality Specificity and Emergent Category Specificity. Journal of Experimental Psychology: General, 120, 339-357

Farah, M.J., & Wallace, M.A. (19920 Semantically Bounded Anomia: Implication for the Neural Implementation of Naming, Neuropsychologia 30, 609-621

Gainotti,G. & Silveri, M.C. (1996) Cognitive and Anatomical Locus of Lesion in a Patient with a Category-Specific Semantic Impairment for Living Beings. Cognitive Neuropsychology, 13, 357-389

Snodgrass, J. & Vanderwart, M. (1980) A standardized set of 260 pictures: Norms for name agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6, 174-215

Warrington, E.K., & McCarthy, R. (1983) Category Specific Access Dysphasia, Brain 106, 859-878

Warrington, E.K., & Shallice, T. (1984) Category-Specific Semantic Impairment, Brain 107, 829-854

'Semantic deficit' implies deficits in higher-order conceptual representation, as distinct from "agnosia", which usually implies damage to a specific type or modality of sensory-functional information. This is a similar usage to that of Barsalou and Simmons (Barsalou & Simmons, 2003). In some sense, individuals who present with semantic deficits have some of the necessary information about the object they cannot recognize, however the higher-order conceptual representation is never realized, perhaps due to an inability of the system to retrieve that information.