Warrington and Shallice's (1984) report on patient J.B.R. [4] was the first and most influential study about the living/nonliving semantic dissociation. In their Critical Review, Semenza and Bisiacchi (1996) concluded:

Has what Semenza and Bisiacchi (1996) admitted was a grossly underspecified account progressed, after a decade, into a sounder and more detailed theory? Has any progress occurred in the definition of relevant semantic dimensions? Semenza and Bisiacchi provided, as examples of interesting dimensions, not only functional and structural attributes, but also relations of class and property, descriptive value, and concreteness/abstractness. They suggested that each of these dimensions was to be considered as orthogonal to the others. Nowadays the multiple semantic system theory, then the major aim of criticism, has lost ground, but it was criticized from the beginning. What is now of more compelling interest is what can still be learned from studying the living/nonliving dissociation and more importantly, from studying category specific disorders of semantic memory.

A huge amount of work has been done on what now can be seen as one of the hottest issues in cognitive neuropsychology of the past decade. Although one may be tempted to conclude that the popularity of the living/nonliving dissociation continued to increase, in exaggerated fashion and in inverse proportion to the importance of the findings, the basic question, “what has to be true of a category to enjoy independent representation and processing in the nervous system?” has not been answered despite undeniable progress. The following is a brief account of true progress made in recent years, and the Mondini et al. (2006, this issue) study is an interesting contribution in several respects.

First, the localization of brain areas that support the processing of the representation of living/nonliving entities is now better understood (Martin & Chao, 2001). Although important, localization only marginally answers basic questions on semantic processing and representations. Indeed, with few exceptions, such as the localization of a process to the right or a left hemisphere, to know precisely where in the brain a given task is performed adds in itself very little information about how the task is carried out.

Second, there is now broad agreement that the living/nonliving effects are not due to confounding factors including unbalanced word frequency, familiarity, or age of acquisition (see Shelton & Caramazza's 2000 for a through account of this point). As a consequence, now one can hardly doubt Shelton and Caramazza's (2000) conclusion that “… at least some patients do show true category-specific effects, although…this is especially difficult to establish in some of the early published cases in which processing difficulty was not controlled (page 427).”

One original contribution is the demonstration that there are gender differences in the representation of semantic knowledge, and these differences may interact with the living/nonliving effects. Laiacona et al. (in press) documented that men performed worse than women for plant, but not animal, categories. Indeed no female patient has ever been reported to have deficits in the plant life category. Laiacona et al. (in press) demonstrate, by controlling for familiarity, that this is not only an effect of a greater acquired familiarity in females with fruit and vegetables, but it is, at least in part, the result of evolutionary pressure, determining a persisting advantage for the plant category in modern females. Even more interesting, and not yet thoroughly explored, is the idea, put forward by Caramazza and co-workers (e.g. Caramazza, 1998; Caramazza & Shelton, 1998), that knowledge is organized into broad domains reflecting evolutionarily salient distinctions in semantic knowledge. Thus categories subsumed under the broadest category of living things, for instance “animals”, “plants”, or “members of the same species (i.e. humans for the human species)” that are important for physical and social needs, might enjoy special processing, and depend on particular neural circuits (Shelton & Caramazza, 2000). It is also possible that, certain nonbiological categories like “tools”, equally important from an evolutionary perspective, might also be processed in particular ways and be dependent (Martin & Chao, 2001) on specific neural regions related to their evolutionary significance (e.g., function). Natural selection would have therefore produced specialized, and hence dissociable, neural circuits dedicated to efficient processing of different categories of objects for which rapid identification conveyed survival advantages. However interesting, this theory says little about the nature of the representation of information within categories. Its promoters (Shelton & Caramazza, 2000) thus added an important corollary: Items within categories share many properties and core properties tend to be intercorrelated from a processing and neural substrate standpoint. This aspect of the theory had been independently developed in the effort to overcome the difficulties that soon appeared for the original Warrington and Shallice (1984) sensory-functional dichotomous theory.

The most serious problem, however, with the evolutional theory is that it is hard to prove empirically. One recent finding, however, may help. Borgo and Shallice (2003) showed in a herpes encephalitis patient with a relative ‘living things’ deficit how concepts representing “mass kinds” (i.e. substances, like water, sand or salt, with no clear perceptual boundaries, that in the lexicon are labelled with “mass”, as opposed to “count” nouns that label compact, enduring objects), associated with living things rather than with nonliving things (e.g., artefacts, such as tools). Although the authors were willing to interpret this finding according to the classical sensory-functional dichotomy (i.e., mass kinds would be best distinguished by sensory features), it could be argued that mass and living things represent the whole of the primitive world when manmade artefacts were not yet present. The introduction of artefacts made a huge change in the life of our primitive ancestors, and a brain that better distinguishes concepts related to artificial objects might be expected to have survival implications in the long run.

A useful distinction has been made between dimensions of conceptual knowledge like frequency, familiarity, typicality and age of acquisition on the one hand and semantic features on the other hand (Sartori & Lombardi, 2004). Semantic features have been shown to contribute to the “core” meaning of a concept and include distinctiveness, dominance and semantic relevance. Distinctiveness is high when a semantic feature is used in defining few concepts. Dominance is high when a semantic feature is frequently mentioned when defining a particular concept; it differs from frequency in the lexicon because it refers to the frequency with which a given feature is mentioned related to a given concept. Semantic relevance is a combination of the first two, and is high when a semantic feature is both frequently mentioned in defining a concept and not mentioned in defining many other concepts. For example, items belonging to living categories have lower average relevance with respect to nonliving categories: Sartori and Lombardi (2004) showed that the identification of living things is more difficult than nonliving only when relevance is not controlled. This dimension yielded the best predictive power in naming-to-description, one of the most frequently used tasks showing category specific effects.

The improvement in recent work on category specific deficits is not limited to a better definition of relevant semantic dimensions. In the present forum it must be emphasized that what positively distinguishes Mondini's et al (2006, this issue) work among contemporary contributions is the attention to the nature of the tasks and to the time course of evolving living/non-living dissociation disorders. The importance of considering performance for several different tasks was first emphasized by Humphreys and Forde (2001). They argued that deficits even for one class of objects cannot be accounted for in terms of a single information processing disorder across all patients. Problems could arise at contrasting loci in different patients, and the same apparent pattern of impairment can be produced by damage to different loci. The observed variety in patterns of deficit could be explained in terms of different levels of processing and different forms of stored knowledge used both for particular tasks and for particular categories of objects. Mondini's et al. (2006, this issue) is one of the few studies where task demands are carefully considered.

The longitudinal approach is a long honoured strategy to reveal dissociation of functions: “a zebra in the high grass can be seen only if moving.” Unfortunately it is seldom used. However, as Mondini et al. (2006, this issue) report, the living/nonliving dissociation has been profitably studied using this approach, with very interesting results. The added value of this study is that the longitudinal approach allows a better understanding of how category specific deficits can be interpreted in terms of a multifaceted relation between category and attribute knowledge. As a result, “attribute” theories and “categorical” theories can be viewed as complementary rather than contradictory with each other in the debate over the living/non living dissociation.

In the case of the living/non living dissociation, the aim of theoretically driven studies has been to uncover truths about the representation and processing of categories in the brain. Unfortunately, the central question of what has to be true of a category in order for it to be independently processed in the brain has not been unanswered. Clearly category specificity in neuropsychology has been observed in more than the living/nonliving dimension. One example is that of body parts, a category that was shown to behave differently from other categories, in both verbal and nonverbal tasks (e.g. in Semenza & Goodglass, 1985). Another interesting category is that of numbers (for a review see Cipolotti & van Harskamp, 2001). Actions and objects, named respectively, by verbs and nouns, have also been extensively investigated (see Druks, 2002, for a review), and the Mondini et al. (2006, this issue) study shows how the contrast between actions and objects may indeed interact with the living/nonliving dimension. Perhaps the most interesting contrasts are those in which semantic categories are represented in multiple ways as exemplified by verbs and nouns, which are represented differently at the lexical level according to special lexical-syntactic rules that are not shared with other categories. Indeed contrasting categories like verbs and nouns, but also mass and count nouns or proper and common names, differ from each other in category-specific morphosyntactic properties, that need to be taken into account in the retrieval of their fully specified phonological form (e.g.: verbs are subject to different inflections with respect to nouns; mass nouns, unlike count nouns, cannot take the plural form). The need for category specific syntax may indeed have contributed to their different representation in the brain. Certainly morphosyntactic differences distinguish these contrasting categories in a more clear-cut way with respect to their semantic differences (see Semenza, 2005, for a brief review). Common names vs. proper names is another extremely interesting contrast, as evidenced by the dissociation of these two categories at several processing levels, allowing a better understanding of the nature of the link between names and their reference (Semenza & Zettin, 1989; Semenza, in press), a long standing issue extending to linguistics and philosophy. All of these categorical contrasts are arguably more important than the living/nonliving contrast for understanding the categorical organization of our knowledge and its processing and representation in the brain, especially when this knowledge is labelled by a single word. Yet little of what has been learned by investigating these domains has been taken into consideration when studying the living/nonliving dissociation, which remains the most popular categorical disorder among neuropsychologists. Whether this is a desirable state of affairs is a matter of debate.