Selective preservation of memory for people in the context of semantic memory disorder: Patterns of association and dissociation

Abstract

A number of single cases in the literature demonstrate that person-specific semantic knowledge can be selectively impaired after acquired brain damage compared with that of object categories. However, there has been little unequivocal evidence for the reverse dissociation, selective preservation of person-specific semantic knowledge. Recently, three case studies have been published which provide support for the claim that such knowledge can be selectively preserved [Kay, J., & Hanley, J. R. (2002). Preservation of memory for people in semantic memory disorder: Further category-specific semantic dissociation. Cognitive Neuropsychology, 19, 113–134; Lyons, F., Hanley, J. R., & Kay, J. (2002). Anomia for common names and geographical names with preserved retrieval of names of people: A semantic memory disorder. Cortex, 38, 23–35; Thompson, S. A, Graham, K. S., Williams, G., Patterson, K., Kapur, N., & Hodges, J. R. (2004). Dissociating person-specific from general semantic knowledge: Roles of the left and right temporal lobes. Neuropsychologia, 42, 359–370]. In this paper, we supply further evidence from a series of 18 patients with acquired language disorder. Of this set, a number were observed to be impaired on tests of semantic association and word-picture matching using names of object categories (e.g. objects, animals and foods), but preserved on similar tests using names of famous people. Careful methodology was applied to match object and person-specific categories for item difficulty. The study also examined whether preservation of person-specific semantic knowledge was associated with preservation of knowledge of ‘biological categories’ such as fruit and vegetables and animals, or with preservation of ‘token’ knowledge of singular categories such as countries. The findings are discussed in the context of a variety of accounts that examine whether semantic memory has a categorical structure.

Introduction

One of the most prominent debates in the cognitive neuropsychological literature in recent years has concerned whether semantic memory has a categorical structure. Such work has focused on the apparent existence of categorical impairments in neurological patients in which particular categories of knowledge are selectively impaired. In most cases it has been found that living categories, such as animals and birds, are more prone to impairment than non-living categories, although there are reports of the reverse dissociation, albeit fewer, in which non-living categories are the more impaired. While it is claimed by some authors that selective impairment, or preservation, can be observed within these categories (such as cases in which animal knowledge is selectively preserved or lost; Caramazza & Shelton, 1998; Hillis & Caramazza, 1991), it also appears that category impairments can cross general ‘living’ and ‘non-living’ boundaries. Thus, patient JBR (Warrington & Shallice, 1984), who appeared to have poorer understanding of exemplars of living compared with non-living categories, also had impaired knowledge of foods and musical instruments, metals, fabrics and precious stones. Several authors have suggested that apparent category differences in fact stem from the nature of the materials used. While it is now accepted that factors such as visual complexity, discriminability, familiarity and frequency can produce spurious effects (e.g. Funnell & Sheridan, 1992; Gaffan & Heywood, 1993; Stewart, Parkin, & Hunkin, 1992), artefactual explanations do not appear to explain adequately all cases (e.g. Farah, Meyer, & McMullen, 1996). Attention has therefore focused on what category-specific disorders might tell us about the structure of semantic memory.

Some accounts have concentrated on the relationship between the development and organisation of semantic memory and differences between the properties or attributes of living and non-living things. Warrington and colleagues noted, for example, that living categories are defined more saliently by sensory or perceptual attributes than by function (Warrington & Shallice, 1984). They suggested that separate modality-based visual-semantic and function-semantic sub-systems underpin perceptual and functional differences between living and non-living things. This view was later elaborated to encompass more fine-grained differences concerning the way in which we interact with objects in the real world, which may require very different contributions from sensory and motor domains (Buxbaum & Saffran, 1998; Warrington & McCarthy, 1987). A related view that is based on the notion of sensory quality, has been put forward by Borgo and Shallice (2001). A different kind of account proposes that exemplars within categories share properties that are more inter-correlated and are therefore represented more closely in ‘semantic space’ than exemplars of other categories (e.g. Caramazza, Hillis, Rapp, & Romani, 1990; Moss & Tyler, 2000; Tyler & Moss, 1997). Damage to an area of semantic space may result in impairment to a particular category or categories because shared properties will be damaged together.

None of these accounts assume that semantic memory has a genuine categorical structure. However, this position is held by Caramazza and colleagues (e.g. Caramazza, 1998; Caramazza & Shelton, 1998; Santos & Caramazza, 2003; Shelton & Caramazza, 2000; Shelton, Fouch, & Caramazza, 1999), who claim that evolutionary adaptation has led to a categorical organisation in which there are specialised semantic domains for the identification of certain categories with obvious survival value, such as animals, foods, plants and tools. They suggest that the only true category-specific disorders involve disruption to categories of knowledge such as these, with other category differences having an alternative explanation. They also argue that this domain-specific hypothesis is consistent with the acquisition of semantic knowledge in children and those aspects of semantic knowledge that are shared with non-human animals (Santos & Caramazza, 2003).

According to some authors (e.g. Gentileschi, Sperber, & Spinnler, 2001; Kay & Hanley, 1999; Miceli et al., 2000), ‘conspecific’ or person-specific knowledge may be considered to be a separable semantic domain under this framework. The importance of representing knowledge of familiar people has evolutionary salience in terms of everyday interaction and communication and clear adaptive value in distinguishing ‘friend’ from ‘foe’. In line with this view, it has been claimed that the ability to distinguish people from other ‘things’, in terms of mechanisms such as body movement patterns, is innate (Meltzoff & Moore, 1993). It has been proposed that there are specialised visual recognition systems for face processing, with dedicated neural structure (Damasio, Damasio, & Van Hoesen, 1982; De Renzi, Perani, Carlesimo, Silveri, & Fazio, 1994; Kanwisher, 2000; Kanwisher, McDermott, & Chun, 1997; Nakamura et al., 2000, but see Gauthier, Skudlarski, Gore, & Anderson, 2000; Gauthier, Tarr, Anderson, Skudlarski, & Gore, 1999, who dispute the existence of a specialised mechanism for faces). It has also been proposed that there are specialised systems for the semantic processing of famous faces (Gorno-Tempini & Price, 2001). An early cognitive model of face processing (Bruce & Young, 1986) suggested that person-specific semantic knowledge is represented in some sense separately from the rest of the semantic system, and more recent accounts are consistent with this view (Barry, Johnston, & Scanlan, 1998; Valentine, Brennen, & Brédart, 1996).

It also appears that person-specific semantic knowledge can be impaired selectively after acquired brain damage. A number of reports of single cases can be found in the literature (see Table 1). Several of these cases involved progressive disease. However, cases with non-degenerative disorder, such as herpes simplex encephalitis and temporal lobectomy, have also been recorded. Several of the neuro-degenerative cases showed a progression from impaired visual recognition of familiar faces to loss of knowledge of familiar people, regardless of modality. The majority are reported to have predominant right temporal lobe damage, including the recent case of JP, who showed focal right temporal lobe atrophy (Thompson et al., 2004), although one patient (APA; Miceli et al., 2000) had a specific left temporal lobe injury caused by a head wound. Other examples of impaired person-specific knowledge in the context of less severe comprehension deficits of other categories have been reported (e.g. Barbarotto, Capitani, Spinnler, & Trivelli, 1995; Hodges & Graham, 1998), as have person-specific deficits associated with familiar face recognition and/or naming impairments (e.g. Edwards-Lee et al., 1997; Gentileschi, Sperber, & Spinnler, 1999; Kitchener & Hodges, 1999; Neuner & Schweinberger, 2000).

It has also been claimed that person-specific semantic knowledge can be selectively preserved after acquired brain damage, demonstrating a double dissociation between comprehension of familiar people and objects. For example, Warrington and McCarthy (1987) and Van Lancker and Klein (1990) describe the performance of globally dysphasic patients with severe comprehension impairments, but in whom recognition and matching of people's names was superior to recognition of objects. However, both Valentine et al. (1996) and Kay and Hanley (2002) have pointed out that the matching-to-sample tests used by these authors may be carried out in the absence of comprehension and do not provide strong evidence for selective preservation of semantic knowledge. More recently, two studies have made use of a variety of semantic comprehension tasks to test the abilities of two neurological patients who appear to have relatively preserved person-specific knowledge in the context of impaired comprehension of object categories (Kay & Hanley, 2002; Lyons, Hanley, & Kay, 2002). Both patients, ML and FH, presented with semantic comprehension impairments after left hemisphere CVA. Both were impaired in semantic association and comprehension tests using object and animal categories, but showed preserved performance on comparable tasks (of the same format and response) with famous people. A further case (patient MA) in whom person-specific knowledge is selectively preserved has recently been reported by Thompson et al. (2004).

While this dissociation is consistent with the domain-specific hypothesis of category organisation of Caramazza and colleagues, it is also compatible with other accounts. Several authors (e.g. Ellis, Young, & Critchley, 1989; Gentileschi et al., 2001) claim that impairment of person-specific knowledge can be observed in the context of a more general impairment of ‘singular object’ or ‘exemplar’ knowledge. Thus, such patients can also have difficulties in comprehension of geographical knowledge, of the location of countries, rivers, buildings and statues, for example, and of famous animals such as Winnie-the-Pooh. These findings suggest that the patients’ semantic difficulty may lie in knowledge that individuates concepts, rather than that which allows categorisation (this is sometimes referred to as a distinction between ‘token’ and ‘type’ knowledge; Jackendoff, 1983). In this context, it is also important to note that while most concrete concepts are identified at the basic level, faces must be differentiated and named at a subordinate (individuated) level. Few studies have attempted to compare faces and types of car, dog or cat, for example (but see Kay & Hanley, 2002; Thompson et al., 2004).

It also remains to be seen whether selective preservation, or impairment, of person-specific knowledge relate to a more general preservation, or impairment, of ‘living things’, or of biological categories. Recent studies by Kay and Hanley (1999) of a person with selectively impaired person-specific knowledge and by Kay and Hanley (2002) of a case in which such knowledge is selectively preserved did not find an association with loss or preservation of ‘biological category’ knowledge more generally.

The finding that person-specific knowledge can be selectively preserved relative to other categories is largely based on results from just three case studies. The present study investigates the extent to which this pattern can be observed in a random selection of neurological patients who present with acquired language disorder in a speech and language therapy setting. A sample of this kind was selected because it is likely that a relatively high proportion of such patients will experience difficulty in the comprehension of objects. We therefore examined the performance of a number of patients on a series of semantic comprehension tests using object, animal and famous people categories (Experiments 1 and 2). To learn more about the nature of person-specific knowledge representation, the patients’ geographical knowledge was also examined, along with their comprehension of other ‘living’ and ‘non-living’ categories (Experiment 3).