Semantic access processing in a supra-modal deficit: A single case

Brain and Cognition 53 (2003) 202–206 www.elsevier.com/locate/b&c

Semantic access processing in a supra-modal deficit: A single case study Francesca Borgo,a Sara Mondini,a,b and Patrizia Bisiacchib b

a University of Trieste, Italy Department of General Psychology, University of Padavo, Padavo, Italy

Accepted 7 May 2003

Abstract The case is presented of a semantic dementia patient, who shows a deficit selective for (i) conceptual class (living things), (ii) attribute processing (visual features) and affecting (iii) input–output modalities at the same processing stage (matching stored representation to attributes). The first experimental part aims at exploring the specific stage/process at which semantic knowledge breaks down through multi-modality tasks, devised to tap different levels within the semantic elaboration flow. The second focuses on the differences between category vs. attribute knowledge across various modalities. The core nature of the patientÕs deficit is investigated through a close comparison of her damage to a specific processing stage across modalities in the light of her class and attribute-specific impairment. The complex pattern of findings is discussed according to current theoretical accounts of semantic memory organization. Finally, the relevance of the adoption of a broad perspective when dealing with semantic memory impairments is highlighted. Ó 2003 Elsevier Science (USA). All rights reserved.

1. Introduction A main issue in the field of semantic memory deficits relates to the observation of patients with selective impairments, such as category-specific deficits affecting different classes of concepts (e.g., living vs. non-living things), modality-specific deficits involving different modalities of input–output (e.g., verbal vs. visual), and attribute-specific deficits, where different types of processes (e.g., visual vs. functional) might be found impaired regardless of the categorical deficit. An influential view has been put forward originally by Warrington and Shallice (1984) and Farah and McClelland (1991): semantic memory is characterised by sub-systems, specialised for analysis of specific attributes (functional/sensory). On this ground, the semantic system is not organised categorically, while the elaboration of different classes of concepts, in this instance living vs. non-living things, relies mainly, though not totally, on a peculiar type of process. Therefore, category-specific deficits emerge from damage to processes based upon the analysis of different attribute types (e.g., sensory ) living things; functional ) non-living things).

As regards modalities of stimulus presentation, conceptual knowledge (organised into sub-systems for attribute analysis) is claimed to be stored separately from different input modalities (Shallice, 1993). According to this theory, modality-specific deficits are explained in terms of access vs. storage problems. The former theoretical account has been challenged by Caramazza and Shelton (1998) and Caramazza, Hillis, Rapp, and Ramani (1990). In Caramazza and SheltonÕs (1998) proposal, category-specific deficits result from damage to brain regions specialised for particular categories (e.g., animals vs. objects) within a unitary semantic system which is organised categorically. According to this theory, in the presence of a category-specific deficit no difference is expected between types of attributes (e.g., visual vs. functional). According to Caramazza et al.Õs (1990) position, modality-specific effects (e.g., visual vs. verbal) can be accounted for in the light of two assumptions. First, the privileged accessibility assumption states that there is a strong link between peculiar attributes (e.g., 3D structural descriptions) and amodal semantic representations: it follows that the type of input can influence task

0278-2626/$ - see front matter Ó 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0278-2626(03)00110-6

F. Borgo et al. / Brain and Cognition 53 (2003) 202–206

performance (e.g., visual better than verbal input). Secondly, the privileged relationship assumption claims the existence of asymmetric relationships between input–output modalities and the type of access to semantic representations required by a task: as a consequence, a difference should be observed between tasks requiring full (e.g., visual ) naming) or partial (e.g., visual ) gesturing) access to semantic representation. A major flaw of these theories of semantic memory might be identified in the attempt to cope with distinct types of semantic impairments, without providing a broad account of the many different, but not necessarily distinct, conceptual deficits. How might a ‘‘supra-modal’’ semantic deficit be explained in terms of current theoretical positions about the organisation of the semantic memory system? More specifically, in a patient who shows a semantic memory impairment, (i) does the impairment involve different input–output modalities? (ii) is it possible to identify a specific process of semantic attributes, which is shown to be affected across input– output modalities? and (iii) can the damaged process account for the supra-modal presence of a categoryspecific deficit? The purpose of this work is to examine whether there is a relation between category-, modality-, and attribute-specific semantic deficits. In the light of the present investigation, a thorough examination of current theoretical proposals will be carried out.

2. Methods and results 2.1. Case report Patient AB is a 64 years-old housewife, with 5 years of education. The MRI scan shows diffuse vascular lesions, and atrophy in the medial temporal lobes, particularly affecting the right hippocampus. A diagnosis of presumed progressive degeneration was confirmed 1 year after the first clinical observation. At the first examination she was alert, motivated and collaborative. In an informal assessment of her spontaneous speech, she was fluent, syntactically correct and adequate to the context, showing however some word-finding difficulties. The neuropsychological assessment highlighted a mild memory deficit, but severe difficulties on phonemic fluency and naming. ABÕs verbal comprehension, repetition, reading and writing skills were however flawless. The patient had a normal performance on attentive and visual–spatial perception tasks. 2.2. Experimental investigation 1 In order to evaluate the semantic memory impairment of patient AB, two dimensions were initially taken into account. First, the different type of processes involved at increasing stages of the semantic analysis were

203

considered: (1) pre-semantic analysis, which does not require access to stored representations; (2a) semantic analysis which involves access to super-ordinate, categorical, information; and (2b) semantic analysis which needs a match between stored semantic representation and (visual/functional) attributes. Secondly, different input–output modalities of stimulus presentation–response type were taken into account: visual–visual; visual–verbal; auditory–verbal; visual–motor; and auditory–motor. This procedure allowed the analysis of the patientÕs performance on a wide range of tasks, commonly used in basic assessment of semantic memory damage, but implying processes at different levels and in different modalities. 2.3. Results of experimental investigation 1 The pre-semantic stage of analysis was assessed in three different input–output modalities: visual–visual (Minimal Feature and Foreshortened View tasks of the BORB); visual–motor (Drawing on Copy and Ideomotor Apraxia test—pantomime gestures performed by the examiner); and auditory–motor (Ideomotor Apraxia task—pantomime gestures verbally asked by the examiner). The patient performed flawlessly on all these tasks (see Table 1a). The semantic stage of analysis requiring access to category information was investigated through two modalities: visual–visual (Item Match task of the BORB) and auditory–verbal (Category Knowledge subtest, by Laiacona, Barbarotto, Trivelli, & Capitani, 1993). In tasks requiring access to super-ordinate information, patient AB was within the normal range in both modalities of stimulus presentation (see Table 1a). The semantic stage of analysis which requires a match between stored representation and visual attributes was tested in a wide range of modalities: visual–visual (Object Decision test from VOSP); visual–verbal (Object Decision from BORB, Silhouettes test from VOSP and Picture Naming); auditory–verbal (Naming on Visual Definition and Semantic Visual Judgements—adapted from Thompson-Schill, Aguirre, DÕEsposito, & Farah, 1999); and auditory–motor (Spontaneous Drawing task). The matching between semantic representation and functional attributes was examined in two modalities: auditory–verbal (Naming on Functional definition and Semantic Functional Judgements—adapted from Thompson-Schill et al., 1999) and visual–motor (Ideational Apraxia test). On both types of semantic matching, regardless of whether visual or functional attribute processing was required, ABÕs performance was highly impaired in all modalities (see Table 1a).

204

F. Borgo et al. / Brain and Cognition 53 (2003) 202–206

Table 1 Experimental investigation 1 and Experimental investigation 2 (1a) Experimental investigation 1 Modality Task Pre-semantic processing Visual–visual Minimal feature (BORB) Foreshortened view (BORB) Visual–motor Ideomotor apraxia Copy of drawing (BORB and others) Auditory–motor Ideomotor apraxia

Result

Comment

23/25 (92%) 23/25 (92%) 12/12 (100%) 8/10 (80%)

Within the range Within the range Within the range Within the range

12/12 (100%)

Within the range

Semantic processing: Matching representation/category information Visual–visual Item match (BORB) 29/32 (91%) Auditory–verbal Category knowledge 111/120 (92%) Semantic processing: Matching representation/visual attributes Visual–visual Object decision (VOSP) Visual–verbal Object decision (BORB) Silhouettes (VOSP) Picture naming Auditory–verbal Naming on visual definition Visual semantic judgement Auditory–motor Spontaneous drawing (BORB and others)

12/20 (60%) 22/32 (69%) 3/30 (10%) 27/137 (20%) 13/30 (43%) 136/192 (71%) 1/6 (17%)

Semantic processing: Matching representation/functional attributes Auditory–verbal Naming on functional 21/30 (70%) definition Functional semantic judgement 148/192 (77%) Visual–motor Ideational apraxia 9/21 (43%)

Within the range Within the range Below Below Below Below Below Below Below

the the the the the the the

range range range range range range range

Below the range Below the range Below the range

(1b) Experimental investigation 2 1st assessment

2nd assessment

Modality

Task

Living

Non-living

Living

Non-living

Visual–verbal Auditory–verbal

Picture naming Naming on functional definitions Naming on visual definitions Semantic functional judgements Semantic visual judgements

14/83 (17%) 7/15 (47%)

43/86 (50%) 14/15 (93%)

4/40 (10%) 2/15 (13%)

15/40 (37%) 12/15 (80%)

3/15 (20%)

10/15 (67%)

2/15 (13%)

3/15 (20%)

62/96 (64%)

86/96 (89%)

34/96 (35%)

84/96 (87%)

60/96 (62%)

76/96 (79%)

22/96 (23%)

66/96 (69%)

The patientÕs good performance in all pre-semantic tasks reveals that the deficit is truly semantic in nature. However, ABÕs knowledge of super-ordinate type of information is spared, implying that this type of knowledge is extremely resistant to damage. AB shows instead a multi-modal deficit which affects the same semantic process in all the modalities examined and suggests that her difficulty is possibly due to degradation of stored representation, while an access problem seems to be ruled out. 2.4. Experimental investigation 2 In order to investigate in depth the nature of ABÕs semantic impairment, it is however crucial to verify

whether or not her supra-modal deficit in matching stored representation to specific types of attributes is related to a selective damage to her knowledge of different conceptual classes. To address this problem, it is relevant to verify whether the knowledge of different classes of concepts (living vs. non-living things) is differentially damaged with respect to attribute types (functional vs. visual), and still affects various input– output modalities. The relation between attribute type and classes of knowledge was then assessed in two separate sessions (1 year between sessions), tapping two different modalities: visual–verbal (Picture Naming task for living and nonliving things) and auditory–verbal (Naming on Visual/ Functional Definition and Semantic Visual/Functional

F. Borgo et al. / Brain and Cognition 53 (2003) 202–206

Judgements—adapted from Thompson-Schill et al., 1999), both for living and non-living things. 2.5. Results of experimental investigation 2 The patientÕs performance in the Picture Naming task (see Table 1b) revealed a significant selective deficit for living compared to non-living things both in the first session (v2ð1Þ ¼ 19:2, p < :001) and in the second (v2ð1Þ ¼ 4:5, p < :05). In the Naming on Visual/Functional Definition task (see Table 1b) AB showed again the relative sparing of non-living things with respect to living, both in the first assessment (v2ð1Þ ¼ 11:3, p < :001) and also a year later (v2ð1Þ ¼ 7:6, p < :01). More importantly, in both sessions she showed a significant advantage for functional compared to visual attribute knowledge (1st session: v2ð1Þ ¼ 4:3, p < :05 and 2nd session: v2ð1Þ ¼ 6:2, p < :05). In the Semantic Visual/Functional Judgements task (see Table 1b) the patientÕs performance evidenced a significant difference between living (damaged) and nonliving things (spared), both in the first (v2ð1Þ ¼ 9:7, p < :01) and the second assessment (v2ð1Þ ¼ 44:1, p < :001). The advantage of functional over visual attribute knowledge was only observed in the second assessment (v2ð1Þ ¼ 9:4, p < :005), being not significant in the first session (v2ð1Þ ¼ 1:9, ns). On closer inspection of ABÕs semantic deficit after a year, the patientÕs deficit turned out to be both categoryspecific (living things worse than non-living) and attribute-specific (visual worse than functional); this pattern was observed across the two input–output modalities tested. 3. Discussion Such a pattern of deficit in different input–output modalities is difficult to explain in terms of privileged access-relationship assumptions (Caramazza et al., 1990). As shown in Experimental Investigation 1, a peculiar type of semantic process, namely, the match between stored representations and attributes, was found to be defective across all the modalities tested. This result challenges Caramazza and colleaguesÕ proposal, insofar as some asymmetry should be found with respect to both different modalities of presentation and different types of attributes. Moreover, the first set of data (Experimental Investigation 1) could not be easily accounted for as a general deficit to a single, amodal, semantic system, on the lines of Caramazza et al. (1990), since no claim is made by the authors about the possibility to disentangle different types of process within the semantic system. ABÕs supra-modal impairment for the matching process could be explained in terms of ShalliceÕs (1993)

205

proposal of a preponderant storage damage. However, no support to ShalliceÕs (1993) theory of multiple semantic sub-systems is given by the results of the first experimental investigation, where no difference is observed with respect to different types of attributes, as both functional and visual attributes are damaged. The data presented in Experimental Investigation 2 actually depict a much more complex picture of the patientÕs semantic deficit. The differential impairment of category knowledge (living worse than non-living) and its relation to the differential damage of types of attributes (visual worse than functional) in two input–output modalities, allow a finer appreciation of the theoretical implications put forward by the contrasting views of Caramazza et al. (1990, Caramazza & Shelton, 1998) on the one hand, and Shallice (1993) and Warrington and Shallice (1984), on the other. The apparent contrast between the absence of difference in the processing of functional and visual attributes found in the first experimental investigation, and the presence of a selective deficit for the visual ones in the second, might be due to the fact that, in the first experimental part, living and non-living things were collapsed together: this procedure was initially chosen to guarantee conformity between tasks frequently used in literature to assess semantic deficits from a broad perspective. This could have contributed to hide a highly relevant effect, which was instead observed in the second experimental investigation, where the main aim was to ascertain the relation between the patientsÕ deficit in matching stored representation to semantic attributes and her category- and attribute-specific impairment. Further comment might be needed to account for the different results found in the second experimental investigation in the Semantic Visual/Functional Judgements test: although the patient showed a category-specific impairment for living things, no significant difference was found in the first assessment between visual and functional attributes, while the effect was present a year later. This result is not consistent with data from the Naming on Visual/Functional Definition task, where a disadvantage for visual attributes was observed both in the first and second assessment. However, the Semantic Judgement task, where the patient had to verify the appropriateness of a given attribute to a target (50% chance), might be a much less demanding task than the Naming on Definition, allowing a better performance. However, with the increasing gravity of the degeneration process, patient AB might have become less sensitive to easy vs. difficult tasks, therefore showing the effect again. The patientÕs supra-modal deficit, interacting with both category- and attribute-specific effects, is hardly in keeping with Caramazza et al.Õs (1990) and Caramazza and SheltonÕs (1998) theories, since, in their view, category-specific impairments cannot be found in interaction

206

F. Borgo et al. / Brain and Cognition 53 (2003) 202–206

with attribute-specific deficits. ABÕs pattern of impairments however may be accounted for in terms of a deficit to a visual–semantic sub-system (Shallice, 1993) which is needed to perform any task relying on the analysis of visual features, regardless of the input–output modality that characterises the task. Moreover, the visual–semantic sub-system is claimed to be massively involved in the processing of living things category, while is considered less relevant in the semantic analysis of the nonliving things category (Warrington & Shallice, 1984; Farah & McClelland, 1991). In the latter view, a category-specific deficit for living things should be observed along with a much more severe impairment of visual attributes, with respect to functional ones, independently of the modality of stimulus presentation. This work does not give an account for the reports of isolated deficits which have often been described in the literature on semantic memory disorders, such as ‘‘pure’’ modality-, category-, or attribute-specific impairments. Indeed, in the literature these three types of semantic memory deficits have been deeply investigated, but quite frequently these dimensions have not been investigated in the same patient. Therefore, the present study stresses the importance of investigating semantic memory deficits from a much broader perspective. Indeed, the presence of isolated deficits to either category- or attribute-specific semantic knowledge has been reported in patients with a degenerative disease (Lambon-Ralph, Howard, Nightingale, & Ellis, 1998), although these patientsÕ performance was not monitored as gravity of disease increased. Furthermore, despite the fact that their results are not matched by ABÕs performance a year later, Gonnerman, Andersen, Devlin, Kempler, and Seidenberg (1997) stressed the importance of monitoring semantic effects in relation to the severity of damage, highlighting a different pattern in the degradation of category knowledge in AlzheimerÕs type dementia patients. Although the importance of previous

research works is acknowledged, single case studies, carried out on patients suffering from deterioration process, might turn out to be extremely critical, as the involution of a complex and dynamic system can be observed over time. Therefore, the relation or interaction of crucial dimensions within the semantic memory system, such as input–output modality, category and attribute knowledge, might provide insightful suggestions about the organisation of the system itself.

References Caramazza, A., Hillis, A. E., Rapp, B. C., & Ramani, C. (1990). The multiple semantics hypothesis: Multiple confusion? Cognitive Neuropsychology, 7, 161–189. Caramazza, A., & Shelton, J. R. (1998). Domain-specific knowledge systems in the brain: The animate–inanimate distinction. Journal of Cognitive Neuroscience, 10, 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. Gonnerman, L. M., Andersen, E. S., Devlin, J. T., Kempler, D., & Seidenberg, M. S. (1997). Double dissociation of semantic categories in AlzheimerÕs disease. Brain and Language, 57, 254–279. Laiacona, M., Barbarotto, R., Trivelli, C., & Capitani, E. (1993). Dissociazioni semantiche intercategoriali: descrizione di una batteria standardizzata e dati normativi [Inter-categorical semantic dissociation: description of a standardised battery and normative data]. Archivio di Psicologia, Neurologia e Psichiatria, 54, 209–248. Lambon-Ralph, M. A., Howard, D., Nightingale, G., & Ellis, A. W. (1998). Are living and non-living category-specific deficits causally linked to impaired perceptual or associative knowledge? Evidence from a category-specific double dissociation. Neurocase, 4, 311– 338. Shallice, T. (1993). Multiple semantics: Whose confusions? Cognitive Neuropsychology, 10, 251–261. Thompson-Schill, S. L., Aguirre, G. K., DÕEsposito, M., & Farah, M. (1999). A neural basis for category and modality specificity of semantic knowledge. Neuropsychologia, 37, 671–676. Warrington, E. K., & Shallice, T. (1984). Category specific semantic impairments. Brain, 107, 829–854.