Semanticrelevance.psy.unipd.it

Category Deficits and Paradoxical Dissociations in Alzheimer’s Disease and Herpes Simplex Encephalitis & Most studies examining category specificity are single-case patient groups were compared with age- and education- studies of patients with living or nonliving deficits. Never- matched healthy controls. The profile in each patient was theless, no explicit or agreed criteria exist for establishing examined for consistency across tasks and across different category-specific deficits in single cases regarding the type of analyses; however, both proved to be inconsistent. One analyses, whether to compare with healthy controls, the striking finding was the presence of paradoxical dissocia- number of tasks, or the type of tasks. We examined two tions (i.e., patients who were impaired for living things on groups of patients with neurological pathology frequently one task and nonliving things on another task). The findings accompanied by impaired semantic memory (19 patients with have significant implications for how we determine cate- Alzheimer’s disease and 15 with Herpes Simplex Encephalitis).
gory effects and, more generally, for the methods used to Category knowledge was examined using three tasks (picture document double dissociations across individual cases in this naming, naming-to-description, and feature verification). Both nonliving items that are named (verified, identified, and Although various tasks and statistical techniques have so on) rather than comparing the patient data with been used to try to document category effects, a critical control data. This may be misleading about the presence and surprisingly unaddressed question concerns what and even direction of category effects. Laws (in press) would be considered acceptable as evidence of a noted several consequences of failing to evaluate patient ‘‘category effect’’ (Laws, 2004, in press). Most category naming against the performance of matched normal effects reflect an initial demonstration of poor picture healthy controls. The most common analysis in over naming for one category and, in some cases, poor two-thirds of all category-specific studies is the within- picture naming is the only evidence presented (e.g., in patient comparison of the absolute scores (using x2: studies of Alzheimer’s patients, see Laws, in press). One Laws, in press); and this reveals quite different dissoci- thing is certain, that is, that picture naming is implicitly ations from those disclosed when a control comparison viewed as the main task for confirming the presence of a is included. Indeed, Laws, Gale, et al. (2005) have category effect (and then that the additional tasks are demonstrated how patients may show false negatives, used to elaborate on that). This assumption reflects the false positives, and even paradoxical dissociations (i.e., fact that most models predict that naming is impaired if impaired for living on one task and for nonliving on semantic memory is impaired (and vice-versa). Never- another task). Hence, the absolute living–nonliving theless, evidence has existed for some time to show that naming difference alone (be it exceptionally large or this is not necessarily the case. For example, Laws, nonexistent) provides an unreliable indicator for both Evans, Hodges, and McCarthy (1995) described patient the presence and the direction of category effects.
SE who had impaired (associative, nonvisual) semantics This raises an important related methodological issue for animals but could name pictures of animals.
concerning how we define a category effects through Recent studies have raised issues about the statistical a test profile. In fact, no a priori test profile exists methods used to assess category effects with naming for documenting a category disorder (e.g., a category data (Laws, in press; Laws, Gale, Leeson, & Crawford, difference for picture naming, attribute verification, 2005). In particular, the common practice of making naming-to-description, fluency, drawing or any com- comparisons between the absolute number of living and bination of these). It also raises a critical issue con-cerning whether such disorders are empirically and/ortheoretically driven. For example, we could make an 1Nottingham Trent University, UK, 2University of Padua, Italy arbitrary empirical decision such that for a category *Now at the University of Hertfordshire, UK effect to be convincing, minimally requires a patient D 2005 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 17:9, pp. 1453–1459 to show the same significant category effect on three matched controls. Two of the Alzheimer’s patients were tasks: A (picture naming), B (drawing), and C (attribute impaired on all three category tasks (see Table 1). Nine verification). Nevertheless, what if another patient is were impaired on picture naming (7 living and 2 non- impaired on A and B, but not C? This patient may still living), 10 on feature verification (all living), and 3 on have a category disorder and tell us something poten- tially more important about the true nature of category Critically, the analyses revealed category inconsisten- effects (e.g., that they might be related to visual knowl- cies across tasks. One patient showed a paradoxical dissociation: patient LZ had a differential deficit for living Finally, although the vast majority of category-specific things on the feature verification task, but for nonliving studies use a case study approach (for a review, see Laws, in press), a minority have used between-groupcomparisons. One reason that group studies have been less popular in this area is because group analyses maycancel out individual category effects. For example, if The comparison with results from x2 points to a number some patients showed a living deficit and some a of both false positives (7) and false negatives (5). In one nonliving deficit, then the overall pattern might simply patient (SL), x2 produced significant dissociations on all reveal poor performance on both (see Gonnerman, three tasks, however, when referenced to control data, Anderson, Devlin, Kempler, & Seidenberg, 1997). Thisis especially likely with severely neurologically impairedcases. The following experiments compare category Table 1. Category Performance in Alzheimer’s Patients knowledge on three tasks in healthy controls with patients with Alzheimer’s disease and Herpes SimplexEncephalitis (HSE); and are designed to explore how category effects might be determined.
The incidence and pattern of category specificity across Alzheimer’s patients as a group (Tippett, Grossman, &Farah 1996; Silveri, Daniele, Guistolisi, & Gainotti 1991) and for individual Alzheimer’s patients (Laws, Leeson, & Gale, 2003; Garrard, Patterson, Watson, & Hodges, 1998; Gonnerman et al., 1997; Mauri, Daum, Satori, Riesch, &Birbaumer, 1994) have been inconsistent.
Most analyses of Alzheimer’s patients have reported living deficits, a minority has reported nonliving deficits, some report both, and still others find no categoryspecificity in Alzheimer’s patients (for a review, see Laws, Gale, et al., 2005). Some of this variability may stem from previously ignored methodological problems.
In particular, Laws, Gale, et al. (2005) note that controls in these studies are often performing at ceiling level, and have shown how this may distort the incidence and even direction of category effects. An additional issue con-cerns the observation that almost all of the studies examining category effects in Alzheimer’s patients have relied exclusively on picture naming as the test ofcategory. The current study addresses these points by testing picture naming, naming-to-description, and fea- ture verification in Alzheimer’s patients and controls.
The criterion for a living–nonliving dissociation was that *L = differential living deficit; NL = differential nonliving deficit; x2 =patient impaired using x2; s = strong dissociation; c = classical a patient must show a discrepancy that would be estimated to occur in <1% of the population of healthy + = living advantage; À = nonliving advantage.
none emerged as significant. Using the criterion of on two of the three tasks. Two HSE patients (DD category-consistent dissociations across all three tasks, and MF) also displayed paradoxical dissociations (see two patients (MS and AE) showed living impairments.
A further six patients were impaired on two tasks; critically, in patient LZ, the deficits were paradoxical,that is, impaired for living on one task (feature verifica- tion) and nonliving (picture naming) on another task.
The case of LZ is of particular importance. Given that Again, a large number of false negatives occur when the features used in the features verification task were using x2 (11 for HSE patients). As with Alzheimer’s all perceptual/visual, and given that semantic activation patients, an absolute (nonreferenced) comparison is from pictures is triggered from visual features of the likely to underestimate rather than overestimate the picture itself, this paradoxical dissociation is particularly incidence of category dissociations. Inconsistency across surprising. The lack of consistency across analyses and tasks for an individual patient again highlights the fact the finding of a paradoxical dissociation across tasks that performance on any one task cannot provide a emphasize why performance on any one task cannot reliable marker for a category effect—this includes provide a reliable indicator of the direction of a category feature verification (which was the most sensitive and reliable task in both groups for detecting significantcategory differences).
Furthermore, the analysis of HSE patients accentuates how absolute scores can be misleading. Consider BM, who showed a significant differential deficit for living things on feature verification in the x2 analysis (76% vs.
90%), but not when compared to controls (although the In contrast to the studies of category effects in groups ofAlzheimer’s patients, many of the reported case studiesof patients with category-specific disorders are people Table 2. Category Performance in HSE Patients (% Living and who have suffered from HSE (22/47 living cases, and not one nonliving case; see anatomical review by Gainotti,2000). The original report of category effects was in a series of four HSE patients (Warrington & Shallice, 1984). As Gainotti (2000) argues, when referring to HSE, that ‘‘the selective impairment of Living things, far from being an occasional phenomenon resultingfrom idiosyncratic factors, is strongly associated with a Following the initial reports of Warrington and Shallice (1984), other patients suffering from the same pathology who also displayed living thing deficits in- creased confidence in the claim that ‘‘real’’ category specificity can be found predominantly in this group ofpatients. Capitani, Laiacona, Mahon, and Caramazza (2003) reported, in their database, 26/61 patients with HSE in the group impaired for living and 1/18 in that impaired for nonliving. Nevertheless, these figures referto the incidence of HSE category-specific patients in the total population of category-specific patients. No data are available regarding the incidence of category speci-ficity among the HSE population. For this reason, the second part of this study examines the incidence and type of category effects that occur in a series of patients The data were analyzed using the same methods out- *L = differential living deficit; NL = differential nonliving deficit; x2 =patient impaired using x2; s = strong dissociation; c = classical lined in Experiment 1. One HSE patient was impaired across all three tasks (SR). Five patients were impaired + = living advantage; À = nonliving advantage.
difference was larger than 6 of 8 patients who showed a conversely, does the failure to find a large absolute significant difference using the Revised Standardized difference across category refute the presence of a cat- Difference Test [RSDT]). The absolute size of the differ- egory deficit. Indeed, several HSE and Alzheimer’s pa- ence is therefore not a definitive marker for the pres- tients showed large absolute differences that were sig- ence or absence of a differential deficit. Moreover, nificant using within-patient x2 analyses, but not when consider patients GE (90 vs. 86%) and ZG (90 vs.
referenced to control data (e.g., HSE patients such as 82%), who showed better absolute feature verification BM, DD, and SG for picture naming; and Alzheimer’s for living than for nonliving things, yet had differential patients such as RN, MR, and VP for naming-to- living deficits (indeed, classical dissociations). So, not description). Conversely and more commonly, x2 analy- only do absolute differences provide no guarantee to the ses led to false negatives in both HSE (11 dissociations) presence of a differential deficit, they provide no guar- and Alzheimer’s (7 dissociations) patients. This reflects antee about the direction of any deficit.
the simple fact that absolute differences are critical fordemonstrating deficits when using x2; however, healthycontrols may show category advantages, and therefore,even a lack of absolute difference in patients may be important. For example, analysis of picture-naming for This study investigated the validity of the criteria that are HSV patients DD and SG indicated no category differ- commonly used to document the presence of category- ence (56 vs. 57% and 76 vs. 78%), yet both showed sig- specific deficits, in particular, the strong dependence nificant dissociations when referenced to control data.
upon within-patient comparisons of absolute living and Conversely, Alzheimer’s patient FR showed a 31% dif- nonliving scores rather than comparing patient perform- ference in living and nonliving naming-to-description ance with that of healthy controls. Additionally, most (19 vs. 50%), but this was not significant when referenced studies have relied upon establishing category disorders on picture-naming tasks; and in some studies (especiallythose examining Alzheimer’s patients), rely solely upon picture naming as evidence of category specificity (Laws, Gale, et al., 2005). The current study illustrates how suchmethods lead to erroneous conclusions about both the As outlined in the Introduction, the extant literature presence and even the type of category deficit exhib- relies heavily upon using impaired picture naming to ited by patients. These findings have implications for a identify category effects; other tests may be used subse- range of questions relating to category disorders and quently to elaborate on this, but picture naming has raise methodological and theoretical issues regarding assumed a role as the sine qua non test of category what might be regarded as acceptable evidence for specificity. Therefore, the failure to find a category dissociations and double dissociations—especially when difference on naming tasks is typically viewed as under- used to make claims about the fractionation of cogni- mining the presence of any category effect (Laws, 1998; Although most cognitive psychologists are comfort- able with the notion of impaired naming and intact semantics (i.e., lexical deficits), the converse—impaired The current study highlights the unreliability of using semantics and intact naming—have been less frequently within-patient comparison of absolute scores. As already reported, perhaps because of the assumption that it mentioned, studies of Alzheimer’s patients frequently ought not to happen within most currently accepted assess picture naming only, and in line with the data models (Laws, Evans, et al., 1995). Nonetheless, the presented here, would be likely to misrepresent not current study documents several Alzheimer’s (2/15: only the incidence, but even the direction of a category 13%)1 and HSE (3/15: 20%) patients who showed a effect. The nature of x2 analyses means, of course, that category effect in semantics (using feature verification), studies always assume the deficit to be reflected in the but no category effect on either of two naming tasks.
smaller of any two scores obtained. The current study Some patients (2/19 Alzheimer’s and 4/15 HSE ) shows how this metric is unreliable and confirms our showed category deficits on tasks other than picture recent findings in two other groups of Alzheimer’s pa- naming, including patients (e.g., Alzheimer’s patient FB and 2 HSE patients DD and TL) with impaired perform- The widespread use of within-patient comparisons to ance on all tasks except picture naming. Therefore, determine category deficits may well have distorted the using picture naming as a critical test of category effects incidence, and therefore, the ratio of living to nonliving would miss some patients who showed consistent def- deficits reported in the literature. Even if patients show a icits on other tasks. Indeed, reliance on any one task patently large absolute difference across category, this may lead to a quite different conclusion from that if does not necessarily denote a category disorder; or another task had been chosen. Moreover, feature veri- fication proved to be a much more sensitive test for We must also consider the possibility that paradoxical detecting category effects (albeit all living deficits) than double dissociations reflect confounding variables. It picture naming or naming-to-description (although might be argued, for example, that f luctuations in there were also differences across pathologies). This attention could impact differentially over the test session suggests that the direction of category effects some- and potentially affect one category more than the other.
times seems to depend upon which test is chosen as the This is unlikely because it would require that the reference test (a prospect that has not been previously confound interacts highly selectively with category. Liv- entertained). In this context, it is worth noting that pa- ing and nonliving stimuli (on all tests) were randomly tients who show living disorders tend to be agnosic, and intermixed when presented, so a factor such as attention therefore, tested with picture naming; however, sev- fluctuation would have to impact only when items from eral nonliving cases have been aphasic, and thus, were one of the two categories were presented. This seems not tested with picture naming, but with tasks such even more implausible in cases when we consider as picture–name matching (see Laws, 2004). Hence, it paradoxical dissociations (i.e., in the opposite direction is common for different category effects to rely upon on a second test). Consider the case of the HSE patient different testing procedures; and as such, again the MF (see Figure 1), who showed a classical double existing literature may be prone to some of the issues dissociation across tasks. His picture naming was below the 1% for living things (and normal for nonlivingthings); and below the 1% for nonliving on featureverification (but normal for living things). In this con- Dissociations and Paradoxical Dissociations text, we would argue that the dissociations are robust to Dissociations often form the basis for speculations about cognitive architecture and modularity especially when Another potential confound concerns the possibility they are doubly dissociated between patients. The cur- that the dissociations reported here (whether consistent rent study shows, however, that dissociations can occur or paradoxical) are chance findings emerging from quite within a patient. Within-patient double dissociations noisy patient data, and that multiple analyses might across tasks (e.g., a living on Task A and a nonliving increase the likelihood of spurious outcomes. Indeed, on Task B) that are believed to have some critical typical statistical/methodological approaches may well processing stage in common, raise questions about the be prone to producing spurious and chance findings in double-dissociation methodology in single-case studies case study analyses. Nevertheless, Monte Carlo simula- and the interpretation of category effects per se. At atheoretical level, many models assume that deficits insemantics will have ‘‘knock-on’’ effects for naming; andso, such models have difficulty accounting for paradox-ical dissociations at the level of semantics and naming.
Paradoxical double dissociations pose problems for double dissociations at a variety of levels of compari-son including: across tasks (as described here), withintasks, and patients (Laws, Gale, et al., 2005); and ofcourse, across patients and across tasks (the typicalapproach in category specificity and cognitive neuro-psychology more generally). Given that paradoxicaldissociations arise, how might we distinguish a paradox-ical dissociation from a real double dissociation (i.e.,one that might be used to ground theories of cognitionor ‘‘carve cognition at its modular joints’’)? How should paradoxical dissociations be interpreted? Of course, it might be argued that paradoxical dissocia-tions are simply unreliable. Indeed, because we did notretest patients, we have no way of confirming whetherparadoxical dissociations are reliable. Nonetheless, thereliability of paradoxical dissociations has to be viewed Figure 1. A classical paradoxical double dissociation between living alongside the fact that reliability is hardly ever examined and nonliving things within one HSE patient (MF ). Note: Patient MF for dissociations in single-case studies. Indeed, follow-up displays a classical double dissociation across category (i.e., impaired analyses of the same patient by same or other research picture naming for living things, but normal nonliving thing naming).
On feature verification, he shows normal living and impaired nonliving groups are rare and sometimes contradictory (Laws, performance. Classical double dissociations (often with weaker 1998). Hence, it is crucial for future studies to examine evidence than here) typically provide the strongest evidence for the the reliability of all dissociations.
separation of cognitive processed (or architecture).
tions show that the RSDT (Crawford & Garthwaite, 2002). Half of the stimuli required a ‘‘yes’’ response and 2005) used in the current study provides excellent half required a ‘‘no’’ response.
control over Type 1 error rate (even when patient data 3. Naming-to-description: Originally published by are heavily skewed through poor performance).
Lambon-Ralph, Howard, Nightingale, and Ellis (1998) The dissociations reported here within one patient as a feature verification task, it was adapted to Italian as a are indistinguishable from those typically reported be- naming-to-description task. Sixteen animals and 12 ob- tween patients in the category-specific literature. De- jects were used. Each concept was described by a per- spite the equivalence of paradoxical and ‘‘regular’’ ceptual and a functional/associative description, thus dissociations (and double dissociations) in terms of their yielding a total of 32 descriptions for living and 24 acceptability as empirical evidence, it may be that both are unreliable. Hence, we should not take any one (oreven some combinations) test as prima facie evidence ofa category effect in one direction. As noted above, reliability, both across task and across time, are criticalissues for future studies to consider in the case study This experiment included 15 (7 women and 8 men)patients with HSE and 12 healthy control subjects (6women and 6 men). Herpes Simplex patients were all recruited in three hospitals located in Veneto (Italy), were native speakers of Italian, and all satisfied thefollowing criteria: all had polymerase-chain-reaction- proven HSE; all showed pathologic changes at CT scan The study included 19 (15 women and 4 men) patients or MRI which were usually bilateral, in the medial- with probable Alzheimer’s dementia and 15 healthy temporal and inferior frontal areas; all underwent treat- elderly control subjects (9 women and 6 men). The 19 ment with Acyclovir; and all were examined between 13 Alzheimer’s patients met the National Institute of Neu- rological and Communicative Disorders and Stroke/Alz- The HSE patients and controls were matched for heimer’s Disease and Related Disorders Association mean (SD) age [53.66 (14.77) vs. 54.47 (12.63): F(1,25) = (NINCDS/ADRDA) criteria for probable Alzheimer’s dis- 0.22, p = .88] and mean (SD) years of education [7.73 ease (McKhann et al., 1984). All 19 patients had (3.17) vs. 10.41 (4.31): F(1,25) = 3.47, p > .05]. They did Hachinski scores of less than 4 (Hachinski et al., differ significantly in their mean [SD] MMSE scores 1975) and a Mini-Mental State Examination (MMSE) [24.66 (3.26) vs. 28.68 (0.72): F(1,25) = 17.31, p < .001].
score below 24/30 (Folstein, Folstein, & McHugh,1975). All underwent CT or MRI scanning, together with a screening battery, to exclude treatable causes ofdementia. Patients with major depression, a history of The same three tasks used for the Alzheimer’s patient stroke or TIA, alcoholism, head injury, or major med- were administered to the HSE patients.
ical illnesses were excluded. Patients were recruited inthree hospitals and in four nursing homes located in Veneto (Italy) and were an unselected series.
The Alzheimer’s patients and controls were matched for age [mean (SD): 77.9 (8.7) vs. 75.33 (4.2): F(1,33) = The data for each individual participant were examined 1.02, p > .05] and years of education [4.73 (0.80) vs. 6.28 by comparing performance with their respective control (4.76): F(1, 33) = 1.5, p > .05]. They did, of course, group using the RSDT (Crawford & Garthwaite, 2005) differ significantly in their MMSE [19.31 (2.45) vs. 27.4 for testing for deficits and dissociations in single-case (1.11): F(1,33) = 126, p < .001].
studies. Of course, it is possible for patients to be im-paired at naming living or nonliving things, but that thedifference between their scores does not reach signifi- cance; equally, a patient may be severely impaired on 1. Picture naming: 32 living (20 animals and 12 fruits both tasks, but still show differential impairment. This and vegetables) and 32 nonliving items matched for method examines whether the discrepancy observed for frequency, familiarity, and visual complexity (Sartori, the patient is significantly different from the discrepan- cies observed for controls and provides a point estimate 2. Feature verification: 80 living and 64 nonliving of the abnormality of the individual’s discrepancy (i.e., it items (e.g., ‘‘a tiger has stripes’’). All features were estimates the percentage of the population that would perceptual. The examinee is required to respond if the obtain a more extreme discrepancy). Patient perform- feature belongs to the concept or not (Sartori et al., ance was classified as impaired if the discrepancy score was estimated to occur in less than 1% of the healthy cognitive state of patients for the clinician. Gainotti, G. (2000). What the locus of brain lesion tells Patients were classified as displaying strong dissocia- us about the nature of the cognitive defect underlying tions if they were (a) impaired at both tasks and (b) showed a significant discrepancy across category; and classified as displaying a classical dissociation if they Garrard, P., Patterson, K., Watson, P. C., & Hodges, J. R.
were (a) impaired in one category, but performed (1998). Category specific semantic loss in dementia ofAlzheimer’s type. Functional–anatomical correlations from normally on the other and (b) showed a significant discrepancy between the two categories (see Crawford Gonnerman, L. M., Anderson, E. S., Devlin, J. T., Kempler, D., & Garthwaite, 2005; Crawford, Garthwaite, & Gray, 2003).
& Seidenberg, M. S. (1997). Double dissociation of These methods of testing for deficits and for differ- semantic categories in Alzheimer’s disease. ences (i.e., dissociations) are to be preferred over the Hachinski, V. C., Iliff, L. D., Zihka, E., Duboulay, G. H., use of z and zD as they treat the statistics of the control McAllister, V. L., Marshall, J., Russell, R. W. R., & Symon, L.
sample as statistics rather than as population parame- (1975). Cerebral blood flow in dementia. ters. Moreover, Monte Carlo simulations show that the RSDT (Crawford & Garthwaite, 2005) provides excellent Lambon Ralph, M. A., Howard, D., Nightingale, G., & control over Type 1 error rate (regardless of whether the Ellis, A. W. (1998). Are living and non-livingcategory-specific deficits causally linked to impaired data are heavily skewed, the control sample is small in perceptual or associative knowledge? Evidence from a size, or the correlation between tasks). We contrasted results from application of this method with those derived from applying, on the same data, those derived Laws, K. R. (in press). ‘‘Illusions of Normality’’: A from using x2—the method most frequently used to methodological review of category-specific naming.
Cortex.
evaluate discrepancies in living/nonliving accuracy in Laws, K. R. (1998). A leopard never changes its spots.
single cases (Laws, in press). Programs to run the RSDT Laws, K. R. (2004). Strange and neglected domains in category specificity: The ‘‘normal’’ and the‘‘nonliving’’. History and Philosophy of Psychology, 6,19–39.
Laws, K. R., Evans, J. J., Hodges, J. R., & McCarthy, R. A. (1995).
Naming without knowing and appearance without We thank Dr. Ivry and the reviewers for insights and helpful associations—Evidence for constructive processes in comments on an earlier version of this article.
Reprint requests should be sent to Dr. Keith R. Laws, School of Laws, K. R., Gale, T. M., Leeson, V. C., & Crawford, J. R.
Psychology, University of Hertfordshire, College Lane, Hat- (2005). When is category specific in dementia of Alzheimer’s field, Hertfordshire AL10 9AB, UK, or via e-mail: keith.laws@ Laws, K. R, Leeson, V. C., & Gale, T. M. (2003). Inflated and contradictory category naming deficits in Alzheimer’sdisease? McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan, E. M. (1984). Clinical diagnosis of Four dementia of Alzheimer’s type patients were not test- Alzheimer disease: Report of the NINCDS-ADRDA work Mauri, A., Daum, I., Sartori, G., Riesch, G., & Birbaumer, N.
(1994). Category-specific semantic impairment in Alzheimer’s disease and temporal lobe dysfunction: Acomparative study. Journal of Clinical and Experimental Capitani, E., Laiacona, M., Mahon, B., & Caramazza, A. (2003).
What are the facts of semantic category-specific deficits? A Sartori, G., Job, R., & Zago, S. (2002). A case of domain-specific critical review of the clinical evidence. semantic deficits. In E. M. E. Forde & G. W. Humphreys (Eds.), Category specificity in brain and mind (pp. 25–49).
Crawford, J. R., & Garthwaite, P. H. (2005). Testing for suspected impairments and dissociations in single-case Silveri, M. C., Daniele, A., Giustolisi, L., & Gainotti, G. (1991).
studies in neuropsychology: Evaluation of alternatives using Dissociation between knowledge of living and non-living Monte Carlo simulations and revised tests for dissociations.
things in dementia of the Alzheimer’s type. Crawford, J. R., Garthwaite, P. H., & Gray, C. D. (2003).
Tippett, L. J., Grossman, M., & Farah, M. J. (1996). The Wanted: Fully operational definitions of dissociations in semantic memory impairment of Alzheimer’s disease: Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975).
Warrington, E. K., & Shallice, T. (1984). Category specific ‘‘Mini-mental state’’: A practical method for grading the

Source: http://semanticrelevance.psy.unipd.it/publications/Laws_Sartori_JCN_2005.pdf