- Email: [email protected]
Lexical knowledge degradation in schizophrenia
Schizophrenia Research 45 (2000) 123–131 www.elsevier.com/locate/schres
Lexical knowledge degradation in schizophrenia Keith R. Laws a, *, Mohammed Al-Uzri b, Ann M. Mortimer c a Department of Psychology, London Guildhall University, Calcutta House, Old Castle Street, London E1 7NT, UK b Department of Psychiatry, University of Leicester, Leicester LE2 7LX, UK c Department of Psychiatry, University of Hull, Willerby HU10 6ED, UK Received 10 September 1999; accepted 10 September 1999
Abstract This study documents a severe picture naming impairment in a series of 22 chronically hospitalised schizophrenics. The pattern and level of naming impairment were comparable in degree and type to that seen neurological patients with left hemisphere lesions. This deficit was further examined to determine whether it conformed to a disorder affecting access to the lexical representations or some degradation of the stored representations themselves. Patient naming was examined twice (18 months apart) for evidence of consistency and for word frequency effects. The pattern of individual patient performance on these criteria showed that the majority had storage disorders; access disorders alone occurred in only a small minority of patients. A subset of 11 of the same patients that were tested on three occasions (across 30 months) showed the same pattern but further indicated that when access problems are present, they reflect difficulty with deliberate, rather than automatic, access to the lexicon. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Access disorder; Naming; Semantics; Storage disorder
1. Introduction Whether schizophrenic patients show cognitive deficits that are consistent across time has been largely ignored, although it often seems to be assumed that they are inconsistent. This dates back to the earliest accounts of the disorder, when Bleuler (1911) observed that ‘‘…at times these patients forget and at other times they know the same fact according to the circumstances involved’’; and that ‘‘The actual amount of knowledge remains preserved…but it is not always available or it is employed in the wrong way…’’ (pp. 73–74). By contrast, Kraepelin (1913) argued * Corresponding author. E-mail address: [email protected] ( K.R. Laws)
that ‘‘Memory, acquired knowledge and expertness remain sometimes fairly well preserved, sometimes they undergo considerable loss’’ (p. 106). These early thoughts suggest that schizophrenic deficits may be inconsistent both across time and across patient. The issue of deficit consistency has important theoretical consequences because it might illuminate certain features of the underlying cognitive dysfunction(s) in schizophrenia. Criteria have been proposed for distinguishing between memory disorders that reflect either a loss/degradation of underlying representations or a problem with accessing intact representations. In the former, knowledge is assumed to be lost/degraded, whereas the latter describes knowledge that is temporarily inaccessible( Warrington and Shallice, 1979). While the validity of this distinction has been
0920-9964/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S0 9 2 0 -9 9 6 4 ( 9 9 ) 0 0 18 4 - X
124
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
occasionally disputed (Rapp and Caramazza, 1993), these two patterns of memory impairment have achieved wide acceptance in the neurological literature. Warrington and Shallice (1979) have proposed that degraded store disorders are characterized by: (1) high consistency with the same items across time; (2) ineffectiveness of priming and cueing techniques; (3) less familiar/frequent items are lost first; and (4) subordinate (or attribute) information is especially affected. In cases of impaired access, representations still exist, but the retrieval process is compromised, and this disorder is characterized by: (1) an inconsistent pattern of performance across different testing times; (2) improved performance with cueing and priming; (3) item familiarity/frequency effects are less apparent; and (4) subordinate attribute knowledge is likely to be less affected (see Fig. 1). The access–store dichotomy has received little direct attention in schizophrenic patients, although several studies might be viewed as supporting a difficulty of accessing representations, rather than a loss or degradation of the representations themselves. All of these studies have examined single criteria (consistency: Allen et al., 1993; cueing: Joyce et al., 1996; priming: Spitzer 1997) and have relied upon group analyses, which must limit the
strength of their conclusions since group profiles on single criteria may be misleading (Rapp and Caramazza, 1993; Laws et al., 1998). More recently, Laws et al. (1998) addressed some of the problems outlined above in a study of 12 schizophrenic patients with impaired familiar face naming. These patients were investigated as a series of case studies to determine whether their memory deficits conformed to either an access or store profile using three of the criteria (consistency, familiarity and cueing). Analysis of the profile of impairment in individual patients conformed well to the criteria outlined by Warrington and Shallice (i.e. patients who were consistent showed familiarity effects and failed to benefit from cueing; or showed the converse profile). Hence, there were heterogeneous performance profiles across patients, with store disorders in some patients and access disorders in others. Notably, those with store profiles were the most severely cognitively impaired (i.e. had the worst naming performance) and were chronically hospitalised, while those with access disorders were less cognitively impaired (had better naming) and were communitydwelling. The current study reports an 18- and 30 month longitudinal study of picture-naming in 22 chroni-
Fig. 1. Criteria for determining disorders of access and store [after Warrington and Shallice (1979)].
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
cally hospitalised schizophrenic patients. Naming performance was examined using two of the criteria outlined by Warrington and Shallice (consistency and frequency), to see whether any naming difficulties reflected problems of access or a degraded store.
125
46 neurological patients with unilateral left hemisphere lesions; McKenna and Warrington, 1983); more recently, the test has been normed against a larger sample of 400 normal subjects ( Warrington, 1997). 3.1. Naming
2. Patients The sample consisted of 22 chronically hospitalised patients meeting DSM-III-R criteria for schizophrenia (who were under the care of one of the authors: A.M.). The mean age of the patients was 46.02 (s.d.=13.94) years, with a mean hospitalisation of 22.77 (11.91) years. The estimated premorbid NART IQ (Nelson, 1982) for the patients was within the normal range (mean NART=99; s.d.=12.81).
3. Method and procedure Picture naming was examined using the Graded Naming Test (GNT; McKenna and Warrington, 1983). This task requires naming of 30 line drawings ranging from well-known items (e.g. kangaroo, corkscrew) to lesser-known items (e.g. yashmak, sextant). This test was originally normed and standardised on 100 normal individuals (and
The mean number of pictures named by the patient group was 12.82±6.6 on first testing and 13.36±6.5 on second testing (which represents no significant change across time: t=−1.45[df=21], p=0.16; similarly the correlation across testing sessions was very high: r=0.96, p<0.000). The combined mean for the patients was significantly lower than the mean for 100 normal subjects (22.54±4.3: Z=2.26, p<0.01) originally tested by McKenna and Warrington (1983), marginally lower than for the 46 patients with left hemisphere lesions (15.15±7.15) and significantly lower than the mean of 20.4±4.1: Z=1.8, p<0.05 for 400 normal subjects ( Warrington, 1997). The distributions of patient and normal performance on the GNT indicate a possible bimodal distribution (see Fig. 2), with about half of the patients falling inside and half falling outside the normal range. Fig. 3 shows the proportion of each group who are naming each of the 30 items — this clearly illustrates that the pattern of naming performance in schizophrenics is different to that of normal
Fig. 2. Distribution of patient and control naming.
126
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
Fig. 3. Proportion of schizophrenics naming each of the graded items (along with normative data from McKenna and Warrington, 1983).
subjects, but similar to that of patients with left hemisphere lesions. Hence, not only do schizophrenics perform poorly on the GNT, but they also show a profile that is similar to left hemisphere lesioned patients.
third of the patients (8/22), the difference was >15 IQ points. Hence, the naming performance of schizophrenics on the GNT is significantly lower than predicted by their achieved NART IQ.
3.2. Naming and IQ
4. Naming consistency across two time periods
Naming performance on the GNT correlates highly with NART IQ (r=0.73: McKenna and Warrington, 1983), and so it is necessary to determine whether the naming ability of patients is disproportionate to their level of intellectual functioning. The GNT provides a regression equation for predicting NART IQ [−9.55+(1.83×GNT score)], and this was used to predict NART IQs for each schizophrenic subject (GNT-derived IQ= 86.18±14.7), which was significantly lower [t= −4.14 (21) p<0.000] than their actual NART scores (96.45±12.8). Indeed, for more than one-
4.1. Method and procedure The naming data for the 22 subjects were analysed using the two parameter stochastic model outlined by Faglioni and Botti (1993) — this provides stochastic probabilities of retrieval [r] and storage [s]. The retrieval and storage of items cannot be regarded as independent — the retrieval of names depends first upon the probability that the names remain stored. In essence, the value of [s] sets an upper limit on the number of items that are available, and [r]
127
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
determines the number of stored items that are accessible. For example, with 30 items and a calculated storage probability [s] of 0.5, this means that the subject has 15 names stored; if the retrieval probability [r] is calculated to be 0.5, then 7.5 names are retrievable at any time (i.e. half of those calculated to be stored ). Values for [s] and [r] were calculated using: retrieval probability [r]=2S /(S 2S ) 2 1+ 2 storage probability [s]=(S 2S )/2r. 1+ 2 Individual patient retrieval [r] and storage [s] probabilities were calculated (see Table 1) and then used to derive estimates of the number of names stored and retrieved by patients — these estimates were compared as z-scores with the mean number of items named by normal controls (see Fig. 4). The mean number of items named from the normative sample (n=100) was used as a conservative indicator of both store size and name accessibility in normal subjects. This showed that 14 had impaired storage and 16 had impaired retrieval (14 had both impaired storage and retrieval );
two had impaired retrieval alone; and six were unimpaired.
5. Naming consistency across three time periods 5.1. Method and procedure Naming data were collected on a third occasion for a subset of 11 patients after 30 months. The data for these subjects were analysed using the three parameter stochastic model (see Faglioni and Botti 1993) — this provides stochastic probabilities of storage [s], automatic retrieval [a] and deliberate retrieval [d ]. One advantage of this analysis is that it provides separate probabilities concerning possible reasons for retrieval failure (i.e. whether it reflects a problem of automatic [a] or deliberate [d ] retrieval ). Faglioni and Botti suggest that [a] retrieval reflects effortless, infallible retrieval (probably determined by algorithms); while [d ] retrieval reflects conscious, fallible retrieval (prob-
Table 1 Storage and retrieval probabilities for schizophrenic patientsa Patient
s
r
Estimated store size
Estimated retrieval size
Z(s)
Z(r)
BR AE JCA DA JCO McB HS MF GS LK MSO MSM PMC CG LW TH RB PC TB AR JCAR SV Mean
0.00**** 0.07**** 0.10**** 0.17**** 0.20**** 0.20**** 0.20**** 0.27*** 0.30*** 0.33** 0.37** 0.43** 0.43** 0.50* 0.53 0.53 0.57 0.57 0.60 0.63 0.67 0.67 0.38
0.00**** 1.00**** 0.60**** 0.83**** 0.92**** 0.75**** 0.63**** 0.89*** 0.95*** 0.91*** 0.88** 0.87** 0.81** 0.86** 0.94* 0.84** 0.92 0.94 0.86 0.90 0.95 0.93 0.83
0 2 3 5 6 6 6 8 9 10 11 13 13 15 16 16 17 17 18 19 20 20 11.36
0.00 2.00 1.80 4.17 5.54 4.50 3.79 7.11 8.53 9.09 9.68 11.27 10.56 12.86 15.06 13.47 15.62 16.06 15.43 17.19 19.05 18.60 10.06
5.24 4.78 4.54 4.08 3.85 3.85 3.85 3.38 3.15 2.92 2.68 2.22 2.22 1.75 1.52 1.52 1.29 1.29 1.06 0.82 0.59 0.59 2.60
5.24 4.78 4.82 4.27 3.95 4.20 4.36 3.59 3.26 3.13 2.99 2.62 2.79 2.25 1.74 2.11 1.61 1.51 1.65 1.24 0.81 0.92 2.90
a *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.
128
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
Fig. 4. Estimated number of names stored and retrieved by schizophrenics tested on two occasions. The continuous line shows the mean number of items named (i.e. stored and retrieved ) by 100 normal subjects; the broken line shows a 2 s.d. cut-off point for impaired retireval.
ably determined by heuristics). To elaborate on the previous example, with a store probability of 0.5, 15 items are stored; if there is an automatic retrieval probability [a] of 0.5, 7.5 items are automatically retrieved; and finally, a number may also be deliberately retrieved from the missing 7.5 items (if the value for [d ] was 0.5, then about three or four items would also be deliberately retrieved. The values of [s], [a] and [d ] were calculated using: storage probability [s]=[3(1−S )S S2]/(3S ) 0 2+ 1 2 automatic retrieval probability [a]=[3S S −S2]/ 1 3 2 (3S S) 1 deliberate retrieval probability [d ]=S /(S S ). 2 1+ 2 Individual patient storage [s], automatic and deliberate retrieval probabilities [a and d ] were calculated. The probability figures were used to derive estimates of the number of names stored, automatically and deliberately retrieved by patients. These were compared with the mean
number of faces named by normal controls as outlined above (see Fig. 5). Five patients again showed normal storage (as with the two-parameter model ). Performance was within the normal range for automatic [a] and deliberate [d ] retrieval in five cases; the remaining five had impaired storage and [a] and [d ] retrieval. Hence, again, pure retrieval problems were rare, but when they did occur (typically in the context of a storage problem), they reflected deliberate [d ] rather than automatic [a] retrieval failures.
6. Frequency analysis Since the GNT is graded by difficulty, it would not be appropriate to compare naming for the first and last 15 items. Rather, we analysed the first 15 items only [which are named by 90%+ of normal subjects: see McKenna and Warrington (1983)],
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
129
Fig. 5. Estimated number of names stored and retrieved by schizophrenics tested on three occasions. The light section of the bars shows the number of names that are automatically retrieved, the dark bar shows the additional number that are deliberately retrieved, and the asterisk shows the estimated number of names stored for each patient. The continuous line shows the mean number of items named by 100 normal subjects; the broken line shows a 2 s.d. cut-off point for impaired retireval.
comparing the number of items named (first five versus last five) for the two testing sessions. This showed that schizophrenics had better naming of the first five [items 1–5: mean=7.77(s.d.=2.9)] than the last five [items 11–15: mean=5.18(s.d.= 2.97): t=5.38(21), p<0.00001]. Since 90%+ of normal subjects name all of the first 15 items, naming does not differ across these two subsets; hence, the schizophrenics show an abnormal pattern and, one that is indicative of a store disorder (i.e. affected by frequency). At the level of individual patients, 19/22 showed worse naming of items 1–5 than 11–15; only three patients showed the reverse profile (BR, JCA and AR). Therefore, of the 14 patients with impaired storage probabilities, 12 also showed the expected frequency effect thought to be characteristic of a storage disorder (and in six cases, this was signifi-
cant even at the individual level: JCO; McB; MF; GS ; LK; MSO).
7. Discussion This paper has documented that chronically hospitalised schizophrenics have a severe picturenaming deficit, and that this is disproportionate to their intellectual level and comparable in degree and type to the naming problems found in neurological patients with left-hemisphere lesions. The analysis of consistency and frequency revealed that store disorders are more common than ‘pure’ access disorders in this sample of chronically ill schizophrenics, i.e. the majority of patients (14/22: 64%) had low storage probabilities, and 12 of these also showed a frequency effect; two patients had
130
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
impaired retrieval without impaired store size, i.e. pure access disorders; and six were unimpaired. These results accord with, and compliment, the one other systematic study documenting that a degrading of lexical-semantic representations is more commonly seen in chronically hospitalised schizophrenics (Laws et al., 1998). These findings have theoretical consequences both for the study of schizophrenia and more generally perhaps for the conceptual distinction between access and storage disorders in neurological patients. While the notions of access and store disorders were developed from studies of inherently stable/deteriorating profiles in neurological patients, it is intriguing that schizophrenics show the same relatively clear-cut patterns of impairment. Although we cannot definitively infer a neurological deficit from poor cognitive performance in schizophrenics (see Laws, 1999), it seems fortuitous that a non-neurological problem results in their showing the same complex profiles of impairment as neurological patients with naming deficits. The documentation of storage deficit profiles in a majority of chronically ill schizophrenic patients must raise the issue of cognitive deterioration in this disorder. Like Laws et al. (1998), and despite covering a 2 year time-span, the current study also found no evidence of decline in patient-naming. Failure to find a change across time does not, however, imply that degeneration does not occur, since the greater part of degeneration may occur earlier in the course of the disorder, i.e. in less chronically ill and/or less cognitively impaired patients (see below). Future systematic longitudinal studies covering greater time spans, along with cross-sectional studies of patients with varied length of illness, are necessary to establish whether there is a slow progression of cognitive deterioration or a more static disease picture in schizophrenia. Certainly, the one previous study of less chronically ill and less cognitively impaired patients suggests a greater prevalence of access disorders in that patient group (Laws et al., 1998). Additional confirmation for this notion also comes from the finding that the length of illness is associated with declining facilitation in semantic priming (and that this occurs independently of age and
medication; see Maher et al. (1996)]. In other words, priming effects are inversely correlated with length of illness and, thus, also consistent with a storage disorder. A small minority of the patients (2/22) in the current study did show pure access problems. Although the incidence might be lower than expected from previous studies (cf. Allen et al., 1993; Joyce, et al., 1996; Spitzer, 1997), this undoubtedly reflects their use of single criteria (which are misleading), along possibly with the examination of different patient samples (i.e. less severely cognitively impaired and less chronically ill; Laws et al., 1998). Nevertheless, the current findings do expand upon previous descriptions of an access disorder in schizophrenia by suggesting that they reflect difficulties with deliberate rather than automatic access. Concerning the latter point, one possibility is that deliberate access difficulties reflect some form of executive dysfunction in those patients, i.e. an inability or failure to develop appropriate retrieval strategies (Laws, 1999). In a recent exposition of the access-store hypothesis in the neurological literature, ( Warrington and Cipolotti, 1996) suggest that access and storage patterns might be related to different pathologies: storage disorders reflecting a loss of neural structures (e.g. degenerative disorders); and access disorders, a reduction of neuronal conductivity (e.g. vascular disorders and tumours). While this anatomical framework may not map easily onto schizophrenia, it might be thought to imply the presence of a degenerative disorder in chronically hospitalised patients. An important caveat to this notion, however, is the finding by Laws et al. (1998) that the degree of deficit severity was related both to the degree of ‘natural’ change in patient cognitive performance across time and to their responsiveness to cueing, i.e. whether they show access or storage deficits. In the current study, the degree of change in naming across time was also inversely correlated with the initial level of naming. Although the correlation was not as large here [r=−0.36, p=0.10], the fact that most patients had storage disorders and were so severely cognitively impaired, probably reduced the range of scores and so, also mitigated substantial change. These studies therefore suggest that access and
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
store disorders — in schizophrenia — probably reflect differences in deficit severity; and so, differences in one underlying neuropathology would seem to offer a more parsimonious explanation than multiple pathologies (cf. Laws et al 1998). One possibility stems from differences in normal local synaptic compensatory mechanisms, that might maintain memories in spite of synaptic degradation. As in normal aging (Bertoni-Fredarri et al., 1990), compensatory processes may preserve deteriorating memory for quite long periods (producing access characteristics); however, beyond a certain limit, they are less able to ‘uphold’ memories (producing store characteristics).
Acknowledgements A version of this paper was originally presented at the 9th International Congress on Schizophrenia Research (ICOSR), Santa Fe, New Mexico, USA, April 1999. K.R.L. wishes to thank the ICOSR organizers for supporting his work with a Young Investigator Award. We would like to thank Professor Manfred Spitzer and an anonymous reviewer for their helpful comments on an earlier draft of this paper.
References Allen, H.A., Liddle, P.F., Frith, C.D., 1993. Negative features retrieval processes and verbal fluency in schizophrenia. Br. J. Psychiatry 163, 769–775. Bertoni-Fredarri, C., Fattoretti, P., Casoli, T., Meier-Ruge, W.,
131
Ulrich, J., 1990. Morphological adaptive response of synaptic junctional zones in the human dendate gyrus during aging and Alzheimer’s disease. Brain Res. 517, 69–75. Bleuler, E., 1911. Dementia Præcox or the Group of Schizophrenias. International Universities Press, New York. trans. J. Zinkin 1950 Faglioni, P., Botti, C., 1993. How to differentiate retrieval from storage deficit: a stochastic approach to semantic memory modeling. Cortex 30, 501–518. Joyce, E.M., Collinson, L., Crichton, P., 1996. Verbal fluency in schizophrenia: a relationship with executive function — semantic memory and clinical alogia. Psychol. Med. 26, 39–49. Kraepelin, E., 1913. Dementia Præcox and Paraphrenia. Livingstone, Edinburgh. Laws, K.R., McKenna, P.J., Kondel, T.K., 1998. On the distinction between access and store disorders in schizophrenia: a question of deficit severity? Neuropsychologia 36, 313–321. Laws, K.R., 1999. A meta-analytic review of Wisconsin Card Sort studies in schizophrenia: general intellectual deficit in disguise? Cog. Neuropsychiatry 4, 1–3. McKenna, P., Warrington, E.K., 1983. The Graded Naming Test. NFER-Nelson, Windsor, UK. Maher, B.A., Manschreck, T.C., Redmond, D., Beaudette, S., 1996. Length of illness and the gradient from positive to negative semantic priming in schizophrenic patients. Schizophr. Res. 22, 127–132. Nelson, H.E., 1982. The National Adult Reading Test (NART ). NFER-Nelson, Windsor UK. Rapp, B., Caramazza, A., 1993. On the distinction between deficits of access and deficits of storage: a question of theory. Cog. Neuropsychol. 4, 131–185. Spitzer, M., 1997. A cognitive neuroscience view of schizophrenic thought disorder. Schizophr. Bull. 23, 29–50. Warrington, E.K., Cipolotti, L., 1996. Word comprehension The distinction between refractory and storage impairments. Brain 119, 611–625. Warrington, E.K., Shallice, T., 1979. Semantic access dyslexia. Brain 102, 43–63. Warrington, E.K., 1997. The graded naming test: A restandardisation. Neuropsychol. Rehab. 7, 143–146.
Lexical knowledge degradation in schizophrenia Keith R. Laws a, *, Mohammed Al-Uzri b, Ann M. Mortimer c a Department of Psychology, London Guildhall University, Calcutta House, Old Castle Street, London E1 7NT, UK b Department of Psychiatry, University of Leicester, Leicester LE2 7LX, UK c Department of Psychiatry, University of Hull, Willerby HU10 6ED, UK Received 10 September 1999; accepted 10 September 1999
Abstract This study documents a severe picture naming impairment in a series of 22 chronically hospitalised schizophrenics. The pattern and level of naming impairment were comparable in degree and type to that seen neurological patients with left hemisphere lesions. This deficit was further examined to determine whether it conformed to a disorder affecting access to the lexical representations or some degradation of the stored representations themselves. Patient naming was examined twice (18 months apart) for evidence of consistency and for word frequency effects. The pattern of individual patient performance on these criteria showed that the majority had storage disorders; access disorders alone occurred in only a small minority of patients. A subset of 11 of the same patients that were tested on three occasions (across 30 months) showed the same pattern but further indicated that when access problems are present, they reflect difficulty with deliberate, rather than automatic, access to the lexicon. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Access disorder; Naming; Semantics; Storage disorder
1. Introduction Whether schizophrenic patients show cognitive deficits that are consistent across time has been largely ignored, although it often seems to be assumed that they are inconsistent. This dates back to the earliest accounts of the disorder, when Bleuler (1911) observed that ‘‘…at times these patients forget and at other times they know the same fact according to the circumstances involved’’; and that ‘‘The actual amount of knowledge remains preserved…but it is not always available or it is employed in the wrong way…’’ (pp. 73–74). By contrast, Kraepelin (1913) argued * Corresponding author. E-mail address: [email protected] ( K.R. Laws)
that ‘‘Memory, acquired knowledge and expertness remain sometimes fairly well preserved, sometimes they undergo considerable loss’’ (p. 106). These early thoughts suggest that schizophrenic deficits may be inconsistent both across time and across patient. The issue of deficit consistency has important theoretical consequences because it might illuminate certain features of the underlying cognitive dysfunction(s) in schizophrenia. Criteria have been proposed for distinguishing between memory disorders that reflect either a loss/degradation of underlying representations or a problem with accessing intact representations. In the former, knowledge is assumed to be lost/degraded, whereas the latter describes knowledge that is temporarily inaccessible( Warrington and Shallice, 1979). While the validity of this distinction has been
0920-9964/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S0 9 2 0 -9 9 6 4 ( 9 9 ) 0 0 18 4 - X
124
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
occasionally disputed (Rapp and Caramazza, 1993), these two patterns of memory impairment have achieved wide acceptance in the neurological literature. Warrington and Shallice (1979) have proposed that degraded store disorders are characterized by: (1) high consistency with the same items across time; (2) ineffectiveness of priming and cueing techniques; (3) less familiar/frequent items are lost first; and (4) subordinate (or attribute) information is especially affected. In cases of impaired access, representations still exist, but the retrieval process is compromised, and this disorder is characterized by: (1) an inconsistent pattern of performance across different testing times; (2) improved performance with cueing and priming; (3) item familiarity/frequency effects are less apparent; and (4) subordinate attribute knowledge is likely to be less affected (see Fig. 1). The access–store dichotomy has received little direct attention in schizophrenic patients, although several studies might be viewed as supporting a difficulty of accessing representations, rather than a loss or degradation of the representations themselves. All of these studies have examined single criteria (consistency: Allen et al., 1993; cueing: Joyce et al., 1996; priming: Spitzer 1997) and have relied upon group analyses, which must limit the
strength of their conclusions since group profiles on single criteria may be misleading (Rapp and Caramazza, 1993; Laws et al., 1998). More recently, Laws et al. (1998) addressed some of the problems outlined above in a study of 12 schizophrenic patients with impaired familiar face naming. These patients were investigated as a series of case studies to determine whether their memory deficits conformed to either an access or store profile using three of the criteria (consistency, familiarity and cueing). Analysis of the profile of impairment in individual patients conformed well to the criteria outlined by Warrington and Shallice (i.e. patients who were consistent showed familiarity effects and failed to benefit from cueing; or showed the converse profile). Hence, there were heterogeneous performance profiles across patients, with store disorders in some patients and access disorders in others. Notably, those with store profiles were the most severely cognitively impaired (i.e. had the worst naming performance) and were chronically hospitalised, while those with access disorders were less cognitively impaired (had better naming) and were communitydwelling. The current study reports an 18- and 30 month longitudinal study of picture-naming in 22 chroni-
Fig. 1. Criteria for determining disorders of access and store [after Warrington and Shallice (1979)].
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
cally hospitalised schizophrenic patients. Naming performance was examined using two of the criteria outlined by Warrington and Shallice (consistency and frequency), to see whether any naming difficulties reflected problems of access or a degraded store.
125
46 neurological patients with unilateral left hemisphere lesions; McKenna and Warrington, 1983); more recently, the test has been normed against a larger sample of 400 normal subjects ( Warrington, 1997). 3.1. Naming
2. Patients The sample consisted of 22 chronically hospitalised patients meeting DSM-III-R criteria for schizophrenia (who were under the care of one of the authors: A.M.). The mean age of the patients was 46.02 (s.d.=13.94) years, with a mean hospitalisation of 22.77 (11.91) years. The estimated premorbid NART IQ (Nelson, 1982) for the patients was within the normal range (mean NART=99; s.d.=12.81).
3. Method and procedure Picture naming was examined using the Graded Naming Test (GNT; McKenna and Warrington, 1983). This task requires naming of 30 line drawings ranging from well-known items (e.g. kangaroo, corkscrew) to lesser-known items (e.g. yashmak, sextant). This test was originally normed and standardised on 100 normal individuals (and
The mean number of pictures named by the patient group was 12.82±6.6 on first testing and 13.36±6.5 on second testing (which represents no significant change across time: t=−1.45[df=21], p=0.16; similarly the correlation across testing sessions was very high: r=0.96, p<0.000). The combined mean for the patients was significantly lower than the mean for 100 normal subjects (22.54±4.3: Z=2.26, p<0.01) originally tested by McKenna and Warrington (1983), marginally lower than for the 46 patients with left hemisphere lesions (15.15±7.15) and significantly lower than the mean of 20.4±4.1: Z=1.8, p<0.05 for 400 normal subjects ( Warrington, 1997). The distributions of patient and normal performance on the GNT indicate a possible bimodal distribution (see Fig. 2), with about half of the patients falling inside and half falling outside the normal range. Fig. 3 shows the proportion of each group who are naming each of the 30 items — this clearly illustrates that the pattern of naming performance in schizophrenics is different to that of normal
Fig. 2. Distribution of patient and control naming.
126
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
Fig. 3. Proportion of schizophrenics naming each of the graded items (along with normative data from McKenna and Warrington, 1983).
subjects, but similar to that of patients with left hemisphere lesions. Hence, not only do schizophrenics perform poorly on the GNT, but they also show a profile that is similar to left hemisphere lesioned patients.
third of the patients (8/22), the difference was >15 IQ points. Hence, the naming performance of schizophrenics on the GNT is significantly lower than predicted by their achieved NART IQ.
3.2. Naming and IQ
4. Naming consistency across two time periods
Naming performance on the GNT correlates highly with NART IQ (r=0.73: McKenna and Warrington, 1983), and so it is necessary to determine whether the naming ability of patients is disproportionate to their level of intellectual functioning. The GNT provides a regression equation for predicting NART IQ [−9.55+(1.83×GNT score)], and this was used to predict NART IQs for each schizophrenic subject (GNT-derived IQ= 86.18±14.7), which was significantly lower [t= −4.14 (21) p<0.000] than their actual NART scores (96.45±12.8). Indeed, for more than one-
4.1. Method and procedure The naming data for the 22 subjects were analysed using the two parameter stochastic model outlined by Faglioni and Botti (1993) — this provides stochastic probabilities of retrieval [r] and storage [s]. The retrieval and storage of items cannot be regarded as independent — the retrieval of names depends first upon the probability that the names remain stored. In essence, the value of [s] sets an upper limit on the number of items that are available, and [r]
127
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
determines the number of stored items that are accessible. For example, with 30 items and a calculated storage probability [s] of 0.5, this means that the subject has 15 names stored; if the retrieval probability [r] is calculated to be 0.5, then 7.5 names are retrievable at any time (i.e. half of those calculated to be stored ). Values for [s] and [r] were calculated using: retrieval probability [r]=2S /(S 2S ) 2 1+ 2 storage probability [s]=(S 2S )/2r. 1+ 2 Individual patient retrieval [r] and storage [s] probabilities were calculated (see Table 1) and then used to derive estimates of the number of names stored and retrieved by patients — these estimates were compared as z-scores with the mean number of items named by normal controls (see Fig. 4). The mean number of items named from the normative sample (n=100) was used as a conservative indicator of both store size and name accessibility in normal subjects. This showed that 14 had impaired storage and 16 had impaired retrieval (14 had both impaired storage and retrieval );
two had impaired retrieval alone; and six were unimpaired.
5. Naming consistency across three time periods 5.1. Method and procedure Naming data were collected on a third occasion for a subset of 11 patients after 30 months. The data for these subjects were analysed using the three parameter stochastic model (see Faglioni and Botti 1993) — this provides stochastic probabilities of storage [s], automatic retrieval [a] and deliberate retrieval [d ]. One advantage of this analysis is that it provides separate probabilities concerning possible reasons for retrieval failure (i.e. whether it reflects a problem of automatic [a] or deliberate [d ] retrieval ). Faglioni and Botti suggest that [a] retrieval reflects effortless, infallible retrieval (probably determined by algorithms); while [d ] retrieval reflects conscious, fallible retrieval (prob-
Table 1 Storage and retrieval probabilities for schizophrenic patientsa Patient
s
r
Estimated store size
Estimated retrieval size
Z(s)
Z(r)
BR AE JCA DA JCO McB HS MF GS LK MSO MSM PMC CG LW TH RB PC TB AR JCAR SV Mean
0.00**** 0.07**** 0.10**** 0.17**** 0.20**** 0.20**** 0.20**** 0.27*** 0.30*** 0.33** 0.37** 0.43** 0.43** 0.50* 0.53 0.53 0.57 0.57 0.60 0.63 0.67 0.67 0.38
0.00**** 1.00**** 0.60**** 0.83**** 0.92**** 0.75**** 0.63**** 0.89*** 0.95*** 0.91*** 0.88** 0.87** 0.81** 0.86** 0.94* 0.84** 0.92 0.94 0.86 0.90 0.95 0.93 0.83
0 2 3 5 6 6 6 8 9 10 11 13 13 15 16 16 17 17 18 19 20 20 11.36
0.00 2.00 1.80 4.17 5.54 4.50 3.79 7.11 8.53 9.09 9.68 11.27 10.56 12.86 15.06 13.47 15.62 16.06 15.43 17.19 19.05 18.60 10.06
5.24 4.78 4.54 4.08 3.85 3.85 3.85 3.38 3.15 2.92 2.68 2.22 2.22 1.75 1.52 1.52 1.29 1.29 1.06 0.82 0.59 0.59 2.60
5.24 4.78 4.82 4.27 3.95 4.20 4.36 3.59 3.26 3.13 2.99 2.62 2.79 2.25 1.74 2.11 1.61 1.51 1.65 1.24 0.81 0.92 2.90
a *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.
128
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
Fig. 4. Estimated number of names stored and retrieved by schizophrenics tested on two occasions. The continuous line shows the mean number of items named (i.e. stored and retrieved ) by 100 normal subjects; the broken line shows a 2 s.d. cut-off point for impaired retireval.
ably determined by heuristics). To elaborate on the previous example, with a store probability of 0.5, 15 items are stored; if there is an automatic retrieval probability [a] of 0.5, 7.5 items are automatically retrieved; and finally, a number may also be deliberately retrieved from the missing 7.5 items (if the value for [d ] was 0.5, then about three or four items would also be deliberately retrieved. The values of [s], [a] and [d ] were calculated using: storage probability [s]=[3(1−S )S S2]/(3S ) 0 2+ 1 2 automatic retrieval probability [a]=[3S S −S2]/ 1 3 2 (3S S) 1 deliberate retrieval probability [d ]=S /(S S ). 2 1+ 2 Individual patient storage [s], automatic and deliberate retrieval probabilities [a and d ] were calculated. The probability figures were used to derive estimates of the number of names stored, automatically and deliberately retrieved by patients. These were compared with the mean
number of faces named by normal controls as outlined above (see Fig. 5). Five patients again showed normal storage (as with the two-parameter model ). Performance was within the normal range for automatic [a] and deliberate [d ] retrieval in five cases; the remaining five had impaired storage and [a] and [d ] retrieval. Hence, again, pure retrieval problems were rare, but when they did occur (typically in the context of a storage problem), they reflected deliberate [d ] rather than automatic [a] retrieval failures.
6. Frequency analysis Since the GNT is graded by difficulty, it would not be appropriate to compare naming for the first and last 15 items. Rather, we analysed the first 15 items only [which are named by 90%+ of normal subjects: see McKenna and Warrington (1983)],
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
129
Fig. 5. Estimated number of names stored and retrieved by schizophrenics tested on three occasions. The light section of the bars shows the number of names that are automatically retrieved, the dark bar shows the additional number that are deliberately retrieved, and the asterisk shows the estimated number of names stored for each patient. The continuous line shows the mean number of items named by 100 normal subjects; the broken line shows a 2 s.d. cut-off point for impaired retireval.
comparing the number of items named (first five versus last five) for the two testing sessions. This showed that schizophrenics had better naming of the first five [items 1–5: mean=7.77(s.d.=2.9)] than the last five [items 11–15: mean=5.18(s.d.= 2.97): t=5.38(21), p<0.00001]. Since 90%+ of normal subjects name all of the first 15 items, naming does not differ across these two subsets; hence, the schizophrenics show an abnormal pattern and, one that is indicative of a store disorder (i.e. affected by frequency). At the level of individual patients, 19/22 showed worse naming of items 1–5 than 11–15; only three patients showed the reverse profile (BR, JCA and AR). Therefore, of the 14 patients with impaired storage probabilities, 12 also showed the expected frequency effect thought to be characteristic of a storage disorder (and in six cases, this was signifi-
cant even at the individual level: JCO; McB; MF; GS ; LK; MSO).
7. Discussion This paper has documented that chronically hospitalised schizophrenics have a severe picturenaming deficit, and that this is disproportionate to their intellectual level and comparable in degree and type to the naming problems found in neurological patients with left-hemisphere lesions. The analysis of consistency and frequency revealed that store disorders are more common than ‘pure’ access disorders in this sample of chronically ill schizophrenics, i.e. the majority of patients (14/22: 64%) had low storage probabilities, and 12 of these also showed a frequency effect; two patients had
130
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
impaired retrieval without impaired store size, i.e. pure access disorders; and six were unimpaired. These results accord with, and compliment, the one other systematic study documenting that a degrading of lexical-semantic representations is more commonly seen in chronically hospitalised schizophrenics (Laws et al., 1998). These findings have theoretical consequences both for the study of schizophrenia and more generally perhaps for the conceptual distinction between access and storage disorders in neurological patients. While the notions of access and store disorders were developed from studies of inherently stable/deteriorating profiles in neurological patients, it is intriguing that schizophrenics show the same relatively clear-cut patterns of impairment. Although we cannot definitively infer a neurological deficit from poor cognitive performance in schizophrenics (see Laws, 1999), it seems fortuitous that a non-neurological problem results in their showing the same complex profiles of impairment as neurological patients with naming deficits. The documentation of storage deficit profiles in a majority of chronically ill schizophrenic patients must raise the issue of cognitive deterioration in this disorder. Like Laws et al. (1998), and despite covering a 2 year time-span, the current study also found no evidence of decline in patient-naming. Failure to find a change across time does not, however, imply that degeneration does not occur, since the greater part of degeneration may occur earlier in the course of the disorder, i.e. in less chronically ill and/or less cognitively impaired patients (see below). Future systematic longitudinal studies covering greater time spans, along with cross-sectional studies of patients with varied length of illness, are necessary to establish whether there is a slow progression of cognitive deterioration or a more static disease picture in schizophrenia. Certainly, the one previous study of less chronically ill and less cognitively impaired patients suggests a greater prevalence of access disorders in that patient group (Laws et al., 1998). Additional confirmation for this notion also comes from the finding that the length of illness is associated with declining facilitation in semantic priming (and that this occurs independently of age and
medication; see Maher et al. (1996)]. In other words, priming effects are inversely correlated with length of illness and, thus, also consistent with a storage disorder. A small minority of the patients (2/22) in the current study did show pure access problems. Although the incidence might be lower than expected from previous studies (cf. Allen et al., 1993; Joyce, et al., 1996; Spitzer, 1997), this undoubtedly reflects their use of single criteria (which are misleading), along possibly with the examination of different patient samples (i.e. less severely cognitively impaired and less chronically ill; Laws et al., 1998). Nevertheless, the current findings do expand upon previous descriptions of an access disorder in schizophrenia by suggesting that they reflect difficulties with deliberate rather than automatic access. Concerning the latter point, one possibility is that deliberate access difficulties reflect some form of executive dysfunction in those patients, i.e. an inability or failure to develop appropriate retrieval strategies (Laws, 1999). In a recent exposition of the access-store hypothesis in the neurological literature, ( Warrington and Cipolotti, 1996) suggest that access and storage patterns might be related to different pathologies: storage disorders reflecting a loss of neural structures (e.g. degenerative disorders); and access disorders, a reduction of neuronal conductivity (e.g. vascular disorders and tumours). While this anatomical framework may not map easily onto schizophrenia, it might be thought to imply the presence of a degenerative disorder in chronically hospitalised patients. An important caveat to this notion, however, is the finding by Laws et al. (1998) that the degree of deficit severity was related both to the degree of ‘natural’ change in patient cognitive performance across time and to their responsiveness to cueing, i.e. whether they show access or storage deficits. In the current study, the degree of change in naming across time was also inversely correlated with the initial level of naming. Although the correlation was not as large here [r=−0.36, p=0.10], the fact that most patients had storage disorders and were so severely cognitively impaired, probably reduced the range of scores and so, also mitigated substantial change. These studies therefore suggest that access and
K.R. Laws et al. / Schizophrenia Research 45 (2000) 123–131
store disorders — in schizophrenia — probably reflect differences in deficit severity; and so, differences in one underlying neuropathology would seem to offer a more parsimonious explanation than multiple pathologies (cf. Laws et al 1998). One possibility stems from differences in normal local synaptic compensatory mechanisms, that might maintain memories in spite of synaptic degradation. As in normal aging (Bertoni-Fredarri et al., 1990), compensatory processes may preserve deteriorating memory for quite long periods (producing access characteristics); however, beyond a certain limit, they are less able to ‘uphold’ memories (producing store characteristics).
Acknowledgements A version of this paper was originally presented at the 9th International Congress on Schizophrenia Research (ICOSR), Santa Fe, New Mexico, USA, April 1999. K.R.L. wishes to thank the ICOSR organizers for supporting his work with a Young Investigator Award. We would like to thank Professor Manfred Spitzer and an anonymous reviewer for their helpful comments on an earlier draft of this paper.
References Allen, H.A., Liddle, P.F., Frith, C.D., 1993. Negative features retrieval processes and verbal fluency in schizophrenia. Br. J. Psychiatry 163, 769–775. Bertoni-Fredarri, C., Fattoretti, P., Casoli, T., Meier-Ruge, W.,
131
Ulrich, J., 1990. Morphological adaptive response of synaptic junctional zones in the human dendate gyrus during aging and Alzheimer’s disease. Brain Res. 517, 69–75. Bleuler, E., 1911. Dementia Præcox or the Group of Schizophrenias. International Universities Press, New York. trans. J. Zinkin 1950 Faglioni, P., Botti, C., 1993. How to differentiate retrieval from storage deficit: a stochastic approach to semantic memory modeling. Cortex 30, 501–518. Joyce, E.M., Collinson, L., Crichton, P., 1996. Verbal fluency in schizophrenia: a relationship with executive function — semantic memory and clinical alogia. Psychol. Med. 26, 39–49. Kraepelin, E., 1913. Dementia Præcox and Paraphrenia. Livingstone, Edinburgh. Laws, K.R., McKenna, P.J., Kondel, T.K., 1998. On the distinction between access and store disorders in schizophrenia: a question of deficit severity? Neuropsychologia 36, 313–321. Laws, K.R., 1999. A meta-analytic review of Wisconsin Card Sort studies in schizophrenia: general intellectual deficit in disguise? Cog. Neuropsychiatry 4, 1–3. McKenna, P., Warrington, E.K., 1983. The Graded Naming Test. NFER-Nelson, Windsor, UK. Maher, B.A., Manschreck, T.C., Redmond, D., Beaudette, S., 1996. Length of illness and the gradient from positive to negative semantic priming in schizophrenic patients. Schizophr. Res. 22, 127–132. Nelson, H.E., 1982. The National Adult Reading Test (NART ). NFER-Nelson, Windsor UK. Rapp, B., Caramazza, A., 1993. On the distinction between deficits of access and deficits of storage: a question of theory. Cog. Neuropsychol. 4, 131–185. Spitzer, M., 1997. A cognitive neuroscience view of schizophrenic thought disorder. Schizophr. Bull. 23, 29–50. Warrington, E.K., Cipolotti, L., 1996. Word comprehension The distinction between refractory and storage impairments. Brain 119, 611–625. Warrington, E.K., Shallice, T., 1979. Semantic access dyslexia. Brain 102, 43–63. Warrington, E.K., 1997. The graded naming test: A restandardisation. Neuropsychol. Rehab. 7, 143–146.