Journal of Clinical and Experimental Neuropsychology
ISSN: 0168-8634 (Print) (Online) Journal homepage: http://www.tandfonline.com/loi/ncen19
Release from proactive interference: Determinants of performance and neuropsychological correlates Christopher Randolph , James M. Gold , Constance J. Carpenter , Terry E. Goldberg & Daniel R. Weinberger To cite this article: Christopher Randolph , James M. Gold , Constance J. Carpenter , Terry E. Goldberg & Daniel R. Weinberger (1992) Release from proactive interference: Determinants of performance and neuropsychological correlates, Journal of Clinical and Experimental Neuropsychology, 14:5, 785-800, DOI: 10.1080/01688639208402863 To link to this article: http://dx.doi.org/10.1080/01688639208402863
Published online: 04 Jan 2008.
Submit your article to this journal
Article views: 19
View related articles
Citing articles: 13 View citing articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ncen20 Download by: [Nanyang Technological University]
Date: 10 November 2015, At: 03:05
Journal of Clinical and Experimental Neuropsychology 1992, Vol. 14, NO.5, pp. 785-800
0168-8634/92/1405-0785$3.00 0 Swets & Zeitlinger
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
Release From Proactive Interference: Determinants of Performance and Neuropsychological Correlates* Christopher Randolph, James M. Gold, Constance J. Carpenter, Terry E. Goldberg, and Daniel R. Weinberger Clinical Brain Disorders Branch, IRP, NIMH, Washington DC
ABSTRACT The Wickens (1 970) modification of the Brown-Peterson short-term memory task has been used to investigate release from proactive interference (PI) in a number of memory-impaired groups. It has been suggested that failure to release from PI is observed only in patients with compromise of both memory and ‘frontal-lobe’ functions. The present study examined performance on this paradigm in patients with schizophrenia (SC), a neuropsychiatric disorder which typically includes both frontal and mnemonic impairments. Patients with SC were found to exhibit significantly less release from PI than normal controls. It was determined through correlational analyses that average Trial 1 performance on this task could predict performance on all subsequent trials, indicating that ‘release’ from PI may measure the same psychological process as the Brown-Peterson task, which does not include a release condition. Trial 1 performance in the SC group was correlated with a wide range of neuropsychological measures, but after the effect of full scale IQ was partialled out, only correlations with measures of memory and measures of frontallobe function remained significant. The results support previous formulations of the neuropsychological concomitants of release from PI, but suggest that failure to release on this paradigm may be secondary to a significant Compromise of the ability to perform the Brown-Peterson task. It is proposed that the experimental design constraints necessary to elicit a failure to release from PI in any patient group may limit the utility of this measure, and that Brown-Peterson performance may be a more reliable index of the neuropsychological functions involved.
Early investigations into the mechanisms of short-term memory included the concurrent finding by Brown (1958) and Peterson and Peterson (1959) that information from three-item strings could be lost over intervals of less than 20 s, provided that subjects were prevented from rehearsing the information during
* This work was partially supported by a grant from the National Alliance for Research on Schizophrenia and Depression to Dr. James Gold. Correspondence to: Christopher Randolph, Ph.D., Experimental Therapeutics Branch, National Institute of Neurological Disorders and Stroke, Bldg. 10, Room 5C104, 9000 Rockville Pike, Bethesda MD 20892 USA. Accepted for publication: January 6, 1992.
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
786
CHRISTOPHER RANDOLPH ET AL.
that time. This finding was interpreted by both investigators as evidence for a short-term memory trace that was subject to rapid decay and this paradigm has been termed the Brown-Peterson task. An alternative explanation was introduced by Keppel and Underwood (1962), who suggested that this loss was due to proactive interference (PI) from previous stimuli. They reported data indicating that normal subjects exhibited little or no forgetting on the first trial of this paradigm, but performed progressively worse across trials. Subsequently, Wickens, Born,and Allen (1963) reported that, if the class of materials employed as stimuli was switched after a few trials (e.g., from consonants to numbers), performance returned to the level of the first trial, indicating a ‘release from PI’. Wickens (1970) continued to use this modification of the Brown-Peterson paradigm to investigate which features of the verbal stimuli subjects were encoding in this task. Wickens concluded that subjects were sensitive to a wide variety of stimulus features, but that changes in semantic dimensions were most effective in producing build-up and release of PI. Most subsequent investigations using this paradigm have therefore employed word triads drawn from distinct taxonomic categories as stimuli. Typically, a subject will be presented four trials of word triads drawn from a single taxonomic category (e.g., animals), switching to a different category on the fifth trial (e.g., fruits). In a widely cited study, Cermak, Butters, and Moreines (1974) used this release from PI paradigm to investigate encoding in patients with alcoholic Korsakoffs syndrome (AK). They reported that the AK patients failed to release from PI on the taxonomic shift, and interpreted this failure as a semantic encoding impairment. Squire (1982) compared a group of AK patients to patients with other forms of amnesia on the release from PI paradigm, also using word mads with taxonomic category shifts. He concluded that, despite similar levels of impaired performance on a recognition memory test, AK patients were the only group that failed to release from PI. This was consideredto be a deficit unique to Korsakoff’s syndrome, and related to frontal-lobe damage superimposed on the basic memory disorder. Within the group of seven AK patients in this study, a rank-order correlation was computed between (1) a score representing the average of the amount of release from PI and performance on a separate recency judgement test; and (2) a score representing the average of performance on three putative tests of frontal-lobe function. A significant positive correlation was obtained, suggesting that poor performance on these three tests was associated with less release from PI and poorer performance at recency estimation. The three frontal-lobe tests were the Wisconsin Card Sort Test (WCST) (Milner, 1963), FAS word fluency (Benton, 1973), and the Embedded Figures Test (Corkin, 1979; Teuber & Weinstein, 1956). Unfortunately, separate correlations were not reported, making it unclear to what degree each of these tests was associated with release from PI. The association between failure to release from PI and ‘frontal’ deficits had been previously reported by Moscovitch (1982), who found a relationshipbetween failure to release from PI and performance on the WCST in patients with unilateral
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
RELEASE FROM PI
787
left frontal lobectomy. This study employed the Craik and Birtwistle (1971) paradigm, using 12-word lists, and noted that the patients who did poorly on the WCST were less likely to release than those who performed at or above normal levels. Freedman and Cermak (1986) compared the performance of AK patients on the Wickens release from PI paradigm to two groups of frontal-lobe-lesion patients. One frontal-lobe group had significant memory impairment, as measured by the Wechsler Memory Scale; the other group had preserved memory function. Both the AK patients and the ‘poor memory’ frontal group failed to release from :’I on the first block of trials, while the ‘good memory’ frontal group demonstrated release. Janowsky, Shimamura, Kritchevsky, and Squire (1989) used the Cermak et al. (1974) design to examine release from PI in patients with frontal-lobe lesions, AK patients, and non-Korsakoff‘s amnesic patients. Only the AK patients failed to exhibit release from PI, providing additional evidence that neither frontal or mnemonic impairments alone are sufficient to impair performance on this paradigm. The Wickens paradigm has also recently been used to study encoding in other patient groups. Beatty and Butters (1986) reported normal release from PI in a group of patients with Huntington’s disease, although Huber and Paulson (1987) reported failure to release in patients with advanced Huntington’s disease. Beatty, Goodkin, Beatty, and Monson (1989) reported normal release from PI in patients with multiple sclerosis who had memory impairment. Goldstein, Levin, and Boake (1989) reported normal release from PI in a group of head-injury patients. They also reported an association between VIQ, WCST performance, and the amount of release. Lower VIQ and more perseverative errors on the WCST were associated with less release from PI in these patients. When VIQ was partialled out, the relationship between WCST performance and release from PI in these patients became nonsignificant. Frontal-lobe lesion volume as measured by MRI failed to correlate with release from PI in this study, and the authors concluded that an association between the amount of release and frontal-lobe functioning was not particularly compelling. Taken together, these studies indicate that failure to release from PI is likely to be observed only in those patients with significant impairment on tests of both memory and frontal-lobe function. This is apparently the case with Korsakoff‘s patients, frontal-lobe-lesion patients with poor memory, and patients with advanced Huntington’s disease. Another neuropsychiatric disorder in which patients often exhibit significant compromise of both mnemonic and frontal-lobe functions is schizophrenia (SC) (Goldberg et al., 1990; Weinberger, in press: Weinberger, Berman, & Zec, 1986). Two previous studies have examined release from PI in patients with SC, with conflicting results. Traupmann, Berzofsky, and Kesselman (1976) reported significant release from PI in a study that lacked normal controls, while Kay (1982) found diminished release from PI in SC patients when compared to normal controls. The designs employed by these two studies were quite different;
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
788
CHRISTopHEjR RANDOLPH ET AL.
Traupmann et al. switched semantic categories after every third trial, while Kay used a four-trial design in which the encoding category was always switched on the fourth trial. One goal of the present study was to further investigate release from PI in schizophrenic patients, using the Wickens paradigm as employed by Cermak et al. (1974), Squire (1982), and Janowsky et al. (1989), in order to allow a more direct comparison to other patient groups. We had two additional goals. First, we wished to more completely characterize the neuropsychological determinants of performance on this task by examiningthe relationship between task performance and scores on a number of standardized neuropsychological tests. The final goal of the study was generated by an observation made in the course of reviewing the neuropsychological literature on release from PI. This observationwas that failure to release from PI on this task by patient groups was invariably accompanied by poorer Trial 1 performance than controls, which raised the possibility that initial level of performance on the test could be predictive of subsequent build-up and release of PI, If this were found to be the case, then release from PI might not be an independent phenomenon, and failure to release could simply be a by-product of extremely poor performance on the basic Brown-Peterson task. To test this hypothesis, correlational analyses were planned to determine whether or not average Trial 1 performance could predict performance on subsequent trials.
METHOD Subjects Sixteen SC patients and 15 normal control subjects participated in the experiment. Patients met DSM-III-R criteria for the diagnosis of schizophrenia and were research patients hospitalized at the NIMH Neuropsychiatric Research Hospital at the time of testing. All patients were tested while on standard clinical doses of neuroleptic medications; some patients also received adjunctive medications. Control subjects were paid volunteers from the housekeeping, secretarial, nursing, and nonprofessional technical staff. Controls were matched to SC patients on the basis of age, sex, and education. There were no significant differences between the groups on either years of education or academic achievement as measured by the Wide Range AchievementTest (WRAT) reading score (Jastak & Wilkinson, 1984). Demographic variables for both groups and neuropsychological data for the SC patients are listed in Table 1. Neuropsychological tests included the Wechsler Adult Intelligence Scale- Revised (WAIS-R; Wechsler. 1981); the Wechsler Memory ScaleRevised (WMS-R; Wechsler, 1987); the Mattis Dementia Rating Scale (MDRS; Mattis, 1973); FAS word fluency (Borkowski, Benton, & Spreen, 1967); and the Wisconsin Card Sorting Test (WCST). The WCST used here was a computerized version (see Weinberger et al., 1986) in which the full 128 cards were always presented, with scoring as per Heaton (1981).
Procedure The procedure generally followed that of Cermak et al. (1974) and Squire (1982). Subjects were presented with 8 blocks of 5 trials. Each trial consisted of the presentation of three words printed on an index card, followed by 15 s of color naming (Stroop Test; Golden, 1978). After the color naming distractor, subjects were requested to recall the previously
789
RELEASE FROM PI
Table 1. Subject Characteristics. Means and SDs.
SC Patients (n=16) 12M. 34.4 11.5 92.3
4F (9.1) (3.3) (15.7)
VIQ PIQ
87.5 90.4 86.2
(1 4.0) (1 5.0)
(11.7)
WMS-R general memory verbal memory visual memory delayed recall att.koncen.
83.6 87.5 82.1 86.5 78.6
(18.7) (16.5) (22.8) (21.8)* (20.5)
Mattis DRS
135.5
(5.9)
FAS fluency
37.9
(1 1.5)
WCST categories % persev. errs
3.6 21 .o
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
Sex Age Education WRAT Reading
Normal Controls (n=15) 11M, 32.5 13.3 89.8
4F (11.7) (1.9) (17.2)
WAiS- R
FSIQ
(3.2)* (W*
* (n=15) seen words. For four of the eight blocks, words on all trials belonged to a single taxonomic category. On the other four blocks, the taxonomic category was shifted on the fifth trial. Alternate forms were created such that the trial five shift (S) items for one subject were the trial five non-shift (NS) items for another, and remained in the same position within the sequence of blocks. One-minute rest periods were given between blocks. Order of blocks was S,NS,NS,S,NS,S,S,NSfor one form and NS,S,S,NS,S,NS,NS,Sfor the other. Taxonomic categories were selected from the Battig and Montague (1969) norms, and included musical instruments, occupations, vegetables, insects, house parts, clothing items, birds, spices, animals, fish, sports, and body parts.
RESULTS AND DISCUSSION
Overall Performance Total recall was calculated for each subject on the shift and non-shift trials across each set of four blocks. Figure 1 depicts performance for both groups on both shift and non-shift trials. The control subjects recalled significantly more words on every trial under both conditions than did the patients (univariate ANOVAs, all p’s < .05). The data
790
CHRISTOPHER RANDOLPH ET AL.
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
1
NON-SWITCH BLOCKS
,
SWITCH BLOCKS
I "\ 1
2
3
4
TRIALS
5
1
2
3
4
5
TRIALS
Fig. 1. Performance of schizophrenic patients (SC) and normal controls (NC) on the release from PI task. There were significant main effects for group and trial on each condition,and a significantgroup by trial interactionon the switch condition
(see text).
were also analyzed with repeated measures ANOVAs. Over the non-shift trials, there was a significant main effect of group, F(1,29) = 15.2,p c .0005 and trial, F(4,116) = 19.4,p c .0001, but no group by trial interaction. On the shift trials, however, in addition to the main effects of group [F(1,29) = 13.0, p c .002] and trial [F(4,116) = 30.2, p < .0001], there was a significant group by trial interaction, F(4,116) = 4.0, p c .005. In order to interpret this significant interaction term, given the consistently significant group differences across trials, recall scores for each of the trials were expressed as percentage of Trial 1 recall for each subject, and individual ANOVAs were calculated in order to determine which trials were the major contributors to the interaction term. It should be noted that these ANOVAs involve proportional changes in performance, rather than the absolute changes analyzed repeated measures ANOVA. Figure 2 depicts recall on Trials 2 through 5 as a percentage of Trial 1 performance. The SC patients recalled fewer words on Trials 2 through 5 as a percentage of their Trial 1 recall than did the controls, and this difference reached statistical significance on Trial 3 [F(1,29)= 10.4,p < .004] and Trial 5 [F(1,29) = 11.0,p < .003].
791
RELEASE FROM PI
h
n cn+
-
100-
-1
-1
a
QO-
a
80-
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
Y
T
r
-1
5
70-
a I-
u.
60-
0
u a w
5040-
0
w
30-
n 2010-
TRIAL 2
TRIAL 3
a : NCvsSC
TRIAL 4
TRIAL 5 (SHIFT)
p
ISSN: 0168-8634 (Print) (Online) Journal homepage: http://www.tandfonline.com/loi/ncen19
Release from proactive interference: Determinants of performance and neuropsychological correlates Christopher Randolph , James M. Gold , Constance J. Carpenter , Terry E. Goldberg & Daniel R. Weinberger To cite this article: Christopher Randolph , James M. Gold , Constance J. Carpenter , Terry E. Goldberg & Daniel R. Weinberger (1992) Release from proactive interference: Determinants of performance and neuropsychological correlates, Journal of Clinical and Experimental Neuropsychology, 14:5, 785-800, DOI: 10.1080/01688639208402863 To link to this article: http://dx.doi.org/10.1080/01688639208402863
Published online: 04 Jan 2008.
Submit your article to this journal
Article views: 19
View related articles
Citing articles: 13 View citing articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ncen20 Download by: [Nanyang Technological University]
Date: 10 November 2015, At: 03:05
Journal of Clinical and Experimental Neuropsychology 1992, Vol. 14, NO.5, pp. 785-800
0168-8634/92/1405-0785$3.00 0 Swets & Zeitlinger
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
Release From Proactive Interference: Determinants of Performance and Neuropsychological Correlates* Christopher Randolph, James M. Gold, Constance J. Carpenter, Terry E. Goldberg, and Daniel R. Weinberger Clinical Brain Disorders Branch, IRP, NIMH, Washington DC
ABSTRACT The Wickens (1 970) modification of the Brown-Peterson short-term memory task has been used to investigate release from proactive interference (PI) in a number of memory-impaired groups. It has been suggested that failure to release from PI is observed only in patients with compromise of both memory and ‘frontal-lobe’ functions. The present study examined performance on this paradigm in patients with schizophrenia (SC), a neuropsychiatric disorder which typically includes both frontal and mnemonic impairments. Patients with SC were found to exhibit significantly less release from PI than normal controls. It was determined through correlational analyses that average Trial 1 performance on this task could predict performance on all subsequent trials, indicating that ‘release’ from PI may measure the same psychological process as the Brown-Peterson task, which does not include a release condition. Trial 1 performance in the SC group was correlated with a wide range of neuropsychological measures, but after the effect of full scale IQ was partialled out, only correlations with measures of memory and measures of frontallobe function remained significant. The results support previous formulations of the neuropsychological concomitants of release from PI, but suggest that failure to release on this paradigm may be secondary to a significant Compromise of the ability to perform the Brown-Peterson task. It is proposed that the experimental design constraints necessary to elicit a failure to release from PI in any patient group may limit the utility of this measure, and that Brown-Peterson performance may be a more reliable index of the neuropsychological functions involved.
Early investigations into the mechanisms of short-term memory included the concurrent finding by Brown (1958) and Peterson and Peterson (1959) that information from three-item strings could be lost over intervals of less than 20 s, provided that subjects were prevented from rehearsing the information during
* This work was partially supported by a grant from the National Alliance for Research on Schizophrenia and Depression to Dr. James Gold. Correspondence to: Christopher Randolph, Ph.D., Experimental Therapeutics Branch, National Institute of Neurological Disorders and Stroke, Bldg. 10, Room 5C104, 9000 Rockville Pike, Bethesda MD 20892 USA. Accepted for publication: January 6, 1992.
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
786
CHRISTOPHER RANDOLPH ET AL.
that time. This finding was interpreted by both investigators as evidence for a short-term memory trace that was subject to rapid decay and this paradigm has been termed the Brown-Peterson task. An alternative explanation was introduced by Keppel and Underwood (1962), who suggested that this loss was due to proactive interference (PI) from previous stimuli. They reported data indicating that normal subjects exhibited little or no forgetting on the first trial of this paradigm, but performed progressively worse across trials. Subsequently, Wickens, Born,and Allen (1963) reported that, if the class of materials employed as stimuli was switched after a few trials (e.g., from consonants to numbers), performance returned to the level of the first trial, indicating a ‘release from PI’. Wickens (1970) continued to use this modification of the Brown-Peterson paradigm to investigate which features of the verbal stimuli subjects were encoding in this task. Wickens concluded that subjects were sensitive to a wide variety of stimulus features, but that changes in semantic dimensions were most effective in producing build-up and release of PI. Most subsequent investigations using this paradigm have therefore employed word triads drawn from distinct taxonomic categories as stimuli. Typically, a subject will be presented four trials of word triads drawn from a single taxonomic category (e.g., animals), switching to a different category on the fifth trial (e.g., fruits). In a widely cited study, Cermak, Butters, and Moreines (1974) used this release from PI paradigm to investigate encoding in patients with alcoholic Korsakoffs syndrome (AK). They reported that the AK patients failed to release from PI on the taxonomic shift, and interpreted this failure as a semantic encoding impairment. Squire (1982) compared a group of AK patients to patients with other forms of amnesia on the release from PI paradigm, also using word mads with taxonomic category shifts. He concluded that, despite similar levels of impaired performance on a recognition memory test, AK patients were the only group that failed to release from PI. This was consideredto be a deficit unique to Korsakoff’s syndrome, and related to frontal-lobe damage superimposed on the basic memory disorder. Within the group of seven AK patients in this study, a rank-order correlation was computed between (1) a score representing the average of the amount of release from PI and performance on a separate recency judgement test; and (2) a score representing the average of performance on three putative tests of frontal-lobe function. A significant positive correlation was obtained, suggesting that poor performance on these three tests was associated with less release from PI and poorer performance at recency estimation. The three frontal-lobe tests were the Wisconsin Card Sort Test (WCST) (Milner, 1963), FAS word fluency (Benton, 1973), and the Embedded Figures Test (Corkin, 1979; Teuber & Weinstein, 1956). Unfortunately, separate correlations were not reported, making it unclear to what degree each of these tests was associated with release from PI. The association between failure to release from PI and ‘frontal’ deficits had been previously reported by Moscovitch (1982), who found a relationshipbetween failure to release from PI and performance on the WCST in patients with unilateral
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
RELEASE FROM PI
787
left frontal lobectomy. This study employed the Craik and Birtwistle (1971) paradigm, using 12-word lists, and noted that the patients who did poorly on the WCST were less likely to release than those who performed at or above normal levels. Freedman and Cermak (1986) compared the performance of AK patients on the Wickens release from PI paradigm to two groups of frontal-lobe-lesion patients. One frontal-lobe group had significant memory impairment, as measured by the Wechsler Memory Scale; the other group had preserved memory function. Both the AK patients and the ‘poor memory’ frontal group failed to release from :’I on the first block of trials, while the ‘good memory’ frontal group demonstrated release. Janowsky, Shimamura, Kritchevsky, and Squire (1989) used the Cermak et al. (1974) design to examine release from PI in patients with frontal-lobe lesions, AK patients, and non-Korsakoff‘s amnesic patients. Only the AK patients failed to exhibit release from PI, providing additional evidence that neither frontal or mnemonic impairments alone are sufficient to impair performance on this paradigm. The Wickens paradigm has also recently been used to study encoding in other patient groups. Beatty and Butters (1986) reported normal release from PI in a group of patients with Huntington’s disease, although Huber and Paulson (1987) reported failure to release in patients with advanced Huntington’s disease. Beatty, Goodkin, Beatty, and Monson (1989) reported normal release from PI in patients with multiple sclerosis who had memory impairment. Goldstein, Levin, and Boake (1989) reported normal release from PI in a group of head-injury patients. They also reported an association between VIQ, WCST performance, and the amount of release. Lower VIQ and more perseverative errors on the WCST were associated with less release from PI in these patients. When VIQ was partialled out, the relationship between WCST performance and release from PI in these patients became nonsignificant. Frontal-lobe lesion volume as measured by MRI failed to correlate with release from PI in this study, and the authors concluded that an association between the amount of release and frontal-lobe functioning was not particularly compelling. Taken together, these studies indicate that failure to release from PI is likely to be observed only in those patients with significant impairment on tests of both memory and frontal-lobe function. This is apparently the case with Korsakoff‘s patients, frontal-lobe-lesion patients with poor memory, and patients with advanced Huntington’s disease. Another neuropsychiatric disorder in which patients often exhibit significant compromise of both mnemonic and frontal-lobe functions is schizophrenia (SC) (Goldberg et al., 1990; Weinberger, in press: Weinberger, Berman, & Zec, 1986). Two previous studies have examined release from PI in patients with SC, with conflicting results. Traupmann, Berzofsky, and Kesselman (1976) reported significant release from PI in a study that lacked normal controls, while Kay (1982) found diminished release from PI in SC patients when compared to normal controls. The designs employed by these two studies were quite different;
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
788
CHRISTopHEjR RANDOLPH ET AL.
Traupmann et al. switched semantic categories after every third trial, while Kay used a four-trial design in which the encoding category was always switched on the fourth trial. One goal of the present study was to further investigate release from PI in schizophrenic patients, using the Wickens paradigm as employed by Cermak et al. (1974), Squire (1982), and Janowsky et al. (1989), in order to allow a more direct comparison to other patient groups. We had two additional goals. First, we wished to more completely characterize the neuropsychological determinants of performance on this task by examiningthe relationship between task performance and scores on a number of standardized neuropsychological tests. The final goal of the study was generated by an observation made in the course of reviewing the neuropsychological literature on release from PI. This observationwas that failure to release from PI on this task by patient groups was invariably accompanied by poorer Trial 1 performance than controls, which raised the possibility that initial level of performance on the test could be predictive of subsequent build-up and release of PI, If this were found to be the case, then release from PI might not be an independent phenomenon, and failure to release could simply be a by-product of extremely poor performance on the basic Brown-Peterson task. To test this hypothesis, correlational analyses were planned to determine whether or not average Trial 1 performance could predict performance on subsequent trials.
METHOD Subjects Sixteen SC patients and 15 normal control subjects participated in the experiment. Patients met DSM-III-R criteria for the diagnosis of schizophrenia and were research patients hospitalized at the NIMH Neuropsychiatric Research Hospital at the time of testing. All patients were tested while on standard clinical doses of neuroleptic medications; some patients also received adjunctive medications. Control subjects were paid volunteers from the housekeeping, secretarial, nursing, and nonprofessional technical staff. Controls were matched to SC patients on the basis of age, sex, and education. There were no significant differences between the groups on either years of education or academic achievement as measured by the Wide Range AchievementTest (WRAT) reading score (Jastak & Wilkinson, 1984). Demographic variables for both groups and neuropsychological data for the SC patients are listed in Table 1. Neuropsychological tests included the Wechsler Adult Intelligence Scale- Revised (WAIS-R; Wechsler. 1981); the Wechsler Memory ScaleRevised (WMS-R; Wechsler, 1987); the Mattis Dementia Rating Scale (MDRS; Mattis, 1973); FAS word fluency (Borkowski, Benton, & Spreen, 1967); and the Wisconsin Card Sorting Test (WCST). The WCST used here was a computerized version (see Weinberger et al., 1986) in which the full 128 cards were always presented, with scoring as per Heaton (1981).
Procedure The procedure generally followed that of Cermak et al. (1974) and Squire (1982). Subjects were presented with 8 blocks of 5 trials. Each trial consisted of the presentation of three words printed on an index card, followed by 15 s of color naming (Stroop Test; Golden, 1978). After the color naming distractor, subjects were requested to recall the previously
789
RELEASE FROM PI
Table 1. Subject Characteristics. Means and SDs.
SC Patients (n=16) 12M. 34.4 11.5 92.3
4F (9.1) (3.3) (15.7)
VIQ PIQ
87.5 90.4 86.2
(1 4.0) (1 5.0)
(11.7)
WMS-R general memory verbal memory visual memory delayed recall att.koncen.
83.6 87.5 82.1 86.5 78.6
(18.7) (16.5) (22.8) (21.8)* (20.5)
Mattis DRS
135.5
(5.9)
FAS fluency
37.9
(1 1.5)
WCST categories % persev. errs
3.6 21 .o
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
Sex Age Education WRAT Reading
Normal Controls (n=15) 11M, 32.5 13.3 89.8
4F (11.7) (1.9) (17.2)
WAiS- R
FSIQ
(3.2)* (W*
* (n=15) seen words. For four of the eight blocks, words on all trials belonged to a single taxonomic category. On the other four blocks, the taxonomic category was shifted on the fifth trial. Alternate forms were created such that the trial five shift (S) items for one subject were the trial five non-shift (NS) items for another, and remained in the same position within the sequence of blocks. One-minute rest periods were given between blocks. Order of blocks was S,NS,NS,S,NS,S,S,NSfor one form and NS,S,S,NS,S,NS,NS,Sfor the other. Taxonomic categories were selected from the Battig and Montague (1969) norms, and included musical instruments, occupations, vegetables, insects, house parts, clothing items, birds, spices, animals, fish, sports, and body parts.
RESULTS AND DISCUSSION
Overall Performance Total recall was calculated for each subject on the shift and non-shift trials across each set of four blocks. Figure 1 depicts performance for both groups on both shift and non-shift trials. The control subjects recalled significantly more words on every trial under both conditions than did the patients (univariate ANOVAs, all p’s < .05). The data
790
CHRISTOPHER RANDOLPH ET AL.
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
1
NON-SWITCH BLOCKS
,
SWITCH BLOCKS
I "\ 1
2
3
4
TRIALS
5
1
2
3
4
5
TRIALS
Fig. 1. Performance of schizophrenic patients (SC) and normal controls (NC) on the release from PI task. There were significant main effects for group and trial on each condition,and a significantgroup by trial interactionon the switch condition
(see text).
were also analyzed with repeated measures ANOVAs. Over the non-shift trials, there was a significant main effect of group, F(1,29) = 15.2,p c .0005 and trial, F(4,116) = 19.4,p c .0001, but no group by trial interaction. On the shift trials, however, in addition to the main effects of group [F(1,29) = 13.0, p c .002] and trial [F(4,116) = 30.2, p < .0001], there was a significant group by trial interaction, F(4,116) = 4.0, p c .005. In order to interpret this significant interaction term, given the consistently significant group differences across trials, recall scores for each of the trials were expressed as percentage of Trial 1 recall for each subject, and individual ANOVAs were calculated in order to determine which trials were the major contributors to the interaction term. It should be noted that these ANOVAs involve proportional changes in performance, rather than the absolute changes analyzed repeated measures ANOVA. Figure 2 depicts recall on Trials 2 through 5 as a percentage of Trial 1 performance. The SC patients recalled fewer words on Trials 2 through 5 as a percentage of their Trial 1 recall than did the controls, and this difference reached statistical significance on Trial 3 [F(1,29)= 10.4,p < .004] and Trial 5 [F(1,29) = 11.0,p < .003].
791
RELEASE FROM PI
h
n cn+
-
100-
-1
-1
a
QO-
a
80-
Downloaded by [Nanyang Technological University] at 03:05 10 November 2015
Y
T
r
-1
5
70-
a I-
u.
60-
0
u a w
5040-
0
w
30-
n 2010-
TRIAL 2
TRIAL 3
a : NCvsSC
TRIAL 4
TRIAL 5 (SHIFT)
p
