Impairments in prospective and retrospective memory following stroke

Authors: Hyun Jung Kim ^a;  Fergus I. M. Craik ^b;  Lin Luo ^b; Jon Erik Ween ^c

Affiliations:	a Department of Rehabilitation Medicine, Eulji General Hospital, Eulji University School of Medicine, Seoul, Republic of Korea
	b Rotman Research Institute, Baycrest and Department of Psychology, University of Toronto, Toronto, Canada
	c Posluns Centre for Stroke and Cognition, Kunin-Lunenfeld Applied Research Unit, Baycrest, Toronto, Canada

DOI: 10.1080/13554790802709039

Publication Frequency: 6 issues per year

Published in:  Neurocase, Volume 15, Issue 2 April 2009 , pages 145 - 156

First Published: April 2009

Abstract

Prospective memory (PM) is the ability to carry out a planned intention at a future time. We studied PM deficits in a group of community-dwelling stroke survivors compared with normal controls. Twelve stroke patients and 12 matched controls performed a series of tests assessing executive function, prospective (PM) and retrospective memory (RM). Patients performed less well than controls on laboratory measures of PM and associative RM; they also showed deficits on standard tests of RM and executive control. The groups did not differ on more structured clinical measures of executive function, RM, PM or self-rated PM and RM. The results are discussed in terms of an impairment in the ability to 'self-initiate' effortful cognitive processes.

Keywords: Stroke; Prospective memory; Cognition; Executive function; Retrospective memory

INTRODUCTION

Prospective memory (PM), the ability to plan future actions and execute them successfully, is a central part of daily living (Cohen, 1996; Kliegel & Martin, 2003; Martin & McDaniel, 2003). PM function has been postulated to depend heavily on frontal lobe areas (Burgess, 2000; Goldstein, Bernard, Fenwick, Burgess, & McNeil, 1993;McDaniel, Glisky, Rubin, Guynn, & Routhieaux, 1999; Okuda et al., 1998; Shallice & Burgess, 1991; West & Covell, 2001; West, Jakubek, & Wymbs, 2002) and has been studied extensively in normal aging (Cohen, West, & Craik, 2001; Einstein & McDaniel, 1990; Henry, MacLeod, Phillips, & Crawford, 2004; Rendell & Craik, 2000) and in traumatic brain injury (Kliegel, Eschen, & Thone-Otto, 2004; Mathias & Mansfield, 2005; Schmitter-Edgecombe & Wright, 2004). PM deficits following stroke (Brooks, Rose, Potter, Jayawardena, & Morling, 2004) have been studied much less frequently, despite the high prevalence of stroke in the adult population (Hachinski, 2007) and the significant impact of stroke on frontal-executive functions (Reed, Eberling, Mungas, Weiner, & Jagust, 2001; Reed et al., 2004).

PM requires the formation of an intention to act following a delay filled with other activities. Typically, the planned action is triggered either by an event (e.g., 'When you meet James, give him this information') or by a specified time (e.g., 'Please phone home at 4:30 pm'). 'Time-based PM' is typically more difficult than 'event-based PM', since no explicit cue is given in the time-based case (Einstein & McDaniel, 1990; Kvavilashvili & Ellis, 1996), with the consequent need for more effortful 'self-initiated processing' (Craik, 1983, 1986; Einstein, McDaniel, Richardson, Guynn, & Cunfer, 1995). Successful PM requires both that the person 'remembers to remember' - the PM component of the overall performance - and also recollects what is to be done. This latter aspect is assumed to involve episodic or 'retrospective' memory (RM). Dissociation of the PM and RM components have been clearly established (Cohen et al., 2001; Kliegel & Martin, 2003; West & Krompinger, 2005).

Studies of PM have utilized laboratory tasks, naturalistic observations and self-rated questionnaires. In the present study, we used all three methods of assessment. We used two laboratory tasks. The first was the Virtual Week (VW) task developed by Rendell and Craik (2000) as a naturalistic laboratory task. VW uses a board game to provide attentional load and occasions for PM cues (times and events). It also allows the embedding of a realistic time-based PM task, and provides a reasonably high number of measurable events for statistical power. The second was the Memory for Intentions (MI) task (Cohen et al., 2001) in which participants are given a series of real-life intentions (e.g., 'make a doctor's appointment') to associate with a series of pictures. When the pictures are re-presented later, embedded in a standard memory task, the subject must identify the original pictures as cues, and recall the associated intentions. In addition to these tasks, we used the Remembering a Belonging subtest from the Rivermead Behavioural Memory Test (RBMT) (Wilson, Cockburn, & Baddeley, 1985) which is a standard clinical event-based PM task. We also used the Prospective and Retrospective Memory Questionnaire (PRMQ) (Crawford, Smith, Maylor, Della Sala, & Logie, 2003) - a 16-item questionnaire measuring self-rated memory failures as a way to quantify an individual's memory complaints. Finally, we also administered a short battery of standard neuropsychological tests, as described in the Methods section.

Recent studies have sharpened the general concept of self-initiation by suggesting that successful performance of PM tasks typically involves the maintenance of an intention or a strategic monitoring set in the interval preceding the PM cue or designated time (Guynn, 2003). This maintenance consumes attentional resources (Smith, 2003), and the ability to maintain the intention to respond appropriately activates brain regions in the frontal pole, lateral frontal, inferior parietal and precuneus (Burgess, Quayle, & Frith, 2001). We hypothesized that PM tasks - particularly time-based tasks - require substantial degrees of self-initiated processing (Craik, 1983) and that patients with anterior lesions should show deficits in tasks requiring such self-initiated processing (Burgess & Shallice, 1996). Our sample of post-stroke patients had lesions that were predominantly anterior (see Subjects) so we had some expectation that they would show PM deficits in at least some tests, despite the fact that they were now living normal lives in the community and reporting no particular cognitive difficulties. We chose patients with high levels of post-stroke functional ability to minimize confounding effects of other cognitive impairments as much as possible.

A second, more exploratory point, concerns the relations between PM and RM - episodic memory for past events. The current literature is divided on the point of whether PM is simply an aspect of RM (Crowder, 1996; Roediger, 1996) or draws on different processes. A recent study of an individual with multiple sclerosis (West, McNerney & Krauss, 2007) reported that the person demonstrated poor performance on a number of PM tasks despite showing superior levels of intelligence and average to superior RM. In this case, at least, it appears that RM was intact whereas PM was negatively affected. The point at issue in the present investigation is whether post-stroke patients will show deficits in both PM and RM, or in one type of memory task and not in the other. Such data could contribute to the debate on the similarity or otherwise of PM and RM. A final purpose of the study was to compare levels of PM functioning assessed by questionnaire and by a simple clinical test on the one hand, with PM performance measured on laboratory tests on the other hand. The hope was that both types of test would yield the same answer, but it also seemed possible that the laboratory measures would be more sensitive, in which case they could be used diagnostically to complement existing clinical batteries.

The aim of the present study was therefore to conduct a comprehensive assessment of how PM is affected by stroke in non-demented chronic stroke survivors who dwell independently in the community. Due to the complex nature of PM performance, we wished to explore the effects of stroke on each type (i.e., event-based, time-based) and component (i.e., PM and RM components) of PM. The two laboratory tasks were used to give separate measures for different types and components of PM. The two neuropsychological assessments (Rivermead and Questionnaire) were included to give a general measure of overall PM functioning. For independently-dwelling stroke patients, a contrast between laboratory task measures and general neuropsychological measures of PM might reveal subtle impairments in stroke patients that are not readily detected by existing assessment tools.

METHOD

 

Subjects

 

Twelve stroke patients and 12 neurologically normal control subjects were included in this study. All stroke patients were recruited from the outpatient Stroke and Cognition Clinic at the Brain Health Centre at Baycrest or the patient database at Baycrest's Rotman Research Institute. The inclusion criteria were: (1) diagnosis of stroke based on clinical assessment and imaging (CT, MRI), (2) fluent English speaker, (3) independent in activities of daily living (ADLs) and living at home, and (4) willing to participate.

The average length of time since stroke was 2.99 years (range 4.8 months to 9 years). There were five patients with right and five with left hemisphere lesions, and two with bilateral lesions. All patients had received either an MRI or a CT scan. The image processing was carried out by a stroke neurologist (JW) with expertise in lesion analysis. All scans were visually rated according to Age Related White Matter Changes (ARWMC) reported to be comparable for CT and MRI (Wahlund et al., 2001). The ARWMC uses a 3-point scale (0 = normal, 1 = isolated white matter hyperintensities, 2 = near confluence of lesions, 3 = confluent lesions) in each of 10 compartments, 5 right and 5 left. The areas are frontal, parieto-occipital, temporal, basal ganglia and infratentorial. Table 1 shows the classification of lesion severity for all 12 patients in each of the 10 brain regions. The majority of lesions are subcortical and in frontal, basal ganglia and parieto-occipital areas. Given the intimate connections between frontal and basal ganglia regions via frontal-striatal circuits (Cummings, 1993), the patients may be described in general terms as having 'frontal' lesions.

TABLE 1 White matter lesion intensity (see text for explanation) for each patient as a function of brain region

	Brain Region
Patient ID	RF	LF	RBG	LBG	RT	LT	RPO	LPO	RIT	LIT
Notes: R and L, right and left respectively; F, frontal; PO, parieto-occipital; T, temporal; BG, basal ganglia; IT, infratentorial.
P1	2	0	0	0	0	0	0	0	0	0
P2	0	1	0	1	0	0	0	0	0	0
P3	0	2	0	1	0	0	0	0	0	0
P4	1	3	1	2	0	0	1	1	0	0
P5	2	2	0	0	0	0	0	1	0	0
P6	3	3	3	3	3	3	3	3	0	0
P7	0	0	3	0	0	0	0	0	0	0
P8	1	0	2	0	0	0	0	0	0	0
P9	1	2	0	3	0	0	0	1	0	0
P10	0	0	0	0	0	3	0	0	0	0
P11	3	3	3	3	0	0	2	2	0	0
P12	2	2	1	1	0	0	2	2	0	0

The exclusion criteria were (1) dementia diagnosed from the medical history or from the Clinical Dementia Rating Scale (Morris, 1993) score of > 0.5; (2) history of alcohol or substance abuse within 5 years of testing, (3) history of head trauma with loss of consciousness, (4) significant other neurological or psychiatric disorders, (5) medications (other than selective serotonin reuptake inhibitors) that affect cognitive function (6) serious, unstable medical illness or neurological impairment that would limit participation or completion of the protocol. Control subjects (n = 12) were recruited from the Rotman Research Institute database. They were all community-dwelling individuals with no history of any neurological or psychiatric conditions. The two groups were matched for age and education (Table 2). The Ethics Committee of Baycrest approved this experiment and each participant gave informed consent.

TABLE 2 Demographics, clinical characteristics and neuropsychological profiles for subject groups

	Patients		Controls		t-Tests
	M	SD	M	SD	t	df	p
Values reported are raw scores. M, mean, SD, standard deviation, df, degrees of freedom. Two-tailed t-tests are reported. See methods section for abbreviations.
Age (years)	69.33	7.02	69.08	4.94	0.10	19.75	.92
Education (years)	14.42	3.73	15.67	3.23	-0.88	21.56	.39
MMSE	28.67	1.07	29.00	1.13	-0.74	21.95	.47
Trail Making Time
A (s)	39.92	13.85	33.83	8.91	1.28	18.77	.22
B (s)	90.00	33.34	78.92	20.92	0.98	18.50	.34
Verbal Fluency
FAS (words)	35.67	10.05	45.25	9.37	-2.42	21.89	.02
Animal (words)	14.17	4.69	18.42	6.61	-1.82	19.83	.08
SART
Reaction Time (ms)	367.33	109.59	438.97	60.54	-1.98	17.14	.06
Error of Commission	10.92	7.19	4.58	2.87	2.83	14.43	.01
Error of Omission	7.42	14.06	3.50	3.68	0.93	12.50	.37
Verbal Paired Associates
Total Recall (words)	9.83	1.85	11.08	3.32	-1.14	17.25	.27
CVLT
Total learning score (1-5)	35.50	10.78	50.75	7.62	-4.08	19.12	.00
Long delay free recall	6.58	3.09	10.75	2.05	-3.89	19.12	.00
Recognition Hits - False alarms	8.50	5.30	11.92	3.87	-1.80	20.14	.09
R-SAT
% Easy Items Completed	90	10	84	17	1.11	17.07	.28

Neuropsychological profile

 

A battery of standard neuropsychological measures was administered to participants to measure executive and immediate memory abilities. First, basic demographics such as age, gender and educational level were recorded. The Mini-mental State Examination (MMSE; Folstein, Folstein, & McHugh, 1975) was administered to measure global cognitive function. The Trail Making Tests A & B (Reitan, 1986) were given to assess speed and cognitive flexibility: The measures of performance taken were the time taken to complete Trails A and Trails B, respectively. Lexical and category verbal fluency abilities were assessed by the Verbal Fluency phonemic (FAS) test and the semantic categories (animal) test (Lezak, 1995). The measures of performance here are the total numbers of items produced for the FAS test and the animal naming test, respectively. A measure of attention and inhibitory capacity was provided by the Sustained Attention to Response Task (SART) (Robertson, Manly, Andrade, Baddeley, & Yiend, 1997). Performance was measured by the mean reaction time (RT), and the number of errors of commission and errors of omission. Memory abilities were assessed by two tests; first the Verbal Paired Associates (VPA) test from the Wechsler Memory Scale-Third Edition (Wechsler, 1997) and second the California Verbal Learning Test-Second Edition (CVLT-II, Delis DC, Kaplan, & Ober, 2000). The measures we took included total correct recall from VPA, and T-score, long delay free recall performance and recognition hits from CVLT-II. Finally, the Revised Strategy Application Test (R-SAT) was administered for the assessment of strategy function (Levine, Dawson, Boutet, Schwartz, & Stuss, 2000; Stuss & Levine, 2002). The main measure of performance was the proportion of easy items completed out of the total number of items in the test. High percentages indicate more efficient strategies.

The characteristics of the study groups are presented in Table 2. The groups did not differ in terms of age and education, but the proportion of men was higher in the stroke sample (Patient: M = 8, F = 4; Control: M = 3; F = 9). The scores on the MMSE did not differ significantly between two groups. Control participants generated more items in both the FAS test and the animal names test, although only the former reached statistical significance. A marginally significant difference between groups was found on SART reaction times (patients were faster than controls), and a significant difference was found on errors of commission (patients made more errors than controls). The stroke patients scored significantly lower on the measures of CVLT-II learning and long-delay free recall but not on recognition memory, as has been noted as a pattern characteristic of vascular cognitive impairment (Roman et al., 2004).

Prospective memory tasks

 

Virtual Week

 

Virtual Week (VW) is a board game developed to assess PM (Rendell & Craik, 2000). Each circuit of the game represents 1 day from 7 am to 10 pm and participants move around the board with the roll of a die. They have to remember to carry out activities either at specified 'times' (as they pass the relevant square on the board) or in response to specified 'events' that are given on event cards picked up in the course of the circuit. The PM tasks were either 'Regular' (the same tasks on all circuits) or 'Irregular' (different tasks on each circuit). The same four 'Regular' tasks were performed on every circuit of the board and were given in instructions before the game started; two were in response to events (taking antibiotics at breakfast and dinner) and two were in response to squares indicating virtual times on the board (taking asthma medication at 11 am and 9 pm). Irregular tasks also included two tasks in response to events and two other tasks in response to virtual time squares. These tasks are however different for each circuit and were given by the 'start card' at the beginning of each circuit. As participants moved around the board, they were required to pick up 'event cards' as they passed 'event squares', make choices about daily activities (e.g., a choice of foods for lunch) and remember to carry out the required activities. All the activities written on cards indicated things that people normally do in a typical day. Participants carried out the specified actions simply by telling the experimenter (e.g., 'The event card tells me that my friend Jane phoned, so I remember to ask her to lunch next week' or 'It is now 4 pm so I remember that I should phone the doctor's office'). The third type of PM task was a time-check task in which the participant had to indicate to the tester when 2 min 30 s and also 4 min 15 s had elapsed from the start of the current circuit, supposedly 'to check virtual lung capacity'. This 'real-life' time was given by a stopwatch placed on the desk facing down and participants were informed that they could check the stopwatch at any time. Therefore, 10 PM tasks were assessed on each day (complete circuit around the board). Each group of 10 consisted of four regular, four irregular and two time-check tasks.

The point of these different tasks was to assess PM for different levels of habit and environmental support. Thus, the regular tasks should be easiest as they occurred predictably on every circuit; the irregular tasks were less habitual as they changed on each circuit; finally, the time-check task required the participant to break out of the board-game mode and remember to check the actual time as given by the stopwatch. Previous results (Rendell & Craik, 2000) have generally confirmed this order of difficulty. Participants were encouraged to tell the examiner the required action later if they failed to carry out the intention at the designated time or event but remembered it later; they were also asked to report if they remembered that they had some task to perform, but were unable to retrieve the content of the task. Although Rendell and Craik attempted to capture the notion of time-based PM in the Virtual Week task by asking participants to respond at various 'times of day', these 'times' are signaled clearly on the board, and are thus not too different from events. Accordingly, and to simplify analyses, the time-based and event-based responses were combined into one measure of PM in the present study. True time-based PM ability is better represented by performance in the clock-checking task.

In this study, participants completed three virtual days per session, and the mean measures over the three circuits were taken as scores. Responses were categorized as 'Correct' or 'Incorrect' on each task. A 'Correct' response on 'Regular' or 'Irregular' trials indicated that the participant had remembered to state the required action before the next roll of the die. A 'Correct' response for the time-check task was within 10 seconds of the target time. Thus, 'Correct' responses indicated correct recognition of the prospective cue and correct recall of the intention. The experimenter recorded types of error as 'Wrong', 'Late', or 'Miss'; 'Wrong' responses indicated that recall of the content was incorrect or the action was recalled simply as 'something' at the correct time. 'Late' responses indicated that recall of the correct content occurred at the incorrect time of the virtual day. A 'Miss' response indicated that the participant failed to recall the target at any time.

Memory for intentions

 

The Memory for Intentions (MI) task developed by Cohen et al. (2001) was modified by constructing a shorter version. Participants were presented with a series of 48 pictures, each paired with a word related to the picture; their task was to remember the pair for a later memory test. Each picture-word pair was presented for 6 sec. This series served as the background 'ongoing activity' for the PM task proper. During this learning phase, participants were shown a further series of 16 pictures, each of which also had a related word to be learned, but had in addition an intended action to be remembered and stated at a later stage in the experiment. These 16 pictures with their accompanying intentions and paired-associate words were each shown for 12 seconds, and were presented randomly intermixed with the 48-item paired associate task. There were 8 intentions in each of two categories (1) highly related and (2) somewhat related to the activity shown in the accompanying picture, and subjects were asked to form the intentions as they would use memory strategies in everyday life. For example, if the intention was 'I need to phone a friend', a highly related cue might be a picture of a person using a telephone; a somewhat related cue might be a picture of two friends hugging. We manipulated the degree of semantic relatedness in order to investigate the efficiency of the prospective and retrospective components of PM in a stroke population. In their aging study, Cohen et al. (2001) concluded that the efficiency of the retrospective component (that is, the ability to recollect the intention when shown the picture at a later time) was influenced by conceptually driven processes, and when the picture and intention were semantically related the efficiency of the retrospective component was similar for older and younger adults.

During the test phase, participants were shown the series of 64 studied pictures plus 16 new pictures as distracters. Participants were instructed to recall the word initially paired with the picture, or to say 'new' if they did not recognize the picture. Additionally, participants were instructed to recall the intentions associated with the 16 PM pictures, which were presented randomly in the test sequence, and not identified as ones with intentions. Participants were also instructed to respond to the experimenter when they detected a PM picture in the test phase even if they could not recall the associated intention. During the test phase, pictures were presented at a subject controlled pace. Task performance was measured in three categories: (1) prospective component, the proportion of correctly recognized PM pictures to total PM pictures, regardless of whether participants explicitly remembered the associated intention, (2) retrospective component, the proportion of intentions remembered correctly as a function of the total number of identified PM pictures, and (3) associative memory, the proportion of words recalled correctly in response to the 48 paired-associate picture cues.

Remembering a belonging: the Remembering a Belonging subtest from the Rivermead Behavioural Memory Test (RBMT, Wilson et al., 1985) was also used to assess PM. At the beginning of the test session, the experimenter requests that the subject give the experimenter a personal belonging (e.g., a watch), which the experimenter then hides from view in an easily accessible place known to the subject. Participants are requested, but not reminded, to ask for the belonging back at the end of the test session and to remember where it was hidden. If subjects do not ask voluntarily, the experimenter prompts them as to whether or not an item had been relinquished. Parameters assessed were (1) remembering to ask for the hidden item and (2) remembering where the item was hidden, either with or without prompting.

Prospective and Retrospective Memory Questionnaire: the Prospective and Retrospective Memory Questionnaire (PRMQ, Crawford et al., 2003), a 16-item questionnaire, 8-items for each memory domain, was also included to measure self-rated memory failure by participants. For the results of the PRMQ, we converted the raw scores to T-scores using the program downloaded from the web site address: www.psyc.abdn.ac.uk/homedir/jcrawford/prmq.htm. The T-scores have a mean of 50 and a standard deviation of 10 in a normal population.

Predictions and statistical analysis

 

If 'frontal' stroke patients have a specific problem with prospective memory, we should see group differences on all PM measures. On the VW task, we expected to see largest effects on the time-check component, given that this aspect requires the subject to break ongoing task set, and thus arguably requires the most 'initiation' and degree of strategic monitoring. Similarly on the MI task we expected to see largest effects on the PM component (remembering which pictures were associated with intentions). With regard to RM, we expected that the stroke group would again have greatest difficulty with tests involving least environmental support (such as free recall), but relatively less difficulty with tests involving more support, such as cued recall and recognition (Craik, 1983, 1986). Statistical analyses were performed using the computer software SPSS version 11.0 for Windows (SPSS, Chicago, IL). Between-group differences in age, education, MMSE, PRMQ, and other neuropsychological tests were examined with independent samples t-tests. Performance on the RBMT was assessed with the Chi-square test. Performance on the Virtual Week and Memory for Intentions tasks in the stroke and control groups were compared by analysis of variance (ANOVA). An alpha level of p < .05 was used for all the above a priori analyses. If significant main effects were found, post-hoc analyses between groups were applied with independent-samples t-tests using the Bonferroni correction.

RESULTS

 

Prospective memory performance

 

Virtual Week: the mean proportions of 'Correct' responses and different types of error are presented in Table 3. The proportions of correct responses were first analyzed with a 2  3 mixed ANOVA using group (patients vs. controls) as the between-subject variable and type of task (regular, irregular, time-check) as the within-subject variable. Both main effects were statistically significant: The control group outperformed the patient group, mean proportions correct: 0.59 vs. 0.39; F(1, 21) = 8.94, p < .01, η_p² = .30; for type of task, F(2, 42) = 11.61, p < .001, η_p² = .36, post-hoc Bonferroni tests revealed that participants were more accurate on the regular than the irregular tasks (p < .001). Furthermore, the interaction between group and type of task approached significance, F(2, 42) = 2.55, p = .09, η_p² = .11, suggesting that the pattern of group differences may differ as a function of task. To further evaluate this possibility, post-hoc t-tests were performed to test for group differences in each task. The results showed that the group difference was significant in the time-check task, t(21) = -2.99, p = .007, r = .55, but not in the irregular tasks, t(21) = -1.87, p = .076, r = .38, or the regular tasks, t(21) = -0.787,p = .44, r = .16.

TABLE 3 Proportions of correct responses and different types of error in the Virtual Week task

	Regular		Irregular		Time-check
	Stroke	Control	Stroke	Control	Stroke	Control
Correct
M	0.61	0.69	0.26	0.41	0.30	0.67
SD	0.17	0.27	0.16	0.22	0.28	0.30
Wrong
M	0.06	0.05	0.21	0.17	0.08	0.00
SD	0.07	0.07	0.08	0.14	0.20	0.00
Late
M	0.05	0.12	0.03	0.03	0.14	0.26
SD	0.08	0.14	0.04	0.04	0.08	0.27
Miss
M	0.27	0.14	0.50	0.38	0.48	0.07
SD	0.18	0.19	0.21	0.21	0.29	0.11

A separate 2  3 mixed ANOVA was performed on the proportions of 'Miss' responses. Both main effects were again statistically significant: for groups, F(1, 21) = 18.58, p < .001, η_p² = .47, stroke patients showed a higher proportion of miss responses than the controls, group means: 0.42 vs. 0.20, for patients and control, respectively; for type of task, F(2, 42) = 8.30, p < .001, η_p² = .28, post-hoc tests revealed that the miss rates were higher in the irregular tasks than the regular and time-check tasks. The interaction between group and type of task was also significant, F(2, 42) = 3.96, p = .03, η_p² = .16. Consequently, post-hoc tests were performed to test for group differences in each task. Again, the results showed a significant difference in the time-check task, t(12.63) = 4.41, p = .001, r = .78 (degrees of freedom corrected for unequal variances), but not in the regular and irregular tasks, t(21) = 1.72, p = .099, r = .35, and t(21) = 1.38, p = .183, r = .29, respectively. The results from the Miss responses point to the same conclusion as with that from the Correct responses and suggests that stroke patients show the greatest deficits in the time-check tasks, and that such deficits are characterized by a combination of lower accuracy rates and higher miss rates. The other two types of error ('Wrong' and 'Late') were not analyzed due to the low numbers of observations.

Memory for intentions: patients correctly recalled fewer words than control subjects in the paired associate task, t(22) = -2.66, p = .014, r = .49. To examine the effects of group and semantic relatedness on the PM task, 2  2 mixed ANOVAs (between: group, patient vs. control; within: semantic relatedness, high vs. somewhat related) were performed on the prospective and retrospective components separately. No main effect of relatedness was found for either the prospective or the retrospective component, nor did relatedness interact with group (all F values < 1.0); the data were therefore collapsed across relatedness and are presented in Figure 1. For the prospective component, a main effect of group was found, F(1, 22) = 21.25, p < .001, η_p² = .491, showing that the stroke group had a significantly smaller proportion of correct responses than the control group. No group difference was found for the retrospective component. Overall then, stroke patients showed a deficit on the relatively difficult paired-associate task (remembering which word was paired with each picture) but performed as well as controls on remembering the activity associated with PM pictures, provided that they recognized the 'intention' pictures as PM cues. The patients did show a deficit in this latter response, however. They were less able to identify the pictures associated with an intention (means = 0.66 and 0.88 for patients and controls, respectively). 

[Enlarge Image]

Figure 1. Mean proportions of correct responses as a function of memory measure in the Memory for Intention task. Error bars represent standard error. *p < .05; **p < .01. PM, Prospective Memory; RM, Retrospective Memory.

Since there was a significant difference in performance on the picture-word paired-associate task between the two groups, performance on this task was included in the analysis to control for the contribution of general associative memory ability. We conducted ANCOVAs on prospective and retrospective components as the dependent measures with group as the between-subject factor, relatedness as the within-subjects factor and associative memory as the covariate. For the prospective component, both associative memory, F(1, 21) = 5.85, p = .025, η_p² = .218, and group, F(1,21) = 10.46, p = .004, η_p² = .332, showed significant effects. That is, patients still showed significant prospective deficits after controlling for their lower performance on the paired-associate task.¹ No significant effects were found for the retrospective component.

Remember a belonging: three control subjects and five stroke patients remembered to ask for their belongings in the Rivermead Test. Ten subjects in each group remembered where the belonging was hidden. Chi-square tests revealed that there were no significant differences between stroke patients and controls on this test (X² = 0.83, p = .67).

Prospective and Retrospective Memory Questionnaire. There were no significant differences in PRMQ T-scores between the groups, though patients scored slightly lower than controls on both components. Mean T-scores for the retrospective memory questions were 47.8 and 51.5 for patients and controls, respectively (t = -1.46, ns); T-scores for the prospective questions were 47.6 and 52.0 for patients and controls, respectively (t = -1.16, ns).

DISCUSSION

 

Neuropsychological tests

 

Group differences in the executive tasks revealed mixed results, such that patients showed deficits in some but not all subcomponents of executive functioning. For example, stroke patients did not show deficits in perceptual speed (Trails A and B), some aspects of sustained attention (SART: error of omission), or strategy application (R-SAT). In contrast, tasks that involve inhibitory control (SART: error of commission) and unguided retrieval (verbal fluency) showed large stroke-related deficits. This pattern of results suggests that impairments in this group of patients are associated with difficulties of 'self-initiation' (Craik, 1983, 1986) and cognitive control, which in turn may reflect inefficiencies of frontal lobe functioning (Grady & Craik, 2000). Despite equivalent levels of education and basic cognitive functions (MMSE), patients showed decreased performance in both immediate memory tasks, VPA and CVLT-II (although only the latter tasks were statistically significant). Thus, the patients exhibited some deficits in associative and item memory, both measured by recall; the groups did not differ on the recognition test of the CVLT-II. Hence, performance holds up well in situations where cognitive processes are supported by the task, the external environment, or well-learned habits, but deficits emerge when decision-making or retrieval processes are not so supported.

Prospective memory

 

Patients made as many correct responses as controls on the regular tasks in the Virtual Week test, suffered a slight deficit on the irregular tasks and a large deficit on the time-check task (r = .16, p = .44; r = .38, p = .076 and r = .55, p = .0007, respectively). Patients also missed more responses than controls in the time-check task, although there were no significant group differences in numbers of misses in regular or irregular tasks. No explicit prospective cues are available in the time-check task; for successful performance participants must therefore break out of the set induced by the Virtual Week task and self-initiate the decision to check the stopwatch. If the post-stroke patients had more trouble in this respect it might be expected that they would check the clock less frequently. In fact, the median number of checks (total for the three 'days') was 8 for the patients and 15.5 for the controls; the difference was not statistically significant, however, owing to the small numbers and the large variance - one patient checked the clock 26 times for example. The fact that there were only two occasions to perform the time-check task per circuit, as opposed to four occasions for both regular and irregular tasks may also have played a part in the results. The observation that no stroke-related deficit was evident under the simplest condition (i.e., the regular tasks) is similar to the findings of Rendell and Craik (2000) who reported no effect of aging in the performance of regular tasks, although age-related differences were found in the irregular and time-check tasks. The authors argued that older adults were able to benefit from the environmental support provided by reoccurrence and by explicit cues in the regular tasks. The present findings suggest that 'frontal' stroke patients are also able to do so, and that this preserved ability may account for the observation that despite impaired performance in a majority of the memory measures and laboratory PM tasks, the patients showed equivalent levels of performance in simple PM tasks and in subjective reports such as the memory questionnaire.

Performance differences between the two groups were larger in the non-repeated irregular task than in the repeated regular task. Compared to the other two tasks, the irregular task is associated with an increased demand to constantly update the intentions held in mind for each trial. However, the performance of stroke patients was poorest in the time-check task despite the fact that this task also repeats. It is suggested that the time-check task is associated with an increased demand for self-initiation as a result of less salient prospective cues relative to the other two tasks. Relative to the easiest condition (regular tasks), controls showed decreased performance only when the prospective demands varied from trial to trial (irregular tasks), whereas patients showed decreased performance both with irregularity and with increased demands for self-initiation. It should be acknowledged that meta-analyses of age-related differences in PM (e.g., Henry et al., 2004) have shown that event-based PM tasks that are high in strategic demand, and also some RM tasks such as free recall, can show as large age-related deficits as those on time-based PM tasks. This suggests that there is nothing particularly unique about PM, but that age-related difficulties, and post-stroke difficulties according to the present results, are more likely to be found on tasks that have a high demand for initiation, strategic monitoring, and executive control.

In the memory for intentions task, we found that patients recalled fewer words than controls in the word-picture paired-associate background task. The patients also missed more prospective memory cues. For those pictures correctly identified as cues, however, patients and controls recalled the associated intentions with equivalent accuracy. The apparent discrepancy in retrospective memory results between the paired-associate recall and recall of intentions may be due partly to the somewhat greater difficulty of the former task (Figure 1) but probably also to the fact that the patient group identified fewer picture cues, so that the conditional recall of intentions was based on fewer (and possibly easier) examples. The ANCOVA analysis showed that pathology has an additional effect (poorer performance on the PM task) even after differences in associative memory were accounted for. Although PM performance and associative memory performance are related, our analysis showed that patients performed more poorly in PM tasks not only because of their lower associative memory functioning. In this MI task, the relatedness of the intention and picture cue had no effect on either the prospective or retrospective component, nor did it interact with any other factor. This may be because both the 'highly related' and 'somewhat related' items were fairly easy to associate. The patient group in this study did not show any deficits in this regard, possibly reflecting their spared semantic functions. The effects of stroke found in this task were very similar to the effects of aging reported by Cohen et al. (2001, Exp. 1). In that study, older adults showed lower PM scores for highly related and somewhat related items, but equivalent RM scores for these items compared to younger adults. Speculatively, it seems possible that both stroke and aging affect the frontal-executive functions responsible for successful PM.

No group differences were found in either the Memory for Belongings task taken from the Rivermead Behavioural Memory Test or in the Prospective and Retrospective Memory Questionnaire. The Memory for Belongings task was obviously difficult for all participants since only 3 control subjects and 5 patients remembered to ask for the object; these are people in their 60s and 70s, however, so perhaps this level of performance on a one-off task is not surprising. Our main point here is that whereas the test is easy to administer, it is obviously rather a blunt instrument, and did not reveal PM difficulties in the patients which were shown on the experimental tasks. Similarly, there were no group differences in either PM or RM as measured by the questionnaire. Both groups scored close to the population average of 50. Our conclusions are that the stroke patients' memory problems are relatively subtle, and are not sufficiently grave to lead them to rate their performance as lower than normal in the questionnaire. Nonetheless, these difficulties were picked up by the VW and MI tests.

Taken together, our assessment of PM functions in stroke patients suggests that they are impaired by some but not all aspect of PM. Specifically, the patients performed as well as the controls on the regular tasks in the VW task and on the RM component of PM in the MI task; they performed less well, however, on the irregular and time-check aspects of VW and on the PM component of the MI task. We therefore suggest that these differences can only be detected by carefully designed tasks that measure these aspects separately. Stroke patients also showed impairments in some RM tasks such as associative memory and free recall. However, our analysis suggests that their deficits in PM functioning clearly involve additional stroke-related deficits beyond general memory ability, possibly deficits related to executive functioning.

CONCLUSIONS AND LIMITATIONS

 

The main limitation of the study is that the group of stroke patients is small. Further studies are clearly needed to confirm and extend the present results. In addition, such further studies should seek to specify the brain areas principally concerned with failures of PM. Our speculative assumption is that these areas are typically frontal in general, as the lesions in our patient group were primarily anterior, but this assumption must be tested by further studies. Importantly, it is also unclear at present whether the present pattern of results is associated with stroke as such, or whether it may be found in a variety of cases involving anterior lesions. Despite the relatively small numbers of subjects in each group, however, the results showed substantial differences between patients and controls in many but not all tests. This differential pattern gives some clues as to the nature of cognitive difficulties experienced by stroke patients with anterior lesions. The present results suggest, first, that difficulties such patients experience arise in situations where the needed action is not well supported by well-practiced habits or by environmental cues. Executive actions or retrieval processes must therefore be 'self-initiated' by the subjects themselves. Second, the stroke patients studied here seemed to have greatest difficulties with spontaneously switching from one (ongoing) activity to another. This is evident in the MI task by their reduced ability to spontaneously recognize pictures as prospective cues rather than associative cues (i.e., the ongoing task), and by their reduced likelihood of spontaneously checking the clock when playing the virtual week board game. In line with this observation, stroke patients had shorter reaction times on the SART and also made more errors of commission, suggesting that they were less able than controls to inhibit a habitual activity. It is also noteworthy that the present sample of patients reported no difficulties in everyday living, and that the groups did not differ in their questionnaire responses. This discrepancy suggests that patients' insight into the nature of their deficits is not complete. Indeed, the disturbances in executive deficits that lead to memory retrieval problems are often 'misinterpreted' by patients as 'short term memory' issues, thus leading them to clinical attention with fears that they suffer from a degenerative dementia. It is quite possible that some of the real-life difficulties experienced by these patients may be embedded in the kind of PM deficits reported here. A larger study would be required to settle this issue.

Acknowledgments

This work was supported by the Korea Research Foundation Grant (KRF-2006-013-E00133) to H. J. Kim and by a grant from the Natural Sciences and Engineering Research Council of Canada to F. I. M. Craik. We also wish to thank HeeSun Lim for research assistance.

REFERENCES

• 1. Brooks, B. M.,  Rose, F. D.,  Potter, J.,  Jayawardena, S. and Morling, A. (2004) Assessing stroke patients' prospective memory using virtual reality. Brain Injury, 18 , pp. 391-401. [informaworld]
• 2. Burgess, P. W. (2000) Strategy application disorder: The role of the frontal lobes in human multitasking. Psychological Research 63 , pp. 279-288. [ csa ]
• 3. Burgess, P. W. and Shallice, T. (1996) Response suppression, initiation and strategy use following frontal lobe lesions. Neuropsychologia 34 , pp. 263-273. [ crossref ]
• 4. Burgess, P. W.,  Quayle, A. and Frith, C. D. (2001) Brain regions involved in prospective memory as revealed by positron emission tomography. Neuropsychologia 39 , pp. 545-555. [ csa ] [ crossref ]
• 5. Cohen, A. L.,  West, R. and Craik, F. I. M. (2001) Modulation of the prospective and retrospective components of memory for intentions in younger and older adults. Aging, Neuropsychology, and Cognition 8 , pp. 1-13. [informaworld]
• 6. Cohen, G. (1996) Memory for intentions, actions, and plans: Memory in the real world pp. 25-53. Taylor and Francis Ltd , Milton Park, UK
• 7. Craik, F. I. M. (1983) On the transfer of information from temporary to permanent memory. Philosophical Transactions of the Royal Society B: Biological Sciences 302 , pp. 341-359.
• 8. Craik, F. I. M. Klix, F. and Hagendorf, H. (eds) (1986) A functional account of age differences in memory. Human memory and cognitive capabilities pp. 409-422. Elsevier , Amsterdam
• 9. Crawford, J. R.,  Smith, G.,  Maylor, E. A.,  Della Sala, S. and Logie, R. H. (2003) The Prospective and Retrospective Memory Questionnaire (PRMQ): Normative data and latent structure in a large non-clinical sample. Memory 11 , pp. 261-275. [informaworld] [Ergonomics Abstracts]
• 10. Crowder, R. G. Brandimonte, M.,  Einstein, G. O. and McDaniel, M. A. (eds) (1996) The trouble with prospective memory: A provocation. Prospective memory: Theory and applications pp. 143-147. Lawrence Erlbaum Associates, Inc , Mahwah, NJ
• 11. Cummings, J. L. (1993) Frontal-subcortical circuits and human behavior. Archives of Neurology 50 , pp. 873-880.
• 12. Delis D. C., K. J.,  Kaplan, E. and Ober, B. A. (2000) California Verbal Learning Test -Second Edition The Psychological Corporation , San Antonio, TX
• 13. Einstein, G. O. and McDaniel, M. A. (1990) Normal aging and prospective memory. Journal of Experimental Psychology: Learning, Memory, and Cognition 16 , pp. 717-726. [Ergonomics Abstracts] [ pubmed ] [ crossref ]
• 14. Einstein, G. O.,  McDaniel, M. A.,  Richardson, S. L.,  Guynn, M. J. and Cunfer, A. R. (1995) Aging and prospective memory: Examining the influences of self-initiated retrieval processes. Journal of Experimental Psychology: Learning, Memory, and Cognition 21 , pp. 996-1007. [ pubmed ] [ crossref ]
• 15. Folstein, M. F.,  Folstein, S. E. and McHugh, P. R. (1975) 'Mini-mental state': A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research 12 , pp. 189-198. [ crossref ]
• 16. Goldstein, L. H.,  Bernard, S.,  Fenwick, P. B.,  Burgess, P. W. and McNeil, J. (1993) Unilateral frontal lobectomy can produce strategy application disorder. Journal of Neurology, Neurosurgery & Psychiatry 56 , pp. 274-276.
• 17. Grady, C. L. and Craik, F. I. (2000) Changes in memory processing with age. Current Opinions in Neurobiology 10 , pp. 224-231.
• 18. Guynn, M. J. (2003) A two-process model of strategic monitoring in event-based prospective memory: Activation/retrieval mode and checking. International Journal of Psychology 38 , pp. 245-256. [informaworld]
• 19. Hachinski, V. (2007) The 2005 Thomas Willis Lecture: Stroke and vascular cognitive impairment: A transdisciplinary, translational and transactional approach. Stroke 38 , pp. 1396-1403. [ crossref ]
• 20. Hamilton, B. L. (1977) An empirical investigation of the effects of heterogeneous regression slopes in analysis of covariance. Educational and Psychological Measurement 37 , pp. 701-712.
• 21. Henry, J. D.,  MacLeod, M. S.,  Phillips, L. H. and Crawford, J. R. (2004) A meta-analytic review of prospective memory and aging. Psychology and Aging 19 , pp. 27-39. [csa ] [ crossref ]
• 22. Kliegel, M. and Martin, M. (2003) Prospective memory research: Why is it relevant?. International Journal of Psychology 38 , pp. 193-194. [informaworld]
• 23. Kliegel, M.,  Eschen, A. and Thone-Otto, A. I. (2004) Planning and realization of complex intentions in traumatic brain injury and normal aging. Brain and Cognition 56 , pp. 43-54. [ csa ] [ crossref ]
• 24. Kvavilashvili, L. and Ellis, J. Brandimonte, M.,  Einstein, G. O. and McDaniel, M. A. (eds) (1996) Varities of intention: Some distinctions and classifications. Prospective memory: Theory and applications pp. 23-51. Lawrence Erlbaum Associates, Inc , Mahwah, NJ
• 25. Levine, B.,  Dawson, D.,  Boutet, I.,  Schwartz, M. L. and Stuss, D. T. (2000) Assessment of strategic self-regulation in traumatic brain injury: Its relationship to injury severity and psychosocial outcome. Neuropsychology 14 , pp. 491-500. [ csa ] [ crossref ]
• 26. Lezak, M. D. (1995) Neuropsychological Assessment-Third Edition Oxford University Press , New York
• 27. Martin, M. K. and McDaniel, M. A. (2003) The involvement of executive functions in prospective memory performance of adults. International Journal of Psychology 38 , pp. 195-206. [informaworld]
• 28. Mathias, J. L. and Mansfield, K. M. (2005) Prospective and declarative memory problems following moderate and severe traumatic brain injury. Brain Injury 19 , pp. 271-282. [informaworld]
• 29. McDaniel, M.,  Glisky, E.,  Rubin, S.,  Guynn, M. and Routhieaux, B. (1999) Prospective memory: A neuropsychological study. Neuropsychology 13 , pp. 103-110. [ csa ] [crossref ]
• 30. Morris, J. (1993) The clinical dementia rating: Current version and scoring rules. Neurology 43 , pp. 2412-2414.
• 31. Okuda, J.,  Fujii, T.,  Yamadori, A.,  Kawashima, R.,  Tsukiura, T.,  Fukatsu, R.,  Suzuki, K.,  Ito, M. and Fukuda, H. (1998) Participation of the prefrontal cortices in prospective memory: Evidence from a PET study in humans. Neuroscience Letters 253 , pp. 127-130. [ crossref ]
• 32. Reed, B. R.,  Eberling, J. L.,  Mungas, D.,  Weiner, M. and Jagust, W. J. (2001) Frontal lobe hypometabolism predicts cognitive decline in patients with lacunar infarcts.Archives of Neurology 58 , pp. 493-497. [ csa ] [ crossref ]
• 33. Reed, B. R.,  Eberling, J. L.,  Mungas, D.,  Weiner, M.,  Kramer, J. H. and Jagust, W. J. (2004) Effects of white matter lesions and lacunes on cortical function. Archives of Neurology 61 , pp. 1545-1550. [ csa ] [ crossref ]
• 34. Reitan, R. M. (1986) Trail Making Test Reitan Neuropsychology Laboratory , Tucson, AZ
• 35. Rendell, P. and Craik, F. (2000) Virtual Week and Actual Week: Age-related differences in prospective memory. Applied Cognitive Psychology 14 , pp. S43-62.[Ergonomics Abstracts] [ crossref ]
• 36. Robertson, I. H.,  Manly, T.,  Andrade, J.,  Baddeley, B. T. and Yiend, J. (1997) 'Oops!': Performance correlates of everyday attentional failures in traumatic brain injured and normal subjects. Neuropsychologia 35 , pp. 747-758. [ csa ] [ crossref ]
• 37. Roediger III, H. L. Brandimonte, M.,  Einstein, G. O. and McDaniel, M. A. (eds) (1996) Prospective memory and episodic memory. Prospective memory: Theory and applications pp. 149-155. Lawrence Erlbaum Associates, Inc , Mahwah, NJ
• 38. Roman, G. C.,  Sachdev, P.,  Royall, D. R.,  Bullock, R. A.,  Orgogozo, J. M. Lopez-Pousa, S. et al. (2004) Vascular cognitive disorder: A new diagnostic category updating vascular cognitive impairment and vascular dementia. Journal of Neurological Science 226 , pp. 81-87. [ pubmed ] [ csa ] [ crossref ]
• 39. Schmitter-Edgecombe, M. and Wright, M. J. (2004) Event-based prospective memory following severe closed-head injury. Neuropsychology 18 , pp. 353-361. [ csa ] [crossref ]
• 40. Shallice, T. and Burgess, P. W. (1991) Deficits in strategy application following frontal lobe damage in man. Brain 114 , pp. 727-741. [ crossref ]
• 41. Smith, R. E. (2003) The cost of remembering to remember in event-based prospective memory: Investigating the capacity demands of delayed intention performance.Journal of Experimental Psychology: Learning, Memory, and Cognition 29 , pp. 347-361. [ pubmed ] [ csa ] [ crossref ]
• 42. Stuss, D. T. and Levine, B. (2002) Adult clinical neuropsychology: Lessons from studies of the frontal lobes. Annual Review of Psychology 53 , pp. 401-433. [ crossref ]
• 43. Wahlund, L. O.,  Barkhof, F.,  Fazekas, F.,  Bronge, L.,  Augustin, M. Sjogren, M. et al. (2001) A new rating scale for age-related white matter changes applicable to MRI and CT. Stroke 32 , pp. 1318-1322.
• 44. Wechsler, D. (1997) Wechsler Memory Scale-Third Edition The Psychological Corporation , San Antonio, TX
• 45. West, R. and Covell, E. (2001) Effects of aging on event-related neural activity related to prospective memory. Neuroreport 12 , pp. 2855-2858. [ crossref ]
• 46. West, R. and Krompinger, J. (2005) Neural correlates of prospective and retrospective memory. Neuropsychologia 43 , pp. 418-433. [ crossref ] [ csa ]
• 47. West, R.,  Jakubek, K. and Wymbs, N. (2002) Age-related declines in prospective memory: Behavioral and electrophysiological evidence. Neuroscience and Biobehavioral Reviews 26 , pp. 827-833. [ csa ] [ crossref ]
• 48. West, R.,  McNerney, M. W. and Krauss, I. (2007) Impaired strategic monitoring as the locus of a focal prospective memory deficit. Neurocase 13 , pp. 115-126.[informaworld]
• 49. Wilson, B.,  Cockburn, J. and Baddeley, A. (1985) The Rivermead Behavioural Memory Test Thames Valley Test Co , Bury St. Edmonds, UK

Notes

¹The group  covariate interaction was significant F(1, 20) = 5.04, p < .05, showing that the assumption of homogeneous regression slope for ANCOVA was violated. However, the violation of assumption has little effect on the conclusion from ANCOVA when the group sizes are equal (Hamilton, 1977). So it seems safe to conclude that the patient and control groups differed after controlling for associative memory performance.

List of Figures

 

[Enlarge Image]

Figure 1. Mean proportions of correct responses as a function of memory measure in the Memory for Intention task. Error bars represent standard error. *p < .05; **p < .01. PM, Prospective Memory; RM, Retrospective Memory.

List of Tables

TABLE 1 White matter lesion intensity (see text for explanation) for each patient as a function of brain region

	Brain Region
Patient ID	RF	LF	RBG	LBG	RT	LT	RPO	LPO	RIT	LIT
Notes: R and L, right and left respectively; F, frontal; PO, parieto-occipital; T, temporal; BG, basal ganglia; IT, infratentorial.
P1	2	0	0	0	0	0	0	0	0	0
P2	0	1	0	1	0	0	0	0	0	0
P3	0	2	0	1	0	0	0	0	0	0
P4	1	3	1	2	0	0	1	1	0	0
P5	2	2	0	0	0	0	0	1	0	0
P6	3	3	3	3	3	3	3	3	0	0
P7	0	0	3	0	0	0	0	0	0	0
P8	1	0	2	0	0	0	0	0	0	0
P9	1	2	0	3	0	0	0	1	0	0
P10	0	0	0	0	0	3	0	0	0	0
P11	3	3	3	3	0	0	2	2	0	0
P12	2	2	1	1	0	0	2	2	0	0

TABLE 2 Demographics, clinical characteristics and neuropsychological profiles for subject groups

	Patients		Controls		t-Tests
	M	SD	M	SD	t	df	p
Values reported are raw scores. M, mean, SD, standard deviation, df, degrees of freedom. Two-tailed t-tests are reported. See methods section for abbreviations.
Age (years)	69.33	7.02	69.08	4.94	0.10	19.75	.92
Education (years)	14.42	3.73	15.67	3.23	-0.88	21.56	.39
MMSE	28.67	1.07	29.00	1.13	-0.74	21.95	.47
Trail Making Time
A (s)	39.92	13.85	33.83	8.91	1.28	18.77	.22
B (s)	90.00	33.34	78.92	20.92	0.98	18.50	.34
Verbal Fluency
FAS (words)	35.67	10.05	45.25	9.37	-2.42	21.89	.02
Animal (words)	14.17	4.69	18.42	6.61	-1.82	19.83	.08
SART
Reaction Time (ms)	367.33	109.59	438.97	60.54	-1.98	17.14	.06
Error of Commission	10.92	7.19	4.58	2.87	2.83	14.43	.01
Error of Omission	7.42	14.06	3.50	3.68	0.93	12.50	.37
Verbal Paired Associates
Total Recall (words)	9.83	1.85	11.08	3.32	-1.14	17.25	.27
CVLT
Total learning score (1-5)	35.50	10.78	50.75	7.62	-4.08	19.12	.00
Long delay free recall	6.58	3.09	10.75	2.05	-3.89	19.12	.00
Recognition Hits - False alarms	8.50	5.30	11.92	3.87	-1.80	20.14	.09
R-SAT
% Easy Items Completed	90	10	84	17	1.11	17.07	.28

TABLE 3 Proportions of correct responses and different types of error in the Virtual Week task

	Regular		Irregular		Time-check
	Stroke	Control	Stroke	Control	Stroke	Control
Correct
M	0.61	0.69	0.26	0.41	0.30	0.67
SD	0.17	0.27	0.16	0.22	0.28	0.30
Wrong
M	0.06	0.05	0.21	0.17	0.08	0.00
SD	0.07	0.07	0.08	0.14	0.20	0.00
Late
M	0.05	0.12	0.03	0.03	0.14	0.26
SD	0.08	0.14	0.04	0.04	0.08	0.27
Miss
M	0.27	0.14	0.50	0.38	0.48	0.07
SD	0.18	0.19	0.21	0.21	0.29	0.11