Simulation of amnesia by normals on Rey's Auditory Verbal Learning Test

Simulation of amnesia by normals on Rey's Auditory Verbal Learning Test

J. Neurolinguistics, Vol. 9, No. 2. pp, 147-156, 19951% Copyright0 19% Elsevier Science Ltd Printed in Great Britain. All rights reserved 0911-6@+4/%...

740KB Sizes 0 Downloads 73 Views

J. Neurolinguistics,

Vol. 9, No. 2. pp, 147-156, 19951% Copyright0 19% Elsevier Science Ltd Printed in Great Britain. All rights reserved 0911-6@+4/%$15.uo+0.00

SIMULATION OF AMNESIA BY NORMALS ON REY’S AUDITORY VERBAL LEARNING TEST K. A. FLOWERS,M. R. SHERIDANand HELEN SHADBOLT Department of Psychology, University of Hull, U.K. Atrstraet-Normal subjects were tested twice on Rey’s Auditory-Verbal Learning Test (AVLT), using a different version of the test each time. For one version subjects pexformed normally, for the other they attempted to simulate a mild to moderate memory impairment. The pattern of scores in the two conditions showed that malingering subjects differed from a typical genuine amnesic performance in three respects: they showed a primacy effect in free recall, they over-performed on the delayed recall trial and they underperformed on the recognition task. These features of Rey’s AVLT give it a good potential to help distinguish between genuine and faked amnesia in naive subjects. Copyright 0 1996 Elsevier Science Ltd

INTRODUCTION Rey’s Auditory-Verbal Learning Test (AVLT) has come to be commonly used for the clinical assessment of memory impairment, not only in such disorders as head injury [ 11, dementia [ 21 and cerebral lesions [ 31, but also in a wide variety of states such as multiple sclerosis [ 41, malnutrition [ 51 and even social drinking [ 61 . One reason for this is that the test yields, in a short space of time, a number of memory scores (immediate word span, learning curve, serial order effect, delayed recall score, recognition score), which have been shown to relate to the various characteristics of different kinds of amnesia. It therefore not only gives a reasonably accurate quantitative measure of any memory deficit [7] but also provides some indication of how it occurs [ 81 and can differentiate between conditions (e.g. [ 91; for review see Peaker and Stewart [ lo]). For the same reason the AVLT has also been used to test for simulated memory impairments in studies of the detection of malingering, and would seem to have considerable potential in this regard. That is, the range of scores it gives can be used to detect a different parrern of abnormality in simulators from that found in genuine amnesic subjects. Previous studies have shown that malingering on memory tests shows up in a number of ways. Subjects feigning amnesia (as with other symptoms) tend to over-exaggerate their difficulties in general, so that they score even lower than genuine amnesics do [ 1l-16)] and often lower than chance [ 17,181. They produce a poor performance even on aspects of memory which remain intact in the genuinely impaired, e.g. implicit memory [ 191, personal identity memory [ 161 or immediate recall without distraction [20] . Also they can often be caught by the fact that the pattern of their responses within the task is different to that of a genuine amnesic subject. Suggested indications of this kind include failing simple items while passing more difficult ones [ 2 11, recognition being as bad or worse than recall [ 11,221, a failure to show any difference between cued and uncued recall [ 231 or between distracted and undistracted recall [20] and a different serial order effect curve [ 16,181. As the Rey AVLT provides measures of undistracted immediate recall (of the last few items at least), of serial order, of recognition and of delayed recall, it is likely to be a very appropriate 147

148

K. A. FLOWERSet

al.

instrument for picking up malingering of amnesia as well as genuine impairments, thus making it doubly useful in clinical assessment. Previous studies by Bernard [ 18,24,25] have suggested that the pattern of scores on the AVLT is quite good as a basis for detecting pretended memory loss, although there is some uncertainty as to whether or not simulators show a primacy effect in the serial order curve. This investigation attempted to replicate these studies systematically so as to see how reliable are the signs of malingering on the AVLT. The four major aspects of interest were the learning curve over the first five trials, the serial order curve (primacy and recency effects), the delayed recall score compared to previous level of learning, and the subsequent recognition score. Also, since the detection of malingering is often claimed to be based on exploiting the subjects’ ignorance of what a true memory performance is, subjects were asked to simulate amnesia either on the first test or on a second test after having performed normally on a previous session. Subjects performed the test twice, each time with a different version (word list), performing normally on one occasion and attempting to simulate a mild to moderate memory deficit on the other. This within-subjects design, using subjects as their own controls (which has not been used before), was adopted in order to see whether malingering is better if the subject has had prior experience of the test. That is, whether having some idea how difficult the test is helps subjects to pitch their level of supposed deficit more accurately than those faced with the test for the first time. It also allowed a check on whether there was any difference in scores between the two word lists of the test.

METHODS Materials Two versions of Rey’s test were used [22]. With version A the list of 15 words was the standard one, the distractor list was Rey’s list C and the recognition passage was the standard one (see Lexak [22]). With version B the list of 15 words was Rey’s list B, and both the distractor list and the recognition passage were designed specifically for the experiment (see Appendix). Subjects

Twenty-eight unpaid volunteer subjects, 15 male and 13 female, took part in the experiment. All of them were either undergraduate students or people in professional occupations. Ages ranged from 18 to 59, with a mean of 26.9 years. Each subject was tested twice, once performing normally and once whilst malingering. The second test was administered no less than 24 hours after the first. Procedure

To measure possible practice effects (see Peaker and Stewart [lo]) subjects were randomly assigned to perform normally or to malinger on their first test. Possible differences in difficulty of the two lists were controlled for by subjects being randomly assigned to take test A or B first. A repeated measures design was used, with subjects receiving the other word list on their second occasion, and changing their performance from non-malingering to malingering or vice-versa. The combination of these variables gave four experimental groups, each with seven subjects: Group Group Group Group

1: 2: 3: 4:

normal with list A first, then malinger normal with list B first, then malinger malinger with list A first, then normal malinger with list B first, then normal

with with with with

list list list list

B. A. B. A.

On agreeing to take part in the experiment, subjects were taken to a quiet private room and told they were to be taking a memory test. Those subjects assigned to malinger were given the following instructions: “This test is used on patients suffering memory ditlicuhies as a result of head injuries or disease of the nervous system. Some people also pretend to be suffering from memory difficulties in order to receive compensation claims etc. Throughout this entire test, I want you to pretend that you are having memory difficulties, and perform on the test in a way you think that someone suffering from a mild to moderate memory loss would.”

AMNESIA SIMULATION ON THE REY

149

AVLT

No specific aetiology was given for the amnesia to be simulated, as subjects simulating different memory problems do not differ in their scores [ 161. After this stage, subjects, whether malingering or not, were tested using exactly the same procedure. Subjects were given the standard instructions for the test (as in Lexak [22]) and asked if they understood. The list of 15 words was then read out at the rate of 1 per Sec. at the end of which the subject was invited to tell the experimenter as many words as they could recall from the list, in any order. This procedure was carried out for four more trials, with the experimenter reminding subjects between each reading that each trial was independent and that any words recalled already in previous trials should be repeated. If a subject recalled all 15 words correctly in less than five attempts, they were credited with a score of 15 for the other trials. Following the first five trials, the distractor list was read out and subjects required to recall freely as with the first five trials. Subjects were then asked to think back to the original list and to tell the experimenter any words from that list they could now remember. Finally, the recognition task was administered-subjects were given a short recognition passage containing all 15 words from the original list and required to circle any words they recognized. Subjects were then asked for feedback. Details of any mnemonics used, or any particular system for malingering were noted. Subjects were then thanked and debriefed.

RESULTS Scores for the five learning trials (tI-V), the delayed recall trial (tV1) and the recognition passage (r) were analysed by a multivariate mixed measures analysis of variance (MANOVA) with trials, malingering/non-malingering and test (1stMnd) as within-subjects factors and list (A/B) as a between-subjects factor. An initial analysis of tests 1 and 2 compared the scores for subjects learning lists A and B. On neither test was there any difference of note between any scores on the two versions of the test. On the first test: trials I-V combined, F(1,26) = 0.21; trial VI (delayed recall), F(1,26) = 0.56; recognition score, F(1,26) = 0.84. On the second test: trials I-V combined, F(1,26) = 0.02; trial VI (delayed recall), F(1,26) = 0.27; recognition score, F(1,26) = 0.58. Also there were no significant interactions in either analysis. For subsequent analyses of group performance, therefore, the scores on the two lists were combined.

Initial learning curves

Table 1 shows mean scores on the first five trials for subjects performing genuinely first or second, and malingering first or second. For the first five trials there were very highly significant main effects of trials [F(4,192) = 117.29, P < 0.0011 and of malingering/nonmalingering [F(1,48) = 59.99, P < O.OOl] , but there was no effect of test [F(1,48) = 0.63, n.s.] and no interaction of test with malingering/non-malingering [F(1,48) = 0.04, n.s.1. Thus neither normal nor malingering scores were affected by whether the subject had performed on the AVLT on a previous occasion. Table 1. Mean number of words recalled and recognized in the two tests Learning

Delayed

Trial I

Trial II

Trial III

Trial IV

Trial V

Recall VI

Recog. score.

Normal 1st Normal 2nd

8.2 7.0

10.2 10.8

12.1 12.1

12.7 12.5

13.3 13.2

11.3 11.1

14.4 14.5

Malingering 1st Malingering 2nd

4.9 5.1

6.5 6.1

7.7 7.1

8.6 7.6

9.0 7.7

7.3 6.7

9.4 6.9

Source: K. A. Flowers, M. R. Sheridan and Helen Shadbolt, Simulation of amnesia by normals on Rey’s Auditory Verbal Learning Test.

150

K. A. FLOWERS et al.

-aMalingerers - - - +- - - Non-malingerers

Fig. 1. Number of words recalled in learning trials.

Two interactions concerning the shape of the learning curve across trials I-V were found to be significant. First there was a two-way interaction between trials and malingering/nonmalingering [F(4,192) = 7.91, P < O.OOl] , which indicates that the learning curve was different when subjects were malingering from that when they were performing normally. This is shown in Fig. 1, which plots the scores for malingering and non-malingering conditions combined across tests and lists. The overall impression is that malingerers fail to improve at the same rate in trials II-V as well as starting from a lower point at trial I. In this they resemble some amnesic subjects quite closely, suggesting that the overall learning score is not a very good detector of malingering. The three-way interaction between trials, malingering/non-malingering and order of test also just reached significance [ F(4,192) = 2.52, P c 0.05 I. Looking at the data in Table 1 it is clear that there is virtually no difference between the learning curves of the two nonmalingering groups, but the curve drops off more for the malingering group tested second. If anything, therefore, a prior experience of the test leads subjects to exaggerate their simulated deficit of memory across trials, but has no effect on genuine performance. This effect was mild, however, and was not strong enough to produce notably different overall scores between first and second test conditions. By contrast the difference between malingering and non-malingering scores was highly significant on each of the five trials (all comparisons P < 0.001). Serial order effect The number of times each word was recalled over the five trials by subjects in malingering and non-malingering conditions was calculated and expressed as a percentage. Figure 2 shows the mean percentage recall in the two conditions on the two lists. To compare the curves for the two versions of the test (lists), two x2 tests were carried out, one for malingerers and the other for non-malingerers, using totalled scores for four sections of the curve (words l-3,4-7, 8-l 1 and 12-15). The result was non-significant in both cases: x2 = 0.83 for malingerers and 0.99 for non-malingerers, showing that there was no difference in shape of the curves for the two word lists. As can be seen in Fig. 2, however, word 6 is frequently remembered on version B of the test in both performance conditions, and appears to be an outlier in the serial position curve. Possible reasons for this are discussed later. Figure 3 shows the combined serial order curve for the malingering and non-malingering conditions. Here it is clear that malingering subjects produced both a primacy and recency

AMNESIA SIMULATION ON THE REY AVLT

151

.

+

i:i A

II

n;B L nmA x nmB .I

2

3

4

5

6

7 Word

Fig. 2. Number of words recalled

8

9

IO

II

12

I3

I4

15

position in each position on lists A and B

--*-

Malingerers

---a.- - - Non-malingerers I

2

3

4

5

6

7

8

9

10

II

12

13

14

IS

Word position Fig. 3. Serial order curves for malingerers

and non-malingerers.

effect similar to that produced when they were performing normally. A x2 test comparing the curves, based on the combined scores for the four sections of the data, did not reach significance (x2 = 0.67, n.s.) indicating that serial order effects were the same in the two conditions. Forgetting

Scores for trial VI, the delayed recall trial, for two conditions are shown in Table 1. Again there was no difference between scores on the two lists [for test 1, F(1,26) = 0.56, n.s. and for test 2, F(1,26) = 0.27, n.s. ] . While not surprisingly there was a strong effect of condition [ F(1,40) = 44.72, P < O.OOl] there was no effect of test order [ F(1,40) = 0.51, n.s.] and the interaction between them did not reach significance either [ F(1,40) = 3.19, P > 0.051. More interesting than the overall level of scores, however, was the relative amount forgotten following the distractor task. To calculate this, the recall score for each subject in each condition was subtracted from the highest score obtained on the initial learning trials. The differences for malingering and non-malingering conditions are shown in Fig. 4. Here nonmalingerers forgot more words (mean 3 S) than malingerers did (mean 2.6), the difference

152

K. A. FLOWERSet al.

Highest recall (h)

Trial VI ([VI)

m

Malingerers

0

Non-malingerers

h-tVI

Fig. 4. Delayed recall scores: relative amounts “forgotten”.

almost reaching significance [ F( 1,27) = 3.95, P = 0.571. The difference is almost certainly minimized by a ceiling effect in the non-malingering condition so that the drop in normal recall is underestimated. What is clear, however, is that malingering subjects do not show the sharp drop in recall after a delay which is typical of amnesic subjects. Recognition The number of words correctly identified in the recognition passages is shown in Table 1. Again there was no difference between the two lists on either test [ F(1,26) = 0.04 and F(1,26) = 0.58, both n.s. ] . Also, again the effect of test order was not significant [ F( 140) = 3.37, n.s.] while that for condition clearly was [F(1,40) = 93.17, P < O.OOl], and for these scores the interaction between them also reached significance [ F( 1JO) = 4.51, P < 0.051. While scores on the non-malingering condition did not differ, those for malingerers on the second test were lower than those for malingerers on the first. Subjects with prior experience of the test, therefore, exaggerate the relative loss on recognition even more than those without. Moreover, when recognition scores were compared with highest recall scores (Fig. 5) malingerers showed a slight drop in score while non-malingerers showed a slight increment. Again a ceiling effect probably minimizes the advantage for recognition in the normal subjects, since they had relatively little room to improve on their highest recall score. Malingerers, however, did not take advantage of their relatively low recall scores to improve on recognition, which again sets them apart from genuine amnesic subjects who commonly improve on recognition compared to their level of recall.

DISCUSSION As subjects showed a learning effect over the five learning trials even when instructed to malinger, this demonstrates their appreciation that improvement does occur in amnesiacs with practice. Their learning curve was flatter than normal, and in this respect too they resembled amnesic subjects quite well. Malingerers, therefore, did not show an obvious difference in total learning scores from that expected of subjects with genuine memory impairment. Detection of malingered amnesia, therefore, is likely to be difficult on an immediate free recall task alone,

AMNESIA SIMULATION ON THE REY AVLT

I

-2 Recognition (r)

m

Malingerel-s

0

Non-malingerers

153

I Highest recall (h)

Recognition increment (r-h)

Fig. 5. Recognition scores: relative increase from recall.

and probably on any test which gives only an overall score with no distinctive pattern of response. There were three aspects of the task, however, where malingering performance gave itself away. First, when the serial position curve was plotted there was a distinct primacy effect, which is unlikely to occur in genuine amnesia. In this the results here agree with those of Wiggins and Brandt [ 161 and Bernard et al. [ 181 rather than Bernard [25]. Many subjects, when explaining how they had proceeded to malinger, said they had employed no particular strategy but just had not concentrated or tried as hard as they would under normal circumstances. This could explain why the serial position curves for the two conditions were the same shape. Subjects (who were unaware that amnesiacs would be likely to forget the first words in the list) simply performed at minimum effort, and hence still recalled the words they found came to mind easiest, which would be those at the beginning and end of the list. It is interesting that the serial position curves for list B show a peak for word 6 under both performance conditions. As this did not occur with list A, it can be assumed that the high percentage of recall was due to the nature of the word on list B-“mountain’‘-as opposed to its position in the list. Many subjects reported using techniques of imagery or word association to remember items, so perhaps this word created a particularly strong mental image in subjects. Alternatively, it might be due to some sort of von Restorff effect if this word stood out in some way from those surrounding it in list B. Clearly individual quirks of memory may occur in the test, and may affect the pattern of scores in particular subjects, as well as general factors such as the speed of presentation of items, the accent of the tester and whether subjects are given feedback about their performance or are warned about the later delayed trial or not. A second sign of malingering was the relatively small drop in recall scores on the delayed trial compared to the best score during learning. Lezak [22] suggests a normal forgetting threshold of 3 words, and in this study the average loss for non-malingerers was about 2 words. Malingerers actually ‘forgot’ less words between trials V and VI than did non-malingerers, perhaps thinking that as they had a lower apparent level of memory initially, they should retain what little they had. Alternatively, individuals who withhold words they can actually remember during trials I-V may then find it difficult to judge how many fewer words to produce on the

154

K. A. FLOWERSet al.

delayed recall task. Whatever the reason, in this respect they gave themselves away, as in genuine amnesia the amount of word loss following a distractor is related to primary learning ability [9] , so they should have ‘forgotten’ more words to simulate amnesia successfully. In contrast, malingerers over-estimated the difficulty of the recognition task, and on average recognised less words than they were able to recall. Given that their mean highest recall score did not exceed 9.0 there was much room for improvement, unlike in the non-malingering condition where a ceiling effect in the initial recall trials accounted for the small increment with recognition. These results reflect the lay person’s misconception of how easy recognition is compared to recall, and follows previous suggestions [ 11,221 that any reduction in recognition over recall is suspicious. In all three aspects of the AVLT which caught out malingerers, the effect would be enhanced if there was less of a ceiling effect on initial learning in non-malingerers. The effect shows up with the M-word task used here, but might be clearer with a 20-word list which was more demanding of normal memory. Amnesic subjects would not necessarily improve their recall scores, but would have a greater chance for improved recognition, and normal subjects would have more to lose on the delayed trial. Primacy and recency effects, moreover, would probably be enhanced as well where subjects are less able to master the whole list. Presenting a task to subjects which contains more information than they can hold or manipulate completely may also make it more difficult for malingerers to fake the pattern of genuine amnesic performance. For example, to produce a particular serial position curve not only requires subjects to remember the words themselves, but also to remember their exact position in the sequence so as to be able to choose accurately which items to report or exclude. So accurate faking requires subjects to have more information available to them than normal, and overloading the system will make it virtually impossible for them to do so. In general, then, tests aimed at detecting malingerers (other than those claiming to remember nothing at all) should involve high information loads which stress the working memory system, so that subjects cannot control their pattern of response easily and are forced to fall back on to their natural way of performing. (A similar proposal for forced choice tasks has been put forward by Pankratz [ 171 who suggested that a long forced choice sequence should be given to make it difficult for subjects to keep track of response probabilities and so produce a ‘chance’ level of performance accurately.) There was no difference in scores on lists A or B of the test, so they appear interchangeable. Nor did prior experience of the form of tbe test affect performance, except that there was some suggestion that those malingering on the second session if anything exaggerated their deficits. The effect, however, was slight. Overall the Rey AVLT, because of its structure, does seem relatively effective in providing evidence to help assessors detect potential malingering of amnesia. Of course the evidence provided here came from subjects who had no real interest in deceiving the tester, but while one cannot be sure that genuine malingerers (so to speak) would perform the same way, there is no evidence that they do not. In fact, subjects in an aroused or anxious state (as malingerers are likely to be) would if anything tend to exaggerate the kind of effects found here in normals, since anxiety adds to cognitive load and would thus make it more difficult for subjects to control the changes in response necessary to fake an amnesic deficit. The effects reported here also rely on the average layperson not knowing the characteristics of amnesic performance on tests of this kind. It would be interesting to see how effectively sophisticated subjects who know about amnesia and its characteristics could simulate a memory deficit on the test.

AMNESIA SIMULATION ON THE REY AVLT

155

REFERENCES 1. Blachstein, H., Vakil, E. and Hoofien, D. Impaired learning in patients with closed head-injuries: an analysis of components of the acquisition process. Neuropsychology 7, 530-535, 1993. 2. Haddad, L. B. and Nussbaum, P. Predictive utility of the Rey Auditory Verbal Learning Test with Alzheimer’s patients. Clinical Geronrologist 9, 53-59, 1989. 3. Query, W. T. and Berger, R. A. AVLT memory scores as a function of age among general medical, neurologic and alcoholic patients. Journal of Clinical Psychology 36, lC@9-1012, 1980. 4. Litvan, I., Grafman, J., Veendrell P. and Martinez, J. M. Multiple memory deficits in patients with multiple sclerosis: exploring the working memory system. Archives of Neurology 45, 607-610, 1988. 5. Sutker, P. B., Allain, A. N., Johnson, J. L. and Butters, N. M. Memory and learning performances in POW survivors with history of malnutrition and combat veteran controls. Archives of Clinical Neuropsychology 7, 431-444, 1992. 6. Waugh, M., Jackson, M., Fox, G. A. and Hawke, S. H. Effect of social drinking on neuropsychological performance. Brirish Journal of Addiction 84, 659-667, 1989. 7. Rosenberg, S. J., Ryan, J. J. and Prifitera, A. Rey’s AVLT performance of patients with and without memory impairment. Journal of Clinical Psychology 40, 785-787, 1984. 8. Mitrushina, M., Satz, P., Drebing, C. E. and Van Gorp, W. The differential pattern of memory deficit in normal aging and dementias of different etiology. Journal of Clinical Psychology 50, 246-252, 1994. 9. Mungas, D. Differential clinical sensitivity of specific parameters of the Rey Auditory Verbal Learning Test. Journal of Consulting and Clinical Psychology 51, 848-855, 1983. 10. Peaker, A. and Stewart, L. E. Rey’s Auditory Verbal Learning Test-a review. In Developments in Clinical and Experimental Neuropsychology, pp. 219-236, J. R. Crawford and D. M. Parker (Editors). Plenum Press, New York, 1989. 11. Rey, A. L’Examen psychologique dans le cas d’encephalopathie traumatique. Archives de Psychologie 37, 126-139, 1941. 12. Rey, A. L&amen Clinique en Psychologie. Presses Universitaires de France, Paris, 1964. 13. Benton, A. L. and Sprecn, 0. L. Visual memory test: the simulation of mental incompetence. Archives of General Psychiatry 4,79-83, 1961. 14. Spreen, 0. and Benton, A. L. The simulation of mental deficiency on a visual memory test. American Journal of Mental Defciency 67, 909-913, 1963. 15. Brandt, J., Rubinsky, E. and Lassen, G. Uncovering malingered amnesia. Annals of the New York Academy of Sciences 444, 502-503, 1985. 16. Wiggins, E. C. and Brandt, J. The detection of simulated amnesias. Law and Human Behavior 12,57-77, 1988. 17. Pankratz, L. A new technique for the assessment and modification of feigned memory deficit. Perceptual and Motor Skills 57, 367-372, 1983. 18. Bernard, L. C., Houston, W. and Natoli, L. Malingering on neuropsychological memory tests: potential objective indicators. Journal of Clinical and Experimental Neuropsychology 49, 45-53, 1993. 19. Williamsen, J. A., Johnson, H. J. and Eriksen, C. W. Some characteristics of post-hypnotic amnesia. Jouml of Abnormal Psychology 70, 123-131, 1965. 20. Baker, G. A., Hanley, J. R., Jackson, H. F., Kimmance, S. and Slade, P. Detecting the faking of amnesia: performance differences between simulators and patients with memory impairment. Journal of Clinical and Experimental Neuropsychology 15, 668-684, 1993. 2 I. Kaufman, A. Medicolegal aspects of head injuries: intellectual impairment and clinico-psychological assessment. Medicine, Science and the Law 18, 56-62, 1978. 22. Lezak, M. Neuropsychological Assessment, 2nd edn. Oxford University Press, New York, 1983. 23. Schacter, D. L. On the relation between genuine and simulated amnesia. Behavioral Sciences and the Law 4, 47-64, 1986. 24. Bernard, L. C. Prospects for faking believable memory deficits on neuropsychological tests and the use of incentives in simulation research. Journal of Clinical and Experimental Neuropsychology 12, 715-728, 1990. 25. Bernard, L. C. The detection of faked deficits on the Rey Auditory Verbal Learning Test: the effect of serial position. Archives of Clinical Neuropsychology 6, 8 l-88, 199 1.

156

K. A. FWWERSet al.

APPENDIX Distracters

and recognition passage used in list B

Warm Fly Leather Brush SloOl Windmill Car Moss Window Purse Light Comer Shampoo Lock Helmet John put the kettle on the stove, laughed as he caught sight of his wet shoes, clothes and towel, and sat at the desk to write to his friend about the morning’s events. Picking up paper and pencil, he put on his glasses and began. There hadn’t been a cloud in the sky and he’d been relaxing in his boat on the lake. Although he hadn’t caught a fish, he’d enjoyed the view of the valley and mountain, and he’d spotted a rare bird and some deer. On seeing the ranger with his gun, John had leapt into the water, swum ashore and run in the direction of the church like a frightened Jamb. Source: K. A. Flowers, M. R. Sheridan and Helen Shadbolt, Simulation of amnesia by normals on Rey’s Auditory Verbal Learning Test.