Article Text
Abstract
OBJECTIVE To test the hypothesis that, during random motor generation, the spatial contingencies inherent to the task would induce additional preferences in normal subjects, shifting their performances farther from randomness. By contrast, perceptual or executive dysfunction could alter these task related biases in patients with brain damage.
METHODS Two groups of patients, with right and left focal brain lesions, as well as 25 right handed subjects matched for age and handedness were asked to execute a random choice motor task—namely, to generate a random series of 180 button presses from a set of 10 keys placed vertically in front of them.
RESULTS In the control group, as in the left brain lesion group, motor generation was subject to deviations from theoretical expected randomness, similar to those when numbers are generated mentally, as immediate repetitions (successive presses on the same key) are avoided. However, the distribution of button presses was also contingent on the topographic disposition of the keys: the central keys were chosen more often than those placed at extreme positions. Small distances were favoured, particularly with the left hand. These patterns were influenced by implicit strategies and task related contingencies.
By contrast, right brain lesion patients with frontal involvement tended to show a more square distribution of key presses—that is, the number of key presses tended to be more equally distributed. The strategies were also altered by brain lesions: the number of immediate repetitions was more frequent when the lesion involved the right frontal areas yielding a random generation nearer to expected theoretical randomness. The frequency of adjacent key presses was increased by right anterior and left posterior cortical as well as by right subcortical lesions, but decreased by left subcortical lesions.
CONCLUSIONS Depending on the side of the lesion and the degree of cortical-subcortical involvement, the deficits take on a different aspect and direct repetions and adjacent key presses have different patterns of alterations. Motor random generation is therefore a complex task which seems to necessitate the participation of numerous cerebral structures, among which those situated in the right frontal, left posterior, and subcortical regions have a predominant role.
- random generation
- perseverations
- brain damage
- frontal lobe
Statistics from Altmetric.com
Random generation is a complex action that demands complete suppression of any rule governed behaviour. It has been studied in healthy subjects with the aim of determining the variables involved in strategy selection.1 However, it has rarely been studied in patients with neurological disorders, despite the fact that it could yield information on their preserved implicit strategies and the brain structures subtending them.
Many findings have shown that when normal subjects are asked to generate a random sequence of numbers or letters, their productions are far from what may be expected for true randomness.2 There are many causes for this. In certain instances, during a random generation of letters, subjects tend to use automatic series (...A, B, C...) or task related implicit strategies (....U, S, A,...).3 For number generation the tendency to follow obvious sequences such as odd-even (8-6-2...), or split randomised has been found.4 Persistence of one action schema, as shown by the discriminating power of the COUNT score, seems to constitute the major source of non-randomness in patients with frontal lobe damage or Parkinson’s disease2 and would be related to an impaired control mechanism. In other cases, a false concept of randomness has been incriminated; during random number generation, normal subjects tend to underestimate the numbers of direct repetitions (6-6 instead of 6-4) and are continuously influenced by the number previously generated. It has also been suggested that a rapid overload of memory resources may lead to errors in evaluation of the group of preceding numbers.1 2 Finally, subjective random generation has also been suggested to depend on factors other than those mentioned above. Variables such as extrasensory perception,5 or intrapersonal psychological modifications,6 or even menstrual cycles5 7 have been claimed to affect generation strategies. Despite these different strategies, the total number of elements will appear in equal proportion,8although the sequences will not reflect what would be expected in a random series—that is, few immediate repetitions of an element will be present.
Studies of subjective randomness have generally been applied to generation of numbers, which are subject to the stimulus related biases described above. The case of random manual pointing seems to be an altogether different situation. Not only are the stimuli non-verbal, as opposed to most other studies, but the strategies themselves would seemingly be dependent on internal action schemes and on external spatial contingencies—that is, the distance between target, distance from the subject, etc. In effect, it is generally proposed that movement behaviour does not depend only on the processing level but also on an interaction between motor structures and the surrounding topographical disposition.9 10 For example, in an experiment designed to study the pattern of random motor sequences in normal subjects, Lincoln et al showed that a systematic bias was present depending on the spatial disposition of the elements to which the subject had to point (square or circle).11Furthermore, brain dysfunction was shown to affect random pointing, at least in Parkinson’s disease2 12 We hypothesised that, during random motor generation, the spatial contingencies inherent to the task would induce additional preferences in normal subjects, shifting their performances farther from randomness. By contrast, perceptual or executive dysfunction could alter these task related biases in patients with brain damage.
Subjects and methods
Twenty five right handed control subjects with no neurological history (13 men, 12 women), recruited from hospital workers and their relatives through local advertisement, carried out the task. Their mean age was 54 (SD 15.3) (range 34 to 78) years. Inclusion criteria consisted of a good knowledge of oral and written French language and absence of present or past psychiatric illness.
A computer keyboard was presented on a table in front of the subject. The keys corresponding to the numbers 1, 2, 3, 4, 5, 6, 7, 8, 9, and 0, were covered by a red sticker so that the subject could not read the numbers, and the keyboard was presented in a vertical position, rotated by 90°, so that key 1 was nearest and 0 farthest from the subject. The remainder of the keyboard was masked. The instruction was: “Here is a set of 10 red buttons, all equal. You are going to press successively 180 times13 on these buttons, randomly, without any strategy. I will tell you when to stop. Speed is not important, so do not hurry. You will do it with the forefinger of your right (left) hand, then with the forefinger of the left (right) hand”. If the task was not clearly understood, further information was given to illustrate the concept, including phrases such as “lack of logical sequences”, “ignoring prior presses”, “lottery-like pressing”. This additional information was necessary mainly for patients with brain injury, especially those with moderate difficulties of comprehension or discourse analysis, to avoid any bias due to insufficient comprehension. A small demonstration of 10 successive trials was given by the examiner, who produced a pseudorandom suite, alternating small movements, large movements, and repetitions on the same button. Then a practice session of 10 trials was given to the subject, and repeated if the examiner thought that the subject did not fully understand the instructions. The subject was then asked to begin. The examiner remained nearby, encouraging the subject by telling him or her how far into the experiment he was, and with the sentence “go on” if the subject stopped before the end of the trial. Half of the subjects began with their right hand on the first trial. The data were recorded on a Macintosh Classic computer for further analysis.
Twenty seven right handed stroke patients (19 men, eight women) with right brain lesions and 23 patients (17 men, six women) with left brain lesions underwent the same task. The patients were recruited either from the acute neurology or rehabilitation wards during the same period as the control subjects. Mean age was 58 (SD 13.8) (range 36 to 89) years for patients with left brain lesions and 61 (SD 12.8) (30-83) years for patients with right brain lesions. Inclusion criteria for the patients were absence of previous neurological events, postonset delay (between one week and one month), manual preference (no left handers), and brain lesion aetiology (ischaemic stroke). The task was carried out provided that they were able to execute a tapping task with their valid hand. Only adults were included in the research. As age (except for childhood), schooling, and IQ do not seem to play any part in a random generation,1 14 these variables were not taken into account. Aphasic patients participated in the sessions if they could be made to understand the task. Only performances of the hand ipsilateral to the brain lesion were taken into account. The patients had a standard clinical neurological examination and a brief clinical neuropsychological evaluation. Spatial neglect was evaluated clinically with two simple graphic tasks, line bisection and circle counting. Language comprehension was tested by the mean of the Montréal-Toulouse β 86 battery15 16 which includes 11 designation tasks using concrete objects, actions, and short sentences (such as: “Out of four figures, point to the figure where the lady walks behind the dog”) Aphasia and neglect were scored 0, 1, or 2 depending on whether symptoms were absent, mild, or medium to severe. The lesion site was established using simple horizontal plates of CT or MRI. This allowed an optimal evaluation of the extent of frontal lobe damage. Sites of lesions were defined with respect to frontal involvement—for example, frontoparietal lesion was called frontal and parietotemporal lesion, non-frontal. Pure subcortical lesions were labelled subcortical. All subjects were tested by the same examiner (JMA).
Measures of randomness analysed at the end of the data collection were: (1) frequency distribution of each of the keys pressed by the group of right hands, then by the group of left hands; (2) the distribution of the intervals between two successive key presses, which included the number of direct repetitions to reflect repetition avoidance and the adjacent key presses to reflect serial counting. The results referred to both hands of the group of control subjects, to the left hand of patients with left brain lesions, and to the right hand of patients with right brain lesions. Statistical analyses were carried out on raw data by comparing the hand ipsilateral to the lesion in patients with brain damage with the same hand in normal subjects. The homogeneity of key presses within the groups was also verified through a non-parametric analysis of reliability across subjects in each group.
Results
Localisation of brain lesions was studied in patients on their brain CT or MRI and compared with axial standard plates, Table 1 shows the different localisations. There were 18 cortical or corticosubcortical lesions with frontal involvement of which 11 were on the right side and seven on the left side. Nineteen patients had cortical or corticosubcortical lesions without frontal involvement; among these, eight were situated on the right and 11 on the left. Finally, there were eight patients with right subcortical lesions and five with left subcortical lesions. Patients with right brain lesions and patients with left brain lesions were comparable in terms of lesion localisation (frontal, parietal, temporal....), but there was a tendency towards a larger lesion size in the patients with right brain lesions. This was due to a greater number of large frontotemporal lesions (eight patients) in right brain lesions than in left brain lesions (four patients). In the left brain lesion group, aphasia was absent in only five patients. In the right brain lesion group, signs of neglect were present in all but eight patients (table 1). The frequencies of presses on each key and the frequencies of the intervals between two presses were considered separately for the right and left hands. The number of times a given key was pressed was computed as a percentage of the total number of presses and the same procedure was carried out for the intervals. Tables 2 and 3 give these percentages along with values expected for randomness.8
Control right hand, control left hand, right frontal brain lesions, non-frontal right brain lesions, subcortical right brain lesions, and frontal left brain lesions were homogenous with respect to distribution of key presses as shown using a Kruskal-Wallis non-parametric test (respectively: H=32.1, df=24, P=0.13; H=32.28, df=24, P=0.12; H=4.5, df=7, P=0.72; H=10.8, df=9, P=0.21; H=12.04, df=7, P= 0.1; H=4.4, df=7, P= 0.74). Only patients with non-frontal lesions and those with subcortical left brain lesions showed a heterogeneous between subjects distribution (H=35, df=11, P<0.001 and H=14.22, df=4, P=0.01).
CONTROLS
Performance did not follow the expected square distribution, but a normal distribution, with the central keys being pressed more often than the peripheral keys (table 2). Comparison of key presses obtained with the right hand and the left hand showed no significant difference in the pattern of performance (χ2 =5.9, df=9, P=0.75).
Analysis of intervals between two successive key presses showed that subjects produced less direct repetitions (successive presses on the same key, corresponding to a 0 interval) than expected in randomness: 1.8% with the right hands and 2.6 % with the left hands compared with an expected 10%. By contrast, successive pointing to adjacent keys (corresponding to the intervals −1 and +1) were much more frequent than predicted: 50% with the right hand and 59% with the left hand. Table 3 shows these two main measures of randomness. Difference between the right hand and the left hand was significant (χ2=111.6, df= 9, P<0.0001).
PATIENTS WITH RIGHT BRAIN LESIONS
Performance of the right hand was analysed (table 2) for the right non-frontal brain lesions, right subcortical brain lesions, and right frontal brain lesions. Pattern of distribution of key presses in patients with right frontal brain lesions was significantly different from that of controls (χ2= 92.1, df=9, P<0.0001). The 10 keys were pressed more equally and the distribution was closer to the expected random distribution (table 2). In patients with right subcortical brain lesions and those with right non-frontal brain lesions, no difference was found compared with controls (χ2 =15.3, df=9, P=0.09; and χ2=7.9, df=9, P=0.54 respectively).
Table 3 shows the frequency of occurrence of every one of the existing intervals which was computed as a percentage for every group separately along with the values expected in theoretical randomness.
Analysis focused on the two main measures of randomness: the number of immediate repetitions (interval=0) and the number of adjacent key presses (interval=1). Between group statistics were carried out separately for these two measures using a Mann-Whitney non-parametric test. Table 4 gives the relative percentages, obtained by subtracting the mean percentage of the right hand control group from the mean percentage of the lesion groups. The number of direct repetitions (interval=0) was significantly increased in the right frontal brain lesions group, compared with the controls (Mann-Whitney:z=−2.40, P=0.017). In this case, the percentage of immediate repetitions (9.8%) approached theoretical randomness (10%). This was not found for the right subcortical brain lesion group (z=−1.37, P=0.17) or the right non-frontal brain lesion group (z=−0.83, P=0.41).
Immediately adjacent key presses (interval=1) were found in greater amounts for patients with right subcortical brain lesions (z=−3.39, P=0.001) and right frontal brain lesions (z=−2.54, P=0.01), but not for those with right non-frontal brain lesions (z=−1.39, P=0.16).
PATIENTS WITH LEFT BRAIN LESIONS
Performance of the left hand was analysed (table 2) for the left non-frontal brain lesion, left subcortical brain lesion, and left frontal brain lesion groups. Distribution patterns of button presses did not show any significant differences from the ones produced by the left hands of control subjects either for the left non-frontal brain lesion group (χ2 =13.1, df=9, P=0.15) or in the left frontal brain lesion group (χ2=12.8, df=9, P=0.15). By contrast, the left subcortical brain lesion group behaved differently from controls (χ2=23.5, df=9, P=0.005) and pressed the keys more equally, with a distribution closer to theoretical randomness.
Again, the frequency of occurrence of every interval was computed as a percentage for every group (table 3) along with the values expected in theoretical randomness.
Between group comparisons were carried out separately on intervals 1 and intervals 0, using a Mann-Whitney non-parametric test. Table 4shows the relative percentages (subtraction of the mean percentage of right hands of the control group from the mean percentage of the lesion groups).The number of direct repetitions (interval=0) was not significantly different from controls either in the left non-frontal brain lesion (z=−0.85, P=0.39), or the left subcortical brain lesion group (z=−0.6 P=0.55), or in the left frontal brain lesion group (z=−1.18, P=0.23).
Adjacent key presses (interval=1) were significantly decreased in the left subcortical brain lesion group, compared with controls (z=−2, P=0.046) and marginally increased in the left non-frontal brain lesion group (z=−1.74, P=0.08).
Discussion
The results obtained with subjective random generation of pointing movements in patients with brain lesions and normal controls led to the following observations:
KEY DISTRIBUTION
In normal subjects and in most patients with brain lesions, the distribution of key presses for both hands tends to be bell shaped, the central keys being pressed more often than the peripheral ones. Patients with lesions involving the right frontal cortical regions or with limited left subcortical brain lesions show a different pattern: the distribution of key presses with their non-paretic hand is nearer to the one expected for randomness—that is, square shaped.
SEQUENTIAL PLANNING
As in random number generation, the number of immediate repetitions of key presses is low among normal subjects.
Further, in controls, both hands tend to make smaller movements (interval=1) than is to be expected in true randomness. Interestingly, strategies differed significantly between hands in control subjects. This prevented any direct comparison between the right hand of patients with right brain lesions and the left hand of patients with left brain lesions. Patients with right frontal brain lesions, however, made more immediate repetitions, reaching the prediction of theoretical randomness (10%). Brain lesions also induced an increase in adjacent key presses in patients with right frontal brain lesions, right subcortical brain lesions, and marginally in those with non-frontal brain lesions but a decrease in patients with left subcortical brain lesions, in whom small movements were less frequent than in controls.
The bell shaped (as opposed to random) distribution of key presses has not been found in other random generation experiments with cognitive tasks. For example, in mental number generation (mental exercise of throwing a dice a successive number of times), the distribution between the six possible numbers has been found to be more even.8In a motor random generation experiment with keys displayed randomly within the visual field, Mittenecker13 did not find a similar distribution. The main difference between the number generation, Mittenecker’s experiment, and our motor generation task is that the disposition of the keys was vertical in our case. Hence, the tendency towards a bell shaped distribution in random generation of movements seems also to be influenced by the way in which the targets are presented. Patients with brain lesions do not behave differently from normal subjects in key distribution, except for patients with frontal cortical involvement and with limited left subcortical lesions, who press the keys more evenly. One explanation could be that patients with right brain lesions, and particularly right frontal brain lesions, have difficulty integrating spatial and visual information, because of visual neglect (table 1). This explanation is not valid for the group with left subcortical brain lesions as none of these patients showed neglect. Another possibility could be that patients with right frontal brain lesions have larger lesions than the other groups and that the effect is related to the size of the lesion. We exclude this possibility because patients with left frontal brain lesions also include those with large lesions and they do not differ from controls. Also, patients with left subcortical brain lesions, whose behaviour is similar to those with right frontal brain lesions, have smaller limited infarcts. The third hypothesis is that the pattern of key distribution, as the repetition avoidance, depends on implicit strategies which necessitate the integrity of frontal-subcortical loops, and that there is a right hemispheric cortical and left subcortical dominance in the control of these mechanisms.17 The fact that patients with lesions affecting the right frontal and left subcortical structures have a more square distribution points to a role for specific cortical and subcortical structures. Because of the between subject variability found in controls, analysis of smaller subgroups of patients was not carried out. The hypothesis, however, would also fit with complementarity and reciprocal inhibition between cortical and subcortical areas.
Analysis of the intervals between two successive key presses suggests that sequential strategies in healthy subjects are governed by the same rules in motor generation as in other random generation tasks. In particular, we have found the same tendency to underestimate direct repetitions. However, some differences related to the motor aspect of the task appear: the choice of adjacent keys between two trials is far more frequent than the theory predicts, especially with the left hand. Control subjects seem to favour small movements. Search for economy of movement cannot be the only cause because it would also favour nearer keys (1, 2,...) as opposed to farther keys (...9, 10). Furthermore, the tendency to favour small relative distances has also been shown for a verbal response.11 Rather than suggesting a single explanation, this pattern of sequences points to a combined role of implicit strategies, task induced behaviours, and spatial contingencies.
These strategies—that is, the tendency to inhibit direct repetitions and to favour small movements, are significantly sensitive to brain lesions, independently of the side and site, as shown in the results, but particularly to lesions involving right frontal. as well as right and left subcortical structures.
Direct repetitions were significantly more frequent in patients with right brain lesions, particularly those with frontal lesions, who showed an overall “random” distribution of direct repetitions: patients with right frontal brain lesions do not seem to inhibit direct repetition. The important spatial aspect of perceptive analysis and motor decision could be taken into account to explain the right frontal dominance compared with the left. However, right frontal lobe function has also been implicated in repetition avoidance in a random number generation task, obviously containing a verbal component.17 Our findings suggest that this right hemispheric dominance is independent of stimuli (number, keys) for cortical frontal lesions, and could be correlated with a greater tendency to make continuous motor perseverations (an abnormal prolongation of a current activity) particularly in patients with right brain lesions.18
Adjacent key presses show a different pattern of alteration compared with direct repetitions depending on the lesioned areas. For cortical lesions, both right frontal and more marginally left posterior cortical lesions increase adjacent key presses meaning that, for this measure, a clear unilateral effect cannot be established. However, left frontal lesions induce no alterations in frequency, suggesting that left frontal structures do not play a significant part in this type of strategy. Subcortical lesions also alter the frequency of adjacent presses confirming previous findings in number2 and motor12 random generation. However, contrary to that found with cortical lesions, there is a strong interhemispheric difference for subcortical lesions. Right subcortical lesions increase adjacent presses, whereas left subcortical lesions decrease them.
It seems then that the pattern of modifications for immediate repetitions is different from that with adjacent key presses. These differences suggest that cognitive processes underlying the production of immediate repetitions and adjacent key presses must be dissimilar, even though they concern the same task. Recurrent perseverations18—that is, the inappropriate and unintentional maintenance of response after disappearance of the stimulus—seemed to be sensitive to left brain lesions, whereas continuous perseverations seemed to be linked to right brain lesions.18 In our results, the pattern of immediate repetitions showed a similarity to continuous perseverations, whereas adjacent key presses resembled recurrent perseverations. However, to test this hypothesis, it would be necessary to compare directly, in patients with cortical and subcortical lesions, the alteration of strategies in random generation and the production of perseverations.
The other essential finding is that cortical and subcortical lesions induce opposite motor behaviour, as shown by informal comparisons. In the right hemisphere, lesions with frontal cortical involvement lead to a more random distribution, more direct repetitions, and less adjacent key presses than subcortical lesions. In left brain lesions the opposite is found; frontal cortical lesions lead to a less random distribution, less direct repetitions, and more adjacent key presses than subcortical lesions. This finding is crucial because it points to opposite and lateralised effects of frontal and subcortical lesions.19 This result goes along with other neurophysiological and clinical findings which indicate a reciprocal inhibitory effect of basal ganglia and frontal structures.19 20
In conclusion, the results suggest that motor random generation is influenced by implicit strategies and false concepts of randomness as in other random generation tasks, but also by the motor aspect of the task and spatial contingencies of the set itself. Analysis of the influence of brain lesions on strategies shows some similarity with the continuous and recurrent motor perseverations. Such a complex task necessitates the participation of numerous cerebral structures, involving mainly right frontal, left posterior, and subcortical areas. Cortical and subcortical structures seem to have different roles, depending on the strategy that is used.
Acknowledgments
We thank Professor M Reicherts for help and advice in the statistical procedures used for this study. The study is partly supported by the Swiss National Science Foundation Grant No 31-42 571 to JMA.