Br. 1. Psycho/. (1978). 69. 9-15
Printed in Great Britain
9
The effect of catch-trials on speed and accuracy among introverts and extraverts in a simple RT task John Brebner and Rosemary Flavel Three predictions from the model of extraversion put forward by Brebner & Cooper (1974) were tested in a simple RT task. In line with the model, extraverts were found to make more commissive errors and to be more affected by an increase in the catch-trial rate. The third prediction -that extraverts would tend to produce longer runs of decreasing RTs than of increasing RTs in the condition with 10 per cent catch-trials, was not borne out. This effectwas, however, seen in the condition with 40 per cent catch-trials although it was not statistically significant, P= 0.06.
In a recent article, Brebner & Cooper (1974) proposed a simple model of introversionextraversion. This model made the assumption that the effects of stimulation impinging upon the individual, and the demands for active responses from him, were independent of each other and that either could have central excitatory or inhibitory effects. From this standpoint, it was suggested that both introverts and extraverts are characterized by an imbalance between the effects of stimulation on the one hand and response organization on the other. In the case of the introvert, stimulation was hypothesized to create an excitatory state (S-excitation)but response preparation to build up an inhibitory state (R-inhibition).The extravert was characterized as a person with the opposite tendency, that is, to generate excitation from the organization and emission of responses (R-excitation)but, in the absence of response demands, the effects of stimulation rapidly become inhibitory (S-inhibition). Those types of ‘response’ which occur as perceptual or cognitive integrations, and the feedback resulting from one’s own responses, were both specifically classified as stimuli within the model. Because the introvert tends to generate excitation from stimulation but inhibition from active responding, Brebner & Cooper described the introvert as ‘geared to inspect’, and his extravert counterpart as ‘geared to respond’ because of his opposite tendency to generate R-excitation but S-inhibition. Thus, the extravert might be more accurately described as ‘response hungry ’ rather than ‘stimulus hungry * a term in current use, even though with sufficiently varied and intense stimulation it is possible to maintain S-excitation even in the extreme extravert. The first evidence supporting this specific model (Brebner & Cooper, 1974) showed that the reaction times of extraverted subjects were more affected by S-inhibition than those of introverted subjects. This result was obtained under conditions which precluded explanation in terms of characteristic arousal levels or feedback-mediated S-inhibition. Moreover, the strength of the effect was shown to vary with time on task, which was also predictable from the model. The experiment reported below, tested the complementary hypothesis derived from the same model that R-excitation would be stronger for extraverted subjects than for introverts. The predictions tested were that, in a simple RT task involving ‘catch-trials’: (1) Extraverted subjects would make more commissive errors than introverted subjects. Since ‘response hungry ’ extraverts are postulated to generate R-excitation from the organization and emission of response, and to depend on R-excitation to maintain their overall state of arousal, they should show a stronger tendency to respond in the absence of the appropriate signal than introverts who are postulated to generate R-inhibition from the organization and emission of responses. (2) The speed and accuracy of extraverted subjects would be relatively more affected than
10
John Brebner and Rosemary Flavel
that of introverts as the proportion of catch-trials increased. This second result is predicted on the grounds that the degree of R-excitation is related to the response rate which is affected by the proportion of catch trials. A further relevant statement by Brebner & Cooper (1974) was that ‘. .where R-excitatory potential is high but S-excitatory potential is low the extraverts’ performance would be relatively better until S-inhibition increased as a function of responding and acted to decrease the overall excitatory potential in the extraverted subjects ’. If in the present experimental situation responding is sufficiently frequent to build up S-inhibition through feedback stimulation, then one might predict (3) an effect akin to repeated ‘involuntary rest pauses ’ either in occasional failures to respond or increased response latencies. These temporary failures differ from the general lowering of responsiveness associated with a continuing high S-inhibition level which results from the lack of opportunity to respond. That effect is expected to reduce speed of performance among extraverts as the catch-trial rate rises. The modulating effect of the ‘involuntary rest pause’ phenomenon should, on the other hand, occur during periods of repeated responding and is most likely to be found where the catch-trial rate is lowest. The parallel between this effect and Eysenck’s (1955) original ‘reactive inhibition’ will be evident.
.
Method Apparatus The stimulus for response was the digit ‘ 1 ’ presented on a Nixie tube located behind a small glass screen set in a timing and display control unit with a matt black painted surface. This unit was linked both to a paper-tape reader, through which programmed tapes controlling the stimulus display were fed, and also to a teletype which encoded the relevant information regarding stimuli and reaction times on to paper tape. The punch was also connected to the response Morse key operated by the subjects. Inter-trial intervals were regular, of 2.3 sec duration, and signals were displayed for 200 msec. Each trial was preceded by a warning light, which became visible in the top left-hand corner of the display screen before the signal was due to occur; the warning signal was of the same 200 msec duration as the signals.
Subjects Sixteen undergraduate subjects from the University of Adelaide, ranging in age from 17 to 24 years (mean age = 19.4 years), took part in the experiment. Selection was made on the basis of E-scores obtained on Form A of the EPI. Those classified as introverts scored in the range 2-5, while the extraverts’ scores ranged from 19 to 23. All subjects had normal vision.
Procedure The subject was seated approximately 1 metre from the display screen with his index finger resting lightly on the response key. To ensure that the subject understood the nature of the task he was given a practice run of five signals to which he responded. Following this, the subject was given a block of 100 trials without catch-trials and was required to respond as quickly as possible to every signal. The purpose of this task was to check for differences in reaction times between the introverts and extraverts not due to the experimental treatments of this study. The experimental session was then begun. In the experimental condition subjects were required to respond as quickly as possible to the digit ‘ 1 ’ but not to respond on any catch-trial. Catch-trials were randomly distributed throughout the series and consisted of the warning light followed by a blank screen. Three blocks of 200 trials were given, representing three differentstimulus conditions. Under condition A, 10 per cent of the total were catch trials, in condition B there were 40 per cent catch-trials. and in condition C 70 per cent. The order of presentation of the three conditions was varied randomly to minimize order effects.
Results Commissive errors Table 1 shows the number of times responses were made on catch-trials by subjects under the three conditions of the experiment. It is clear from the table that extravert subjects tended to make more of these commissive errors, and that the frequency of such errors is inversely related
Introverts, extraverts and efect on catch-trials
II
Table 1. Errors committed by responding on catch trials
Condition Subject
E-score 21 23 22 21 20 19 22 19
B (40%)
C (70% catchtrials)
Total
8 4 0 4 9 6 5 0 36
5 2 0 3 11 5 3 0 29
20 11 4 16 34 21 12 2 120
0
1
2 2 2 2 2 2
0
1 0
A (10%)
7 5 4 9 14 10
4 2 55
9 10 11 12 13 14
5 4 2 4 4
15
5 4
16
4
4 16
1
1
6 2 0
0 0 0 0
2 2 4 8 4 2 3
1 3
6 31
1 1
12
to the proportion of catch-trials. 2 confirms the first prediction that extraverts do make significantly more errors (xz= 52.4, d.f. = 1, P < 0.01). The difference between the three conditions is also significant (2= 15.3, d.f. = 2, P < 0.01). Given that the data in Table 1 are evidently unsuitable for parametric statistics, and that the power of most published distribution-freeanalyses of variance is extremely low (Wilson, 1956), the interaction between personality types and the three conditions proposed in the second prediction was tested by performing a Mann-Whitney U test on the difference in errors between the two extreme conditions A and C. This gave U = 14.5 which is significant ( P < 0.05) for the appropriate one-tailed test. While this result is in line with the prediction made, nevertheless, since introverts made so few errors at all, it is possible it merely reflects the fact that the closer to the ‘zero error’ bound the introvert group is, the smaller must be the change in errors across the conditions of the experiment for these subjects. Further research under conditions in which the commissive error rate of introverfs was higher than in this study would be useful provided the line between the different S-inhibition effects above was retained. A second source of commissive errors lies in the number of anticipatory responses made by the subject. By this is meant simply that the subject would have responded whether a signal appeared or not, but that where a signal did occur, a response is not an error unless clearly programmed before the occurrence of the signal. The limit of 80 msec was accepted as the criterion for an RT being defined as anticipatory. All longer RTs were accepted as genuine responses to the signal. This category of anticipatory commissive errors proved illuminating. Table 2 shows the number of such errors in each condition for both groups. The fact that none of the introverts made such errors is striking and makes any statistical test between the two groups unnecessary. A significant difference across conditions in the predicted direction is obtained from the extraverts, x2=21.1, d.f. = 2 , P 0.05). which argues against any general tendency to trade accuracy for speed among the extraverts. But the two groups do differ in the effect which raising the catch-trial rate has on their RT. The interaction conditionsxpersonality type is significant (F=8.5, P < O a O l ) , and again supports prediction (2) that extraverts will be more affected than introverts as the catch-trial rate increases. The ‘involuntary rest pause’ phenomenon The temporary performance decrement predicted for extraverts with low catch-trial rates was tested in the following way. The decrement should show itself either in missing responses or in slow RTs. There were only three failures to respond in the whole experiment, and although th? extraverts accounted for twice as many of these as the introverts, such a result is not quite 100 per cent conclusive. Rather better evidence is provided by changes in the speed of responding. Since it is a short-lived decrement in performance which was predicted, the mean length of runs of RTs * Due to a mechanical failure, the R T data from one extravert subject (6) was unreliable. His results were omitted from R T analysis. To equate the numbers in the two groups for Anova one introvert subject (14). chosen randomly, was also dropped from the
R T analysis.
Introverts, extraverts and efect on catch-trials
13
which decreased in latency was compared with that for RTs which increased in latency. If extraverts generate R-excitation from organizing responses, but S-inhibition from the feedback effects of responding, their performance should tend to increase in speed as R-excitation increased until the rise in S-inhibition was sufficiently strong to oppose it at which point RT should lengthen. Responses providing more feedback, as in more effortful or more complicated responses would be expected to build up S-inhibition at faster rates than easier responses. Since, in this experiment the same response is repeated, the build up of S-inhibition from feedback is assumed to proceed at a constant rate during responding. In the present context what is expected is that for extraverts only, where the catch-trial rate is low, the mean length of runs of RTs in which response latency decreased should be greater than the mean length of runs of increasing RTs. Table 4 shows the mean lengths of increasing and decreasing runs of RTs. Where the sign is positive, the mean length of runs in which RT became progressively faster exceeded the mean length of runs in which RT became progressively slower. Table 4. Mean run lengths
Condition B (40%)
Condition A (10%)
RT incr.
Diff.
Condition C (70%)
RT decr.
RT incr.
Diff.
RT decr.
RT incr.
1.60 1.46 1-44 1.48 1.71 1.48 1.58
(=)
(+) (-) (+) (t) (=) (-)
1*60 1.66 1.34 1*58 1.79 1.48 1.34
1.34 1.30 1.34 1.47 1.40 1.51 1.68
(+) (+) (+) (+) (+) (+) (-)
1*50 1*54 1.79 1.86 1.52 1 *56 1 *45
1.55 1a45 1-55 1-61 1a65 1*67 1 -53
1.58 1.50 1.40 1-43 1.39 1.61 1.58
(-) (+) (-) (-) (+) (-) (+)
1.52 1.66 1-34 1.41 1*43 1.59 1.61
1-42 1.51 1.42 1.59 1-47 1.58 1.62
(+) (-) (+)
1a56 1 *50 1.55 1-81 1a 6 6 1.54 1-42
1 *58 I a35 1*67 1*37 1-62 1*30 1a38
(t) (t) (-) (-)
Diff.
RT decr.
By inspection of Table 4 the prediction (3) above is not supported in condition A. However, in condition B, the results are in the predicted direction although, on a binomial test, they fail to reach statistical significance;on the appropriate one-tailed test P= 0.06. Why the predicted effect should be at its strongest in condition B becomes a matter for speculation. Possibly since mean RT was at its fastest in condition A, ‘regression toward the mean * occurs more often in that condition. This would reduce the mean length of those runs on which RT continued to decrease. A slightly more complex alternative interpretation is that although R-excitation was highest for extreme extraverts in condition A, S-inhibition deriving from feedback also exerts its strongest influence in that condition. It is in condition A that responses are most frequent. S-inhibition from feedback builds up over repeated responding but tends to dissipate in the period between responses. S-inhibition from feedback, thus opposes R-excitation in condition A whefe its effect is stronger than in other conditions, though for most of the time it is weak in relation to the
14
John Brebner and Rosemary Flavel
strength of R-excitation in the condition. The overall speed of responding, therefore remains highest in condition A, but there is an attenuating effect on runs of decreasing RT which is due to S-inhibitionfrom feedback. Following this line through, in condition B, R-excitation is lower, but the reduction in S-inhibition from feedback allows relatively longer runs of decreasing RT. In condition C,the high degree of S-inhibition for extraverts resulting from low response demands tends, if anything, to produce runs of increasing RT for these subjects. The ad hoc nature of this interpretation guarantees that it matches the findings, but further research seems to be indicated to establish the different action of the two types of S-inhibition postulated. The theory outlined above is an attempt to amalgamate some of the general features of earlier theories which have been substantiated experimentally. For example, there is by now ample evidence to show that, in some circumstances, extraverts generate response-mediated inhibition faster than introverts do, as Eysenck’s (1955) reactive inhibition theory argued. There is also support for the later suggestion that extraverts and introverts differ in their arousability. Unfortunately, however, these explanations seem to be regarded at best as independent of one another, and at worst as alternatives. In some studies it seems that holding too narrowly to one or other of these explanations produces predictions which are unrealistic when the actual experimental task is considered. Thus, the finding by Nocita (1973) that extraverts showed repetitive, perseverative behaviour in an insoluble choice task, ran counter to the hypothesis in that study that responding creates an inhibitory state in extraverts who should, therefore, tend not to repeat responses. Also, the prediction by Buckalew ( 1973) that extraverts would have longer simple RTs since introverts have lower sensory thresholds, was not supported, and it was the extraverts who produced shorter RTs. The present theory predicts that ‘response hungry ’ extraverts will be faster on their initial response and respond more frequently, encompassing both these experimental outcomes. An experiment by Brebner & Cooper (unpublished) has confirmed precisely these predictions within an ‘inspect or respond’ task using coloured slides. Other studies of the behaviour of the two personality types under sensory deprivation and vigilance conditions seem to fit the view that extraverts are prone to R-excitation but S-inhibition, so that they tend to be impulsively active and depend upon responding to maintain their alertness, and, conversely are less aroused by stimulation in the absence of activity. Miyashiro & Russell’s (1974) finding that extraverts did not seek longer durations of a sound stimulus during sensory deprivation, but did seek the stimulus more frequently, points to the extraverts’ need to respond actively. Morgenstern, Hodgson & Law’s (1974) finding that extraverts’ performance at a learning task improved under distraction while that of introverts worsened, is accompanied by the observation that the exploratory movements of the two groups, which were made to improve the artificiallydistorted character of the stimulus, differed from one another. The movements of the introverts were few apd slow in comparison to the large extravagant movements of the extraverts. Here again the need for active responding emerges as does the tendency among extraverts to vary their movements more than introverts in an experiment by Hill (1975). Finally, a relationship between impulsivity and inability to maintain attention in a vigilance task has been reported (Thackray, Jones & Touchstone, 1973), using items from Form A of the EPI to measure impulsiveness. No full review would be appropriate here, but the studies outlined fit the same pattern of extraverts seeking activity to maintain arousal rather than more passive forms of self-stimulation such as visual inspection. This does not mean that extraverts do not seek stimulation. With sufficiently intense and varied stimulation it is possible to maintain arousal even in the extreme extravert, but it would appear from the experimental evidence that active responding is more effective.
Introverts, extraverts and effect on catch-trials
I5
Conclusion The findings of this study support the main hypothesis that R-excitation is stronger in extraverted individuals. From these results together with the data of the previous study (Brebner & Cooper, 1974) a coherent picture of the extravert is emerging which shows him to be prone to both types of S-inhibition and dependent upon R-excitation to maintain his responsiveness. The distinction between S-inhibition generated by feedback from responding and S-inhibition which builds up in the extravert when response demands are low, was drawn in the original statement of the model. The data of this experiment warrant further research into these different types of S-inhibition. References BREBNER, J. & COOPER,C. (1974). The effect of a low rate of regular signals upon the reaction times of introverts and extraverts. J. Res. Person. 4, 263-276. BREBNER,J. & COOPER,C. (1976). Stimulus-or response-induced excitation, a comparison of the behaviour of introverts and extraverts. J. Res. Person. (in press). L. W. (1973). Relationship between a BUCKALEW, physiological and personality index of excitability. Physiol. Psychol. 1, 158-160. EYSENCK, H. J. (1955). A dynamic theory of anxiety and hysteria. J. Ment. Sci. 101, 28-51. HILL,A. B. (1975). Extraversion and variety-seeking in a monotonous task. Br. J. Psychol. &(I), 9-13. MIYASHIRO, C. M. & RUSSELL,D. L. (1974). Experimental participation as a source of stimulation in sensory and perceptual deprivation studies of stimulus-seeking behaviour by introverts and extraverts. Percept. Mot. Skills 38, 235-238.
MORGENSTERN,F. S., HODGSON, R. J. & LAW,J. (1974). A comparison of introverts and extraverts exposed to conditions of distraction and distortion of stimulus in a learning task. Ergonomics 17(2), 21 1-220. NOCITA,F. E. (1973). The relationship of frustration, neuroticism and extraversion to repetitive behaviour. Diss. Abstr. Int. 33(8B), 3990-3991. THACKRAY, R. I., JONES,K. N. & TOUCHSTONE, R. M. (1973). Personality and physiological correlates of performance decrement on a monotonous task requiring sustained attention. Civil Aeromed. Inst., Oklahoma, Fed. Aviat. Agency, AD. 777, 825, 13 pp. WILSON, K . V. (1956). A distribution-free test of analysis-of-variance hypotheses. Psychol. Bull. 53, 96-10].
Received 8 June 1976; revised version received 22 August 1976 Requests for reprints should be addressed to Dr J. Brebner, Department of Psychology, University of Adelaide, Adelaide, South Australia 5001. Rosemary Flavel is at the same address.
Printed in Great Britain
9
The effect of catch-trials on speed and accuracy among introverts and extraverts in a simple RT task John Brebner and Rosemary Flavel Three predictions from the model of extraversion put forward by Brebner & Cooper (1974) were tested in a simple RT task. In line with the model, extraverts were found to make more commissive errors and to be more affected by an increase in the catch-trial rate. The third prediction -that extraverts would tend to produce longer runs of decreasing RTs than of increasing RTs in the condition with 10 per cent catch-trials, was not borne out. This effectwas, however, seen in the condition with 40 per cent catch-trials although it was not statistically significant, P= 0.06.
In a recent article, Brebner & Cooper (1974) proposed a simple model of introversionextraversion. This model made the assumption that the effects of stimulation impinging upon the individual, and the demands for active responses from him, were independent of each other and that either could have central excitatory or inhibitory effects. From this standpoint, it was suggested that both introverts and extraverts are characterized by an imbalance between the effects of stimulation on the one hand and response organization on the other. In the case of the introvert, stimulation was hypothesized to create an excitatory state (S-excitation)but response preparation to build up an inhibitory state (R-inhibition).The extravert was characterized as a person with the opposite tendency, that is, to generate excitation from the organization and emission of responses (R-excitation)but, in the absence of response demands, the effects of stimulation rapidly become inhibitory (S-inhibition). Those types of ‘response’ which occur as perceptual or cognitive integrations, and the feedback resulting from one’s own responses, were both specifically classified as stimuli within the model. Because the introvert tends to generate excitation from stimulation but inhibition from active responding, Brebner & Cooper described the introvert as ‘geared to inspect’, and his extravert counterpart as ‘geared to respond’ because of his opposite tendency to generate R-excitation but S-inhibition. Thus, the extravert might be more accurately described as ‘response hungry ’ rather than ‘stimulus hungry * a term in current use, even though with sufficiently varied and intense stimulation it is possible to maintain S-excitation even in the extreme extravert. The first evidence supporting this specific model (Brebner & Cooper, 1974) showed that the reaction times of extraverted subjects were more affected by S-inhibition than those of introverted subjects. This result was obtained under conditions which precluded explanation in terms of characteristic arousal levels or feedback-mediated S-inhibition. Moreover, the strength of the effect was shown to vary with time on task, which was also predictable from the model. The experiment reported below, tested the complementary hypothesis derived from the same model that R-excitation would be stronger for extraverted subjects than for introverts. The predictions tested were that, in a simple RT task involving ‘catch-trials’: (1) Extraverted subjects would make more commissive errors than introverted subjects. Since ‘response hungry ’ extraverts are postulated to generate R-excitation from the organization and emission of response, and to depend on R-excitation to maintain their overall state of arousal, they should show a stronger tendency to respond in the absence of the appropriate signal than introverts who are postulated to generate R-inhibition from the organization and emission of responses. (2) The speed and accuracy of extraverted subjects would be relatively more affected than
10
John Brebner and Rosemary Flavel
that of introverts as the proportion of catch-trials increased. This second result is predicted on the grounds that the degree of R-excitation is related to the response rate which is affected by the proportion of catch trials. A further relevant statement by Brebner & Cooper (1974) was that ‘. .where R-excitatory potential is high but S-excitatory potential is low the extraverts’ performance would be relatively better until S-inhibition increased as a function of responding and acted to decrease the overall excitatory potential in the extraverted subjects ’. If in the present experimental situation responding is sufficiently frequent to build up S-inhibition through feedback stimulation, then one might predict (3) an effect akin to repeated ‘involuntary rest pauses ’ either in occasional failures to respond or increased response latencies. These temporary failures differ from the general lowering of responsiveness associated with a continuing high S-inhibition level which results from the lack of opportunity to respond. That effect is expected to reduce speed of performance among extraverts as the catch-trial rate rises. The modulating effect of the ‘involuntary rest pause’ phenomenon should, on the other hand, occur during periods of repeated responding and is most likely to be found where the catch-trial rate is lowest. The parallel between this effect and Eysenck’s (1955) original ‘reactive inhibition’ will be evident.
.
Method Apparatus The stimulus for response was the digit ‘ 1 ’ presented on a Nixie tube located behind a small glass screen set in a timing and display control unit with a matt black painted surface. This unit was linked both to a paper-tape reader, through which programmed tapes controlling the stimulus display were fed, and also to a teletype which encoded the relevant information regarding stimuli and reaction times on to paper tape. The punch was also connected to the response Morse key operated by the subjects. Inter-trial intervals were regular, of 2.3 sec duration, and signals were displayed for 200 msec. Each trial was preceded by a warning light, which became visible in the top left-hand corner of the display screen before the signal was due to occur; the warning signal was of the same 200 msec duration as the signals.
Subjects Sixteen undergraduate subjects from the University of Adelaide, ranging in age from 17 to 24 years (mean age = 19.4 years), took part in the experiment. Selection was made on the basis of E-scores obtained on Form A of the EPI. Those classified as introverts scored in the range 2-5, while the extraverts’ scores ranged from 19 to 23. All subjects had normal vision.
Procedure The subject was seated approximately 1 metre from the display screen with his index finger resting lightly on the response key. To ensure that the subject understood the nature of the task he was given a practice run of five signals to which he responded. Following this, the subject was given a block of 100 trials without catch-trials and was required to respond as quickly as possible to every signal. The purpose of this task was to check for differences in reaction times between the introverts and extraverts not due to the experimental treatments of this study. The experimental session was then begun. In the experimental condition subjects were required to respond as quickly as possible to the digit ‘ 1 ’ but not to respond on any catch-trial. Catch-trials were randomly distributed throughout the series and consisted of the warning light followed by a blank screen. Three blocks of 200 trials were given, representing three differentstimulus conditions. Under condition A, 10 per cent of the total were catch trials, in condition B there were 40 per cent catch-trials. and in condition C 70 per cent. The order of presentation of the three conditions was varied randomly to minimize order effects.
Results Commissive errors Table 1 shows the number of times responses were made on catch-trials by subjects under the three conditions of the experiment. It is clear from the table that extravert subjects tended to make more of these commissive errors, and that the frequency of such errors is inversely related
Introverts, extraverts and efect on catch-trials
II
Table 1. Errors committed by responding on catch trials
Condition Subject
E-score 21 23 22 21 20 19 22 19
B (40%)
C (70% catchtrials)
Total
8 4 0 4 9 6 5 0 36
5 2 0 3 11 5 3 0 29
20 11 4 16 34 21 12 2 120
0
1
2 2 2 2 2 2
0
1 0
A (10%)
7 5 4 9 14 10
4 2 55
9 10 11 12 13 14
5 4 2 4 4
15
5 4
16
4
4 16
1
1
6 2 0
0 0 0 0
2 2 4 8 4 2 3
1 3
6 31
1 1
12
to the proportion of catch-trials. 2 confirms the first prediction that extraverts do make significantly more errors (xz= 52.4, d.f. = 1, P < 0.01). The difference between the three conditions is also significant (2= 15.3, d.f. = 2, P < 0.01). Given that the data in Table 1 are evidently unsuitable for parametric statistics, and that the power of most published distribution-freeanalyses of variance is extremely low (Wilson, 1956), the interaction between personality types and the three conditions proposed in the second prediction was tested by performing a Mann-Whitney U test on the difference in errors between the two extreme conditions A and C. This gave U = 14.5 which is significant ( P < 0.05) for the appropriate one-tailed test. While this result is in line with the prediction made, nevertheless, since introverts made so few errors at all, it is possible it merely reflects the fact that the closer to the ‘zero error’ bound the introvert group is, the smaller must be the change in errors across the conditions of the experiment for these subjects. Further research under conditions in which the commissive error rate of introverfs was higher than in this study would be useful provided the line between the different S-inhibition effects above was retained. A second source of commissive errors lies in the number of anticipatory responses made by the subject. By this is meant simply that the subject would have responded whether a signal appeared or not, but that where a signal did occur, a response is not an error unless clearly programmed before the occurrence of the signal. The limit of 80 msec was accepted as the criterion for an RT being defined as anticipatory. All longer RTs were accepted as genuine responses to the signal. This category of anticipatory commissive errors proved illuminating. Table 2 shows the number of such errors in each condition for both groups. The fact that none of the introverts made such errors is striking and makes any statistical test between the two groups unnecessary. A significant difference across conditions in the predicted direction is obtained from the extraverts, x2=21.1, d.f. = 2 , P 0.05). which argues against any general tendency to trade accuracy for speed among the extraverts. But the two groups do differ in the effect which raising the catch-trial rate has on their RT. The interaction conditionsxpersonality type is significant (F=8.5, P < O a O l ) , and again supports prediction (2) that extraverts will be more affected than introverts as the catch-trial rate increases. The ‘involuntary rest pause’ phenomenon The temporary performance decrement predicted for extraverts with low catch-trial rates was tested in the following way. The decrement should show itself either in missing responses or in slow RTs. There were only three failures to respond in the whole experiment, and although th? extraverts accounted for twice as many of these as the introverts, such a result is not quite 100 per cent conclusive. Rather better evidence is provided by changes in the speed of responding. Since it is a short-lived decrement in performance which was predicted, the mean length of runs of RTs * Due to a mechanical failure, the R T data from one extravert subject (6) was unreliable. His results were omitted from R T analysis. To equate the numbers in the two groups for Anova one introvert subject (14). chosen randomly, was also dropped from the
R T analysis.
Introverts, extraverts and efect on catch-trials
13
which decreased in latency was compared with that for RTs which increased in latency. If extraverts generate R-excitation from organizing responses, but S-inhibition from the feedback effects of responding, their performance should tend to increase in speed as R-excitation increased until the rise in S-inhibition was sufficiently strong to oppose it at which point RT should lengthen. Responses providing more feedback, as in more effortful or more complicated responses would be expected to build up S-inhibition at faster rates than easier responses. Since, in this experiment the same response is repeated, the build up of S-inhibition from feedback is assumed to proceed at a constant rate during responding. In the present context what is expected is that for extraverts only, where the catch-trial rate is low, the mean length of runs of RTs in which response latency decreased should be greater than the mean length of runs of increasing RTs. Table 4 shows the mean lengths of increasing and decreasing runs of RTs. Where the sign is positive, the mean length of runs in which RT became progressively faster exceeded the mean length of runs in which RT became progressively slower. Table 4. Mean run lengths
Condition B (40%)
Condition A (10%)
RT incr.
Diff.
Condition C (70%)
RT decr.
RT incr.
Diff.
RT decr.
RT incr.
1.60 1.46 1-44 1.48 1.71 1.48 1.58
(=)
(+) (-) (+) (t) (=) (-)
1*60 1.66 1.34 1*58 1.79 1.48 1.34
1.34 1.30 1.34 1.47 1.40 1.51 1.68
(+) (+) (+) (+) (+) (+) (-)
1*50 1*54 1.79 1.86 1.52 1 *56 1 *45
1.55 1a45 1-55 1-61 1a65 1*67 1 -53
1.58 1.50 1.40 1-43 1.39 1.61 1.58
(-) (+) (-) (-) (+) (-) (+)
1.52 1.66 1-34 1.41 1*43 1.59 1.61
1-42 1.51 1.42 1.59 1-47 1.58 1.62
(+) (-) (+)
1a56 1 *50 1.55 1-81 1a 6 6 1.54 1-42
1 *58 I a35 1*67 1*37 1-62 1*30 1a38
(t) (t) (-) (-)
Diff.
RT decr.
By inspection of Table 4 the prediction (3) above is not supported in condition A. However, in condition B, the results are in the predicted direction although, on a binomial test, they fail to reach statistical significance;on the appropriate one-tailed test P= 0.06. Why the predicted effect should be at its strongest in condition B becomes a matter for speculation. Possibly since mean RT was at its fastest in condition A, ‘regression toward the mean * occurs more often in that condition. This would reduce the mean length of those runs on which RT continued to decrease. A slightly more complex alternative interpretation is that although R-excitation was highest for extreme extraverts in condition A, S-inhibition deriving from feedback also exerts its strongest influence in that condition. It is in condition A that responses are most frequent. S-inhibition from feedback builds up over repeated responding but tends to dissipate in the period between responses. S-inhibition from feedback, thus opposes R-excitation in condition A whefe its effect is stronger than in other conditions, though for most of the time it is weak in relation to the
14
John Brebner and Rosemary Flavel
strength of R-excitation in the condition. The overall speed of responding, therefore remains highest in condition A, but there is an attenuating effect on runs of decreasing RT which is due to S-inhibitionfrom feedback. Following this line through, in condition B, R-excitation is lower, but the reduction in S-inhibition from feedback allows relatively longer runs of decreasing RT. In condition C,the high degree of S-inhibition for extraverts resulting from low response demands tends, if anything, to produce runs of increasing RT for these subjects. The ad hoc nature of this interpretation guarantees that it matches the findings, but further research seems to be indicated to establish the different action of the two types of S-inhibition postulated. The theory outlined above is an attempt to amalgamate some of the general features of earlier theories which have been substantiated experimentally. For example, there is by now ample evidence to show that, in some circumstances, extraverts generate response-mediated inhibition faster than introverts do, as Eysenck’s (1955) reactive inhibition theory argued. There is also support for the later suggestion that extraverts and introverts differ in their arousability. Unfortunately, however, these explanations seem to be regarded at best as independent of one another, and at worst as alternatives. In some studies it seems that holding too narrowly to one or other of these explanations produces predictions which are unrealistic when the actual experimental task is considered. Thus, the finding by Nocita (1973) that extraverts showed repetitive, perseverative behaviour in an insoluble choice task, ran counter to the hypothesis in that study that responding creates an inhibitory state in extraverts who should, therefore, tend not to repeat responses. Also, the prediction by Buckalew ( 1973) that extraverts would have longer simple RTs since introverts have lower sensory thresholds, was not supported, and it was the extraverts who produced shorter RTs. The present theory predicts that ‘response hungry ’ extraverts will be faster on their initial response and respond more frequently, encompassing both these experimental outcomes. An experiment by Brebner & Cooper (unpublished) has confirmed precisely these predictions within an ‘inspect or respond’ task using coloured slides. Other studies of the behaviour of the two personality types under sensory deprivation and vigilance conditions seem to fit the view that extraverts are prone to R-excitation but S-inhibition, so that they tend to be impulsively active and depend upon responding to maintain their alertness, and, conversely are less aroused by stimulation in the absence of activity. Miyashiro & Russell’s (1974) finding that extraverts did not seek longer durations of a sound stimulus during sensory deprivation, but did seek the stimulus more frequently, points to the extraverts’ need to respond actively. Morgenstern, Hodgson & Law’s (1974) finding that extraverts’ performance at a learning task improved under distraction while that of introverts worsened, is accompanied by the observation that the exploratory movements of the two groups, which were made to improve the artificiallydistorted character of the stimulus, differed from one another. The movements of the introverts were few apd slow in comparison to the large extravagant movements of the extraverts. Here again the need for active responding emerges as does the tendency among extraverts to vary their movements more than introverts in an experiment by Hill (1975). Finally, a relationship between impulsivity and inability to maintain attention in a vigilance task has been reported (Thackray, Jones & Touchstone, 1973), using items from Form A of the EPI to measure impulsiveness. No full review would be appropriate here, but the studies outlined fit the same pattern of extraverts seeking activity to maintain arousal rather than more passive forms of self-stimulation such as visual inspection. This does not mean that extraverts do not seek stimulation. With sufficiently intense and varied stimulation it is possible to maintain arousal even in the extreme extravert, but it would appear from the experimental evidence that active responding is more effective.
Introverts, extraverts and effect on catch-trials
I5
Conclusion The findings of this study support the main hypothesis that R-excitation is stronger in extraverted individuals. From these results together with the data of the previous study (Brebner & Cooper, 1974) a coherent picture of the extravert is emerging which shows him to be prone to both types of S-inhibition and dependent upon R-excitation to maintain his responsiveness. The distinction between S-inhibition generated by feedback from responding and S-inhibition which builds up in the extravert when response demands are low, was drawn in the original statement of the model. The data of this experiment warrant further research into these different types of S-inhibition. References BREBNER, J. & COOPER,C. (1974). The effect of a low rate of regular signals upon the reaction times of introverts and extraverts. J. Res. Person. 4, 263-276. BREBNER,J. & COOPER,C. (1976). Stimulus-or response-induced excitation, a comparison of the behaviour of introverts and extraverts. J. Res. Person. (in press). L. W. (1973). Relationship between a BUCKALEW, physiological and personality index of excitability. Physiol. Psychol. 1, 158-160. EYSENCK, H. J. (1955). A dynamic theory of anxiety and hysteria. J. Ment. Sci. 101, 28-51. HILL,A. B. (1975). Extraversion and variety-seeking in a monotonous task. Br. J. Psychol. &(I), 9-13. MIYASHIRO, C. M. & RUSSELL,D. L. (1974). Experimental participation as a source of stimulation in sensory and perceptual deprivation studies of stimulus-seeking behaviour by introverts and extraverts. Percept. Mot. Skills 38, 235-238.
MORGENSTERN,F. S., HODGSON, R. J. & LAW,J. (1974). A comparison of introverts and extraverts exposed to conditions of distraction and distortion of stimulus in a learning task. Ergonomics 17(2), 21 1-220. NOCITA,F. E. (1973). The relationship of frustration, neuroticism and extraversion to repetitive behaviour. Diss. Abstr. Int. 33(8B), 3990-3991. THACKRAY, R. I., JONES,K. N. & TOUCHSTONE, R. M. (1973). Personality and physiological correlates of performance decrement on a monotonous task requiring sustained attention. Civil Aeromed. Inst., Oklahoma, Fed. Aviat. Agency, AD. 777, 825, 13 pp. WILSON, K . V. (1956). A distribution-free test of analysis-of-variance hypotheses. Psychol. Bull. 53, 96-10].
Received 8 June 1976; revised version received 22 August 1976 Requests for reprints should be addressed to Dr J. Brebner, Department of Psychology, University of Adelaide, Adelaide, South Australia 5001. Rosemary Flavel is at the same address.
