‘ PHYSIOLOGICAL AND COGNITIVE CORRELATES OF SIGNAL DETECTION IN AN AUDITORY VIGILANOE TASK Thesis for the Degree of M. A MICHIGAN STATE UNIVERSITY ' ANTOINETTE KRUPSKI 1969 ABSTRACT PHYSIOLOGICAL AND COGNITIVE CORRELATES OF SIGNAL DETECTION IN AN AUDITORY VIGILANCE TASK BY Antoinette Krupski This study was undertaken to examine the relation- ship between autonomic activation and performance on a sustained vigilance task and also to examine the inter- relationships between physiological, personality and cognitive measures that have been previously employed to study attention. Thirty-one college males took a series of tests which included the Hidden Figures Test, the Stroop Color-Word Test and the Eysenck Personality Inventory. Each subject also completed a 48 minute auditory vigilance test during which time continuous recordings of heart rate, skin conductance and GSR were made. The physiological results indicated that a high level of physiological activity, or arousal, improves vigilance performance in that highly aroused subjects showed little or no detection decrement. These physio - logical results were statistically significant for high and low heart rate groups, while GSR magnitude and GSR amplitude results were in the predicted direction. The physiological results were discussed in the context of activation theory. None of the other measures related meaningfully to either vigilance performance or to the physiological measures. An explanation of this lack of relatedness was that the sustained attention measured in the vigilance situation is a quite different process from the short term attention measured in the other tests. i? _ :7 Approved: (Ajax/g 11:11 vie/("71; ti ,7 Date: cziznbgcflévoj /7 ,/f5 f ,7 , Thesis Committee: Bakan, Chairman P. D. Co Raskin J. Uleman PHYSIOLOGICAL AND COGNITIVE CORRELATES OF SIGNAL DETECTION IN AN AUDITORY VIGILANCE TASK BY Antoinette Krupski A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTERS OF ARTS Department of Psychology 1969 ACKNOWLEDGEMENTS I would like to thank Dr. Paul Bakan who served as chairman of my thesis committee. His encouragement and understanding as well as his criticisms were immensely valuable. Dr. David Raskin must also be acknowledged for the generous offering of his equipment and his time. His comments and criticisms lent a great deal of direction to this endeavor. I am indebted to Dr. James Uleman for reviewing the final manuscript and for the encouragement he offered during the initial phases of this research. I would also like to extend sincere thanks to Steve Porges who,unselfishly and enthusiastically, offered invaluable technical assistance and moral support. A final thanks goes to Jim Bever who was instrumental in preparing the data analysis routines. ii TABLE OF CONTENTS Page ACKNOWLEDGEMENTS. . . . . . . . . . . . . . . . . . . . ii LIST OF TABLES. . . . . . . . . . . . . . . . . . . . . iv LIST OF FIGURES. . . . . . . . . . . . . . . . . . . . . v LIST OF APPENDICES. .'. . . . . . . . . . . . . . . . . vi Chapter I. INTRODUCTION. . . . . . . . . . . . . . . . .1. . 1 II. METHOD. . . . . . . . . . . . . . . . . . . . . .21 III. RESULTS. . . . . - . . . . . . , . . . . . . . . 43 Iv. DISCUSSION. . . . . . . . . . . . . . . . . . . .61 REFERENCESO O 0 O 0 n O O O O O O 0 O O O O O 0 O O O O 68 iii Table LIST OF TABLES Summary of the analysis of variance of detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the three test periods, V1, V2, V3. . . . . . . Summary of the analysis of variance of detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the practice period. . . . . . . . . . . . . . . Summary of the analysis of variance of detections as a function of test periods for high and low GSR magnitude groups. . . . . . Summary of the analysis of variance of detections as a function of test periods for high and low GSR amplitude groups. . . . . . Summary of the analysis of variance of mean GSR responses as a function of vigilance periods for high and low GSR amplitude groups. . . . . . . . . . . . . . . . . . . Factor Analys is. O O O O O O C O I O O O O 0 iv Page . . .46 o o 50 . . 50 . . 54 O O o 54 . . 59 LIST OF FIGURES A sample record during signal presentation. Mean vigilance detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the three test periods, V1, V2, V3. 0 O O I C O O O O O O O O O O O O O O 0 Mean vigilance detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the practice period. . . . . . Mean vigilance detections as a function of test periods for high and low GSR magnitude groups I O O O O O O O O O O O O O O O I O I Mean vigilance detections as a function of test periods for high and low GSR amplitude groups 0 O a O O O O O 6 0 O O O a O O O O 0 Mean GSR responses as a function of vigil- ance periods for high and low GSR amplitude grouPS................... Page .30 .45 .48 .49 .52 .53 LIST OF APPENDICES Appendix page A. Instructions for Tests. . . . . . . . . . . .74 B. Tests and Answer Sheets. . . . . . . . . . . 83 C. Correlation Matrix. . . . . . . . . . . . . .91 D. Raw Data. . . . . . . . . . . . . . . . . . .97 vi INTRODUCTION Vigilance tasks have been used by a number of investigators as a measure of attention (Buckner and McGrath, 1963: Bakan_ g£_§l., 1963: Eysenck, 1964). In such a task the subject is required to detect signals which occur infrequently and at irregular intervals. These signals are generally interspersed among more numerous nonsignal stimuli. The making of successful detections, then, requires the subject to attend to the task constantly, as a lapse in attention quite often results in a missed signal. Hence, number of detected signals is generally used as the measure of attention. Most investigators have found that the number of correct detections generally decreases as the task progresses. This implies a general decrease in attention. After 45 minutes, decrement usually becomes quite marked, with the average subject detecting 24%.fewer signals than he did initially (Bakan, et_§l., 1963). Although the decremental trend is typically found for groups, the amount of decrement as well as total detections are subject to wide individual differences. Previous investigations have attempted to relate these individual differences in attention to physiological, personality and cognitive measures with moderate success (Stern, 1964: Stern, 1966: Bakan .§E_§l., 1963: Claridge, 1966; Eysenck, 1964). However, these investigations have dealt with quite fragmentary aspects of the attention process. The present study was designed to define the attention process more completely in terms of physiological, cognitive and personality measures. The diversity of the measures employed in this study necessitates separate reviews of the literature. The physiological background will be dealt with initially by reviewing the general arousal theories which consider arousal or activation to be intimately related to attention. Theoretical implica- tions will be considered along with research conducted outside the theoretical framework. The second part of this review is concerned with other measures related to attention. The relationship between these variables along with some additional physiological distinctions will be examined. Physiological Background An adequate level of physiological arousal is thought to underly good vigilance performance, or attention. The most popular theories dealing with this assumption include activation theory (Lindsley, 1951: Hebb, 1955: Duffy, 1962: Malmo, 1959), and the theory of John Lacey (1963, 1967). Lindsley (1951) was the first to coin the term, 'activation theory' and his work was elaborated by a number of theorists including Hebb (1955), Malmo (1959), and Duffy (1962). Basically, activation theory depicts a continuum of arousal ranging from deep sleep to excited states of behavior where arousal is measured by magnitude of autonomic or EEG activity to a stimulus or stimulus situation. Activation or arousal processes are thought to reveal the intensive rather than directional aspects of behavior and presumably correlate with measured adequacy of performance in an inverted-U fashion where highest performance levels correspond to moderate physiological activation. Malmo (1965) cites a number of relevant studies that support activation theory. He reports rising gradients in skeletal muscle tension, heart rate and respiration during mirror drawing and tracking tasks, both of which demand sustained attention. Activation theory, however, fails to explain an apparent decrease in skin conductance while gradients in other subsystems were rising. EEG measures, too, were found to be gradated only under special conditions. Ignoring these EEG and skin conductance findings, Malmo concludes that 'tonic' background level of physiological activity is required by the organism for sustaining a relatively even level of attention from the beginning to the end of a task. Stern (1964) verified that a high level of physiological activity is related to a task requiring attention. He found that subjects under conditions of sensory deprivation, but engaged in a vigilance task, sustained a higher level of physiological activity than a group instructed to ignore the stimulus used in the vigil and simply relax. Stern's task-involved group maintained their pretest heart rate while the control group's heart rate fell to 85% of its original level. Although base conductance decreased for both groups, the decrease was significantly greater for the controls. The two groups did not differ in breathing rate although the respiration amplitude of the controls fell to below 80% of their original level while the experimental group maintained their initial depth as reflected in respiration amplitude. Sokolov (1963) proposes a theoretical approach which shares many similarities to activation theory. He states that increased sympathetic activity has an excitatory and facilitating effect on some types of perfor- mance in that it serves to provoke or maintain cortical activation which results in increased receptor sensitivity. The facilitating effect of increased sympathetic activity is readily apparent in work done on the orienting reflex (OR). The OR is a centrally organized system of autonomic, somatic, neural and neuromotor reactions to any novel stimulus. Properties of a novel stimulus would include any increase, decrease or qualitative change in the stimulus field. Specifically, responses include alpha desynchronization, pupil dilation, decreased skin resistance and vasomotor change consisting of cephalic vasodilation and peripheral vasoconstriction. Heart rate acceleration is also implied by the Sokolov model (Sokolov, 1963; Eynn, 1966: Razran, 1961: Graham and Clifton, 1966). The function of an OR is thought to be facilitation of the reception of stimuli resulting from the increase of discriminatory power of sensory systems. And, as Maltzman and Raskin (1965) point out, the Operations which define an OR correspond in part to the conscious—centered concept of attention, i.e., stimulus change, increased cerebral blood volume and decreased blood volume in the limbs. Evidence from studies of individual differences appear to support this in that a number of investigators have found autonomically active subjects are more 'aware' or attentive than their more inactive counterparts (Maltzman and Raskin, 1965: Israel, 1966: Courter et al., 1965). Although the OR patterned reaction does appear to facilitate sensory discriminations, it is not directive in the consummatory sense. It merely 'Opens the organism up' and does not really allow for 'management' of the stimulus (Razran, 1961). YOu will note the similarity again to the intensive, undirected aspects of behavior ascribed by 'irousal' in the activation framework. In spite of the similarities between Sokolov and the activation theorists, procedural and methodological differences make the two schools of thought far from synonymous. Patterns of response discussed by the activationists usually rely on evidence from tracking or mirror drawing tasks. These tasks are usually 30 seconds to 10 minutes in length. The response measures taken are generally known as tonic responses, or responses which change slowly, e.g., skin conductance base level. Sokolov, on the other hand, is mainly concerned with phasic responses, or responses which occur immediately after stimulus presentation. The phasic response is brief and discrete, such as a GSR response. 80, although both theorists would prObably agree that high levels of physiological activity accompany states of greater awareness or attention, the common conclusion is based on quite divergent methodologies. The work of Lacey (1967) points to a substantial amount of evidence which is in direct opposition to activation theory. He cits pharmacological and lesion work that give evidence of simultaneous physiological arousal and behavioral somnolence as well as the reverse. He also refers to work done by Malmo (1966), Elliott (1964) and Mirsky and Garden (1962), all of whom were unable to detect changes in all systems paralleling performance. Lacey concludes from this evidence that arousal is not a unidimensional process, but can be viewed as several processes, each one being a type of arousal in itself. More specifically, electrocortical, autonomic and behavioral indices are separate arousal systems which are intimately related, but not always operating simultaneously or congruently. Instead, increases in these systems depend to a large extent on the situation as well as the individual subject: the aim or goal of behavior is represented as well as the intensive function. For example, it is known that subjects have idiosyncratic patterns of response in that individuals do not respond with equal increments or decrements in all subsystems. These idiosyncratic patterns of response are produced in subjects under different stimulus conditions. He has found conductance increases under all environmental conditions while heart rate response varies with the type of situation. Decelerations occurred when the subject was required to attend to visual and auditory inputs while acceleration occurred when the subject was required to resort to internal activities which involve storage, retrieval, and recombination of information. Lacey concludes that conductance is a generalized response while heart rate is a response with more specific correlates. He proposes that cardiac deceleration accompanies, and perhaps even facilitates, environmental intake while cardiac acceleration accompanies and facilitates environmental rejection. Obrist (1963) was successful in replicating Lacey's findings utilizing different tasks. Campos and Johnson (1966), however, found evidence which conflicted with the Lacey theory. Using Lacey's original tasks, they found heart rate deceleration only when subjects did not have to verbalize their observations. If a subject was required to verbalize either at that time or at a later temporal period there were highly significant increases in both heart rate and skin conduc- tance. These accelerations were apparent even in those tasks dealing with environmental intake where Lacey predicts deceleration. The theories both have face validity in that each is supported by a substantial amount of evidence. However, neither explains all the observations that have been reported. Again, procedural and methodological consider— ations are quite important. Lacey, unlike either Sokolov or Malmo, characteristically uses the minute preceedin and the minute following stimulus presentation as the basis for response measurement. His tasks are also quite diverse, ranging from mental arithmetic to the cold pressor test. Obviously the nature of Lacey's tasks, as well as his units of measurement, are quite different from the 10 minute tasks of Malmo and the immediate responses 10 examined by Sokolov. It is therefore not surprising that the conclusions of these theorists are not identical. In spite of the differences, all three theorists appear to be dealing with the same general problem: autonomic concomitants of various cognitive tasks, including 'attention' or what can be inferred (e.g., Sokolov, OR) to be at least part of the attention process. The basis for each supposition however, is mainly short term tasks or brief stimulus presentations. Whether the physiological concomitants are maintained in tasks of greater length, such as a vigilance task, is still open to question. Stern (1964) found that subjects engaged in a vigilance task maintained initial levels of response while subjects who were not task-involved showed decreasing physiological gradients. His results lend strong support to an activation’ position. Other work done outside the theoretical frame- work with sustained vigilance is not as clear cut. The most consistent findings are electrodermal results. Dardano (1962), Eason,.g5_al. (1965), Ross, 2;;é$,,(1959), and Stern (1966) all found a decrement in conductance base level to be related, to some degree, to performance decrement. ll Dardano's was a visual task where the subject was required to detect a sine wave .5 inch higher in amplitude than the background sine wave which was continually blinking on and off. The conductance level of the subjects showing least decrement decreased to 88%Iof its original level during the three hour task. The high decrement group, on the other hand, showed a 67%.decrease in skin conductance. Eason, et a1. (1965) also used a visual task. Here, the signal consisted of a light which stayed on for .8 seconds rather than the nonsignal duration of .5 seconds. Although only group data are reported, a significant base level decrement accompanied the group's performance decrement. Rate of signal presentation (240/2 hours vs. 60/2 hours) was found to be unrelated to either performance or physio= logical responsiveness. Conductance levels were found to be more tenuously related to performance in the Ross__gt_al. (1959) study. The subject was instructed to detect a double jump in a clock watching task while skin conductance readings were taken at five minute intervals. Results of a cluster analysis appeared to suggest a relationship between high basal conductance and better performance. The results were in the predicted direction, but statistically nonsignificant. 12 Stern (1966), too, found group trends of decreasing arousal as reflected in rising resistance base levels during a visual vigilance task. He also examined signal frequency where one group of subjects was exposed to 120 signals per hour and the other group to 60. The high frequency group detected a higher percentage of signals but had higher base resistance (lower arousal) than the infrequent signal group. Heart rate and EMG were also investigated by both Stern and Eason §£_al. Neither found a change in heart rate over the vigil while EMG level results were inconsistent. Eason, et al's data revealed an overall increase in neck muscle tension while Stern's data showed differences for groups in frequent and infrequent signal conditions. The subjects who received frequent signals showed decreasing neck muscle tension while the infrequent signal group showed much greater variability with no clear cut trend. Two other variables, skin potential response and EEG amplitude,have also been related to vigilance perfor_ mance and appear to support an activation hypothesis. Skin potential work was done by Surwillo (1965). He found a larger number of spontaneous skin potential responses before 13 detections of double jumps of a clock than before undetected jumps or missed signals. Frequent spontaneous responses are believed to accompany states of higher sympathetic activation. The study dealing with EEG responses was performed by Haider. et a1. (1964) who utilized a visual task requiring the subject to detect dim light flashes that were interspersed among more numerous bright flashes. Their results revealed a decrease in amplitude of average visual cortical evoked responses that paralleled performance decrement. Since size or amplitude can be considered an indication of strength of response, or arousal level, the decreasing size would predictably parallel attention decrement according to the activation model. Activation theory appears to be supported thus far in that conductance base level (excepting Stern), skin potential response frequency and EEG amplitude parallel performance measures. However, the heart rate stability in the face of other changing systems is difficult to explain in the activation framework. The conflicting EMG results, too, pose a difficult problem for either theory. Lacey's theory, on the other hand, would most likely predict the conductance results and could probably explain the heart 14 rate data within his theoretical framework. A vigilance task could be viewed as what Lacey calls an 'intermediate stimulus' which involves both paying attention to incoming stimulation and the "internal manipulation of symbols and retrieval of stored information". The prediction of no. significant heart rate change in a situation where response tendencies cancel each other out would be well supported. The studies reveiwed so far have treated physiolog- ical activity as the dependent variable. A number of other studies have reversed the picture and have investigated performance on vigilance while manipulating sympathetic activity through the use of drugs. Generally, the results have favored an activation hypothesis in that increased arousal has generally improved performance by reducing or eliminating decrement while administration of sympathetic depressants have impaired performance (Callaway and Dembo, 1958; Callaway and Band, 1958: Mackworth, 1965: Bakan, 1961). It is quite difficult to draw any hard or fast conclusions concerning arousal levels and vigilance performance from the work reviewed thus far. Procedural and methodological differences make any such comparisons quite tenuous. For example, although all the studies have used visual tasks, there were differences in the specific 15 task utilized as well as in signal rate, intensity of signal, length of vigil and probably numerous other variables. Except for the Haider,‘g£_§l. (1964) studyIall the main hypotheses under consideration were concerned with variables other than simply physiological measures of vigilance. Signal frequency, background noise and intensity were all examined in the above studies. In fact, in most studies reviewed here the physiological findings were secondary to other interests. The effects of other conditions such as differing signal presentation rate and background distraction could very possibly have had a confounding effect on the physiological observations. It is also apparent that a limited number of measures have been utilized, base conductance being the most popular. The Haider. §t_al. (1964) finding of decreasing EEG amplitude gives promising direction to a study of amplitudes in other systems such as electro- dermal responses, particularly since 'tonic' and 'phasic' responses are believed to have different centers of control (Sokolov, cited in Graham and Clifton, 1965). If this is the case, GSR amplitude measures might provide some very interesting findings in an area of rather limited explorations. 16 One of the purposes of the present study is to more thoroughly examine the specific physiological concom- itants of attention, or vigilance performance. Specifically, the role of arousal or autonomic activation will be examined in relation to vigilance performance. Substantial evidence from arousal theorists as well as from outside a theoretical framework suggest that arousal or physiological activation plays some part in the attention process. The specific relationship however is not clear for tasks that involve sustained attention. This study is different from earlier work in that only one stimulus condition will be utilized so that the possibility of confounding variables will be reduced. The measures will also be expanded upon-—heart rate as well as basal conductance, GSR magnitude and GSR amplitude will be examined. Hopefully, a clearer relationship between the physiological measures and vigilance performance will emerge. Cognitive and Personality Measures The Hidden Figures Test as well as the Stroop Color- Word Test have been used by investigators as measures of attention (Witkin et al., 1962; Karp, 1963). The Hidden 17 Figures Test requires the subject to specify which of five designs is "hidden" within a more complex geometric form. The form in which the figure is enclosed is very distracting so that good performance is thought to reflect the ability to overcome distraction (Witkin g£_al., 1962). Tests like the Hidden Figures have been found to load highly on attention and concentration factors (Karp, 1963). The Stroop Color-WOrd Test also requires the subject to overcome a distracting context, but of a some- what different nature. The test consists of three cards. The first is a card of color names which are printed in black letters on a white card (WeCard). The second card consists of rectangular blocks of colors (C—Card), while the third is composed of the color names which are printed in incongruous colors (CW-Card), e.g., the word BLUE is printed in red, the word YELLOW is printed in green. The subject's task is to read each card as fast as he can. The CW card is most difficult in that subjects are instructed to ignore the word and read only the color in which the word is printed. All subjects have some difficulty in doing this. A measure of the difficulty is computed by subtracting C-Card time from CWACard time. This is known 18 as the interference score: the higher the interference score, the more susceptible is the subject to distraction. It has generally been found that performance on the Stroop Color-word test is improved by high drive conditions (Agnew and Agnew, 1963) and by stimulant type drugs (Callaway, 1959). There is also evidence that depressants impair performance (Jenson and Rower, 1966). This line of evidence supports the activation hypothesis in that rising levels of sympathetic activation accompany better performance and vice versa. In a somewhat different approach, Eysenck (1964) attributes differences in attention and arousal level to differing rates and degrees of cortical inhibition build- up in individuals. He claims that introverts build inhibition more slowly, to a lesser degree, and dissipate it more quickly. Extraverts, on the other hand, build up inhibition quickly, show high degrees of inhibition and dissipate it very slowly. As evidence for his theory, Eysenck cites the poorer performance of extraverts on tracking tasks in that they make more involuntary rest stops. These rest stops are due, he claims, to a slower dissipation of inhibition in extraverts. Extravert's perfor- mance on vigilance is also predicted to suffer for the same l9 reason. Bakanl'g£_§l. (1963) found this to be the case. They found that extraverts had significantly greater decrement in an auditory vigilance task as compared to introverts. The measures discussed up to this point, i.e., vigilance, physiological activity, the Hidden Figures Test, the Stroop Color—WOrd Test and the Eysenck Personality Inventory, all appear to share some basic similarities. The sustained attention measured by a vigilance test, for example, as well as the ability to overcome distraction on the Stroop Color—WOrd Test both seem to be aided to some degree by a higher level of autonomic activity. The personality traits of introversion-extraversion also seem to be related in part to physiological processes which are quite similar to those described as ”arousar , as well as to vigilance performance. These underlying physiological assumptions as well as the experimental evidence imply a relationship between arousal level and performance on these tests. It appears as though higher arousal levels improve performance in an activation type manner. The evidence also suggests common basic processes underlying sustained as well as short term attention processes. A greater understanding of the inter—relationships between these variables might 20 further clarify the attention process and also provide evidence which might reveal sources of individual differences in attention. Hence, the second purpose of this study is to examine the inter—relationships of the physiological, personality, and cognitive measures that have been previously employed to study attention. To summarize, the purposes of this study include: 1. The examination of the relationship'between autonomic activation and performance on a sustained attention task. 2. The examination of the inter—relationships of physiological, personality and cognitive measures that have been previously employed to study attention. The carry out these purposes, 31 male subjects took the following tests: a 48 minute auditory vigilance test during which time continuous heart rate and skin conductance recordings were made, the Hidden Figures Test, the Stroop Color-WOrd Test and the Eysenck Personality Inventory. Analyses of variance as well as correlational and factor analyses were performed on the resulting data. METHOD Subjects—~Subjects were 31 male volunteers from an introductory psychology course. Their ages ranged from 18 to 21. Apparatus—-Zinc cup electrodes, 5/8 inch in diameter, were used for recording. Cotton pads soaked in a 1%,zinc sulfate solution were placed within the cups and served as the electrolyte. GSR electrode placement was on the base of the left thumb and on the inside of the left forearm. Heart rate electrodes were placed on the right bicep and on the inside of the left ankle. The right forearm was the site of the ground electrode. Response recording was made on a 2—channel Beckman Type RS Dynograph. GSR measures were fed into a Beckman Type 462 amplifier through a Beckman Type 9892A PGR coupler which utilizes a simple Wheatstone bridge circuit. A constant current of 2 microamperes was passed through the subject. Heart rate recording was done with a Beckman cardiotachometer coupler Type 9857 where heart rate in beats per minute was represented by output amplitude. 21 22 Vigilance Task--A copy of the vigilance tape used by Bakan,.g£_al. (1963) was employed. This tape is 64 minutes long and consists of digits which are spoken at the rate of one per second. The original tape was constructed by .splicing so as to be uniform throughout. Each digit was recorded onetime, played on an endless tape, and recorded to produce multiple recordings of the same digit. Pieces from these multiple recordings were spliced together to produce a tape in which the sound of any digit was constant throughout the tape. This original tape was duplicated on a Viking 433 recorder. It was stereophonically recorded on one track while a tone was recorded on a second track at the onset of each signal. Although the tone was not audible to the subject,it served to close a sound activating switch which set off an event marker on the dynograph at the onset of each signal. The tape began with a female voice saying, "The numbers will begin in 30 seconds", which was followed by 30 seconds of silence and then the presentation of the first number. The onset of the voice and the onset of the first number served as the two stimuli for initial OR measures. 23 The tape was played back through a Dyna SCA—35 Stereo Control Amplifier to Sharpe stereo earphones which were worn by the subjects. Subjects were instructed to detect odd-even-odd combinations of digits that were all different and successive, e.g., 943, 725. The tape was divided into four periods of 16 minutes and there were ten signals, or odd-even-odd combinations, presented during each period. Although the task appeared continuous to the subject, there were actually four equivalent 16 minute periods, differing only in the specific signals to be detected. The distribution of signals over time was the same in each period, the time between signals being: 69, 152, 23, 181, 108, 102, 44, 13, 141 and 144 seconds. subjects were to press a button located in front of their right hand when they detected a signal. The button was a modified doorbell that was mounted on a wooden block for stability and convenience. It was connected to a second marker pen on the dynograph so that each detection response was recorded on the paper alongside the actual physiological responses. Consequently each subject's record consisted of continuous recordings of skin resistance, heart 24 rate, detection responses, and critical detection points or signals. The paper speed was set at l mm/sec. Procedure--Subjects were tested individually. Each subject was administered the Hidden Figures Test followed by the Stroop Color-WOrd Test. Upon completion of these tests, the subject was led to the experimental room where the electrodes were attached. subjects were asked to remove their watches and place them in their pockets. The experimental room was adjacent to the room where the recording was done, so the subject was alone during the entire task. Care was taken to make the room as free from distracting stimuli as possible. A fan was used to mask noises from the relay and other sources. While attaching the electrodes, the E explained what the electrodes were and briefly informed the subject what responses were to be recorded. When electrode place- ment was complete the subject was asked to read instructions for the vigilance task. The last page of instructions consisted of a practice set where the subject looked at a series of digits and was told to write down the appropriate odd-even-odd combinations. When the subject appeared to have completed the visual practice set, the E asked if he 25 had any questions. If the subject had no questions the following instructions were read to the subject: The first part of this task is a practice period followed by a short rest. During the rest period you will be allowed to stretch and move around if you like. However, try to remain as still as possible while the practice and the actual test are going on. The electrodes, particularly those on the left side, are very sensitive to movement. So please try to keep movement, other than pressing the button, at a minimum. Get comfortable before the experiment begins. The actual test will follow the rest period. It will be identical in nature to the practice. Earphones were placed on the subject and the E left the experimental room and took her place in the recording room. The dynograph was turned on and if all appeared to be in working order, the tape was begun. After the first 16 minute period, the recording apparatus and tape were turned off and the earphones taken off the subject. He was told the practice period was over and' asked if he had any questions. If there were no questions, he was given about five minutes to stretch and relax before resuming the task. The subject was not given specific knowledge of results, although he was told whether his performance was adequate or not. subjects who responded very often or did not respond at all during the practice were asked to verbally repeat the instructions. 26 If the instructions were repeated correctly nothing more was said concerning the subject's performance. If however, the subject's report of the directions was inaccurate, the E would explain the procedure once again. This was found necessary in about five cases. Each subject was also offered a glass of water during the rest period. After the rest period, the earphones were replaced and recording resumed as in the practice period, but for 48 minutes instead of 16; Continuous recordings of skin resistance and heart rate were made during this time as well as during the practice. At the end of the 48 minutes, the earphones and electrodes were removed and the Eysenck Personality Inventory was administered. When the subject completed the test, the following questions were asked: What is your age? Are you taking any drugs? How much sleep did you get last night? Do you smoke? How much? How long would you estimate the numbers task took, from the end of the rest period on? No subject admitted to taking any drugs for at least two weeks prior to the experimental session. The subject was then asked not to tell other members of his class what went on in the experiment so as to maintain the naivete of other potential subjects. 27 The E showed the subject his physiological record and explained in more detail the significance of the various responses. This explanation was of an extremely general nature intending to simply familiarize the subject with psychophysiological recordings and also to alleviate any undesirable emotional effects of the experimental situation. The experimenter felt this procedure desirable as a number of students were obviously annoyed upon completion of the vigilance test. Scoring--The Hidden Figures Test was scored in two ways: total number correct and per cent correct of those answered. The per cent correct score was found desirable in that subjects varied widely in the number of items attempted. Three scores were obtained for the Stroop Color- WOrd Test: Color Card reading time, Color-word Card reading time and an interference score. Interference was the difference between Color-WOrd reading time and Color reading time. The higher this score, the more interference in reading the Color-WOrd Card. The Eysenck Personality Inventory yielded three scores: Extraversion, Neuroticism, and Lie. These tests were scored in the manner advised in the test manual (Eysenck and Eysenck, 1963). 28 Number of correct detections for each period constituted the vigilance score. An upper limit of five seconds was placed on response latency: responses made later than five seconds after signal cessation were not counted. Conductance base level and GSR amplitude were computed at each detection point. Any response was scored if it began from one to six seconds after signal cessation. These resistance scores were then transformed into log conductance units. The transformation used was a modification of the Haggard method (1945) and is explained in detail by Raskin (in press). Conductance base level, GSR magnitude, and GSR amplitude scores were averaged for each 16 minute vigilance period for each subject. In addition, the overall mean GSR magnitude and GSR amplitude for the entire test were computed for each subject. GSR magnitude scores are response measures that incorporate zero entries for trials on which there is no response, i.e., if there is no response, a score of zero is averaged with the changes recorded on othr trials. Amplitude scores are also response measures, however, zero responses are not incorporated; it is a measure of GSR size given that a response occurred (Prokasy, 1967). 29 Basal heart rate was computed for each subject by averaging the first three beats following signal onset at each detection point during the test periods. The scored beats always occurred prior to signal completion. This resulted in 30 scores for each subject. Mean basal heart rate for each period, as well as the mean basal heart rate for the entire test, were computed from these scores. The actual quantification of the machine output was accomplished by determining the amplitude of each : response to the tenth of a millimeter. The amplitude of a response is a measure of heart rate in beats per minute. Figure 1 is a graphic explanation of the scoring method. Figure l. A sample record during signal presentation. Paper speed was set at l mm/sec. The pen deflection in the top margin signifies the subject's detection response. The penciled mark to the left of the detection response marks the onset of the signal. Consequently, the subject in this example responded 5 seconds after signal onset. The GSR channel is pictured with the response interval marked. The area between 0-4 is counted as signal onset. Any response which began in the 4—10 second interval, or 4-10 seconds after signal onset, was scored as a GSR response. Base level and peaks were scored by measuring the distance in mm from the top margin line to (B) the point at which the line first turned in a downward (-) direction, and the distance in mm from the top margin line to the point (P) at which the line changed direction again. This time in an upward direction (+) or to a slope of 0 (flat). The marked line in the heart rate channel signifies signal onset. The three beats following signal onset were averaged at each detection point. The three beats in this example are marked by arrows. The scored beats always occurred prior to signal completion. Each beat was scored by measuring the distance in mm from the botton margin line to the extreme left—hand corner of each beat. The pen deflection in the botton margin signifies signal onset. 3O 31 Analysis--Analysis of the data included a correlation matrix of 40 variables and factor analysis as well as several analyses of variance which were computed on the conductance and heart rate data. The correlation matrix included the following variables: '.1. Per cent correct on Hidden Figures (%HF) is the per cent correct of those attempted on the Hidden Figures Test. 2. Hidden Figures Score (HE) is the total number of items answered correctly on the Hidden Figures Test. 3. Stroop Interference Score (Stroop Interf.) is calculated by subtracting subject's time in reading the Color Card from the time taken to read the Color—word Card. The resulting difference is thought to indicate word interference, or difficulty in attending to the colors and ignoring the words. High interference scores are thought to be an indication of greater susceptibility to distraction. 4. Extraversion Score (Ex.). A high score on the Eysenck Personality Inventory Extraversion Scale indicates high extraversion. 32 5. Neuroticism Score (g). A high score on the Eysenck Personality Inventory Neuroticism Scale indicates high neuroticism. 6. Lie Scale Score (L). A high score on the Eysenck Personality Lie Scale indicates a high number of items reflecting dishonesty were answered. 7. .21 is the total number of correct detections made during the first vigilance test period. This is actually the second 16 minutes of vigilance, the first 16 minutes being considered the practice period (VP). 8. ‘22 is the total number of correct detections made during the second vigilance test period. 9. ‘23 is the total number of correct detections made during vigilance period 3, or the third test period. 10. Absolute Change V3-l (VB—l) is a difference score computed by subtracting the number of detections made in period 1 (first test period) from those made in period 3 (last test period). Positive difference scores indicate an increment in detections over time whereas negative scores indicate a decrement in detections over time. For example, a subject who detected 8 signals in period 1 (V1) and 2 signals in period 3 (V3) would have a difference score of —6, i.e., he shows a decrement in 33 the number of signals detected from period 1 to 3. An increment in detections over time would be exemplified by a subject who detected 2 signals in period 1 and 8 signals in period 3. His score would be +6. 11. Total Vigilance Score (Tot. V) is the total number of detections for the three test periods, V1, V2, V3. 12. EBA is the initial GSR response amplitude to the voice on the tape which occurred prior to the vigilance practice period. 13. .932 is the initial GSR response amplitude to the first number on the tape which occurred 30 seconds after the voice and marked the beginning of the vigilance practice period (VP). 14. Commission Errors (CE) result from the subject responding in the absence of a signal. 15. i Mag V1 is the mean GSR magnitude for the first 16 minute test period, or V1. This score was computed by taking the mean of the response amplitudes at each detection point for the first test period. Since magnitude scores incorporate zero entries for trials on which there is no response, a score of zero was averaged with the changes recorded at other points when a response did not occur. 34 16. i Mag V2 is the mean GSR magnitude for the second 16 minute test period, or V2. This score was computed by taking the mean of the response amplitudes at each detection point for the second test period. Since magnitude scores incorporate zero entries for trials on which there is no response, a score of zero was averaged with the changes recorded at other points when a response did not occur. 17. i Mag V3 is the mean GSR magnitude for the third 16 minute test period, or V3. This score was computed by taking the mean of the response amplitudes at each detection point for the third test period. Since magnitude scores incorpprate zero entries for trials on which there is no response, a score of zero was averaged with the changes recorded at other points when a response did not occur. 18. Absolute Change 2 Mag V3—l (i Mag3-l) is a difference score computed by subtracting the mean GSR magnitude of period 1 (first test period) from the mean magnitude of period 3 (last test period). Positive difference scores indicate an increment in magnitude from period 1 to 3 whereas negative scores indicate a decrement in magnitude from period 1 to 3. For example, a subject 35 whose mean GSR magnitude in period 1 was 1.000 log conductance units and 0.500 in period 3 would have a differ- ence score of -.500 log conductance units, i.e., he shows a decrement in GSR magnitude from period 1 to 3. An increment in GSR magnitude over time would be exemplified by a subject who had a mean GSR magnitude of .500 log conductance units in period 1 and 1.000 in period 3. His score would be +.500. 19. i LBC V1 is the mean log base conductance for the first 16 minute test period, or V1. The basal conductance level was determined at each detection point and averaged for the first test period. 20. X LBC V2 is the mean log base conductance for the second 16 minute test period,,or V2. The basal conductance level was determined at each detection point and averaged for the second test period. 21. i LBC V3 is the mean log base conductance for the third test period, or V3. The basal conductance level was determined at each detection point and averaged for the third test period. 22. Absolute Change i LBC v3—1 (x LBC3—l) is a difference score computed by subtracting the mean log base conductance of period 1 (first test period) from the 36 mean conductance of period 3 (last test period). Positive difference scores indicate an increment in base conductance from period 1 to 3 whereas negative scores indicate a decrement in conductance from period 1 to 3. For example, a subject who had a mean base conductance of 2.500 log units in period 1 and 2.000 in period 3 would have a difference score of -.500, i.e., he shows a decrement in log base conductance from the first to the third period. An increment in basal conductance over time would be exemplified by a subject who had a mean base conductance of 2.000 in period 1 and 2.500 in period 3. His score would be +.500. 23. Absolute Change 2 Mag V3-2 (i Mag3-2) is a difference score computed by subtracting the mean GSR magnitude of period 2 (second test period) from the mean magnitude of period 3 (last test period). Positive difference scores indicate an increment in magnitude from: period 2 to 3 whereas negative scores indicate a decrement in magnitude from period 2 to 3. For example, a subject whose mean GSR magnitude in period 2 was 1.000 log conductance units and 0.500 in period 3 would have a difference score of -.500 units, i.e., he shows a decrement 37 in GSR magnitude from the second to third periods. An increment in GSR magnitude would be exemplified by a subject who had a mean GSR magnitude of .500 units in period 2 and a magnitude of 1.000 log units in period 3. His score would be +.500. 24. Absolute Change i V3-2 (i V3-2) is a difference score computed by subtracting the number of detections made in period 2 (second test period) from those made in period 3 (last test period). Positive difference scores indicate an increment in detections from period 2 to 3 whereas negative scores indicate a decrement in detections from period 2 to 3. For example, a subject who detected 8 signals in period 2, and 2 signals in period 3 would have a difference score of -6, i.e., he shows a decrement in the number of signals detected from period 2 to 3. An increment in detections over time would be exemplified by a subject who detected 2 signals in period 2 and 8 signals in period 3. His score would be +6. 25. C~Card score is the reading time in seconds of the Stroop Color-Card. 26. CW—Card score is the reading time in seconds of the Stroop Color-WOrd Card. 38 27. .Timg; This is the subject's estimate of how long the vigilance task took. 28. Smoke. The subject was scored on smoking if he admitted to smoking % pack of cigarettes per day or more. The scoring here was discrete: l=smoker, 0=non- smoker. 29. Coffee. The subject was scored as a coffee drinker if he admitted to drinking one cup of coffee per day or more. Scoring here was discrete: 1=coffee drinker, 0=nonecoffee drinker. 30. X Amp V1 is the mean GSR amplitude for the first 16 minute vigilance test period, or V1. This score was computed by averaging the*response size at each detection point in the period, given that a response occurred. No zero responses were incorporated. 31. 2 Amp V2 is the mean GSR amplitude for the second 16 minute vigilance test period, or V2. This score was computed by averaging the response size at each detection point in the period, given that a response occurred. No zero responses were incorporated. 32. 2 Amp V3 is the mean GSR amplitude for the third 16 minute vigilance test period, or V3. This score 39 was computed by averaging the response size at each detection point in the period, given that a response occurred. No zero responses were incorporated. 33. %»Change Amp 3 is the per cent change in GSR amplitude from vigilance periods 1 to 3, or the first and last test periods. The per cent change scores were computed by dividing the mean GSR amplitude in period 3 by the mean GSR amplitude in period 1. Scores under 100% indicate a decrement. For example, a subject whose mean GSR amplitude was 10.000 log conductance units in the first test period (V1) and 5.000 units in period 3, or the last test period would have a % Change score of 50%. This subject showed a decrease in GSR amplitude from the first to last test period. An increment in GSR amplitude, or scores over 100%, would be exemplifed by a subject whose mean GSR amplitude was 5.000 units in period 1 and 10.000 in period 3. His % Change score would be 200%. 34. % Change Amp 2 is the per cent change in GSR amplitude from periods 2 to 3, or the second and third test periods. The per cent change scores were computed by dividing the mean GSR amplitude in period 3 by the mean GSR amplitude in period 2. Scores under 100% indicate a 40 decrement. For example, a subject whose mean GSR amplitude was 10.000 log units in V2 (second test period) and 5.000 in V3 (last test period), would have a %.Change score of 50%. This subject showed a decrease in GSR amplitude from the second to third test period. An increment in GSR amplitude, or a score over 100%wwou1d be exemplified by a subject whose mean GSR amplitude was 5.000 log units in period 2 and 10.000 units in period 3. His % Change score would be 200%. 35. Absolute Change 2 Amp V3-l (i Amp V3-l) is the difference score computed by subtracting the mean GSR amplitude of period 1 (first test period) from the mean GSR amplitude of period 3 (last test period). Positive difference scores indicate an increment in amplitude from period 1 to 3 whereas negative scores indicate a decrement in amplitude from period 1 to 3. For example, a subject whose mean GSR amplitude in period 1 was 10.000 log conduc- tance units and 5.000 in period 3 would have a difference score of -5.000 log units, i.e., he shows a decrement in GSR amplitude from period 1 to 3. An increment in GSR amplitude over time would be exemplified by a subject who had a mean GSR amplitude of 5.000 log units in period 1 and of 10.000 units in period 3. His score would be +5.000. 41 36. %.Change LBC 3 is the per cent change in log base conductance from vigilance periods 1 to 3, or the first and last test periods. The per cent change scores were computed by dividing the mean log base conductance in period 3 by the mean log base conductance in period 1. Scores under 100% indicate a decrement. For example, a subject whose mean LBC was 2.500 in V1 (first test period) and 2.000 in V3 (last test period), would have a per cent change score of 80%. This subject showed a decrease in basal conductance from the first to the last test period. An increment in conductance, or scores over 100%; would be exemplified by a subject whose mean conductance was 2.000 in period 1 and 2.500 in period 3. His % Change score would be 125%. 37. % Change LBC 2 is the per cent change in log base conductance from periods 2 to 3, or the second and third test periods. The per cent change scores were computed by dividing the mean base conductance in period 3 by the mean LBC in period 2. Scores under 100% indicate a decrement while scores over 100% signify an increment in basal conductance. (For an example, see variable #36). 38. Overall Mean Magnitude (i Mag) is the mean GSR magnitude computed from all three test periods, V1, V2, V3. 42 39. Overall Mean Base Level (Z_§B§) is the mean log base conductance computed from all three test periods, V1, v2, v3. 40. Overall Mean Amplitude (2 Amp) is the mean GSR amplitude computed from all three test periods,Vl, V2, V3. RESULTS Physiological and Analysis of Variance Results Overall test period means for base level, GSR magnitude and GSR amplitude were found to correlate highly with each of their respective period means. These correl— ations all exceeded .94. Due to these high inter-correl- ations within the physiological measures, only the overall mean for the three test periods for each variable will be utilized in further analyses. Similarly, only the OR to the first number (0R2) will be referred to as the correlation of the OR to the voice and the OR to the first number was .85 (P<.01). In order to determine how physiological differences relate to detection performance, several analyses of variance were performed. In these analyses, subjects were divided into high and low groups for each of the physiological measures. This resulted in high and low GSR magnitude, high and low GSR amplitude, as well as high and low heart rate groups. The group divisions were obtained from distributions of each subject's overall mean score for 43 44 each variable. The median was calculated and served as the division point. Those subjects who scored above the median were designated as the 'high' group for that variable, while those falling below the distribution median were considered part of the '1ow' group. Thus, the mean physiological scores served as independent variables while detections over trials served as the dependent variables. The trends of subjects divided into high and low groups on the basis of basal heart rate which was averaged during the actual test are shown in figure 2. High heart rate subjects start off detecting fewer signals than do the low groups and show a decrement during the second 16 minute test period but come back during the last period to a rate of signal detection which is slightly higher than their initial level. Low heart rate subjects, on the other hand, initially detect more signals but exhibit a consistent decremental trend for the last two periods. The interaction, groups x test periods, was found to be significant, F(2,48)=3.66, p<.05. (See table 1) A similar difference was obtained when subjects were divided into high and low groups on the basis of their mean basal heart rate during the practice period. 45 g 7000' .3 \ Hi h H R 4’ 6.0.. \\::§\\\¢///////. g . . O B \. a) 5.0-‘r- \\\~. Q LOW H. R. g 4.0... (D 2 Vigilance Test Periods Figure 2.--Mean vigilance detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the three test periods, VI, V2, V3. Table 1. Summary of the analysis of variance of detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the three test periods, V1, V2, V3. Source SS df Ms F Groups (G) 3.71 l 3.71 - Error (S's(G)) 265.95 24 11.08 Test Periods (T) 11.49 2 5.75 2.95 G X T 14.26 2 7.13 3.66* Error (S's(G) X T) 93.62 48 1.95 Total 389.03 77 *p<.05 46 47 unfortunately, four subject's records were not scorable during the practice period as the heart rate channel was turned off for the initial 16 minutes. Figure 3 shows the curves for the two groups where the N=22. It is very similar to figure 2 except that high heart rate subjects detect more signals than the low group in the initial test period. Again, the high group shows some decrement in period 2, but exceeds the initial performance level in period 3. Low heart rate subjects show a consistent decrement in detections over the three test periods. The interaction again, groups x test periods, was found to be significant, F(2,40)=10.1l, p<.01._ (See table 2) Although not statistically significant, trends in a similar direction were obtained with GSR amplitude and GSR magnitude groups. Figure 4 depicts the high and low GSR magnitude groups. High magnitude subjects appear to detect a greater number of signals consistently over time and show no decrement. The low magnitude subjects, on the other hand, start off detecting fewer signals and show a sharp drop in the number of signals detected after the first test period. The low subjects never recover from this drop. The ANOV revealed differences between groups, Mean Detections 48 Hi h H. R. 7.0 . g 6.0 4- 5.0 + \\ 4.0 _ Low H. R. i . . . I I I l 2 3 Vigilance Test Periods Figure 3.--Mean vigilance detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the practice period, VP. Mean Detections 49 7.0 “ 6'0 w- r—' —L' FF°High Mag. \\ 5.0 u \ \ \\ 4 4.0 h \v—‘ Low Mag. 4’ I I I~ l 2 3 Vigilance Test Periods Figure 4.--Mean vigilance detections as a function of test periods for high and low GSR magnitude groups. Table 2. Summary of the analysis of variance of detections as a function of test periods for high and low basal heart rate groups where basal heart rate was averaged for the practice period, VP. l§ource SS df MS F Groups (G) 29.34 1 29.34 2.77 Error (S's(G)) 211.94 20 10.60 Test Periods (T) 7.39 2 3.70 2.50 G X T 29.94 2 14.97 10.11*** Error (S's(G) X T) 59.37 40 1.48 Total 337.98 65 ***p(.01 Table 3. Summary of the analysis of variance of detections as a function of test periods for high and low GSR magnitude groups. Source SS df MS F Groups (G) 40.00 1 40.00 3.69# Error (S's(G)) 303.15 28 10.83 Test Periods (T) 8.96 2 4.48 2.50# G X T 10.07 2 5.04 2.82# Error (S's(G) X T) 100.32 56 1.79 Total 462.50 89 #p (. 10 50 51 F(1,28)=3.69, p<;10, over test periods, F(2,56)=2.50, p(.10, and an interaction between groups and test periods, F(2,56)=2.82, p<.10. Table 3 summarizes these results. Figure 5 depicts the detection trends of high and low GSR amplitude subjects. High amplitude subjects start off at a lower rate of detection than do the low amplitude subjects, however they maintain this rate throughout the task. Low amplitude subjects, on the other hand, initially detect more signals, but show a large decrement after the first 16 minute test period from which they nevery fully recover. In this analysis, the groups x test period interaction revealed differences, F(2,56)=2.57, p<;10. (See table 4) An additional finding was that high amplitude subjects responded more frequently at detection points whether they detected the signal or not. These curves are shown in figure 6. High amplitude subjects gave more detectible GSR responses during a greater number of signal presentations than did the low amplitude subjects. These responses increased in number during the second test period and decreased slightly during the third period. The low group, on the other hand, showed a slight decrease in response frequency from period 1 to 2 and an increase from Mean Detections 52 7 o 0 'I" \ 6.0 ‘P ‘\\ h ‘ 4Hig Amp. 5.0 -- vflfld;¥<::.-r""Low Amp. 4.0 ‘- 4 I : I l 2 3 Vigilance Test Periods. Figure 5.--Mean vigilance detections as a function of test periods for high and low GSR amplitude groups. 53 8.0“" High Amp. U) m 3 7 o o ' '7 n. U) 0) ’1' 3‘, 6.0+ “‘“w/ o C (U .I g % Q = i i l 2 3 Vigilance Test Periods Figure 6.--Mean GSR responses as a function of vigilance periods for high and low GSR am— plitude groups. Table 4. Summary of the analysis of variance of detections as a function of test periods for high and low GSR amplitude groups. Source SS df MS F Groups (G) 0.00 1 0.00 - Error (S's(G)) 325.56 28 11.63 Test Periods (T)S 5.95 2 2.98 - G X T 10.45 2 5.26 2.57# Error (S's(G) X T) 114.54 56 2.05 Total 460.77 89 #p<.10 Table 5. Summary of the analysis of variance of mean GSR responses as a function of vigilance periods for high and low GSR amplitude groups. Source SS df MS_ F Groups (G) 48.45 1 48.45 6.13** Error (S's(G)) 221.37 28 7.91 Test Periods (T) 2.50 2 1.25 - G X T 3.20 2 1.60 - Error (S's(G) X T) 107.13 56 1.91 Total 382.65 89 **p (.025 54 55 period 1 to 2 and an increase from period 2 to 3. The group effect was found to be significant, F(1,28)=6.13, p(.025. (See table 5) Correlation Results Detection Correlates--Detections for the three test periods, V1, V2, V3, all correlated significantly with total detections (r=.87, .87, .85, p<.01). Mean GSR magnitude and total detections were significantly related (r=.44, p(.05); high detection scores are reliably accompanied by high GSR magnitudes. Vigilance decrement, or a decrease in detections over time, was positively related to a decrease in GSR magnitude over time (r=.45, p<;05). No such trend was found for change in base conductance or GSR amplitude change. Per cent correct of those attempted on the Hidden Figures Test correlated significantly with total vigilance detections (r=-.4l, p<105). The negative correlation indicates that a higher per cent correct of those attempted on the test related to lower detection scores. These results are quite puzzeling and appear to lack a logical explanation. 56 Subject's time estimates are also related to vigilance performance (r=-.37). A negative correlation with total vigilance indicates that good vigilance performance is related to a low time estimate. Detection decrement was found to be reliably related to neuroticism score and coffee drinking. High scoring neurotics show little decrement over time (r=.50, p<§01), while coffee drinkers are apparently more susceptible to detection decrement (r=-.39, p<305). Correlations Between the Physiological Measures--The GSR and conductance measures were found to be highly inter- related. Correlations between magnitude and base level (r=.73), magnitude and amplitude (r=.92), and base level and amplitude (r=.72) were all significant beyond the .01 level. The correlations between OR and base level (r=.39, p<205), OR and magnitude (r=.55, p<201), as well as OR and amplitude (r=.59, p<.01) were also found to be significant. Changes over time between the GSR and conductance measures were also significantly related. Difference scores computed from period 1 to 3 resulted in significant correl- ations between base conductance and GSR magnitude (r=.58, p<§01), magnitude and amplitude (r=.45, p<105), as well as base level and amplitude (r=.63, p<301). 57 The relationship between heart rate and the other physiological variables is not reported as complete heart rate records were obtained from only 26 of the subjects. Mechanical difficulties and artifacts made it impossible to obtain scorable records from the others. As a result, heart rate was not included in the correlation matrix to avoid a further reduction of the N. Miscellaneous Correlations--Number of commission errors are sizably and also significantly related to the conductance and GSR measures. Correlations with initial OR (r=—.40, p(305) and GSR amplitude (r=-.50, p