GENETIC COMPONENTS OF AUTONOMIC STIMULUS- RESPONSE AND IN’DIVIDUALV- RESPONSE SPECIFICITY; A TWINS STUDY Dissertation for the Degree Of Ph. D. MICHIGAN STATE UNIVERSITY ROBERT STUART BUNDY 1975 'q 1v “ This is to certify that the ’ thesis entitled response and inclivi»V A twins study. Date July 25, 1975 0-7639 III III III IIII III IIIIII III IIII IIII II III IIIIII IIIIII : _ 6 3‘4 0 31793 ABSTRACT GENETIC COMPONENTS OF AUTONOMIC STIMULUS- RESPONSE AND INDIVIDUAL-RESPONSE SPECIFICITY: A TWINS STUDY By Robert Stuart Bundy The genetic components of autonomic nervous system activity were investigated in fifteen pairs of monozygotic and fifteen pairs of dizygotic twins. Twins were tested during a mental arithmetic task, a reaction time task and a rest period. The dependent variables were heart rate, skin conductance, and respiration. Results were analyzed for the presence of stimulus- response and individual-response specificity. Twin pairs tended to remain in the same relative point in the distri- bution from one stimulus condition to another, supporting an individual-response specficity interpretation. Herit- ability estimates were fairly high for most dependent measures. However, for many of the dependent measures differences in the distributions of the two populations made comparisons difficult. The differences in distribu- tion were most likely a result of sampling error due to the small number of subjects used in the study. -~ :vm‘a .. _ “gr GENETIC COMPONENTS OF AUTONOMIC STIMULUS- RESPONSE AND INDIVIDUAL-RESPONSE SPECIFICITY: A TWINS STUDY By Robert Stuart Bundy A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1975 TABLE OF CONTENTS Page LIST OF TABLES . iii LIST OF FIGURES V INTRODUCTION . METHOD . 13 RESULTS 23 DISCUSSION . 55 APPENDICES A. Twin questionnaire . . 60 B. Factors affecting heritability estimates . 63 C. Subject instructions . . 77 D. Means and sums of squares for all data . . 82 LIST OF REFERENCES 90 ii " 1W Table 10. ll. 12. 13. LIST OF TABLES Summary of twins studies using electrodermal measures Summary of twins studies using heart rate measures Descriptive statistics for heart period Correlations between stimuli for heart period . Descriptive statistics for heart period variability Correlations between stimuli for heart period variability Descriptive statistics for electrodermal frequency Correlations between stimuli for electro- dermal frequency . Descriptive statistics for electrodermal frequency Correlations between stimuli for electro- dermal total height Descriptive statistics for skin conductance level Correlations between stimuli for skin conductance level Descriptive statistics for breathing rate . - iii Page 10 27 28 33 34 34 39 43 44 46 47 49 I All .I [I II [III III II \‘ Ill-I'll! If] Table 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Correlations between stimuli for breathing rate . Descriptive statistics for reaction time data . Estimates of H2 given true H2 and a percent- age of misclassification . Means and sums of squares Means and sums variability Means and sums frequency Means and sums total height . Means and sums Means and sums Means and sums data . of of of of of of ductance level . . squares squares squares squares squares squares iv for for for for for for for heart period . heart period electrodermal electrodermal skin con- breathing rate . reaction time Page 50 54 67 83 84 85 86 87 88 89 ‘2 -Lu'r Figure LIST OF FIGURES Page Heart rate variance and heart rate averaged across subjects for all scoring periods . . . 29 Electrodermal total height and electrodermal frequency averaged across subjects for all scoring periods . . . . . . . . . . . . . . . 36 Skin conductance level and breathing rate averaged across subjects for all scoring periods . . . . . . . . . . . . . . . . . . . 42 Second by second heart rate variance and heart rate averaged across reaction time trials 5-15 and across all subjects . . . . . 52 IIIIIfllII li{[({{. lull-II OI‘IIIII .l‘l‘l‘ulll'f‘l,l‘[| INTRODUCTION Low correlations among various autonomic response measures have been a constant source of interest to psychophysiologists. Intrasubject correlations appear to be well below individual reliabilities of the various response measures (Lacey, 1956). For example, Elliott (1964) found correlations between heart rate and skin conductance of .46 for adults and .12 for children. Lazarus (1966) has reported similar figures even though several mathematical techniques were employed to increase the correlations. Low correlations among autonomic response mea- sures are particularly troublesome for activation theorists who tend to rely on a unitary concept of arousal. Whereas activation theorists (Duffy, 1972; Malmo, 1959; Selye, 1950) do not require that all physi- ological measures show perfect intercorrelations they have some difficulty explaining autonomic patterns which show stimulus-response specificity. In other words, different stimulus situations will often cause different patterns of physiological activity. Darrow (1929) noted that "sensory stimuli" caused an increase in electrodermal activity and cortical arousal but caused a decrease in heart rate. In spite of the fact that Darrow's observa- tion seems to contradict Cannon's (1928) notion of auto- nomic activation, psychologists continue to operate from an activation hypothesis. Although there have been several reports of stimulus-response specificity with such dimensions as fear and anger (Ax, 1953; Funkenstein, 1956; Schachter, 1957; WOlf & Wolf, 1947), simple stimulus properties (Davis, Buchwald, & Frankman, 1955), require- ments for environmental intake or rejection (Lacey, Kagan, Lacey, & Moss, 1963; Obrist, 1963), and hunger and pain (Engle, 1959), not until Lacey (1967) argued. that activation theory was in need of revision did psycho- physiologists begin to seriously investigate the behavioral correlates of particular autonomic response patterns. Lacey suggested that situations which require attention to the environment in the absence of cognitive processing were accompanied by an increase in electro- dermal activity and a decrease in heart rate. This stimulus-response specificity has stimulated a consider- able amount of research activity in recent years although investigators have tended to neglect electrodermal measures while concentrating on heart rate components of attention, often using heart rate deceleration (Graham & Jackson, 1970) and reduction in variability (Porges, 1972) as measures of attention. Whereas the specific psychological dimensions that are related to heart rate deceleration have been debated (see Hahn, 1973) the empirical fact remains that during some kinds of attentional activities heart rate deceleration is accompanied by an increase in electrodermal activity. Activities such as mental arithmetic are associated with a high heart rate and high electrodermal activity while other activities such as rest are associated with low levels of both measures. Another approach to explaining difference in patterns of autonomic activity is to examine individual differences in response patterns. This individual- response Specificity has particular relevance to psycho- somatic medicine since it is sometimes assumed that patients with psychosomatic complaints are overresponsive in a particular organ system. Several investigators have reported that people with psychosomatic complaints are more responsive in the affected organ than in other organs (Engle & Bickford, 1961; Malmo & Shagass, 1949; M003 & Engle, 1962). Lacey, Bateman, and Van Lehn (1963) were the first investigators to examine individual- response specificity in normal populations. They applied a number of stimuli to more than 200 subjects. Response levels for three different measures -- heart rate, heart rate variability, and skin conductance -- were then rank ordered for each subject. The investigators found that the rank order for each response for each subject tended to remain the same in each stimulus situation. For example, a subject who had a heart rate in the 80th per- centile during one stimulus situation would tend to have a heart rate in the 80th percentile in another stimulus situation while another response might rank consistently in the 30th percentile. These results were replicated and extended in a study in which blood pressure measure- ments were included (Lacey & Lacey, 1958). Thirty-nine of the 42 subjects showed statistically significant coefficients of concordance indicating that subjects tended to show the same pattern of autonomic activity even during different stimulus conditions. The studies by Lacey and his associates only looked at level scores during the stimulus situations and did not look at change scores. That is, levels were not compared with pre-stimulus or baseline levels during rest to see if there were patterns to the change as well as to the level that is reached. Schnore (1959) extended the Lacey studies by looking at levels of several autonomic and muscle activity scores as well as the changes in the activity levels attributable to the stimulus presentation. Results indicated high indi- vidual—response specificity. Coefficients of concordance for level ranged from .34 to .99 with a median of .80. The coefficients of concordance for the change scores were somewhat lower, .23 to .80 with a median of .51. In the studies by Lacey and his associates and by Schnore there was no evidence for stimulus-response specificity despite the fact that several different tasks were used. Similar studies by Engle (1960) and Engle and Bickford (1961) did find both individual—response and stimulus- response specificity when analysis of covariance designs were used. It should be pointed out that subjects in these experiments generally show individual-response stereotypy but there are many subjects who do not show stereotypy. Sternbach (1966) has suggested that the degree of stereotypy that a subject shows is itself an individual characteristic such that subjects can be classified according to the rigidness or randomness of their response patterns. There is little evidence indicating the stability of individual-response specificity over time. Lacey and Lacey (1962) measured several autonomic responses in children to a cold pressor task and found reasonably stable response patterns over a 4-year-period. Oken et a1. (1963) found very little relationship between response patterns measured a few days apart but the sti- muli were different for the two testing sessions. During the first day the stressor was a psychiatric interview and during the second day the stressor was simply an unpleasantly hot room. Thus, sessions were not entirely comparable and it is difficult to assess the effects of the different kinds of stimuli. Both of the longitud- inal studies tested subjects on only two different occasions so it is difficult to determine whether changes in pattern were due to habituation or whether the patterns were, in fact, unstable. It is interesting to note that the study which showed the highest stability of response patterns (Lacey & Lacey, 1962) had the longest time between testing sessions and the subjects were children, a population that would usually be expected to have the greatest amount of change, especially over a four year period. It may be that separate tests con- ducted a few days apart is an inappropriate method to test for the stability of response patterns since any stabilities that exist may not show up until after several tests. It is very likely that much of the difference between the first session and the second session are due to habituation rather than instability. In the experiment by Lacey and Lacey (1962) each test session would be like a first session since four years had elapsed between sessions. Engle (1960) has given three different but related definitions of stimulus-response specificity: "(1.) Maximal change occurs in the same function to a given stimulus in a set of subjects, (2.) Consistent rank orders of responses to a given stimulus occur in a set of subjects, and (3.) Consistent inter-response correlations to a given stimulus occur in a set of sub- jects." He has also given a set of parallel definitions for individual-response specificity: "(1.) Maximal change occurs in the same function within each subject to a set of stimuli, (2.) Consistent rank orders of responses occur within the same subject to a set of stimuli, and (3.) Consistent interresponse correlations occur within the same subject to a set of stimuli." According to Engle's viewpoint stimulus-response spe- cificity is a population variable rather than an indi- vidual variable. It would not matter that individual subjects would show response patterns unlike the popula- tion as a whole as long as there were a reasonably reliable pattern for the entire population. Individual idiosyncratic response patterns are of little interest since there would be no way of telling in a single test whether the pattern was due simply to normal variation or whether the pattern was elicited reliably by a par- ticular stimulus for a particular subject. Test-retest should reveal whether individuals can show different patterns of stimulus response specificity but thus far no such studies have been conducted. None of these studies have looked at heritabil- ities of the patterns although it is often assumed that such patterns are genetically influenced since psycho- somatic disorders often run in families (Sternbach, 1966). To date, the twins method is the only technique used to study the heritability of autonomic activity. Psychophysiological studies of twins There have been several twin studies of auto- nomic activity with varied purposes and consequently varied paradigms. Tables 1 and 2 summarize these studies. In these tables, "r" refers to the intra- class correlation for the twin pairs. Intraclass correlations can be derived from an analysis of vari- ance or by the standard Pearson product-moment correla- tion in which every pair is entered twice, once in each order. The effect of this procedure is to produce a regression equation with a slope of ”r" and an intercept of 0.0. "F" refers to the ratio of the M2 to DZ within twin variances. Some of the studies have employed only MZ twins and many have measured only resting levels of autonomic activity. All of the studies have claimed to find heritable factors although such claims would be om.H HH. we an. cc Hammond vHou ou uncommon OH.H NH. we mm. mm maHnaHum umH cu uncommon MH.H 9H. he aH. He N poHuma .ummu um Hm>mH mH.H OH. mu mm. Hg H pOHumn .umou um Hm>mH Hmnuamuoa mama ..Hm um Oweaumao mm. HH muoom :OHumnanm: pm>HHmv «m. HH manaHum umH ou mpnuHawna mmcommmu magnuospaoo wGOH oomH .wSHB w Hanna es. HN manuaawma Anmv I A+mv cams mo. Hm wpauHawma mmcoamwu came Hm. Hm Hw>mH maHnsHum coma Nw. HN Ho>mH mnHaaHumwum amma :oHumaHaHuopr pmcoHquaoo .ooamumHmmH momH .HUOHm oo.H om cm nope umaame mm. cm qm m HHmn mm.H on qm N HHmp «m. cm cm H HHon 0¢.H mm «m uanH mo amMHm ob mpsuHawma uncommon moamumHmmu momH ..Hm um wumncmpam> no. n momdommmuaoa m ou mHmHuu ma. 5 manaHum :oHumnqum: ou hocmuMH mmcoammu cannumHmmu ooaH .amaaomm mm. oH ummu Hw>wH monouUSUnoo qqu .wmuaom a anon m u z u z maHsaHum pan whamnmz hpaum ND NE .mmpsmmmfi HmahmpouuomHm waHm: mmanum maHSu mo humaaamII.H MHmma ooafl .waas a “menu ow. NH oo. NH mwcmno .aHma oHamnomH no. NH we. NH maHaaHum .aHma UHEmsomH NH. NH no. NH maHsaHummun .cha oHamnomH no. NH ow. NH wwamso .ummu uoHou moouum Hm. NH Nm. NH manaHum .ummu uOHou noouum mH. NH 0n. NH manaHummum .ummu MOHoo aoouum womH ..Hm um ouHmmzm ow.H mm. mm He. 0N Hommmua pHoo cu mwcmno mm.H Ne. mm mm. mN mnHaaHum umH 00 mo «0 mwsuwamms cm.~ as. me no. mm N voauma ummu Hm.m mm. we no. mm H pOHumm ummu mmmH ..Hm um mmpHumHo $0. HN Almv I A+mv come no. HN unaanwma mmcommmu amwa mm. HN Hm>mH manaHumumoa amma mm. HN Hm>mH msHaaHummum name :OHumcHaHuome pmaoHquaoo NcmH .xUOHm mm.N oH NN noun “mean: mo.H 0H NN m HHmn om.H 0H NN N HHmn ON.H 0H NN H HHmn mm.~ 0H Nu uewfia mo :mmHm ou mpSuchma mmaoawmu momH ..Hm um wumncmpam> om. om. ad mm. am ummu sapwoaeumuoumaamn Homa ..Hm um museum: #0. oH uwmu qemH .wmuaom w umOH m u z u z mnHaaHum zpsum Na N2 .wmhfimmwa mumh UHMMS wfifimfi mwfififium may“ NO uhhmae—HmlloN Wan—gun. 11 difficult to justify in those studies using only MZ twins. Some of the studies used other psychophysiological mea- sures but these measures are not reported in Table 1 and 2 since heart rate and electrodermal activity are of primary interest in the current study. Despite the differences in the studies some general patterns emerge. MOst of the studies show higher concordance in MZ than in DZ twins. Although heritability estimates are not usually reported, F ratios based upon the reported data are generally higher for the heart rate measures than for the electrodermal measures. The latency measure reported by Rachman (1960) is of little psychological interest since it is thought to primarily reflect the conduction rate of the sweat gland effector fibers and the migration of acetylcholine to the sweat glands (Edelberg, 1972). The generally lower heritabil- ities for electrodermal measures may reflect the wider variation of the measurement techniques which are employed and perhaps a lower reliability of the measure. All of the studies seem to be operating from an activa- tion assumption since none of the studies looked for stimulus-response specificity. Moreover, none of the studies tested for heritable factors in individual- response stereotypy. The present study was designed to examine heri- tabilities of stimulus-response and individual-response 12 specificity by assessing heart rate, electrodermal and respiration measures in M2 and DZ twins. The specific tasks employed were mental arithmetic, reaction time, and rest. Mental arithmetic has previously been demon— strated to elicit high heart rate and high electrodermal activity (Engle, 1960; Lacey, 1959). The reaction time paradigm yields reliable temporal changes in heart rate (Allen, 1973; Chase, Graham, & Graham, 1968; Fitzgerald & Porges, 1970; Obrist, Webb, Sutterer, & Howard, 1970) which are related to changes in respiration and heart rate variability (Headrick & Graham, 1969; Porges, 1972). The reaction time paradigm also requires attention which normally produces heart rate deceleration and increased electrodermal activity (Lacey, 1959; Obrist, 1963). Rest normally produces a low heart rate and low elect— rodermal activity (Lacey, 1959). A genetic factor in individual-response stereo- typy would be indicated by overall patterns of responses which, for any given stimulus condition, are more similar from MZ pairs than for DZ pairs. A genetic factor in stimulus-response patterns would be indicated by differ- ences in patterns of responses across stimulus conditions which are more similar for M2 pairs than for DZ pairs. METHOD Subjects The subjects were 15 M2 and 15 DZ twins. The sample was predominantly female. There were 6 male MZ pairs and 4 male DZ pairs. The twins were recruited from a list of all Michigan State University students who had identical last names and birth dates. Names were provided by the registrar's office. Zygosity for most pairs was determined by the Nichols and Bilbro (1966) questionnaire procedure. (See Appendix A). (See Appendix B for a discussion of zygosity determin- ation). The height and weight of each twin was also measured at the time of the experiment. According to this procedure, twins are diagnosed at two different levels. If the twins fit any of the descriptions at the first level they are classified at that level. If none of the first level descriptions fit, the MZ and DZ points are added up according to the descriptions at the second level and the classification with the highest number of points determines the assignment of the twins. The diagnostic rules were as follows: 13‘ 14 First Level Diagnosis of Dz Distinctly different hair color or curliness Distinctly different eye color Height differences of 3 inches or more Both twins report that they are never mistaken by teachers Diagnosis of MZ Both twins report they were frequently mistaken by parents when young Both twins report that they were frequently (or one frequently and the other occasionally) mis- taken by parents recently Both twins report that they are frequently (or one frequntly and the other occasionally) mis- taken by close friends Second Level point towards diagnosis of D2 Slight differences in hair color, curliness, or texture Slight differences in eye color Height difference of one and one half inches or more Weight differences of fifteen pounds or more Either twin reports that they are never mistaken by casual friends 15 Twins agree that they are fraternal One point towards diagnosis of M2 Either twin reports that they were occasionally or frequently mistaken by parents when young Either twin reports that they were occasionally or frequently mistaken by parents recently Either twin reports that they are frequently mistaken by teachers Either twin reports that they are occasionally mistaken by close friends Either twin reports that they are frequently mistaken by casual friends Twins agree that they are identical One pair of twins was classified as MZ because they had previously participated in another twins study in which blood typing determination revealed that they were MZ's. Another pair who claimed to be MZ were classified as DZ because they said that they could not give each other blood transfusions since one was Rh+ and the other Rh-. The questionnaire data confirmed the classification of these two sets of twins. In three cases there was either a tie or only one point of difference between the MZ and DZ classifi- cations at the second level of the questionnaire so these pairs were classified by other criteria. One pair was I’ll-Ill :l' {fl-lull III"! .II [Ill (Ill [[1 [1" ll. |l III II .III I l 5' Irl IIlt1.'\I I ll.‘ l l6 classified as MZ because they had birth marks of exactly the same shape and size. Another pair was classified as DZ because one had a fingerprint ridge count of 119 while the other had a count of 94. This was greater than any other MZ pair. The third pair was classified as DZ because they had entirely different birth marks and their dentist reported that their teeth were entirely different. The correlations for height suggest that there were no gross errors in classification. The M2 twins correlation was .97 and the DZ correlation was .64. These figures are fairly close to those published by Lykken (1974) who reported correlations of .91 and .54 and those published by Newman, Freeman and Holzinger (1937) who reported correlations of .93 and .64 respec- tively. Apparatus Skin conductance, electrocardiogram, and respiration were recorded on a four channel Grass model 7 polygraph. For the skin conductance measure a con- stant voltage (0.5V) bridge was used which has an out- put of 1.0mV per 1.0 micromho of input. The polygraph channel was operated in the DC mode with the output reading directly in conductance units. The electro- cardiograph channel was frequncy limited to provide 17 ‘maximum output of the R wave and minimum output of the P and T waves, movement artifact, electrodermal signals and electromyographic signals. The output of the electrocardiograph channel then drove a recording pen and a beat interval counter. The beat interval counter provided a display of the interval between the last two R waves. Each second a printer printed out the number being displayed on the counter. Respiration was recorded by a bellows which was strapped around the subject's chest. The output of the bellows was attached by a plastic tube to a pressure transducer which in turn was attached to a DC channel of the polygraph. A total of three active electrodes were attached to the subject. Two skin conductance electrodes were placed about 1.5 cm apart on the hypothenar eminence of the left hand. These two electrodes also served as the left arm electrode for the electrocardiogram. A third electrode attached to the volar surface of the right wrist served as the right arm electrode for the electro- cardiogram. A ground electrode was also attached to the volar surface of the left wrist. Appropriate tests were performed to assure that there was no interaction between the heart rate and electrodermal measurements. All of the electrodes were of the silver-silver chloride type constructed according to Venables and Martin (1967). The electrolyte for the two skin conductance 18 electrodes was a Unibase preparation (Lykken & Venables, 1971). The electrolyte for the right arm and ground electrodes was Beckman electrode paste. Prior to applying each electrode the sites were cleaned with 70% ethanol and allowed to dry. Design and Procedure Immediately after arriving at the laboratory a' shortexplanation of the experiment was given and the two conductance electrodes were attached. The subject was then given a copy of the instructions and asked to read the instructions. (See Appendix C for a copy of the instructions.) After the experimenter answered any questions, the subject was then seated in a sound attenuated booth, the remaining two electrodes were attached, and two practice trials of the reaction time task were given. A.minimum of 10 minutes was allowed between the attachment of the conductance electrodes and the start of the first task to allow for skin hydration (Edelberg,l972). The subject was allowed to relax in the booth with the door open until this 10 minute period was completed. When 10 minutes had passed the subject was told that the first task would begin in about a minute and to wait for specific instructions over the speaker. 19 The door of the booth was closed and after a minute the tape recorder started. The voice on the tape recorder said "Okay, when I tell you to start you are to count backwards from 800 by 7's as fast as you can. You are to count to yourself. Everytime I say 'number' tell me what number you are on and then continue counting backwards to yourself. Remember you are to count back- wards from 800 by 7's and speak only when I say 'number'. Okay, you may start - NOW." After 30 seconds and again after 60 seconds the voice on the tape recorder said "number." Fifteen seconds after the last number was requested the voice on the tape recorder said "Okay, you may stop counting now. Please pick up the thumb operated reaction speed switch, hold it in your right hand and get ready for the reaction time test. Remem- ber that this is a test of speed. You are to press the switch as quickly as possible after the ready light goes off and the go light goes on. The first trial will start in about a minute." The ready light was on 16 seconds for each trial with a randomly determined inter- trial-interval of 20, 25, or 30 seconds (i=25 seconds). Immediately after the ready light went off the go light came on. The go light went off when the subject pressed the switch. There was a total of 15 trials. 20 Following the last reaction time trial the sub- ject was instructed via the tape recorder: "That com- pletes the reaction.time task. All you have to do for the remainder of this experiment is sit back and relax for approximately 5 minutes. Following this rest period the experimenter will come in and disconnect the sensors.‘ Data Scoring The analysis proceeded from the data collected during predetermined time periods. The data were derived from the three different measures; heart rate, skin conductance and respiration. Mental arithmetic and rest. Data were collected from 20 second time periods during these two stimulus conditions. There were three sample periods during the arithmetic task, one ten seconds after the onset of the task, one ten seconds after the first "number" was requested, and one ten seconds after the last number was requested. The subjects did not verbalize during any of these periods. There were four sample periods during the rest condition. They were during the latter half of the second through the fifth minutes of the condition. The data which were analyzed included: 1. heart period 2. heart period variability 21 3. breathing rate 4. breathing depth in mm of pen deflection 5. electrodermal frequency, the number of positive pen deflections of the skin conductance measure 6. the sum of the heights of the positive deflections of skin conductance measure 7. the skin conductance level at the beginning and end of each sample period Reaction time. Data were collected during 32 second time periods for each of the last 10 trials. The scoring period started 8 seconds before the start of each trial and was divided into four 8 second periods respectively designated prestimulus, orienting response, attend, and response. The same 7 variables were analyzed as in the mental arithmetic and rest periods but for the heart rate data the periods for analysis were separated for each of the 8 second periods. In addition the follow- ing variables were analyzed: 8. the height of the skin conductance response to the onset of the READY LIGHT 9. the height of the skin conductance response to the respond signal 10. 22 reaction time, the time from the offset of the READY LIGHT to the button press ' ' ' .1213! RESULTS Data analysis The data for each variable and each stimulus condition were submitted to an analysis of variance routine (McNemar, 1962, 322-329). Sums of squares were used to find the variances, covariances and intra- class correlations. (See Appendix D for a summary of the means and sums of squares for each of the variables and stimulus conditons.) Means and sums of squares as well as variances and covariances are listed in all tables in unconverted scoring units. However, data reported in the figures are in normal units. Change scores were also computed for each subject by subtract- ing the mean during one stimulus condition from the mean of another stimulus condition. These data also were analyzed by the analysis of variance noted above. The data were then converted to logarithmic units and the same analyses were performed. The data were converted to log units for two different reasons. The first reason was to counter the possibility that change score hereitabilities may have been influenced by the scaling procedure. If the amount of change from 23 24 one stimulus condition to another is a multiplicative function of the level, then the log transformation should equate the change scores for subjects who initially start at different levels. Log transformation of the data was selected since this is the most common trans- formation applied to physiological dependent variables. The second reason for using log transformation was that examination of the two different heritability estimates suggested that some of the data were heter- oscedastic. Log transformation should have reduced heteroscedasticity. Heritabilities first were computed by the gen- eral formula Hi=2(er'rDz) where "r" represents the intraclass correlation. For much of the data, variances of the M2 and DZ populations were quite different. Under these conditions it is difficult to compare the two correlations. Therefore, heritabilities were also computed by the formula H§=(VwDZ -VwMz) Vt which was derived by Dr. John Hunter for the purposes of this study. Vw represents the variance of the differ- ences between the twin pairs and Vt represents the variance of the entire population. This formula is 25 equivalent to the first formula when the variances of the two populations are equal. The second formula assumes that the variance of the difference is unrelated to level (i.e. homoscedasticity). The variance of the difference was computed from the formula Vw=2(Var. - Cov.). Finally, each twins score on one task was paired with the co-twin's score on another task and the corre- lation coefficient was computed. A typical example of. these correlations is shown in Table 4. For example, in heart period the twin by co-twin correlation between mental arithmetic and reaction time was .41. Since each twin is entered twice, once for reaction time and once for mental arithmetic, each of these correlations is based on 30 pairs of data. For comparison purposes the intraclass correlations which are based upon 15 pairs of data points, are entered along the diagonal. In addition, each twin's score on one stimulus condition was paired with his score on another stimulus condition and the correlation was computed. These correlations, based upon 60 pairs of data points, are listed as "Total, subject by subject" correlations. The alpha coefficients are listed in parentheses along the diagonal. 26 One way to obtain an estimate of the relative genetic contributions of individual-response specificity and stimulus-response specificity is to compare the twin by co-twin correlations with the intraclass correlations. If the twin by co-twin correlations and intraclass corre- lations are approximately the same, no evidence for stimulus-response specificity would be obtained. However, if the heritabilities for the two stimulus conditions were high and the twin by co-twin correlations between stimuli were low, we would have evidence for a heritable factor in stimulus-response specificity. Each twin would have to remain in approximately the same point in the distribution during the two stimulus conditions for these correlations to be equal. This would indicate individual response specificity. Heart period The mean trial-by-trial heart rate is shown in Figure 1. Heart rate is fairly high for the mental arithmetic task but fairly low for the reaction time task. This is consistent with the observation that attention such as that required by a reaction time task is associated with relatively low heart rates. Table 3 summarizes the descriptive statistics for heart period. Generally there is little difference between the correlations obtained for the untransformed 27 om. wH.H ooo.| woo. qo.u Noo. coo. mm. MMIHm om. «a. Hoo. NHo. «o. moo. NHo. Hm. mmaoo .um> u .>oo .Hm> u whammmz N N No NZ .uOHHma upon: you moHumHumum m>HudHuommonl.m MHm pOHuma uuwmn on .mmamno mm. om. NHo.I owN. qo.l oHH. NNN. oq. malam om.H oo. Noo.: omo. NH.I omH. MHm. om. mmu oOHumm uummn .mwamnu NH. .mm.l oqe. «NH.H oq. HNH. NoN. NN. ummm mN. HN.! qu. mow. Ne. on. mmm. Nm. maHu cOHuommm wN. No. HwH. mmm.H NH. owm. mQN.H Ne. oHumanuHum Hmuamz NuHHHanum> voHumo undo: NOH NN. mo.l owe. mmo.H me. ooN. mHm. oq. ummm mm. co. omm. Hoo. om. omH. ooq. mm. mEHu aOHuummm «o. No.1 ooH. on. mm. HoH. NmN. .qm. oHumanuHum Hmucmz NuHHHanum> oOHumm pummm pm m: .>oo .um> u .>oo .Hm> u nuance: N N N9 N2 .NuHHHanum> voHumm undo: How moHumHumum m>HumHuommoll.m NHm ooHHQNqum a wOH NuHHHanHm> oOHuwaqummm .NuHHHQMHum> oOHumo ammo: How HHnaHum ammsumn chHuMHmuuooll.o MHmoo .um> u .>ou .um> u nuance: N N9 N2 .Nuamsvmum HmaumpouuUMHm How moHumHumuw o>HunHuommonl.N mHmoo .um> u .>oo .um> u whammmz N N .N9 N2 .ustm: Hmuou HmaumvouuomHo How moHumHuMum m>HuaHuomonII.m MHHmH moamuonoaoo MOH .mmdmno nH. om. ooN. NHm.H MH. NNN. «oN.H mm. mMIHM mo. mm. ooH. mNN.m mo. Hmo. «m«.« HN. mMImH cannuonoaoo .mwcnao oo. «N.I Ho«. omo. NN. NoN. H««. Ho. ummm mN. NH.I com. No«. oo. ooH. NoN. oo. maHu :OHuummM mo. oN.: «MN. o«m. No. «MH. oMN. Nm. oHumESHHum Hound: Hm>mH moamuoovsoo on o«.: oo. wmo. «oH. mm. me. mNm. Nm. Ammo ummm o«.l mo.l coo. oNH. Nm. NmH. mNN. on. AHMV waHu GOHuommm 3.- 3. m8. m2. 3. NS. mmm . mm. 22v #3533 H35: Hm>mH moamuonpaoo am am .>oo .um> u .>oo .Hm> u munmmmz N N N9 N2 .Hm>MH moamuonpaoo ame How mmoHumHumum 0>HunHuommnll.HH mHmmH monsoonpcookon Hm>mH mocmuonocoo .Hm>wH moamuunpaoo cme How HHnaHum ammaumn maOHumHmuuooul.NH MHmeH 48 stimulus conditions and therefore may be of more interest. Unfortunately there were differences in the variances of the two populations and the log transoformations often yielded different results. At present, the safest con- clusion is that the heritability estimates are generally positive although they are too highly divergent to make any good estimate of the true heritability. The twin by co-twin correlations shown in Table 12 are almost exactly the same level as the intraclass correlations. There was virtually no difference in the distribution from one stimulus to another, therefore providing no evidence for stimulus-response specificity. Breathinggrate The trial by trial breathing rate illustrated in Figure 3 was relatively high for the mental arithmetic task but lower for the other two tasks. Breathing rate is rarely reported in psychophysiological research so there was no reason to expect one pattern of responding over another. The intraclass correlations for M2 twins are generally higher than for DZ twins but the differences are vey small and in a few cases in the opposite direc- tion. The variance for the DZ population is somewhat higher than for the M2 population yielding higher Hg 49 No. HN. Noo. oNo. «N. HHo. «mo. «m. MMIHM «H. mN.I HHo. Noo. NH. Noo. «mo. «o. mmnoo .Hm> u .>oo .um> u spawns: N N No N2 .mumu wcHaummun How moHumHumum w>HumHuummoll.MH MHH¢H 50 H«m.o oo. Awm.o «N. am. Ham.o HN. NN. ms. mo. NN. HN. «m. 9H. om. mm. em. as. Amm.o so. Aaa.o am. am. Ham.o Hm. mm. as. NH. HN. mo. as. am. am. mm. as. as. ummm aOHuomom oHumazuHum Hmucmz uomnnsm Np oomflpam .Hmuoa ummm maHu coHuommm oHumazuHum Hmucmz cHauloo Na :Hau .No ummm maHu aoHuummm oHumanuHum Haucoz nHauloo No aqu .N2 maHu UHumanuHum ummm coHuummm Hound: mumu wcHnumwun wOH maHu UHumanuHum uwmm coHuomwm Hauamz mumu wdHnummum coHumHmuuoo .mumu wcqummun you HHnaHum consume mnoHumHmuuooll.«H MHmoo .um> u .>oo .um> u whammmz a No N: .kummu QEHU HHOHUUQQH HON mUHumflumum 0>HUQHHUWQQII.MH MQM