12 pages

Approach-Withdrawal and Cerebral Asymmetry- Emotional Expression and Brain Physiology

of 12
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  Journal of Personality and Social Psychology1990, Vol. 58, No.  2 330-341 Copyright 1990 by the American Psychological Association, Inc.0022-3514/90/ 00.75 Approach-Withdrawal and Cerebral Asymmetry:Emotional Expression and Brain Physiology I Richard J. Davidson University of Wisconsin—Madison Clifford  D.  Saron and Joseph A. Senulis University of Wisconsin—Madison Paul Ekman Human Interaction LaboratoryUniversity of California, San Francisco Wallace V. Friesen University of California, San Francisco In this experiment,  we  combined the measurement of observable facial behavior with simultaneousmeasures of brain electrical activity to assess patterns of hemispheric activation in different regionsduring the experience of happiness and disgust. Disgust  was  found to be associated with right-sidedactivation in the frontal and anterior temporal regions compared with the happy condition. Happi- ness was  accompanied  by  left-sided activation in the anterior temporal region compared with disgust. No  differences  in  asymmetry  were  found between emotions  in  the central and parietal regions. Whendata aggregated across positive films were compared to aggregate negative film data, no reliabledifferences in brain activity were found. These  findings  llustrate the utility of using facial behaviorto verify the presence of emotion, are consistent with the notion of emotion-specific physiologicalpatterning, and underscore  the  importance of anterior cerebral asymmetries for emotions associatedwith approach and withdrawal. This and the accompanying report (Ekman, Davidson, &Friesen, 1990) represent the convergence of two independentyet overlapping lines of research. One is theory and evidencefrom Ekman's laboratory, which, from an evolutionary per-spective, takes the position that each emotion is characterizedby unique patterns of expressive behavior and physiological ac-tivity (e.g., Ekman, 1977, 1984). The second is research fromDavidson's laboratory on cerebral asymmetry and emotion,which indicates the differential involvement of the two cerebralhemispheres  in  the control of certain positive and negative emo-tions (e.g., Davidson, 1984a, 1984b, 1987; Davidson &Tomar-ken, 1989).Psychophysiological Specificity and EmotionThe degree to which different emotions are accompanied byunique patterns of physiological activity is a question as old asthe study of emotion  itself.  Two  diametrically opposed positionson this issue have been advanced, with ardent supporters ofeach view. Cognitive theorists (e.g., Mandler, 1975; Schachter& Singer, 1962), following Cannon (1927), have claimed thatdifferent emotions are accompanied by the same pattern of un-differentiated physiological arousal. The second position, con- This research  was  supported by grants from the National Institute ofMental Health (MH 40747 and MH 43454) and the Graduate Schoolof the University of Wisconsin to Richard J. Davidson and by ResearchScientist Award MH 06092 from the National Institute of MentalHealth to Paul Ekman.We wish to thank Andrew Tomarken for valuable comments on anearlier draft of this article, and Linda Kinney, Barry Cohen, Barry Gor-don, and Andrea Straus for assistance with the data analysis.Correspondence concerning this article should be addressed to Rich-ard J. Davidson, Department of Psychology, University of Wisconsin,1202 West Johnson Street, Madison, Wisconsin 53706. sistent with the theoretical writings of Darwin (1872/1955) andJames (1890), asserts that different emotions are accompaniedby unique patterns of physiological activity (e.g., Ax, 1953). Ek-man (1984) has specifically suggested that unique central or au-tonomic patterns will differentiate among the primary emo-tions of happiness, sadness, anger, fear, disgust, and surprise.There are now data available that are beginning to illuminatethis debate. Ekman and his colleagues (Ekman, Levenson, &Friesen,  1983;  Levenson, Ekman, & Friesen, in press) have un-covered evidence for unique patterns of autonomic activity thatdifferentiate  among the  negative emotions of fear,  anger,  disgust,and sadness, and differences between these negative emotionsand happiness. Other investigators have also reported reliabledifferentiation among some negative emotions on the basis ofmeasures of autonomic physiology (e.g., Schwartz, Weinberger,   Singer,  1981).To date, there has been little research that has explored thehypothesis of distinctive central nervous system patterns thatdifferentiate among the primary emotions. A number of ele-mentary circuits for certain constellations of emotional behav-ior have been described (see Papez, 1937, and MacLean, 1949,for early analyses of this problem and Panksepp, 1982, for amore modern view), although the degree to which these maponto discrete emotions in humans is not entirely  clear.  Becauseunique expressive signals and subjective experience character-ize each of these primary emotions, one would expect them tobe associated with distinctive, invariant patterns of central ner-vous system activity at some level of the neuraxis. One impor-tant goal of the present research is to advance knowledge re-garding this unexplored topic.Cerebral Asymmetry, Approach-Withdrawal,and EmotionIt is the second body of research that provides a useful theo-retical  guide  for specific hypotheses about  the  distinctive central 33  BRAIN ASYMMETRY AND APPROACH-WITHDRAWAL 331 nervous system activity for  some  of the emotions. Davidson andothers have amassed a variety of evidence over the past  10  yearsthat indicates that the anterior regions of the two hemispheresof the brain (frontal and anterior temporal) are differentiallyinvolved in certain positive and negative emotions (see reviewsby Davidson, 1984a, 1984b; Davidson & Tomarken, 1989;Kinsbourne   Bemporad, 1984; Leventhal   Tomarken, 1986;Silberman & Weingartner, 1986; Tucker & Frederick, 1989).Davidson (1984a, 1987), following an early suggestion of Kins-bourne's (1978) concerning the relation between approach be-havior and the left hemisphere, has theorized that the funda-mental continuum along which the anterior cortical regions arelateralized  is  approach-withdrawal, with the left anterior regionsubserving  an  approach system and the homologous right hemi-sphere region subserving a withdrawal system. Davidson hasfurther argued that to the degree that approach and withdrawalare components of different emotions, such emotions shoulddifferentially activate the anterior regions of the two cerebralhemispheres. For example, several emotion theorists (e.g., Ek-man   Friesen,  1975;  Plutchik,  1980)  have noted that both fearand disgust often include behavioral components of with-drawal, although in somewhat different fashions. For fear, Ek-man proposes that withdrawal entails escaping from the threat-ening stimulus, whereas in disgust the withdrawal entails termi-nating the input, whether it  be  olfactory, oral, or visual.In this experiment,  we  sought to examine the pattern of brainelectrical activity derived from multiple scalp loci during thesetwo withdrawal emotions, fear and disgust. The brain activityduring  these  emotions  was to be  compared to an approach emo-tion, happiness, and a baseline condition.  We  used short, 1-minpositive and negative film clips to arouse these different emo-tional states.Methodological Desiderata for PsychophysiologicalResearch on EmotionBelow  we  note eight characteristics that ideally should be in-corporated into all research on the biological substrates of hu-man emotion and that apply to studies of both autonomic andcentral nervous system components of emotion.  We  then illus-trate how the present research conforms to these desiderata.1.  Emotion must  be actually  elicited.  Although this require-ment may seem trivial, many experiments that purport  to  studyemotion may not actually involve the production of emotion insubjects. Some studies focus more on the perception of emo-tional information, whereas others examine the language thatis used to describe emotion. If the goal of the research is tocharacterize the psychophysiology of emotion, then there mustbe evidence, apart from the dependent variable of interest, thatemotion  was  actually produced. 2.  Adequate  procedures  must  be used to verify  the  presence of the  intended  emotion.  One of the noteworthy characteristics ofemotion  is the  lack of an isomorphic relation between an elicitorand  a  particular emotion  (see  Ekman, 1984). In other words, thesame elicitor will often produce an array of different emotionsacross subjects. Even in response to  elicitors  that are specificallychosen to target  a  particular discrete emotion, subjects typicallywill report  a range  of different emotions if given the opportunity(e.g., Ekman, Friesen, & Ancoli, 1980). Sometimes these otheremotions are experienced at different points in time over thecourse of an eliciting event and, at other times, emotions mightbe produced in blends, with different emotions experienced si-multaneously. Whatever the time course of the different emo-tions,  we  believe that the use of self-report  is  insufficient to ver-ify the presence of  the  emotion intended by the investigator, ifa self-report instrument asks only about the targeted emotion.Such a measure will not  be  sensitive to the possibility that otheremotions might have been experienced in addition to the targetemotion.We believe that one of the major methodological inadequa-cies in previous psychophysiological research on emotion hasbeen the failure to verify with precision that the intended emo-tion  was  elicited in every subject. Typically, it  is  only the investi-gator's presumption that supports the possibility that the emo-tion elicitor produced the intended emotion. The physiology inresponse to the elicitor is usually examined without regard toverifying the investigator's presumption. 3.  Epochs of  different discrete  emotions must be  separable. Given the likelihood that any given elicitor will produce morethan one emotion that may change sequentially over time, it isimperative to identify separate epochs during which differentdiscrete emotions are present. Only in this  way  can the physiol-ogy that accompanies different emotions be compared. Notethat  this  requirement necessitates  a  method, such  as the  analysisof facial behavior, that provides a continuous or near-continu-ous measure of emotional state. Using such a measure, an in-vestigator can then extract, post hoc, those periods duringwhich different discrete emotions were present. Because mostphysiological studies aggregate data  over the  entire length of theeliciting stimulus, emotions other than the intended one wouldcontribute to the physiological changes observed. Previous fail-ures to find physiological differentiation among emotions may,at least in part, be attributable to the failure to compare differ-ent discrete emotions. Most investigators unwittingly comparebetween two or more blends of emotion. 4.  Behavioral  and  physiological  measures of  emotion  must be appropriately  synchronized.  If a continuous behavioral mea-sure of emotion is used to flag epochs during which differentdiscrete emotions are present, the behavioral and physiologicaldata streams must be accurately synchronized. This necessi-tates the use of common, simultaneously produced signals onthe videotape record and the computer  in  which the physiologi-cal data are stored. 5.  The physiological measures  chosen for study must  have  asufficiently fast time  constant  to  reflect  brief  periods  of emotion. Given the relatively  fleeting  nature of emotion, with most epi-sodes lasting less than 4 s (Ekman, 1984), it is clear that onlythose physiological processes that have a relatively fast timeconstant are logical candidates to examine the physiologicalsubstrates. Certainly, other affective phenomena related to emo-tion, such as mood, have longer durations and would thereforebe expected to involve other physiological systems whose re-sponse properties are more enduring.6.  At least two emotions and a baseline  condition  must be compared.  Many experiments on the psychophysiology of emo-tion, including some of our own previous studies (e.g., David-son, Schwartz, Saron, Bennett, & Goleman, 1979), have onlyincluded two emotion conditions. The problem with this strat-egy is that no nonemotion reference condition is included.Thus, it would be impossible to conclude which, if any, of theemotion conditions would differ from a baseline period. In astudy involving a comparison of two emotions, it is possible  332 DAVIDSON, EKMAN, SARON, SENULIS, FRIESENthat the two conditions could differ from each other, but notfrom a baseline condition. It  is  also possible that only one of theemotion conditions would differ from baseline. In studies thatcompare only one emotion condition with a baseline, it is im-possible to conclude whether the physiological changes ob-served during the emotion period are unique to that specificemotion or are nonspecific changes associated with any emo-tion.  For this  reason, at least  two  emotion conditions and  a  base-line are required.7.  The intensity of  the  elicited emotion must be matched among  conditions.  When two or more emotions are compared,it  is  imperative to match the intensity of emotion so that differ-ences in intensity do not confound the emotion-specific com-parisons. Many investigators who have compared two or moreemotions have not used any procedure to match the intensityof the elicited emotions (e.g., Schwartz et  al.,  1981). If  two  emo-tions differ in some parameter of physiology, but also differ inintensity, it  is  not possible to disentangle whether  the  physiolog-ical differences are simply a function of the intensity of emo-tion-related activation per se, or rather are associated with thespecific emotion that  was  elicited.8.  The data must be of  sufficient  duration for  each  emotionunder study.  Precisely what constitutes  a  sufficient duration willvary as a function of  the  physiological measures of  interest.  Ifstable estimates of physiological activity are to be obtained, acertain minimum amount of artifact-free data is required.Within-subject aggregation is performed to arrive at a singleindex of  a  particular physiological variable for a certain emo-tional state for  each  subject. For example, if physiological activ-ity during smiling was of interest, all instances of the artifact-free physiology during the target  smiles  would be extracted, an-alyzed, and then aggregated for each subject. If  the  dependentmeasure was the electroencephalogram (EEG), a minimum ofapproximately 10 s (across all instances of the target emotion)would be required to obtain a stable estimate of spectral power(Davidson, 1988). Note that although 10 total seconds are re-quired, each individual epoch of smiling can be quite  brief,  asshort  as 1  s per expression period.The study reported here  was  designed to accommodate eachof these eight methodological desiderata. Short film clips wereused to elicit emotion. These clips have been used extensivelyin  previous  emotion research and  have  been found to elicit bothself-report and facial  signs  of positive and negative emotion thatare comparable in intensity (e.g., Ekman et  al.,  1980).  To  verifythe presence of the target emotion, we used a combination offacial behavior and self-report criteria. The facial behavior wasused to flag the onset and offset of different facial signs of emo-tion. We used self-report criteria to ensure that subjects werereporting emotions whose valence was consistent with the film clip  used and  to  ensure that  the  intensity of the reported positiveand negative affect was matched. We extracted happy epochsfrom positive film clips and disgust epochs from the negativefilm clips. We had expected to be able to also extract fear ep-ochs, but happy and disgust  were the  only facial expressions thatoccurred with sufficient frequency.The EEG measures were precisely synchronized with the fa-cial behavior by having the same triggers produce event marks on  the  video  record and  the  computer data  base  simultaneously.The use of EEG, and the analytic method employed to quantifythe EEG, ensured that a sufficiently accurate time resolutionwas achieved. Chunks of EEG 1.02 s in duration served as theepoch length for analysis. These chunks were overlapped by 75%.  The analysis epoch started at the precise time of the onsetof the facial expression. Our study compared three periods—happy expressions, disgust expressions, and baseline—thus en-abling  us  to specify both the degree to which EEGs during eachexpression type differed from each other and from a restingbaseline. Finally,  we  required  a  minimum of  10 s  of artifact-freeEEG during each expression type (aggregated across multipleinstances of the expression) to use a subject's data for that ex-pression. This ensured that a sufficient amount of EEG wasavailable to compute stable estimates of spectral power (seeMocks & Gasser, 1984). Finally, we included a control proce-dure to evaluate the efficacy of our elaborate data extractionprocedure. EEG was analyzed in a manner similar to what istypically done in studies on the psychophysiology of emotion.Data from both positive films were aggregated and comparedwith the data from the two negative  films,  rrespective of facialbehavior. We predicted little difference in anterior asymmetrybetween the positive and negative film conditions because thedata for  this  comparison are aggregated across  all  nonfacial andfacial expression periods, including those that are inconsistentwith the target emotion. To  summarize, the goal of this study  was  to compare epochsof brain activity coincident with the expression of differentemotions. Facial signs of happiness and disgust occurred withsufficient frequency to compare EEG activity coincident witheach. We recorded EEG over the left and right hemispheres inthe frontal, anterior temporal, central, and parietal regions andvideotaped subjects unobtrusively while they watched shortemotional film clips. We specifically predicted that activationasymmetry would differentiate between happy and disgustemotions in the frontal and anterior temporal regions. The cen-tral and parietal regions were assumed to be relatively unin-volved in the generation of emotion. Brain electrical activitywas measured from these sites to serve as a comparison withthe more anterior sites and to evaluate the degree to which thevalence-dependent asymmetries  were  specific  to  the anterior  re- gions.The following specific hypotheses were tested in this experi-ment: Hypothesis  1.  EEG asymmetry from the entire film period,independent of facial behavior, will not discriminate betweenpositive and negative  film  conditions. Hypothesis 2.  Frontal and anterior temporal activationasymmetry will discriminate between happiness and disgust.Specifically, disgust will be associated with greater right-sidedanterior activation compared with happiness. Conversely, hap-piness will be associated with more left-sided activation com-pared with disgust. We specifically offer no predictions regard- ing  between-hemisphere differences within emotion conditions.The rationale for not making such predictions is based on thefact that substantial individual differences in baseline asymme-try exist, upon which are superimposed task-dependentchanges (see Davidson   Tomarken, 1989, for a review). Thus,for a subject with tonic extreme right frontal activation, wewould not necessarily expect that happiness would be associ-ated with absolute left frontal activation  (i.e.,  greater activationin the left compared with the right frontal lead). However, wewould still predict the between-condition, within-hemispheredifferences described above. Hypothesis  3.  Both disgust and happiness will be discrimi-  BRAIN ASYMMETRY  AND  APPROACH-WITHDRAWAL   nated from baseline  on  measures  of  anterior activation asym-metry. We specifically predicted that disgust would  be  associ-ated with  a  significant increase in right-sided anterior activationcompared with baseline and that happiness would be accompa-nied  by a  significant increase  in  left-sided anterior activationcompared with baseline.Method Subjects A total  of  37 right-handed (assessed with  the  Edinburgh HandednessInventory; Oldfield, 1971) women between  the  ages  of  17  and  41 yearswere tested. The sample was restricted  to  right-handed subjects becausehemispheric specialization is known  to  differ in left-handed subjects.  Of these  37  subjects,  26 had at  least  one  instance each  of  a happy  and  dis-gust expression during the positive  and  negative film clips. Of these  26, 17  had at  least  one  instance  of  each  of  these expressions that were  ac- companied  by  artifact-free  EEG  (EEG was scored  for  artifact prior  to any analysis;  see  below). Two  of  these  17  subjects were eliminated  be- cause of failure  to  meet the duration criteria, which was  a  minimum  of 10 total seconds of artifact-free EEG during each expression type. Fourother subjects were eliminated because they reported negative affect (i.e.,  fear, sadness, disgust,  or  anger)  at a  level  of  3  or  more  on a 0-8- point scale during at least one of the positive films from which the happyexpressions were extracted. We were thus left with  a  final sample  of 11 subjects.  The  percentage  of  subjects  who  were excluded  is  similar  to previous studies in which EEG data were extracted during facial expres-sion periods (Fox   Davidson, 1988). Procedure Subjects were tested individually. Prior  to  commencement  of the study, subjects were told that the experiment was concerned with subjec-tive and physiological reactions to short emotional film clips. After sign-ing  a  consent form that indicated that film clips designed  to  elicit bothpositive  and  negative emotion would  be  presented,  an  experimenter ap-plied electrodes  for  the measurement  of  the EEG. The subject was toldthat  no  intercom existed between  the  rooms  and  that  if the  subjectneeded  to  speak with  the  experimenter  for any  reason during  the  ses-sion,  she  could press  a  button mounted  on one arm of  the chair thatwould ring  a  signal  in  the control room, at which point the experimenterwould enter.  We  specifically designed  the  situation  in  this fashion  to maximize the degree to which subjects perceived themselves  to  be view-ing the film clips privately, with little experimenter contact.To further increase  the  degree  to  which subjects believed that theywere  not  being observed during the film viewing periods,  the  room wasdarkened  at the  time  the  experimenter departed.  The  subject was toldthat  the  purpose  of  the darkened room was  to  mimic  a  movie theater.Two small  red  lights provided low-level ambient illumination, whichwas required for subjects  to  view the number pad  on  which their ratingswere made (see below).The experimenter answered  any  general questions that  the  subjecthad and explained that the remainder of the instructions would  be  pre-sented  on the  rear-projection screen toward which they faced. Instruc-tions were presented  in a  self-paced fashion.  A  screenful of instructionswas presented  at a  time,  at the  bottom  of  which was  the  phrase pressto continue. The  subjects advanced through  all of  the instructions  at their own pace.The experiment began with baseline recordings  of  physiology, afterwhich  the  film clips were presented. Following  the  presentation  of  thefilm clips, another  set  of baseline trials was presented. Each  set  of base-line periods consisted of the presentation of one eyes-open and one eyes-closed trial, each  30 s in  duration.  The  order  in  which these trials werepresented was counterbalanced within  and  between subjects. The onsetand offset of the baseline trials were marked  by  the presentation of tonepips. Emotion-Arousing Stimuli There were five film trials, each comprising  a  different short film  of approximately  60 s in  duration.  The  first clip  was  used  to  acclimatesubjects to the procedure. The next two were intended  to  evoke positiveemotions,  and the  last  two  were designed  to  evoke negative emotions.Prior research with these films (Ekman  et  al., 1980; Ekman   Friesen,1974)  had  found that subjects reported strong feelings  of  amusementand happiness and showed smiling expressions during the positive films.Feelings of fear, sadness, disgust, and pain and  a  variety of negative emo-tional expressions occurred  in  response  to  the negative films.All  the  films were silent  and in  color. Silent films were desirable  for our purposes because different auditory patterns might conceivablyelicit different patterns  of  hemispheric activation solely as  a  function  of the acoustic variation among the clips (e.g., Carmon & Nachson, 1973).One of the positive films showed  a  puppy playing with flowers. The sec-ond  was a  clip  of  monkeys playing  and a  gorilla taking  a  bath  in the zoo.  The  order  in  which  the  two positive film clips were presented  was counterbalanced across subjects.The  two  negative film clips always followed  the  positive clips.  The rationale  for  this  was  based  on  both previous work  by  Ekman  et al. (1980)  and  our own pilot work, which indicated that the negative affectelicited  by the  negative films tended  to  persist longer than  the  positiveaffect elicited  by  the positive films. If we had counterbalanced the orderof positive  and  negative films, the persisting negative mood would haveinterfered with  the  intended effect  of  the positive films, decreasing  the number  of  positive emotional expressions that would have occurred  in response  to  these clips. The negative film clips were taken from trainingmovies used  in the  teaching  of  nurses. One clip depicted  a  leg amputa-tion  and the  other  was the  scene  of a  third-degree burn victim. Bothwere quite gruesome. The order  in  which the amputation and burn clipswere presented was counterbalanced across subjects.Films were presented with  a  Lafayette Model 925 Analyst film projec-tor, which generated  a  frame count pulse for automated control. Digitallogic counted these pulses  for  precise timing  of  film onset and offset. Subjective Ratings  of  Emotion After each of the baseline  and  film trials, subjects rated the emotionsthey  had  experienced during the preceding trial  on a  series  of  unipolarscales. Separate scales were included  for  interest, happiness, amuse-ment, contentment, excitement, fear, sadness, anger, disgust, pain,  and arousal. The instructions informed the subject that zero represented  no emotion  and 8 the  most intense feeling  of  that emotion. These ratingscales were projected  one at a  time  on the  rear projection screen.  The subjects entered their rating by pressing  a  number on their key pad.For  the  subjects  who  were retained  in the  analyses  (see  above),  the intensity  of  amusement  (the  dominant emotion reported during  the positive clips) during  the  positive clips  and  disgust during the negativeclips was comparable.  In  addition, there were  no  differences  in the in- tensity  of  any  of  the rated emotions between  the  subjects who were  re- tained  for  the EEG analysis  and  those who were excluded. Table  1  pre- sents the mean ratings of the positive  and  negative emotions  in  responseto the positive and negative film clips for the subjects who were retainedand those who were excluded. Video Recordings During each of the film clips, subjects were videotaped unobtrusivelythrough  a  wire mesh screen that served as the border of the rear projec-tion screen. The camera (G.E. Site Guard  II)  was mounted in the projec-tion room, which was adjacent  to  the subject room, along with the filmprojector.  The  camera was absolutely invisible  to the  subject,  and not one subject suspected that  she  was being videotaped. After  the  experi-
Related Documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!