Basic Emotions: Can Conflicting Criteria Converge? Terence J. Turner
Andrew Ortony
University of Kentucky
Northwestern Uni\,ersity
fc . 10. er! in dt
The authors discuss some of the key points raised by Ekman (1992), Izard (1992), and Panksepp (1992)in their critiquesofortony and Turner's(l990)suggestion that there are and probably can be no objective and generally acceptable criteria for what is to count asa basic emotion. A number of studies are discussed that are relevant to the authors' contention that a more promising approach to understanding the huge diversity among emotions is to think in terms of emotions being assemblages of basic conlponents rather than combinations of other basic etnoriot~s.The authors stress that their position does not deny that emotions are based on 'hardwired" biological systems. On the other hand, the existence ofsuch systems does not mean that some emotions (such as those that appear on lists of basic emotions)have a special status. Finally, the authors note that Ekman, Izard, and Panksepp, in adopting different starting points for their research, arrive at rather different conclusions as to what basic emotions are and which emotions are basic. It is concluded that converging resolutions of these questions are improbable.
-
We start by emphasizing that our article (Ortony & Turner, 1990) had four main goals. First, we wanted to examine the evidence in favor of the claim that there exists a small set of "basic" emotions. Second, we wanted to ask what the criteria for membership'in sucp a privileged set should be. Third, we wanted to question the theoretical utility of the notion of basic emotions. Fourth, we wanted to sketch an alternative way of thinking about epotions in which the basic e!ements would be not basic emotions, but components of emotions. Rather than trying to deal one by one with the critiques of Ekman (19921, h a r d (1992), and P a n k s e ~(19921, ~ m'e shall focus on Some of the central issues of Content that they raise. ~ h u swe , tackle first the notion, suggested by both Ekman and Izard, that the existence of universal and unique facial expressions for some emotions supports the idea that these emotions are biologically basic. We then turn to Ekman's suggestion that the emotionsasbasic is supported by ~hysiological differentiation among the emotions. Next nre discuss Panksepp's position that emotions are mediated by integrative brain systems. Finally. we address an issue raised by Izard, namely that basic emotions have unique and unchanging feeling states associated with them.
P3
.'
*
research-of Ekman and lzard focuses primarily on'prototypicai expressions of a small set of allegedly basic emotions (especial!\. in their cross-cultural work that established the hniversalit) of certain emotions).For us, the fact th&some integrated facial expressions are prototypical does not give them a special slatus in terms of reflecting mental states, particularly since such expressions nowhere near exhaust the full range of displays that are associated with mental states. In many cases, only a feu. components (and in some cases only one componect) of facial displays are needed either to signal the corresponding feeling to an observer (Ekman, 1992, p. 551) or to evoke the feeling via facial feedback (Strack, Martin, &Stepper, 1988). We therefore think that adopting a componential approach may be a useful research strategy for studying emotions (even for those n.hosc interests lie in the study d f facia] expressions) becauje it can provide a a?d subtle description of mental well as
-
Facial Expressions of Emotions We d o not (and did not) dispute the fact that there are universal facial expressions associated with certain emotions. onthis issue, we differ from Ekman and lzard only in terms of how we interpret this evidence and in terms of which aspects of facia] displays we consider to be of greatest scientific interest. The Preparation oft his article was supported in part by Nationalscience Foundation Grants BNS 83 18077 and BNS 872 1853. Correspondence concerning this article should be addressed to Terence j.Turner, Department o ~ p s y c ~ o ~ oKnrtIe g y , Hall, Universityof Kentucky. Lexington, Kentucky 40505-0044. Electronic mail may be sent to
[email protected].
(
.
T
an interpretati1.e framework for fabial expressions that are no: prototypical or that are mixes ofcomponcnts found in a n,..i. of facial displays, Primarilyas awxample of what an alternative, compor.catin our aniclc (Orial approach might look like, m,e tony & Turner, 1990) a way in which facial displays of anEer might be decomposed into hardwired components, and we cited a feu studies that have examined such components (Cacioppo. Pettg & Morris, 1985; smith, 1989). As the paucityof our references indicate. there is relatively little direct evidence about how such components might be i n t e ~ e t e dbecause research has tended to focus on integrated facial displays. However, there is some indirect evidence. We present the evidena in the form of answers t o four questions that relate to central assumptions of the basic emotions view of facial expressions. ~ i ~that some ~ ~emotionsareassociated , with p r o t o ~ ~ i cal expressions, should research concentrate on the relationhtween these emotions and such expressions, or is it also wad examining the components of emotions, perhaps espenal' those that are shared by emotions with different prototypical facial expressions? Consider two possible components ofe-'eiP '
;.$
566
.,
de en tiv: ant fou an( Da pla suc fie( Iza SYS fac; ily but ten chi. wa! acti alsc Pro
.
.
...
nen shatha: ior~ cur assc tha: exP em( leve emc It emc expi and the ; ing (199 Izar. asso the 1 and joy ; thal tern: tem Uri. Posi
One component would be a tendency to approach an ,iic~t. the other ~ ~ o ube l dthe contrasting tendency to avoid an .. .sirct. It seems plausible to suggest that protot>,pical cases of ...,.. ...lr pobably involve an alroidance tendencx whereas those of 3.,\c~rhappiness probably involve an approach tendency. Some such as anger, might have the tendency to approach ,I some instances (as in an angry attack), and in others, a ten;.-,liy to avoid (as in angry flight). It has been proposed that ... ,,,tterns of brain activity discriminate between these two ten:.-,ties. Thus, Davidson (1984) has suggested that when an ..r,otion in\,olves an approach component, a pattern of rela. : $ ~ ]greater y left anterior activation occurs, whereas ifan avoid:?;? component is involved, then right anterior activation is How do these components relate to "basic" emotions ,j to their associated facial expressions? A study by Fox and advidson (1988) suggests an answer. In this stud?.; infants dis: , ~ ~ emotional ~ed facial expressions to a number of stimuli, r j ~ has being separated from their mothers. The authors classin:d a number of the expressions as angry or sad according to lrar&s Maximally Discriminative Facial Movement Coding 5\stem (Izard, 1979). Fox and Davidson found that when these !acial expressions occurred, the pattern of regional brain activdeaended not on which expression it was (anger or sadness), haton whether the child cried. When the child cried, the pat::m typical of avoidance tendencies occurred, and when the :!,;Id did not cry, the pattern typical of an approach tendency ~.350bserved.Inter6stingly. the same pattern of regional brain i:~ivityfound in anzer and sadness in the absence ofcrying was :bo associated with the child's expression of joy at the a p ;:oach of the mother. These data suggest to us that there are at least two compo;:nts of emotions (approach and avoidance tendencies) that are .-arsd by such diverse emotions as anger, sadness, and jo); and :'at these components are distinct, identifiable, and behav?:ally relevant. Most important. the same component can oc: i ~ rin the presence of a variety of facial expressions supposedly 5sociated with very different "basic" emotions. We conclude :;a1 by focusing on "basic" emotions and their associated facial '!uressions. it is quite easy to miss important commonalities in :,:otions, and that there is much to be gained by moving the .,:I of analysis down from basic emotions to components of. :::otions. I:might be argued in response to this that the tarounderlying :@otionalstates that Fox and Davidson (1988) postulated to ::?lain their data are simply partsofthe same emotion system, 3 that the facial expression of sadness, for example, indicates -:activation of the sjvstei?lrather than of some particular feelIf state. This approach appears consonant with Panksepp's :j2) position on emotion systems, although not with that of zrd (1992)who claims that there is a single, invariant feeling Qciated with each "basic" emotion. However, the fact that ::Pattern ofresults found for sadness was also found for anger . . J o y is not encouraging for this view because, presumably, ': and anger involve the activation of very different systems "Sadness. Therefore, the notion of integrated emotion sys"5 cannot explain the data unless one argues that such syss: ' g1i.e rise to dissociable components that can occur in a 'lety of different emotions. This, of course, is essentially the 'ni50n we proposed in our article.
.,ns. ~
L.
tions (especial$ :universalil! c'
1a
spec~sl5 : ~ : . ly since suri .. ofdispla)~rC. ases, onl! a ' . ponent) off.. . ~ n d i n gfeelini t-, .e the feeling \ .. 38). 1% the:.:''. ,ma!; be z LI..:'. for those \:.! s) becaase 1: r 11 states as ur:! L . ;ions that aarc : , found in z r.
-'
?
itive, camp::' :. n our artiil. I;!' displays of :.: ponents? a"? " component' { ( z . As the pauci! pr le direa evid:::' lreted becar' ': ial displa)'S. f!:.' sent the evidf";' it relate to &'a' express!,' z ' ed with pr~h?::; the relatior.t. o r is it also l' .. ,erhaps erpr:la trent protO1>P:;" mponenb 01 Ct;: ' -'-
' -
