Psychological Review 1986, Vol. 93, No. 2, 119-135
Copyright 1986 by the American Psychological Association, Inc. 0033-295X/86/$00.75
A Triangular Theory of Love Robert J. Steinberg Yale University This article presents a triangular theory of love. According to the theory, love has three components: (a) intimacy, which encompasses the feelings of closeness, connectedness, and bondedness one experiences in loving relationships; (b) passion, which encompasses the drives that lead to romance, physical attraction, and sexual consummation; and (c) decision/commitment, which encompasses, in the short term, the decision that one loves another, and in the long term, the commitment to maintain that love. The amount of love one experiences depends on the absolute strength of these three components, and the kind of love one experiences depends on their strengths relative to each other. The three components interact with each other and with the actions that they produce and that produce them so as to form a number of different kinds of loving experiences. The triangular theory of love subsumes certain other theories and can account for a number of empirical findings in the research literature, as well as for a number of experiences with which many are familiar firsthand. It is proposed that the triangular theory provides a rather comprehensive basis for understanding many aspects of the love that underlies close relationships.
What does it mean "to love" someone? Does it always mean the same thing, and if not, in what ways do loves differ from each other? Why do certain loves seem to last, whereas others disappear almost as quickly as they are formed? This article seeks to answer these and other questions through a triangular theory of love. This tripartite theory deals both with the nature of love and with loves in various kinds of relationships. The presentation of the theory will be divided into three main parts. In the first part, the main tenets of the theory will be explained and discussed, and the theory will be compared with other theories of love. In the second part, the implications of the theory for close relationships and satisfaction in them will be described. In the third part, the theory will be shown to account for many of the empirical phenomena that have been observed with regard to love.
The intimacy component refers to feelings of closeness, connectedness, and bondedness in loving relationships. It thus includes within its purview those feelings that give rise, essentially, to the experience of warmth in a loving relationship. The passion component refers to the drives that lead to romance, physical attraction, sexual consummation, and related phenomena in loving relationships. The passion component thus includes within its purview those sources of motivational and other forms of arousal that lead to the experience of passion in a loving relationship. The decision/commitment component refers to, in the short term, the decision that one loves someone else, and in the long term, the commitment to maintain that love. The decision/commitment component thus includes within its purview the cognitive elements that are involved in decision making about the existence of and potential long-term commitment to a loving relationship. In general, the intimacy component might be viewed as largely, but not exclusively, deriving from emotional investment in the relationship; the passion component as deriving largely, although not exclusively, from motivational involvement in the relationship; and the decision/commitment component as deriving largely, although not exclusively, from cognitive decision in and commitment to the relationship. From one point of view, the intimacy component might be viewed as a "warm" one, the passion component as a "hot" one, and the decision/commitment component as a "cold" one. The experience of love can be partitioned in a number of ways, and so it is important to note at the outset that the present partitioning into intimacy, passion, and decision/commitment is not the only one possible, nor is it even valid for all possible purposes. Nevertheless, the argument will be made that the pro-
The Triangle of Love Three Components1 The triangular theory of love holds that love can be understood in terms of three components that together can be viewed as forming the vertices of a triangle. These three components are intimacy (the top vertex of the triangle), passion (the left-hand vertex of the triangle), and decision/commitment (the right-hand vertex of the triangle). (The assignment of components to vertices is arbitrary.) Each of these three terms can be used in many different ways, so it is important at the outset to clarify their meanings in the context of the present theory. I am grateful to Michael Barnes, Susan Grajek, and Sandra Wright for their collaborations in my empirical research on love, and to Ellen Berscheid, Keith Davis, Elaine Hatfield, Martin Hoffman, and George Levinger for their excellent comments on an earlier version of this article. Correspondence concerning this article should be addressed to Robert J. Sternberg, Department of Psychology, Yale University, Box 11A Yale Station, New Haven, Connecticut 06520.
1 My use of the term components in this article differs from my use of the term in my theorizing about intelligence (e.g., Sternberg, 1985), where the term is used to refer to a mental process.
ROBERT J. STERNBERG
posed partitioning is particularly useful for understanding the elements of love, and how they function in close relationships. Although love, like other psychological phenomena, can be partitioned into various kinds of components, it is important not to lose sight of the whole in the analysis of its parts. Love is a complex whole that appears to derive in part from genetically transmitted instincts and drives but probably in larger part from socially learned role modeling that, through observation, comes to be denned as love. To a large extent, then, love is prototypically organized (Rosch, 1978), such that certain feelings, drives, thoughts, and behaviors appear as more highly characteristic of love as it is socially denned, whereas others appear as less characteristic.2 Indeed, one way to study love would be through the examination of people's conceptions or implicit theories of love (Barnes & Sternberg, 1986, are currently involved in such an investigation). Such an investigation capitalizes on principles of descriptive psychology in order to provide a framework for loverelated phenomena (Davis & Roberts, 1985; Ossorio, 1985). A theory of love, therefore, can help one understand the range and composition of the phenomenon of love but should not result in the whole's being lost in its parts. The similarities and differences among the three components of love may be better understood by examining their respective properties, some of which are summarized in Table 1.
Properties of the Components of Love The three components of love differ with respect to a number of their properties. For example, the emotional and other involvement of the intimacy component and the cognitive commitment of the decision/commitment component seem to be relatively stable in close relationships, whereas the motivational and other arousal of the passion component tends to be relatively unstable and to come and go on a somewhat unpredictable basis. One has some degree of conscious control over the feelings of the intimacy component that one experiences (if one is aware of them), a high degree of control over the commitment of the decision/commitment component that one invests in the relationship (again, assuming awareness), but very little control over the amount of motivational and other arousal of the passion component one experiences as a result of being with or even looking at another person. One is usually quite aware and conscious of the passion component, but one's awareness of the intimacy and decision/commitment components can be highly variable. Sometimes one experiences warm feelings of intimacy without being aware of them or without being able to label them. Similarly, one is often not certain of how committed one is to a relationship until people or events intervene to challenge that commitment. The importance of each of the three components of love differs, on the average, as a function of whether a loving relationship is short-term or long-term. In short-term involvements, and especially romantic ones, the passion component tends to play a large part. The intimacy component may play only a moderate part, and the decision/commitment component may play hardly any part at all. In contrast, the intimacy component and the decision/ commitment component typically play relatively large parts in a long-term close relationship. Indeed, it is difficult to sustain
Table 1 Properties of Triangle Vertices Component Property Stability Conscious controllability Experiential salience Typical importance in short-term relationships Typical importance in long-term relationships Commonality across loving relationships Psychophysiological involvement Susceptibility to conscious awareness
Moderately high Moderate Variable Moderate
Low Low High High
Moderately high High Variable Low
such a relationship without at least some degree of involvement and commitment. In contrast, the passion component typically plays only a moderate part, and its role may decline somewhat over time. The three components of love also differ in their commonality across loving relationships. The intimacy component appears to be at the core of many loving relationships (Sternberg & Grajek, 1984), whether that relationship is toward a parent, a sibling, a lover, or a close friend. The passion component tends to be limited to just certain kinds of loving relationships, especially romantic ones, whereas the decision/commitment component can be highly variable across the different kinds of loving relationships. For example, commitment tends to be very high in one's love for one's children, but relatively low in one's love for those friends that come and go throughout the span of one's lifetime. The three components also differ in the amount of psychophysiological involvement they offer. The passion component is highly dependent on psychophysiological involvement, whereas the decision/commitment component appears to involve only a modest amount of psychophysiological response. The intimacy component involves an intermediate amount of psychophysiological involvement. In sum, the three components of love have somewhat different properties, which tend to highlight some of the ways in which they function in the experiences of love as they occur in various kinds of close relationships.
Composition of the Triangle The intimacy component. In the context of the triangular theory, the intimacy component refers to those feelings in a relationship that promote closeness, bondedness, and connectedness. Our research indicates that it includes, among other things, feelings of (a) desire to promote the welfare of the loved one, (b) !
I am grateful to Keith Davis for making this observation.
TRIANGULAR THEORY OF LOVE
experienced happiness with the loved one, (c) high regard for the loved one, (d) being able to count on the loved one in times of need, (e) mutual understanding with the loved one, (f) sharing of one's self and one's possessions with the loved one, (g) receipt of emotional support from the loved one, (h) giving of emotional support to the loved one, (i) intimate communication with the loved one, and (j) valuing the loved one in one's life (Sternberg & Grajek, 1984). These feelings form only a subset of the possible ones that can be experienced in the intimacy component of love, and moreover, it is not necessary to experience all of these feelings in order to experience love. To the contrary, our research indicates that one experiences the intimacy component of love when one samples a sufficient number of these feelings, with the number that is sufficient probably differing from one person to another. The feelings are usually not experienced independently; to the contrary, they may be experienced as one overall feeling. Nevertheless, they appear to be at least partially decomposable, as in the listing here. Sternberg and Grajek (1984) actually tested three alternative theories of the nature of love, focusing upon its intimacy component. They referred to the three theories as Spearmanian, Thomsonian, and Thurstonian. The nature of the three theories is illustrated in Figure 1. All three theories are based on structural models of intelligence. The Spearmanian theory is based on Spearman's (1927) theory of general intelligence (g). In terms of a structural model of love, one might conceptualize love partly in terms of a single g, which would be an undifferentiated "glob" of highly positive feelings that is essentially nondecomposable. To experience love would be to experience this glob of highly positive feelings. The Thomsonian model is based on Thomson's (1939) theory of the "bonds" of intelligence. In terms of a structural model of love, one might conceptualize love partly in terms of feelings that, when sampled together, yield the composite experience that we label love. On this view, though, the composite is not an undifferentiated unity; rather, it can be decomposed into a large number of underlying bonds that tend to co-occur in certain close relationships and that in combination result in the global experience that we view as love. The Thurstonian theory is based on Thurstone's (1938) theory of primary factors. In terms of a structural model of love, one would emerge with a theory viewing love partly in terms of a small, consistent set of feelings that have approximately equal importance and salience in the overall experience we describe as love. Love is not one main thing, whether decomposable (Thomsonian model) or not (Spearmanian model). Rather, it is a set of primary structures that are best understood separately rather than as an integrated whole. All contribute simultaneously to the experience of love. According to this notion, global experiences such as love can be decomposed into multiple overlapping (correlated) factors, and one could essentially combine factor scores to obtain an overall index of the strength of the love. Sternberg and Grajek (1984) used factor- and cluster-analytic methods to distinguish among these three theories. These methods were applied to the Rubin Loving and Liking Scales as well as to the Scale of Interpersonal Involvement used by Levinger, Rands, and Talaber (1977). The data were analyzed not only for
' Spearmanian " Model
" Thomsonian" Model
Oj~ affects, cognitions, motivations
affects, cognitions, motivations
Figure 1. Three alternative models of love.
the measures of loving and liking for one's lover, but also for measures of loving and liking for one's mother, father, sibling closest in age, and best friend of the same sex. Subjects in the study were 35 men and 50 women in southern Connecticut, ranging in age from 18 to 70 years, with a mean of 32 years. Factor analysis of the data of these subjects for each of the close relationships supported a Thomsonian model: A general factor emerged even after varimax rotation of the principal-axis solution (which tends to obscure rather than to highlight a general factor), but the general factor proved to be decomposable through hierarchical cluster analysis. In other words, the factor analysis supported either the Spearmanian model or the Thomsonian model, both of which are consistent with a general factor, but not the Thurstonian model, which is not consistent with a general factor (at least at the first order of analysis). The decomposability of the general factor supported the Thomsonian model but not the Spearmanian one, in that Spearman's model does not allow for the decomposability of the general factor. In the Sternberg-Grajek (1984) study, the Thomsonian model was viewed as applying to the three components of love considered jointly. However, a subsequent examination of the contents of the Rubin and Levinger et al. scales revealed that they focus primarily on the intimacy component of close relationships rather than on passion or decision/commitment. Hence, the analysis of these scales is seen as applying most directly to the intimacy component. An interesting and, to some extent, surprising finding of the Sternberg-Grajek (1984) study was that the structure of intimacy in love does not appear to differ consequentially from one loving
ROBERT J. STERNBERG
relationship to another. In other words, the general factor and ensuing clusters that were obtained for each relationship were about the same. This finding suggests that the intimacy component of love forms a common core in loving relationships. In other words, whereas the passion and decision/commitment components appear to be unique to loving relationships with certain classes of individuals, the intimacy component does not appear to be limited to just certain loving relationships. Consider, for example, loves for a mother, a father, a sibling, a best friend of the same sex, and a lover. According to the present point of view, the intimacy component forms a common core in each of these loving relationships. However, the passion and decision/commitment components are experienced more selectively. For example, the passion component probably plays a major part in love for a lover, but only a minor part, if any at all, in love for a parent, especially a same-sex parent. Similarly, the decision/commitment component is likely to play an important role in certain loving relationships, especially those with members of one's nuclear family (e.g., the mother, father, and siblings, if any). However, commitment over the long term need not play an important role, or any role at all, in love for a lover. Indeed, many romantic loves are short term and are never intended to be anything else. (Note that the term commitment is used here and elsewhere in this article to refer to long-term investment in a loving relationship, not to refer to the degree of responsibility one feels for another in a loving relationship.) Although the structure of the intimacy component of love may be roughly the same from one loving relationship to another, the amounts of love one feels toward various individuals may differ considerably. For example, in the Sternberg-Grajek (1984) study, we found that men tend both to love and to like their lover more than their mother, father, sibling closest in age, or best friend. Women, on the other hand, were found to love their lover and their best friend of the same sex about equally, but to like their best friend of the same sex somewhat more than they like their lover. For the women, as for the men, loving and liking of the lover and best friend exceeded that of the mother, father, and sibling closest in age. For both men and women, the sibling closest in age was loved and liked least of all from among this group of individuals. Our pattern of results is generally comparable to that of Swensen (1972), who used a different set of measures in order to obtain his results. Thus, both of these sets of results suggest that there are consistent differences in typical amounts of love across different close relationships. Sternberg and Grajek (1984) also found that the predictability of the amount of love one feels for one individual from the amount of love one feels for other individuals differs across relationships. In particular, they found that the amount of love one experiences for one member of one's nuclear family (mother, father, sibling closest in age) tends to be predictable from the amount of love one feels for another member of that nuclear family. However, amounts of love experienced toward members of the nuclear family do not predict amounts of love one experiences for individuals outside the nuclear family. In other words, whereas the amount of love one experiences for one's mother, father, and sibling closest in age are mutually predictive, these amounts of love are not predictive of the amount of love one feels for one's lover or one's best friend of the same sex. Nor is the amount of love one experiences for one's lover predictable
from the amount of love one experiences for one's best friend of the same sex. In other words, amounts of love tend to be predictable within but not outside of the nuclear family. As noted above, the Sternberg-Grajek (1984) study focused on the intimacy component of love. However, there is more to love than just the intimacy component. Consider in turn the passion and decision/commitment components. The passion component. The passion component of love comprises those motivational and other sources of arousal that lead to the experience of passion. It includes what Hatfield and Walster (1981) refer to as "a state of intense longing for union with the other" (p. 9). In a loving relationship, sexual needs may well predominate in this experience. However, other needs, such as those for self-esteem, succorance, nurturance, affiliation, dominance, submission, and self-actualization, may also contribute to the experiencing of passion. The strengths of these various needs will almost certainly vary across persons, situations, and kinds of loving relationships. For example, sexual fulfillment is likely to be a strong need in romantic relationships but not in filial ones. The manifestations of these needs are through psychological arousal and physiological arousal, although these two kinds of arousal are not easily separable. Indeed, psychological arousal will almost inevitably interact with physiological arousal, with arousal of one kind leading to arousal of the other kind. The passion component of love will almost certainly be highly and reciprocally interactive with intimacy. One will feel, for example, intimacy in a relationship in large part as a function of the extent to which the relationship meets one's needs for passion. Conversely, passion may be aroused by intimacy. In some close relationships with members of the opposite sex, for example, the passion component develops almost immediately, and it is only after a while that the intimacy component develops. The passion component is what may draw the individual to the relationship in the first place, but the intimacy component helps sustain closeness in the relationship. In other close relationships, however, the passion component, especially as it applies to physical attraction, develops only after the intimacy one. Two close friends of the opposite sex may find themselves developing a physical attraction for each other that did not develop immediately, and indeed did not develop until they achieved a certain level of intimacy with each other. The intimacy and passion components need not always covary positively. In certain kinds of relationships, for example, those with prostitutes, individuals may seek out another who maximizes fulfillment of needs for passion while purposefully minimizing intimacy. Negative covariation between the intimacy and passion components can be a function of person as well as of situation: Some people find that the attainment of emotional closeness and intimacy actually interferes with their attainment of sexual fulfillment. The point to be made, quite simply, is that although the form of interaction between the intimacy and passion components will vary across persons and situations, the two components of love will almost certainly interact in close relationships, in one way or another. The decision/commitment component. The decision/commitment component of love consists of two aspects, a short-term one and a long-term one. The short-term one is the decision that one loves a certain other. The long-term aspect is the commitment to maintain that love. These two aspects of the decision/com-
TRIANGULAR THEORY OF LOVE
mitment component of love do not necessarily go together. The decision to love does not necessarily imply a commitment to love. Oddly enough, commitment does not necessarily imply decision. Many people are committed to the love of another without necessarily even admitting that they love or are in love with the other. Most often, however, decision will precede commitment both temporally and logically. Indeed, the institution of marriage represents a legalization of the commitment to a decision to love another throughout one's life. It is important not to neglect the decision/commmitment component of love just because it does not have the "heat" or "charge" of the intimacy and passion components of love. Loving relationships almost inevitably have their ups and downs, and there may be times in such relationships when the decision/commitment component is all or almost all that keeps the relationship going. This component can be essential for getting through hard times and for returning to better ones. In ignoring it or separating it from love, one may be missing exactly that component of loving relationships that enables one to get through the hard times as well as the easy ones. The decision/commitment component of love interacts with both the intimacy and the passion components. For most people, it results from emotional and other involvement of the intimacy component or the motivational and other arousal of the passion component. However, intimate involvement or passionate arousal can follow from commitment, as would be the case in certain arranged marriages or in close relationships in which one does not have a choice of partners. For example, one does not get to choose one's mother, father, siblings, aunts, uncles, cousins, or the like. In at least some of these close relationships, one is likely to find that whatever intimacy or passion one experiences results from one's cognitive commitment to the relationship, rather than the other way around. Thus, love can start off as a decision, and whatever else follows may follow from that decision. The decision is not always one that promotes involvement or arousal. For example, a married individual may meet another with whom he or she falls in love. Whereas it can be difficult to control the intimacy component of love and exceedingly difficult to control passion, the decision/commitment component is one over which one has considerable control, and this control may prevent the further development of the relationship into a fullfledged romance. Of course, the decision can also go the other way. The point to be made, simply, is that the decisional aspect can control the other aspects of the relationship. It is important to distinguish the decisional aspect from the commitment aspect, however. In the example of the married individual who meets another with whom he or she falls in love, the decision to pursue that relationship does not necessarily imply a commitment to it. Husbands and wives who discover that their spouses are having affairs often leap immediately to conclusions on the basis of this knowledge about the decision of the spouse to have an affair. The more important information, however, might be the commitment of the spouse to that affair and to the relationship that generated it. In sum, the three components are all important parts of loving relationships, although their importance differs from one relationship to another. Moreover, the importance of these components of love may differ over time within a relationship as well as across relationships at a given time.
Table 2 Taxonomy of Kinds of Love Component Kind of love
Nonlove Liking Infatuated love Empty love Romantic love Companionate love Fatuous love Consummate love Note. + = component present; — = component absent. These kinds of love represent limiting cases based on the triangular theory. Most loving relationships will fit between categories, because the various components of love are expressed along continua, not discretely.
Kinds of Love The components of love and their interrelationships can better be understood by considering the kinds of love to which they may give rise in different combinations. These various kinds of love are summarized in Table 2. There are eight possible subsets of the various components of love. Each of these subsets differs in the kind of loving experience to which it gives rise. Consider the limiting cases. 1. Nonlove. Nonlove refers simply to the absence of all three components of love. Nonlove characterizes the large majority of our personal relationships, which are simply casual interactions that do not partake of love at all. 2. Liking. Liking results when one experiences only the intimacy component of love in the absence of passion and decision/ commitment. The term liking is used here in a nontrivial sense, not merely to describe the feelings one has toward casual acquaintances and passers-by in one's life. Rather, it refers to the set of feelings one experiences in relationships that can truly be characterized as friendships. One feels closeness, bondedness, and warmth toward the other, without feelings of intense passion or long-term commitment. Stated in another way, one feels emotionally close to the friend, but the friend does not "turn one on," nor does the friend arouse the thought that "one loves the friend" or that one plans to love the friend for the rest of one's life. It is possible for friendships to have elements of passionate arousal or long-term commitment, but in such cases, the friendship goes beyond mere liking and is best classified in one of the categories below. A test that can distinguish mere liking from love that goes beyond liking is the absence test. If a typical friend whom one likes goes away, even for an extended period of time, one may miss the friend, but one does not tend to dwell on the loss. One can pick up the friendship some years later, often in a different form, without even having thought much about the friendship during the intervening years. When a close relationship goes beyond liking, however, one's reaction to the absence test is quite different. One actively misses the other person and tends to dwell on or be preoccupied with that person's absence. The other is actively rather than passively missed, and the absence
ROBERT J. STERNBERG
has a substantial and fairly long-term effect both on one's life and on one's reactions to one's life. When the absence of the other arouses strong feelings of intimacy, passion, or commitment, it is best to classify the relationship as going beyond liking; thus, to classify it in one of the categories described below is appropriate. 3. Infatuated love. Infatuated love is "love at first sight." Infatuated love, or simply, infatuation, results from the experiencing of passionate arousal in the absence of the intimacy and decision/ commitment components of love. Infatuations are usually rather easy to spot, although they tend to be somewhat easier for others to spot than for the individual who is experiencing the infatuation. Infatuations can arise almost instantaneously and dissipate as quickly under the right circumstances. They tend to be characterized by a high degree of psychophysiological arousal, manifested in somatic symptoms such as increased heartbeat or even palpitations of the heart, increased hormonal secretions, erection of genitals (penis or clitoris), and so on. Infatuation is essentially the same as what Tennov (1979) calls "limerence," and like Tennov's limerence, it can be quite lasting in duration under certain circumstances. 4. Empty love. This kind of love emanates from the decision that one loves another and has commitment to that love in the absence of both the intimacy and passion components of love. It is the kind of love one sometimes finds in stagnant relationships that have been going on for years but that have lost both the mutual emotional involvement and physical attraction that once characterized them. Unless the commitment to the love is very strong, such love can be close to none at all, because commitment can be so susceptible to conscious modification. Although in our society we are most accustomed to empty love as it occurs as a final or near-final stage of a long-term relationship, in other societies, empty love may be the first stage of a long-term relationship. For example, in societies where marriages are arranged, the marital partners may start with the commitment to love each other, or to try to love each other, and not much more. Such relationships point out how empty love need not be the terminal state of a long-term relationship. Indeed, it can be the beginning rather than the end! 5. Romantic love. This kind of love derives from a combination of the intimacy and passion components of love. In essence, it is liking with an added element, namely, the arousal brought about by physical attraction and its concomitants. According to this view, then, romantic lovers are not only drawn physically to each other but are also bonded emotionally. This view of romantic love seems to be similar to that found in classic works of literature, such as Romeo and Juliet and Tristan and Isolde. This view of romantic love differs, however, from that of Hatfield and Walster (1981), who argue that romantic love does not differ from infatuation. 6. Companionate love. This kind of love evolves from a combination of the intimacy and decision/commitment components of love. It is essentially a long-term, committed friendship, the kind that frequently occurs in marriages in which the physical attraction (a major source of passion) has died down. This view is captured in the title of Duck's (1983) book, Friends for Life. This view of companionate love is also essentially the same as that of Berscheid and Walster (1978).
7. Fatuous love. Fatuous love results from the combination
of the passion and decision/commitment components in the absence of the intimacy component. It is the kind of love we sometimes associate with Hollywood, or with whirlwind courtships, in which a couple meets on Day X, gets engaged two weeks later, and marries the next month. It is fatuous in the sense that a commitment is made on the basis of passion without the stabilizing element of intimate involvement. Although the passion component can develop almost instantaneously, the intimacy component cannot, and hence relationships based on fatuous love are at risk for termination and, in the case of shot-gun marriages, for divorce. 8. Consummate love. Consummate, or complete, love results from the full combination of the three components. It is a kind of love toward which many of us strive, especially in romantic relationships. Attaining consummate love can be analogous in at least one respect to meeting one's target in a weight-reduction program: Reaching the goal is often easier than maintaining it. The attainment of consummate love is no guarantee that it will last. Indeed, its loss is sometimes analogous to the gain of weight after a weight-reduction program: One is often not aware of the loss of the goal until it is far gone. I do not believe that all manifestations of consummate love are necessarily difficult either to develop or maintain. For example, one's love for one's children often carries with it the deep emotional involvement of the intimacy component, the satisfaction of motivational needs (e.g., nurturance, self-esteem, selfactualization) of the passion component, and the firm commitment of the decision/commitment component. For many but not all parents, formation and maintenance of this love is nonproblematical. Perhaps the bonding between parents and children at birth renders this love relatively easier to maintain, or perhaps evolutionary forces are at work to ensure that parent-child bonding survives at least those formative years in which the child must depend very heavily on the parent's love and support. Whichever of these may be the case (and it may be more than one), consummate love can be easier or more difficult to form and maintain, depending on the relationship and the situation in which it is developed and maintained.
Relations of Triangular Theory to Other Theories of Love The framework for understanding love generated by the triangular theory seems to make intuitive sense in terms of people's everyday experience and also seems to capture some of the kinds of love that are perhaps missed by frameworks that are not theoretically generated. For example, the Berscheid-Walster (1978) distinction between romantic and companionate love is useful, but according to the present framework, it is incomplete and not quite correct in that it does not distinguish between infatuated and romantic love. Similarly, Maslow's (1962) distinction between D-love (Deficiency love) and B-love (Being love) seems incomplete in light of the framework presented above. D-love is closest to what is referred to here as infatuated love, whereas B-love is closest to consummate love. However, there seem to be many other kinds of love as well. As noted earlier, Tennov's (1979) concept of limerence deals only with what is referred to here as infatuated love. Concepts similar to Maslow's D-love and Ten-
TRIANGULAR THEORY OF LOVE
nov's limerence derive from other clinical psychologists such as Reik (1944), who viewed love as the search for salvation, and Freud (1922), who viewed love largely in terms of striving for an ego ideal. According to the present framework, though, love in at least some of its forms is much more than the search for salvation or an ego ideal. Similarly, it potentially comprises more than the decisional and commitment aspects emphasized by Peck (1978). The taxonomy above also differs in spirit from some recent theories that have emanated from or at least have been closely associated with questionnaire studies. For example, Rubin (1970, 1973) has used psychometric methods to derive what he has called a Love Scale, which he has distinguished from a Liking Scale. The Love Scale is based on a three-component theory of love: affiliation or dependent need, predisposition to help, and exclusiveness and absorption. Rubin's Love Scale measures some elements of all three vertices of the love triangle but probably measures most reliably and validly the vertex of the intimacy component. It is perhaps because of this concentration on the measurement of this vertex that scores on the Rubin Liking and Loving Scales are fairly highly correlated. In our own data, we have obtained a correlation of .72 between the two scales for liking and loving of a lover and higher correlations for liking and loving of a mother (.73), father (.81), and sibling (.80), but a slightly lower correlation for liking and loving of a best friend (.66; Sternberg & Grajek, 1984). Rubin (1970) obtained somewhat lower correlations between the two scales for lovers, but his lower correlations were based on a college-student sample of couples who answered a newspaper advertisement directed at "dating couples." This sample was probably somewhat restricted in range in a number of respects. Our own sample consisted of adults (not necessarily tested in couples) ranging in age from 18 to 70 years, with a mean age of 32, all of whom were presently or recently involved in love relationships. Thus, it is proposed that the Rubin Love Scale is differentiated from the Rubin Liking Scale to the extent that it measures the passion and decision/ commitment components of love as well as intimacy. Another recent theory, that of Davis (1985), is logically rather than factor analytically derived but has been tested using questionnaire data. Davis has proposed that love differs from liking by the addition of two clusters, a physical attraction cluster and a caring cluster. Whereas the triangular theory would view physical attraction as separating infatuated or romantic love from liking, it would not separate caring from the liking involved in a friendship. According to the triangular theory, caring is typically a part of the liking in a friendship, and indeed, Davis's own data may not clearly support his separation of the caring cluster from the liking involved in good friendships. A taxonomy that is related in spirit, although perhaps not in content, to that generated by the triangular theory is that of Lee (1977), who has proposed what he refers to as "a typology of styles of loving." His taxonomy includes (a) eras, the love style characterized by the search for a beloved whose physical presentation of self embodies an image already held in the mind of the lover; (b) Indus, which is Ovid's term for playful or gamelike love; (c) storge, a style based on slowly developing affection and companionship; (d) mania, a love style characterized by obsession, jealousy, and great emotional intensity; (e) agape, which is altruistic love in which the lover views it as his or her duty to
love without expectation of reciprocation; and (f) pragma, a practical style involving conscious consideration of the demographic characteristics of the loved one. Although Lee's theory is related to the triangular theory in spirit, its content is quite different. For example, eros would be regarded in the triangular theory as fairly close to infatuated love, whereas mania would be regarded as infatuated love gone berserk. Ludus would not be viewed as a kind of love but rather as a style of interrelating that people can use in various kinds of loving relationships. For example, infatuated lovers, romantic lovers, and companionate lovers, as well as lovers of the other kinds, are all capable of playing games with one another. Storge would be viewed as quite close to companionate love. Agape would be viewed as a concomitant to the love that characterizes the loving relationships of persons with an altruistic disposition in their personalities. Finally, pragma would not be viewed as a kind of love at all but rather as a pragmatic style of search for a lover, as its name implies. Indeed, an overly pragmatic style can get in the way of ever finding any kind of love at all. Those who exhibit pragma may be searching for physical, financial, or other forms of comfort rather than love. Lasswell and Lobsenz (1980) used Lee's theory as the basis for the construction of a Love Scale Questionnaire. Their questionnaire was designed to measure each of the six kinds of love in Lee's theory. We administered the Lasswell-Lobsenz questionnaire to the 85 subjects in our own experiment on the nature of love, but our factor-analytic results failed to uphold the typology proposed by Lee. However, the triangular theory has never been tested against Lee's theory, and so the issue of the relative empirical validities of the two theories remains an open question. Indeed, the triangular theory is at present being tested as a whole for the first time (Sternberg, 1986). Whereas the triangular theory seeks an integration of a number of relationship-based phenomena into love, other theories seek more of a separation. For example, some would view infatuation as wholly distinct from love (e.g., Peck, 1978). Others would view commitment as distinct from love (e.g., Kelley, 1983; Lund, 1985). Yet, both clinical and empirical data suggest the difficulties of making clean separations. Exhaustive reviews of the literature (e.g., Brehm, 1985; Duck, 1983; Hinde, 1979) show how intricately woven together are concepts and feelings of love and romance, or infatuation, in contemporary western civilization, and how difficult it is statistically even to separate romantic love from love in general. Moreover, the data of Lund indicate a high correlation between measures of love and commitment, even after items with very high correlations have been weeded out of a commitment scale. As Kelley (1983) notes, even though he "has drawn a distinction between love and commitment, . . . [he] has recognized the considerable overlap between the two" (p. 312). The data of those who have studied the various phenomena of close relationships strongly suggest the wisdom of retaining conceptual distinctions among these phenomena (as in the three components of the triangular theory), while at the same time recognizing their strong correlation in loving relationships. Although pure, limiting cases of separation among components of love can be conceptualized and identified, it is perhaps better to view these components as interactive aspects of love rather than as independent phenomena to be conceived of and studied in isolation from each other.
ROBERT J. STERNBERG
Respective Courses of the Components of Love Each of the three components of love has a different course, and the differing temporal courses of the components almost inevitably result in changes in the nature of a given loving relationship over time. Consider each component of love in turn. The intimacy component. The course of the intimacy component of love, as presented here, is based on Berscheid's (1983) theory of emotion in close relationships, which is itself based on Mandler's (1980) more general theory of emotion. Although the intimacy component is not synonymous with an emotional component, it is largely composed of emotional elements and seems to function in ways quite akin to those of emotions as conceptualized by Berscheid. Thus, the theory is viewed as relevant for consideration of the intimacy component, or at least the emotional aspect of it. According to Berscheid, emotion in close relationships is experienced only as the result of interruption of paired action sequences, or what might be referred to as scripts (Schank & Abelson, 1977). As two individuals get to know each other, they form increasingly large numbers of these paired action sequences, or scripts. Early during a relationship, there will be a high degree of uncertainty in the relationship, because one has not yet become able to predict the other's actions, as well as emotions, motivations, and cognitions. There will be frequent interruptions and disruptions in interpersonal relations as the two individuals get to know each other. As time goes on, the frequency of interruptions is likely to decrease because the individuals become more predictable to each other and dependent on each other for predictable interactions. According to Berscheid's theory, as the amount of interruption decreases, so will the amount of experienced emotion. Eventually, one might find oneself experiencing little or no emotion at all. It is perhaps this course of emotion in close relationships that led Livingston (1980) to refer to love as a process of uncertainty reduction. The decreasing experiencing of intimacy in a close relationship, especially a romantic relationship, has both a positive and a negative side. The positive side is that the decrease in experienced intimacy is the result of an increased amount of interpersonal bonding. In other words, it results from the relationship's becoming closer, not more distant. Thus, one might view the relationship as having a large amount of latent intimacy, even though that intimacy is not manifest. The negative side is that it often becomes difficult to distinguish the close relationship from no relationship at all because of the lack of observable intimacy. This situation is represented in Figure 2, which shows both latent and experienced levels of intimacy as a function of the temporal course of the relationship. The failed or failing relationship will differ from the successful relationship primarily in terms of the latent intimacy rather than in terms of the experienced or manifest intimacy. Fortunately, there are ways of distinguishing a live relationship from one that is dying or dead. The most obvious way is to generate some kind of interruption in order to observe the amount of intimacy that this interruption generates. For example, the lover's going away, even for a brief period of time, can help one ascertain the amount of feeling one still has left for the lover. Or, changing established routines, as on a joint vacation, can be useful in assessing the state of intimacy in a relationship. Sometimes it is only through extreme interventions, whether
o o E
Duration of Relationship Latent Level Manifest Level of Intimacy of Intimacy —-^ Successful Relationship ^^^ — Failed Relationship Figure 2. The course of intimacy as a function of duration of relationship.
intentional or unintentional, that one learns of the amount of intimacy one has or has had in a relationship. For example, when a partner dies, one is often surprised, as are others, by the amount of grief and distress that is experienced. Following Berscheid (1983), even couples that argued and never seemed to get along can have considerable amounts of intimacy invested in the relationship, whatever the nature of that intimacy may be or have been. The death of a spouse is one of the surest ways of finding out the amount of intimacy one had invested in the relationship. Similarly, individuals who divorce are often surprised by the amount of postdecisional regret, or at least emotion, they experience. Often they had no idea of the amount of intimacy they had in the relationship until they forceably ended the relationship. Indeed, the divorce may have stemmed in part from their unawareness of their own and the other's intimate investment. This view of the course of intimacy in close relationships renders it essential that interpersonally involved individuals create minor interruptions in order to discern their levels of intimate involvement before they create major interruptions, wherever possible. A perfectly good relationship may be destroyed for lack of knowledge about the nature of intimate involvement in close relationships. The passion component. The course of the passion component in close relationships is quite different from that of the intimacy component. The view presented here is based on Solomon's (1980) opponent-process theory of acquired motivation. The passion component probably does not draw exclusively from motivational arousal for its substance, but it appears to draw very heavily on such arousal. Moreover, its course bears a close resemblance to that predicted by Solomon's theory. Thus, this theory is viewed as providing a characterization of the temporal course of the passion component, or at least its motivational aspects. According to Solomon's theory, experienced motivation for a person or an object is a function of two underlying opponent processes. The first, positive process, is quick to develop but also
TRIANGULAR THEORY OF LOVE
quick to fade. The second, negative or opponent process, is slow to develop and also slow to fade. The result of the two processes working in conjunction is a motivational course somewhat like that depicted in Figure 3. According to the theory as used here, a surge in passion can be experienced almost instantaneously upon meeting another individual to whom one is attracted, whether physically or otherwise. This passionate arousal increases quickly but also peaks fairly rapidly. The peak of arousal corresponds to just that point before the opponent process of passion begins to recruit. Once this opponent process begins to recruit, the experienced level of passion decreases, or habituates, as the positive force remains constant and the negative force results in decreasing levels of arousal. Eventually, one reaches a more or less stable and habituated level of arousal toward the individual or object. At this point, both the positive and the negative forces are stable and in equilibrium. Should one lose the individual (or object), one does not merely go back to baseline, that is, the null level of passionate arousal one felt before ever having encountered the individual or object. Rather, one sinks below baseline, resulting in likely feelings of depression, remorse, and extreme discomfort. The transition to below-baseline status of positive arousal results from the loss of the positive passionate force (the person or object is gone) but the continuance of the negative force (the effects of the absence continue to be felt). It is only gradually that the effects of the negative force, which is slow to disappear, begin to moderate and one eventually returns to a state at or close to baseline. It is useful to think of the motivational model in terms of addictions to various kinds of substances. Indeed, it is the analogy of the passion component of love to the motivational aspect of addictions that has led Peele and Brodsky (1976) to refer to love as an addiction. Consider, for example, addictive substances such as drugs, cigarettes, or coffee (choose your poison!). Initially, one has no particular motivation toward or need for the addictive substance. When one starts use of the addictive substance, one feels a "high" as a result. One is then likely to increase use of the substance. However, as one increases use of the substance, one starts to habituate: A given amount of the substance no
Experienced Level Positive Drive 1 . _ , r Opponent Processes Negative Drive J
Figure 3. The course of passion as a function of duration of relationship.
ll Jf T3 O
•S.I 0) Q.-7J ,*! 0> LU o
Duration of Relationship Successful Relationship —— Flagging Relationship Failed Relationship Figure 4. The course of decision/commitment as a function of duration of relationship.
longer has the same effect or produces the same high that it once did. Eventually one reaches a habituated state where one needs to continue use of the substance merely to prevent entry into a state of withdrawal, with its resulting symptoms of depression, irritability, and craving for the substance. Should one cease use of the substance, there will be a difficult withdrawal period in which one will experience a variety of unpleasant psychological and somatic symptoms. After the withdrawal period has ended, one can, at best, return to baseline. The decision/commitment component. The course of the decision/commitment component of love over the duration of a close relationship depends in large part on the success of that relationship (and vice versa). Generally, this level starts at a zero baseline before one meets or gets to know the individual and then starts increasing. Usually, if the relationship is to become a long-term one, the increase in level of commitment in the decision/commitment component will be gradual at first and then speed up. If the relationship continues over the long term, the amount of commitment will generally level off, yielding an S-shaped curve. If the relationship begins to flag, the level of commitment will begin a period of descent, and if the relationship fails, in the sense of approaching an ending, the level of commitment may go back down to baseline. As always, the smoothness of the hypothetical curve does not take into account the rockiness of many relationships. Even the most successful relationships will have their ups and downs, with the commitment curve varying accordingly. Figure 4 shows an idealized curve of decision/commitment over the duration of a relationship, without the bumps that almost inevitably occur along the way. To conclude this section, it can be noted that the respective curves representing amounts of intimacy, passion, and decision/ commitment show somewhat different forms. The differences can be even greater than those shown here because of individual
ROBERT J. STERNBERG
differences in close relationships. Because of the different forms of the curve, relationships will change over time. It is necessary to have some way within the theory of conceptualizing the nature of these changes. A way of conceptualizing such changes is considered in the next section.
Beyond the Basic Triangle To this point, the discussion has proceeded as though all triangles of love are the same, and as though there is only one such triangle. However, both of these assumptions are oversimplifications. It is now necessary to extend the triangular theory to take into account the greater complexity of love in close relationships.
Geometry of the Love Triangle
The geometry of the love triangle depends upon two factors: amount of love and balance of love. Amount of love: Area of the triangle. Figure 5 shows three different triangles differing only in area. These differences in area represent differences in amounts of love experienced in three hypothetical relationships: the larger the triangle, the greater the amount of experienced love. It is actually possible to specify coordinates for the three components of love, with higher absolute values of coordinates representing greater amounts of each of the three hypothetical constructs. Balance of love: Shape of the triangle. Figure 6 shows four distinct triangles that are dissimilar in shape. The equilateral triangle at the top represents a balanced love in which all three components of love are roughly equally matched. The second, a scalene triangle pointing to the left side, represents a relationship in which the passion component of love is emphasized over the others. This relationship is likely to be one in which physical attraction plays a large part but in which the intimacy and decision/commitment components play smaller parts. The third, an isosceles triangle, represents a relationship in which the inIntimacy Less Love
UNBALANCED TRIANGLES Intimacy
Figure 6. Shape of triangle as a function of kind of love.
timacy component plays a large part and the passion and decision/ commitment components play smaller parts. This triangle represents a relationship in which the two lovers are very good friends and are close to each other but the physical aspects and commitment to the future are more marginal. The fourth, a scalene triangle pointing to the right side, represents a relationship in which the decision/commitment component predominates over intimacy and passion. This triangle represents the highly committed relationship in which intimacy and physical attraction have waned or in which those components were never there in the first place. By varying both the area and the shape of the triangle of love, it becomes possible to represent a wide variety of different kinds of relationships, and particularly to represent the course of a close relationship over time. It should be noted that the triangle is only a gross representation of the subtleties of love in a relationship. As mentioned earlier, the constituents of the intimacy component experienced in a loving relationship are not a single entity but rather a union of many different entities (as per the Thomsonian model). Similarly, many different sources of passion may enter into love in a close relationship, and a variety of cognitions in the decision/commitment component yield the decision to love someone and the decision to remain committed to that love. Hence, a detailed diagnosis of the state of a relationship would necessitate going beyond looking only at the area and shape of the triangle. Nevertheless, these elements of the geometry of the triangle are useful in conceptualizing different kinds of relationships that can evolve as loving relationships. Multiple Triangles of Love
More Love Figure 5. Area of triangle as an index of amount of love.
Love does not involve only a single triangle. Rather, it involves a great number of triangles, only some of which are of major theoretical and practical interest. The main triangles will be considered here. Real versus ideal triangles. One not only has the triangle representing his or her love for the other in a close relationship, but also a triangle representing an ideal other for that relationship. This ideal may be based in part on experience in previous re-
TRIANGULAR THEORY OF LOVE
lationships of the same kind, which form what Thibaut and Kelley (1959) refer to as a "comparison level," and in part on expectations of what the close relationship can be. These expectations may or may not be grounded in reality. Figure 7 represents four of the possible relations between real and ideal triangles. The first panel shows real and ideal triangles as coincident: In other words, the actual relationship corresponds essentially perfectly to the ideal for that relationship. The second panel shows underinvolvement: The person's triangle in the actual relationship is at lower levels of the three components than the person ideally would like. The third panel shows overinvolvement: Levels of the three components are greater than the individual would like. The fourth triangle shows misinvolvement: In this particular instance, levels of intimacy and passion are less than what ideally might be desired, but the level of decision/commitment is greater than what ideally might be desired. Note that whereas the second and third triangles involve mismatches primarily in area, the fourth triangle involves mismatch primarily in shape. Of course, it is possible to have mismatches in both area and shape or to have mismatches in neither, as in the first triangle. Our research suggests that the overlapping area between the real and ideal triangles is associated with satisfaction in close relationships, whereas the nonoverlapping area between the two triangles is associated with dissatisfaction (Steinberg & Barnes, 1985). The greater the degree of discrepancy between the two triangles, the less satisfied an individual will be in a loving re-
Intimacy Perfectly Matched Involvements
Decision / Commitment
Closely Matched Involvements
Moderately Mismatched Involvements
. \ V
Severely Mismatched Involvements
Figure 8. Relations between involvements of two individuals in a relationship.
Figure 7. Relations between real and ideal levels of involvement.
lationship. If the discrepancy from the ideal is too great, the relationship can be in serious jeopardy. Self versus other triangles. There are always at least two people involved in interpersonal love relationships, and each of them experiences a triangle of love. Hence, one can conceptualize the degree of match or mismatch between the triangles of the partners in the loving relationship. Figure 8 shows four of the possible relations between the triangles of two individuals. In the first panel, the triangles are almost perfectly matched. Such a relation between triangles is unlikely in practice, of course. The second panel shows closely matched involvements, the third shows moderately mismatched involvements, and the fourth shows severely mismatched involvements. Involvements can differ both in area and shape of the respective triangles. Again, our research suggests that overlapping area between the two triangles is associated with satisfaction in loving relationships, whereas nonoverlapping area is associated with dissatisfaction (Sternberg & Barnes, 1985). Self-perceived versus other-perceived triangles. Finally, it is possible to distinguish between self-perceived and other-perceived triangles. In a loving relationship, one has a triangle that represents one's love for the other. However, there is no guarantee that this triangle will be experienced by the other in the same way that it is experienced by the self. In other words, the partner in a loving relationship may not perceive one's levels of the three components of love in the same way that the levels are perceived
ROBERT J. STERNBERG
Figure 9. Relations between perceptions of two individuals in a relationship.
by the self. Hence, there can be discrepancies between an individual's triangle as experienced by the self and as experienced by the other. Figure 9 shows two possible levels of discrepancy, one with a minor discrepancy and one with a major discrepancy between self- and other-perceived triangles. Interactions among the love triangles. The three pairs of triangles generate eight (23) different triangles in a single close relationship. Of course, one can generate further such triangles, for example, how one perceives the other perceiving oneself. However, these further triangles become rather esoteric and not terribly interesting. Sternberg and Barnes (1985) sought to study the effects of these eight triangles on satisfaction in close relationships. They had 48 subjects—24 Yale undergraduate and graduate student couples—fill out the Rubin Love Scale, the Rubin Liking Scale, and the Levinger et al. Scale of Interpersonal Involvement in four different ways. In particular, they had subjects produce ratings for (a) how one feels about the other, (b) how one believes the other feels about oneself, (c) how one would wish to feel about an ideal other, and (d) how one would wish an ideal other to feel about oneself. Because both members of each loving couple were tested, this procedure produced eight different scores, four for each individual. Moreover, it was possible to look at a variety of difference scores as well as simple scores. The simple scores are simply the four scores generated by the four different kinds of ratings provided. The difference scores are generated by subtracting certain scores from other scores. For example, it is possible to compute the difference between how one feels about the other and how one would ideally like to feel about the other, or between how one feels about the other and how one perceives the other to feel about oneself. Both of these scores are within-person difference scores in that they are generated from the data of a single subject. It was also possible to compute between-subjects difference scores that were generated by subtracting across rather than within subjects' scores. For example, it is possible to compute the difference between the way one feels about the other and the way the other feels about
oneself, or between the way one would ideally like the other to feel about the self and the way the other actually feels about the self. Figure 10 depicts geometrically the various kinds of scores that were generated by the data. In addition to filling out the Rubin and Levinger et al. scales, subjects also filled out a questionnaire querying feelings about the quality of the relationship, assigning ratings of 1 through 9 to (a) satisfaction with the relationship, (b) success of the relationship, (c) closeness of the relationship, (d) exclusivity of the relationship, (e) degree to which they feel "in love" with the partner, (f) communication in the relationship, (g) predicted duration of the relationship, (h) extent to which needs are met in the relationship, (i) extent to which the subjects believe their partner's needs are met in the relationship, (j) extent to which the subjects believe they measure up to their partner's ideal, (k) extent to which the partner measures up to their own ideal, (1) their commitment to the relationship, and (m) the partner's perceived commitment to the relationship. These ratings were all highly intercorrelated, with the exception of the exclusivity rating, and so they were combined into a single score representing overall relationship satisfaction. Both absolute and signed difference scores were computed, although the absolute difference scores proved to be more revealing than the signed ones. A number of interesting findings emerged from the Sternberg-Barnes (1985) study, only some of which will be discussed here. First, consider the question of whether ideal others matter for satisfaction in a romantic relationship, or matter for satisfaction
(Male or Female) DIFFERENCE SCORES Within Subject
(Male or Female) Between Subjects
Female Male S-*-0P Figure 10. Geometric representations of simple and difference scores.
TRIANGULAR THEORY OF LOVE
as much as do real others. Five of six correlations of ideal others with satisfaction were statistically significant but of relatively modest magnitude (.28 to .41). These correlations did not even overlap the range of the six correlations concerning real others (.55 to .75). Thus, looked at in their own right, feelings about ideal others seemed to matter, but they mattered less then feelings about actual others. Second, consider the question of whether perceptions of how the other feels about oneself matter for satisfaction in romantic relationships, or matter as much as do one's own feelings toward the other. We found that one's perceptions of the other's feelings toward oneself clearly do matter for satisfaction, and at a level roughly comparable to that of one's feelings toward the other. Thus, in evaluating a relationship, one takes into account one's perceptions of the feelings of the other toward oneself at about the same level as one takes into account one's feelings toward the other. Third, consider the question of kinds of comparison levels. Whereas a single comparison level could be defined in terms of a difference between how one feels about the other and how one might feel about an ideal other (the Thibaut-Kelley, 1959, comparison level for an ideal other), other comparison levels can be defined as well by taking the various possible difference scores into account. In a sense, then, our set of difference scores represented a generalization of Thibaut and Kelley's (1959) concept of a comparison level. As it turns out, five of the six comparison levels defined within person by these difference scores related significantly and substantially to satisfaction within romantic relationships. The correlations ranged in magnitude from .66 to .73 for the Rubin Love Scale, from .34 to .59 for the Rubin Liking Scale, and from .70 to .80 for the Levinger et al. Interpersonal Involvement Scale. The standard (Thibaut-Kelley) comparison-level score is no more powerful than the five other meaningful ones, and indeed it may be less powerful than the strongest predictor of relationship satisfaction. Fourth, consider the question of just what is the strongest predictor of relationship satisfaction. We found it to be the difference between how the other is perceived to feel about the self and how the ideal other would feel about the self. In other words, one is satisfied when the way the other is perceived to feel corresponds to the way one ideally would want the other to feel, and one is dissatisfied as the discrepancy between the real and ideal increases. It is interesting to note, therefore, that the strongest predictor of relationship satisfaction was not one's feelings for the other but rather a function of the way the other is perceived to feel and the way the other ideally would be perceived to feel about the self. It is possible to speculate on why this difference score might be such a powerful predictor of relationship satisfaction. When the other is perceived to be overinvolved, this perception often results in one's drawing away from the other in order to establish the desired distance. However, this withdrawal often leads to an intervention on the other's part to bring one closer. This intervention in turn leads to a further drawing away by the self, and so on. In other words, the asymmetry between feelings can lead to progressively greater asymmetry, and eventually to the relationship's falling apart. Apparently, one can tolerate varying degrees of both self- and other-involvement in relationships, but this tolerance depends on the perception of these two levels of involvement as being relatively similar.
Fifth, consider the question of how differences between the ways two people feel about each other compare, in their effects on relationship satisfaction, with differences between the ways each individual feels, on the one hand, and perceives the other to feel, on the other hand. In other words, consider the difference between each individual's perception of the way the other feels versus the way the other actually feels. Our study indicated clearly that it is the perceived rather than the actual feelings that best predicted satisfaction in romantic relationships. The median magnitude of correlation with satisfaction for within-person difference scores was .66, whereas the median magnitude of correlation with satisfaction for between-persons difference scores was only .34. Sixth, consider the interrelationships between the four basic ratings of self to other, other to self, self to ideal other, and ideal other to self, as computed from the two Rubin scales and the Levinger scale. The intercorrelations tended to be quite high, with a median of .68. Clearly, there tended to be a high degree of correspondence among the various feelings and perceptions of feelings. Seventh, consider the question of whether there are differences in the magnitudes of ratings of feelings concerning real versus ideal others. Analysis of variance clearly revealed that ideal others received more favorable ratings than did actual others. Eighth, consider the question of whether there are differences in the magnitudes of ratings from the self to the other, on the one hand, and ratings of the perceived other to the self, on the other. Analysis of variance revealed no mean difference between these two ratings. Thus, actual relationships fall below ideal standards, on the average, but they are not perceived as being asymmetrical, on the average. Finally, consider the question of whether it is possible to predict satisfaction in romantic relationships on the basis of rated feelings from and to real and ideal others. The results of simple and multiple correlational analysis indicated that it is possible to make such predictions, and at a high level of accuracy. The use of just two ratings (difference between perception of actual other to self and ideal other to self; self to other) yielded multiple correlations with satisfaction in excess of .8. To conclude this section, the multiple-triangles framework proves to be viable for understanding loving relationships and for predicting satisfaction in these relationships, or at least in the romantic ones we have studied so far. By looking at all of the various triangles in combination, it is possible to understand in some depth the dynamics of two people's feelings about each other in such relationships. Action Triangle It was noted earlier that the triangle representing the way an individual feels toward another may not be perceived by the other in the same way that it is perceived by the self. There can be any number of sources of this discrepancy in perceptions, but almost certainly one of the most powerful sources is the failure of many individuals to express their love fully in action. It is one thing to feel a certain way but another thing altogether to express these feelings, and often the feelings fail to be communicated because of the inability or unwillingness of the individual to translate the three components of love into actions. Hence, it is necessary to
ROBERT J. STERNBERG
think in terms of an "action triangle" that represents the three components of love as translated into action. The actions that convey each of the three components of love differ. For example, some of the ways in which one might express the intimacy component are by (a) communicating inner feelings; (b) promoting the other's well-being; (c) sharing one's possessions, time, and self; (d) expressing empathy for the other; and (e) offering emotional and material support to the other. Some ways of expressing the passion component include (a) kissing, (b) hugging, (c) gazing, (d) touching, and (e) making love. Some ways of expressing the decision/commitment component include (a) pledging, (b) fidelity, (c) staying in a relationship through hard times, (d) engagement, and (e) marriage. Of course, the actions that express a particular component of love can differ somewhat from one person to another, from one relationship to another, and from one situation to another. Nevertheless, it is important to consider the triangle of love as it is expressed through action, because action has so many effects on a relationship. What are some of the effects of action on love in a close relationship? First, one's actions can actually affect one's levels of the three components. Self-perception theory (Bern, 1972) indicates that one's feelings and thoughts can be affected by one's actions just as one's actions can be affected by one's feelings and thoughts. In other words, the way people act shapes the way they feel and think, possibly as much as the way they feel and think shapes the way they act. Second, certain actions lead to other actions. In other words, acting in certain ways tends to produce acting in other ways and, thus, to build up a network of actions. Expressing one's love through action can lead to further expression of this love through action, whereas failure of self-expression can lead to further failure of this kind. Third, the way one acts is likely to affect the way the other feels and thinks about oneself. In other words, one's actions can be expected to have an effect on the other's triangle of love for oneself. Fourth and finally, one's actions will almost inevitably have an effect on the other's actions, thereby leading to a mutually reinforcing series of paired action sequences. The point to be made is that a theory of love should not get lost within the individuals involved in the relationship. It is necessary to take into account the ways in which individuals express their love. Without expression, even the greatest of loves can die. The triangle serves as a useful geometric metaphor for conceptualizing the interrelations among the three components of love and for conceptualizing relations between the various instantiations of these three components of love: for the self, for the other, for the ideal self, for the ideal other, and for action. In the triangular theory, the locations of points represent coordinates for each of the three components of love, but there is no intention for the distances between points to represent, in any sense, the distances between the various components, or for the cosines of the angles at the vertices to represent correlations between the three components. In other words, the triangle is used as a heuristic, not as a full-fledged geometric model partaking of all of the properties of analytic geometry. The triangular metaphor is useful only to the extent that it serves as a worthwhile heuristic. There are an infinite number of other possible geometric and nongeometric metaphors that would be isomorphic to the triangular metaphor, and the only property that truly serves to distinguish the infinite number of representations is heuristic gen-
erativity. Moreover, other nonisomorphic geometric representations might be considered as well. In sum, it is important to distinguish between the claims of the theory and the particular metaphor used to represent these claims. The theory could be represented by other metaphors, but at least so far the triangle has proved to be a useful one. This section concludes the presentation of the basic elements of the triangular theory of love. This presentation has included not only basic elements of the theory but also a summary of data that are consistent with the theory. Although none of these data were originally collected in order to test the theory, in that the development of the theory followed the various studies described, the data suggest that the triangular theory has at least some merit in helping us understand love in close relationships. However, it is necessary to look beyond one's own data in assessing the empirical viability of a psychological theory. In the next section, empirical findings from the literature on interpersonal attraction in close relationships are considered in terms of how they would be accounted for by the triangular theory of love.
Empirical Phenomena as Viewed Through the Lens of the Triangular Theory The triangular theory of love can account for a number of the main empirical phenomena in the literature on love and close relationships. It is possible to provide here only a brief review of findings and their interface with the triangular theory. Nevertheless, such a review helps show how the theory can be used to understand various kinds of data in the literature and in people's experiences. Some of the main empirical and anecdotal data have already been dealt with in one way or another in this article. For example, the range in kinds of love that one can experience is dealt with in theory primarily by the different possible combinations of intimacy, passion, and decision/commitment, as shown in Table 2. The role of comparison levels, and especially of ideals, is dealt with by the various love triangles discussed in the section on the multiple triangles of love, and the role of action is also discussed in this section. There are a number of other phenomena, not discussed above, that can also be understood in terms of the triangular theory. One such phenomenon is the finding by Walster, Aronson, Abrahams, and Rottman (1966) that on initial dates, physical attractiveness is about all that matters for satisfaction with the date. According to the triangular theory, the passion component of love is the quickest to recruit; the other two components take more time. As a result, there may be relatively little basis for judgment of a partner as suitable for a loving relationship—after a first date—other than passion criteria, such as physical attractiveness. Yet another related finding is that of Dutton and Aron (1974), who found that individuals who are physiologically aroused are more likely to take a romantic interest in a member of the opposite sex whom they meet during the period of that arousal than are individuals who are not so aroused. In particular, these investigators had their subjects walk across either a bridge that swayed from side to side as one walked across it or a bridge that was more stable and closer to the ground. Men who walked across
TRIANGULAR THEORY OF LOVE
the unstable bridge were more likely to be romantically interested in a confederate who was at the scene of the crossing. Peele and Brodsky (1976) have referred to love as an addiction, and it is easy to understand this reference in terms of the triangular theory. According to the theory, only one of the three components of love behaves in a manner comparable to that of addiction, but this component—passion—follows a pattern very similar to that of addiction. The course of acquired motivations described earlier, after the theory of Solomon (1980), applies at least as well to dependencies on things such as drugs, alcohol, cigarettes, and coffee as it does to dependency on other persons. One might argue that these other dependencies are physiological, whereas dependency on another person is psychological. This argument seems incomplete, however. Substance dependencies have a major psychological component as well as a physiological component. It is for this reason that readdiction to the substances is so likely, even after the physiological dependency has been conquered. Moreover, dependencies on other people probably acquire physiological as well as psychological properties. (Indeed, one might argue that psychological states always have physiological substrates.) When one is jilted by a lover, for example, there can be massive psychological effects, and massive physiological effects as well. Symptoms such as irritability, loss of appetite, depression, and inability to concentrate have physiological correlates that correspond to the psychological manifestations. One of the most common observations in everyday life is that people want what they cannot have. In the domain of interpersonal relationships, the phenomenon is simply that of the attraction of the individual who is "hard to get." The status of the hard-to-get phenomenon is not totally clear. For example, Walster, Walster, Piliavin, and Schmidt (1973) found that people tended to be attracted not to those who were hard to get, in general, but to those who were hard to get for others but relatively easier to get for themselves. However, the interpretation of this study appears to be open to at least some question (Wright & Contrada, 1983). The Walster et al. (1973) findings notwithstanding, the hard-to-get phenomenon appears to be one that is well entrenched in people's experience as well as in literature and even in a musical, The Fantasticks. In the psychological literature, there is a theory—reactance theory—that seeks to explain psychologically why people should want what they have difficulty getting (Brehm, 1966; Brehm & Brehm, 1981). In terms of the triangular theory, the locus of reactance in interpersonal relationships would be in the passion component of love. The inability to attain a desired goal state coupled with the belief that the desired goal state is not beyond attainment would lead to increasing the level of the passion component and, usually, behavioral attempts to attain that goal state. After a certain point, these attempts can start to feed on themselves and to persist, even in the absence of realism regarding the likelihood that the goal state will be attained. One of the most robust findings in the literature on attraction in close relationships is the finding that similarity works in close relationships. In other words, people are more likely to form relationships with and later marry people who are more similar to themselves, and also to be happier in relationships with such people (Burgess & Wallin, 1953; Byrne, 1971). In the triangular theory, greater similarity in each of the three components of love will lead to triangles with more overlapping area and corre-
spondingly less nonoverlapping area. Such relationships are predicted to be more satisfactory. Hence, to the extent that greater similarity in background, attitudes about life, and attitudes about the particular relationship affect the love triangles of the two individuals so as to make them more similar, the couple is more likely to be happy in their relationship. One of the odder findings in the literature is what is sometimes referred to as the "exposure effect" (Saegert, Swap, & Zajonc, 1973). It has been found that mere exposure to another individual can foster liking, although it is much less clear that mere exposure fosters loving. This finding fits in with the triangular theory. Whereas mere exposure is not likely to generate physical attraction in and of itself, it is likely to generate at least some elements of emotional connectedness. It is difficult to be with a person over an extended period of time and not to form some kind of emotional bond. And it is the emotional bond that is responsible for liking in the triangular theory. Hence, the exposure effect is likely to promote liking, but not passionate or necessarily committed love. Once a relationship is attained, it goes through a certain course. Theorists have different ideas about what this course is, and a number of them posit stage models of the development and, in some cases, dissolution of relationships (e.g., Kerckhoff & Davis, 1962; Levinger, 1983; Murstein, 1976; Reiss, 1960). The triangular theory predicts that relationships will almost inevitably have a course that will result in qualitative shifts over time. The reason for such shifts is the different courses over time of the three components of love. One of the frequent findings, both in people's experience and in the literature on interpersonal attraction, is that it is difficult to maintain romantic love over a long period of time (Berscheid & Walster, 1978). This aspect of the course of relationships is predicted by the rapid rise but also the relatively rapid fall of the motivational curve in close relationships. Habituation of romance, as well as of other motivated states, can be relatively rapid to develop. However, the rate at which habituation develops will depend on the relative strength of the positive and negative forces in the opponent-proccess account of motivation, and the relative strengths of these two forces are likely to differ as a function of the particular motivational needs involved. For example, the motivational needs that lead us to desire sexual fulfillment may last long beyond the needs that lead us to desire sexual fulfillment from any one particular person. The needs that lead many of us to feel unconditional love for our children also seem to be remarkably persistent, for reasons that are not at present altogether clear. The general point is that relationships will go through different states and possibly stages as a function of the course of the three components, and although there will be differences across persons, relationships, and situations in the exact shapes of the respective curves, there will always be changes in the nature of the relationship with changes over time in the three components of love. One aspect of development in virtually all successful relationships will be what Altman and Taylor (1973) refer to as "social penetration." Social penetration refers to the increasing depth and breadth that characterize relationships as people get to know each other over time. In the triangular theory, social penetration has its most immediate effects on the intimacy component of a relationship. Indeed, the results of the Sternberg-Grajek
ROBERT J. STERNBERG
(1984) study suggest that ability to communicate effectively is almost a sine qua non of a successful loving relationship. In traditional conceptions of sex roles, women tend to stress intimacy and social penetration more in their lives than do men, and one might speculate that the Sternberg-Grajek (1984) finding that women like their best friend of the same sex more than they like their lover stems from the women's success in finding greater communicational intimacy in closeness with other women than with men. The divorce rate today is approaching 50% in the United States, and it seems fitting to conclude this article with some comments on what kinds of things can sustain each of the three components of love in close relationships. First, consider the intimacy component of love. If we accept Berscheid's (1983) view that emotion in close relationships is felt when there is some kind of interruption or disruption of a paired behavioral chain between two people, then it would appear that the worst enemy of the intimacy component of love is stagnation. Although people want some predictability from a loving relationship, too much predictability can probably undermine the amount of intimacy experienced in a close relationship. Hence, it is necessary always to introduce some elements of change and variation—to keep the relationship growing. Obviously, there will be different ways in which change and growth might take place. For some people, the elements of change will be through vacations. For others, it will be through developing new mutual interests. For still others, it will be through experimenting with new behavioral patterns in the relationship. The means of growth and change must be individualized to the relationship, but the need for these two elements is probably common across long-term relationships. Second, consider the passion component of love. In some sense, this component is probably the most difficult to sustain, because it is least subject to conscious control and most subject to habituation. It is well known from conditioning theory that intermittent reinforcement is probably the best maintainer of behavior that results from acquired motivation. However, intermittent reinforcement in the context of a long-term close relationship can potentially take on a rather sinister character. In some cases, the administration of intermittent reinforcement can border on the manipulative or actually become manipulative. Perhaps the best way to maximize the passion component of love over the long term is, first, to analyze the needs the relationship is fulfilling and to do what one can to make sure that these needs continue to be fulfilled and, second, to analyze what needs the relationship is not fulfilling and to try to develop the relationship so that it can meet these needs as well. Again, the particular set of needs and the ways in which they are best met will probably differ somewhat from one relationship to another. Third, consider the decision/commitment component of love. This is the component in which intervention is easiest because it is most subject to conscious control. The best way to maintain commitment in a relationship is probably both to maintain the importance of the relationship in the couple's lives and to maximize the happiness one achieves through the relationship. Doing these things entails working on the intimacy and passion components of love, and especially expressing these components as well as one's commitment to the relationship through action. If one can attain the consummate love that results from high degrees
of the three components in a loving relationship, then under suitable situational circumstances, that relationship seems likely to be one that will last and thrive. To conclude, a triangular theory of love has been presented that attempts to explain and characterize a variety of love-related phenomena. The theory analyzes love in terms of three components—intimacy, passion, and decision/commitment—attempting to provide at the same time both a descriptive and an explanatory framework for how these components can combine into different forms of love (see Ossorio, 1985; also Shweder & Miller, 1985). Although the theory remains at this point an incomplete statement, it provides at least one step toward understanding the nature of love in everyday life. References Altaian, I., & Taylor, D. A. (1973). Social penetration: The development of interpersonal relationships. New \brk: Holt, Rinehart & Winston. Barnes, M., & Sternberg, R. J. (1986). Implicit theories of love. Manuscript in preparation. Bern, D. J. (1972). Self-perception theory. Advances in Experimental Social Psychology, 6, 1-62. Berscheid, E. (1983). Emotion. In H. H. Kelley et al. (Eds.), Close relationships (pp. 110-168). New \brk: Freeman. Berscheid, E., & Walster, E. H. (1978). Interpersonal attraction (2nd ed.). Reading, MA: Addison-Wesley. Brehm, J. W. (1966). A theory of psychological reactance. New \brk: Academic Press. Brehm, S. S. (1985). Intimate relationships. New \brk: Random House. Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory of freedom and control. New \brk: Academic Press. Burgess, E. W., & Wallin, P. (1953). Engagement and marriage. Philadelphia: Lippincott. Byrne, D. (1971). The attraction paradigm. New York: Academic Press. Davis, K. E. (1985, February). Near and dear: Friendship and love compared. Psychology Today, pp. 22-30. Davis, K. E., & Roberts, M. K. (1985). Relationships in the real world: The descriptive approach to personal relationships. In K. J. Gergen & K. E. Davis (Eds.), The social construction of the person (pp. 145163). New York: Springer-Verlag. Duck, S. (1983). Friends for life. New York: St. Martin's. Dutton, D. G., & Aron, A. P. (1974). Some evidence for heightened sexual attraction under conditions of high anxiety. Journal of Personality and Social Psychology, 30, 510-517. Freud, S. (1922). Certain neurotic mechanisms in jealousy, paranoia, and homosexuality. In Collected Papers (Vol. 2). London: Hogarth. Hatfield, E., & Walster, G. W. (1981). A new look at love. Reading, MA: Addison-Wesley. Hinde, R. A. (1979). Towards understanding relationships. London: Academic Press. Kelley, H. H. (1983). Love and commitment. In H. H. Kelley et al. (Eds.), Close relationships (pp. 265-314). New \brk: Freeman. Kerckhoff, A. C, & Davis, K. E. (1962). Value consensus and need complementarity in mate selection. American Sociological Review, 27, 295303. Lasswell, M., & Lobsenz, N. M. (1980). Styles of loving. New York: Ballantine. Lee, J. A. (1977). A typology of styles of loving. Personality and Social Psychology Bulletin, 3, 173-182. Levinger, G. (1983). Development and change. In H. H. Kelley et al. (Eds.), Close relationships (pp. 315-359). New York: Freeman. Levinger, G., Rands, M., & Talaber, R. (1977). The assessment of involvement and rewardingness in close and casual pair relationships
TRIANGULAR THEORY OF LOVE (National Science Foundation Tech. Rep. DK). Amherst: University of Massachusetts. Livingston, K. R. (1980). Love as a process of reducing uncertainty. In K. S. Pope (Ed.), On love and loving (pp. 133-151). San Francisco: Jossey-Bass. Lund, M. (1985). The development of investment and commitment scales for predicting continuity of personal relationships. Journal of Social and Personal Relationships, 2, 3-23. Mandler, G. (1980). The generation of emotion: A psychological theory. In R. Plutchik & H. Kellerman (Eds.), Emotion: Theory, research and experience. Vol. 1: Theories of emotion (pp. 219-243). New York: Academic Press. Maslow, A. H. (1962). Toward a psychology of being. Princeton, NJ: Van Nostrand. Murstein, B. I. (1976). Who will marry whom? Theories and research in marital choice. New York: Springer. Ossorio, P. G. (1985). An overview of descriptive psychology. In K. J. Gergen & K. E. Davis (Eds.), The social construction of the person (pp. 19-40). New York: Springer-Verlag. Peck, M. S. (1978). The road less traveled: A new psychology of love, traditional values and spiritual growth. New York: Simon & Schuster. Peele, S., & Brodsky, A. (1976). Love and addiction. New York: New American Library. Reik, T. (1944). A psychologist looks at love. New York: Farrar & Rinehart. Reiss, I. L. (1960). Premarital sexual standards in America. New York: Free Press. Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categorization (pp. 27-48). Hillsdale, NJ: Erlbaum. Rubin, Z. (1970). Measurement of romantic love. Journal of Personality and Social Psychology, 16, 265-273. Rubin, Z. (1973). Liking and loving: An invitation to social psychology. New York: Holt, Rinehart & Winston. Saegert, S. C, Swap, W., & Zajonc, R. (1973). Exposure, context, and interpersonal attraction. Journal of Personality and Social Psychology, 25, 234-242. Schank, R., & Abelson, R. (1977). Scripts, plans, goals, and understanding. Hillsdale, NJ: Erlbaum.
Shweder, R. A., & Miller, J. G. (1985). The social construction of the person: How is it possible? In K. J. Gergen & K. E. Davis (Eds.), The social construction of the person (pp. 41-69). New York: Springer-Verlag. Solomon, R. L. (1980). The opponent-process theory of acquired motivation: The costs of pleasure and the benefits of pain. American Psychologist, 35, 691-712. Spearman, C. (1927). The abilities of man. New York: Macmillan. Sternberg, R. J. (1985). Beyond IQ: A triarchic theory of human intelligence. New York: Cambridge University Press. Sternberg, R. J. (1986). Construct validation of a triangular theory of love. Manuscript in preparation. Sternberg, R. J., & Barnes, M. (1985). Real and ideal others in romantic relationships: Is four a crowd? Journal of Personality and Social Psychology, 49, 1586-1608. Sternberg, R. J., & Grajek, S. (1984). The nature of love. Journal of Personality and Social Psychology, 47, 312-329. Swensen, C. H. (1972). The behavior of love. In H. A. Otto (Ed.), Love today (pp. 86-101). New York: Association Press. Tennov, D. (1979). Love and limerence. New York: Stein & Day. Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York: Wiley. Thomson, G. H. (1939). The factorial analysis of human ability. London: University of London Press. Thurstone, L. L. (1938). Primary mental abilities. Chicago: University of Chicago Press. Walster, E., Aronson, V, Abrahams, D., & Rottman, L. (1966). Importance of physical attractiveness in dating behavior. Journal of Personality and Social Psychology, 4, 508-516. Walster, E., Walster, G., Piliavin, J., & Schmidt, L. (1973). "Playing hardto-get": Understanding an elusive phenomenon. Journal of Personality and Social Psychology, 26, 113-121. Wright, R. A., & Contrada, R. J. (1983). Dating selectivity and interpersonal attractiveness: Support for a "common sense" analysis. Unpublished manuscript, University of Texas, Austin.
Received April 1, 1985 Revision received November 6, 1985