You are on page 1of 14

Journal of Personality and Social Psychology 1991, Vol. 60, No.

3, 425-438

Copyright 1991 by the American Psychological Association, Inc. 0022-3514/91/$3.00

The Concept of Love Viewed From a Prototype Perspective
Beverley Fehr
University of Winnipeg Winnipeg, Manitoba, Canada

James A. Russell
University of British Columbia Vancouver, British Columbia, Canada

Even if superordinate concepts (such as fruit, vehicle, sport) are prototypically organized, basic-level concepts (such as apple, truck, hockey) might be classically defined in terms of individually necessary and jointly sufficient features. A series of 6 studies examined 1 basic-level concept in the domain of emotion, love, and found that it is better understood from a prototype than a classical perspective. The natural language concept of love has an internal structure and fuzzy borders: Maternal love, romantic love, affection, love of work, self-love, infatuation, and other subtypes of love can be reliably ordered from better to poorer examples of love. In turn, each subtype's goodness as an example of love (prototypicality) was found to predict various indices of its cognitive processing. Implications for a scientific definition and typology of love are discussed.

Prototype theory has provided insights into concepts central to psychology, including behavioral act (Buss & Craik, 1983), personality trait (Cantor & Mischel, 1979), intelligence (Neisser, 1979), social situation (Cantor, Mischel, & Schwartz, 1982), and environmental setting (Tversky & Hemenway, 1983). The theory has inspired important new approaches to psychiatric diagnosis (Cantor, Smith, French, & Mezzich, 1980) and personality assessment (Broughton, 1984). The studies reported in this article are part of a larger project designed to explore the applicability of prototype theory to the domain of emotion concepts. We had two related purposes. The first was to use the tools of prototype theory to throw light on the elusive concept of love. We shall have something to say about both the definition of and typologies of love. The second purpose concerned prototype theory itself. Prior demonstrations of a prototype structure in concepts focused on superordinate concepts, such as fruit, vehicle, or emotion. Would the same results occur, we asked, for basic-level concepts, such as apple, truck, or love? We shall begin with the second issue.

Prototype Approach
Traditionally, the general terms of a language were thought to denote categories of objects or events, each member of which possessed features that were each necessary and together sufficient to define membership in that category. According to this classical view, to know the meaning (sense) of a general term (i.e., to have the concept associated with it) was to know at least implicitly these necessary and sufficient features. A definition could therefore be formulated by philosophical discussion or— because the defining features were also in the objects or events —by empirical investigation. Although some writers continue to defend the classical view (Armstrong, Gleitman, & Gleitman, 1983; Harnad, 1987; Osherson & Smith, 1981), considerable psychological research reinforces a growing skepticism over its plausibility as an adequate account of most concepts used in everyday speech and thought (Medin, 1989; Mervis & Rosch, 1981; Smith & Medin, 1981). Rosch's (1975,1977,1978) proposal of a prototype account as an alternative to the classical view was followed by various nonclassical accounts (Kahneman & Miller, 1986; Lakoff, 1987; Medin, 1989; Neisser, 1987; Smith & Medin, 1981). Indeed, in one account, concepts are thought of not only as ill-defined but as varying from one person to the next, and, for the same person, from one time to the next (Barsalou, 1987). We make no attempt in this article to differentiate within this family of nonclassical accounts. Rather, our purpose is to continue to specify through empirical means the properties of emotion concepts. In previous research, we have compared the classical with the prototype view in the domain of emotion (Fehr, 1982,1988; Fehr & Russell, 1984; Fehr, Russell, & Ward, 1982; Russell, 1991; Russell & Bullock, 1986), but we are not alone in this interest. Indeed, writers as far back as William James (1890/ 1950,1902/1929) have viewed emotion concepts in a way more compatible with a prototype than a classical perspective. Averill

Portions of this research were presented at the Annual Convention of the Canadian Psychological Association, Montreal, Quebec, Canada, June 1988. This research was supported by a University of Winnipeg Group II grant (4100-000) awarded to Beverley Fehr and by a grant from the Social Sciences and Humanities Research Council of Canada to James A. Russell. We thank Pat Keelan and Sandra Wolfram for their assistance in data coding and Jim Averill, Michael Mascolo, and Phil Shaver for helpful comments on a draft of this article. Correspondence concerning this article should be addressed to Beverley Fehr, Department of Psychology, University of Winnipeg, Winnipeg, Manitoba, Canada, R3B 2E9, or James A. Russell, Department of Psychology, University of British Columbia, 2136 West Mall, Vancouver, British Columbia, Canada V6T1Y7.

Evidence of prototype structure found at one level of a hierarchy does not show that concepts at other levels of even that same hierarchy must be so structured. not by virtue of possessing all of a set of denning features. it was Rosch and her associates who drew attention to the psychological significance both of prototypes and of the basic level (Rosch. Mervis. As Brehm (1985. happiness. This level is probably basic in Rosch's sense. At the basic level. types of aggression may be represented as prototypic. which sharply distinguishes superordinate-. vegetable. Schwartz. Shaver. 1972. At the topmost. and Dobbs (1989) considered aggression a superordinate concept. More generally.1982) has argued forcefully for viewing emotions as syndromes. Many of the middle-level categories may be further divisible. and yet there is insufficient evidence on Concept of Love Love can be studied as a relationship. or exemplary. Newport and Bellugi (1978) analyzed American Sign Language. we had hypothesized that they do—such is a corollary of viewing anger. Ortony. Again. but these same features are also shared with members of contrasting categories. no one feature of which is denning. Thus. 1 . we hope to offer evidence that should address the concerns of those who continue to defend a classical view in the domain of emotion (Johnson-Laird & Oatley. emotion. or superordinate. Lysak. The number of categories at this level is indefinite because prototypes shade gradually into nonprototypes. Resemblance being a matter of degree.. 1987. Mascolo & Mancuso. forming a subordinate level. That fruit is organized around prototypes and has fuzzy borders does not show that apples and oranges are so organized. and so on. having a fight with your wife a basic-level concept. 90) commented: Lysak. 1977. the borders of the concept of emotion were found to be fuzzy: Subjects could not decide whether or not certain peripheral members. The prototype analysis to be presented here thus bears a family resemblance to other analyses of the emotion domain. vehicle. At the middle level. Tiller & Harris. however.1985) drew explicitly on the idea of prototypes when developing paradigm cases of friendship and love relationships. fear. 1990. and love as scripts (Fehr & Russell. the number of categories at the middle level is indeterminate: Events vary in the extent to which they are emotions. Rule. Indeed. features are shared by all or most members. anger is divided into wrath. 1987). 1973) early work on focal points of color categories. the superordinate category of emotion was found to have what Rosch called an internal structure—gradedness in membership—that predicted various indices of cognitive processing involving emotion concepts. is a member of the superordinate concept. Kirson. In applying the prototype perspective to the domain of emotion categories. these cases are mentally represented as individual. Their analysis suggested that only superordinate concepts lack common features shared by all members.g.g. Averill and Boothroyd (1977) denned an ideal or paradigm case of romantic love and measured the degree to which actual cases resemble it. what is important here is not so much the correct particular level in the hierarchy but the idea that concepts at different levels of the hierarchy may not all be structured in a prototypic way. and other highly general terms. love. We suspect that because of its relatively long name. Members resemble each other in overlapping and criss-crossing ways that vary in kind and number. on other accounts. Clore. having a fight with your wife or fighting in a hockey game may each show internal structure and fuzzy borders. In this article. and no sharp boundary separates emotions from nonemotions. basic-. & Foss. p. RUSSELL whether basic-level concepts are themselves prototypically or classically organized. rage. What is at the heart of the prototype perspective. Earlier. In any case. Gray. which shade gradually into nonemotions with no sharp boundaries to be found. 1989. As hypothesized. for the notion of script extends to episodes the notion of prototype. and subordinate-level categories. furniture. were genuine examples of emotion. is the notion of resemblance. Davis and Todd (1982. although sometimes forgotten—it has even been omitted from some psychologists' lists of the emotions (Ekman. Other investigators of emotion have also drawn on prototype theory (Conway & Bekerian. & O'Connor. 1976). as an experience. Izard. Ironically. Rosch's (e. it is helpful to begin by picturing that domain as an inclusion hierarchy. The evidence gathered so far pertains to the superordinate concept. Johnson. as an attitude.. Tomkins. we examine not cases of love. For example. on the other hand. At the subordinate level. but the natural language concept of love. features are likewise shared by all or most members. Love has been defined in a variety of ways. Love is divided into romantic love. & Boyes-Braem.426 BEVERLEY FEHR AND JAMES A. they are represented as a more generalized schema. such as pride and respect. At the subordinate level. They argued that the basic-level concept of aggression is more consistent with a classical than a prototype conceptualization. having a fight with your wife is more likely a subordinate-level category. 1984). cases of emotion. and throwing down your gloves a subordinate-level concept. Rule. On some accounts. emotion is divided into fear. and Dobbs (1989) offered another hypothesis in which the nature of concepts varies with level in the hierarchy. Indeed. Moreover. as far as we know. but by sufficient resemblance to the prototypical. 1987. The important exceptions are Labov's (1973) study of cups and Rosch's (e. and so on. annoyance. and other prototypical and less prototypical emotions from anger to zest. Love is an important emotion. we focus on love as an emotion. Available evidence supports this line of thinking. In this article. actual experiences. and that throwing down your gloves is more likely a feature than another level of categorization. almost all of the studies to date offered as support for prototype theory involve concepts at a superordinate level. and so on. one purpose of the present research was to begin to examine empirically whether basic-level categories have internal structure and fuzzy boundaries. 1975) later work pertained to fruit. A middle-level category. emotion. filial love. or a particular event.1 Therefore. the number of subcategories of love is indefinite. love is a prototypical emotion—it received the highest prototypicality rating of any emotion in the Fehr and Russell (1984) study. 1984). 1984). (1980. level is the word emotion.

Failure to agree on a definition or a typology of love suggests that psychologists are unsure what to include under the heading of love. subjective ratings of how natural or peculiar it sounds in sentences about love (Study 5). bean bag chair. Stop after a few minutes or 20 items. familial love. lawn chair. Love has been said to consist of two (Hatfield & Walster. theories on love. The gradual change in availability is consistent with the idea of internal structure—that prototypical members shade gradually into nonprototypical members and then into nonmembers with no sharp boundary to be found. Twenty of these subcategories were then selected for analysis in the remaining studies. recliner. and the family resemblance pattern of features listed for the types of love (Study 6). or subordinate. 123 were mentioned by only one subject and 93 by more than one. subjects directly rated the goodness-of-example (prototypicality) of each of the 20 target subtypes of love chosen in Study 1. and so on. the subtypes of love so generated varied widely in how readily they came to mind. Remember that there are no right or wrong answers—just give us your opinion. 1984. From one sample. After describing that research. you might write: rocking chair. aimed at establishing that subcategories of love vary in their degree of membership in love. we shall then discuss the relationship of our findings to a scientific definition and typology of love. 1924). reaction time to verify sentences concerning its membership in the category love (Study 3). that degree of membership predicts important indices of cognitive processing of members.g. three (Kelley. Maslow. involving a predisposition to think. comparison of one with another is postponed until a final section. and that members share a family resemblance to one another. We have books on love." Close your eyes and imagine a true red. 1955). ideas. All together. Between these extremes was no clear break between those available and those unavailable from memory. the probability of its inclusion in the category love (Study 4). if asked to list types of the category CHAIR. Study 3): This study has to do with what we have in mind when we hear and use words. These were the 10 terms listed most frequently. There may be no fixed number of subtypes into which love can be divided. Of these. the inability to achieve consensus is understandable: There may be no small set of criterial features common to all and only instances of love. 1975). The category we're interested in is LOVE. feel. family love. Now imagine an orangish red. listed by over 60% of the subjects. Swensen's (1972) definition of love focused on behaviors such as shared activities. In the first study. with a range of 17 to 47. In this study. From a prototype perspective. Yet no one has a single. roughly half were men. subjects were asked to list whatever subtypes of love they could think of.6 years. along with another 10 chosen from across the entire range of frequency. 1986) subtypes. 1978). As can be seen. 104). Method Subjects (N = 92) read the following instructions (adapted from Fehr & Russell. Because internal structure is best demonstrated through a convergence of these operations. At the other extreme were the 123 items each listed by only one subject. we treated syntactic variants (e. 1922/1951). 1984. along with the frequency with which they were mentioned. disclosing intimacies. 427 Winnipeg who volunteered their participation. Imagine a pur- . and research on love. Please list as many types of LOVE as come to mind. or eight (Sternberg. This grouping left a total of 216 different items. these six studies yielded eight separate ways of mapping the internal structure of the concept of love. Method Subjects (N= 84) read the following instructions (adapted from Fehr & Russell. At one extreme was friendship. For example. Skolnick (1978) defined love as "a constructed experience built with feelings. the 20 items are listed in Table 2. 1988). In each study. Rubin (1970) defined love as an attitude held by one person toward another. psychologists have yet to agree on a typology of love. In this way. seven (Kemper. simple definition that is widely accepted by other social scientists. Subjects in these studies were students at the University of Study 2: Prototypicality Ratings In this study. Results and Discussion Subjects generated an average of 8.CONCEPT OF LOVE Social scientists have had as much trouble defining love as philosophers and poets. and love of family) as identical responses. Study 1: Free Listing of Subtypes of Love Subjects were presented with the concept of love and asked to list as many subtypes as came to mind. and acts that achieve reproductive success (Buss. In summarizing the responses. and behave in certain ways toward that person. Similarly. 1978.. and so on. level in the hierarchy of emotion. Overview Six studies were conducted. Twenty target subtypes of love were selected from Table 1 for further analysis in the subsequent studies. and cultural symbols" (p. Different writers have pointed to different features of love as denning: frustrated desire (Freud. One purpose of our research was to explore empirically just this line of thinking. 1983). subjects generated a portion of the next lower. we estimated their mean age as 20. The ease with which subcategories of love came to mind also provided one measure of their degree of membership within the category of love. kitchen chair. erogenous stimulation (Watson. that no sharp boundary separates members from nonmembers.69 responses each. Consider the word "red. These 93 items appear in Table 1. we're concerned with types within a general category. These studies explored other predicted consequences of graded membership: direct ratings of the degree to which each subcategory is an example of love (Study 2). Study 1): This questionnaire is part of a larger project on the sorts of things we have in mind when we hear and use words. rewarding interactions (Centers. half women. stool.

1987)." when X is the type listed." they are not as good examples of "red" (not as clear cases of what red refers to) as the clear. plish red.34 .84 4.61 . "Simone de Beauvoir once wrote that men and women have different concepts of love. such as Table 2 Prototypicality Ratings. but still recognize which is the better example of "red.78 . Notice that to judge how good an example something is. You will be asked how good an example of love the various types or instances of love are. Item Mutual Of sports Agape Ofart Casual Compassionate Dependent Offood Of girlfriend Of grandparents Heterosexual Of humanity Long term Of money Occupational Of self Surface Trust Unselfish Aesthetic Ambition Of beauty Of books Brotherhood Compulsive Fantasy Free Between friends of the same sex Helpful Hopeless Hurtful Incestual Inner Kinship Love-hate Love-30 (tennis) Natural Requited Serious Shallow Sharing Spontaneous Transitory Understanding Of work Young No. men more romantic ones (Peplau. I don't doubt that this is true" (Solomon.65 . Consistent with this idea. 1988.55 .46 .428 Table 1 Free Listing of Types of Love Item Friendship Sexual Parental Brotherly Sibling Maternal Passionate Romantic Familial Puppy Paternal Lust Infatuation Of pets Physical True Emotional Platonic Respectful Sisterly Everlasting Of life Marital Of nature Obsessive Patriotic Unrequited Affection Of animals Erotic Intimacy Selfish Conditional First Intense Possessive Spiritual Unconditional Admiration Blind Committed Deep Giving Of God Homosexual Honesty Materialistic No.55 .76 4. errors 0 0 0 1 0 0 1 0 1 0 0 0 0 1 1 0 1 1 2 2 Note. 1983). For half the subjects.57 .47 4.53 .47 . and (b) the intraclass correlation coefficient (which is equivalent to the average of all possible split-half reliability coefficients) was . Reaction Times. Number of errors refers to the number of subjects reponding "false" when presented with the statement "X is a type of love. true red.95. Of course. individual prototypicality ratings were not as reliable as these aggregates (Barsalou. For example.14 2. previous studies have found sex differences in beliefs and feelings about love.60 4. p.57 . and Errors in Verification for 20 Types of Love Nfean reaction time . N = 84. RUSSELL worry about why you think something is or isn't a good example. r= .98 2. times mentioned 51 25 23 22 22 20 19 19 18 18 17 14 13 10 10 10 9 9 9 9 8 8 7 7 7 7 7 6 6 6 6 6 5 5 5 5 5 5 4 4 4 4 4 4 4 4 4 BEVERLEY FEHR AND JAMES A. Just give us your opinion. the 20 terms were presented in reverse order. Two indices showed the reliability of these means: (a) Separate means calculated for the two orders of presentation were highly correlated. Prototypicality ratings were made on a scale ranging from 1 (extremely poor example of love) to 6 (extremely good example of love).47 .51 . 69).79 3. The mean correlation between two raters across the 20 terms was .98 3." The word we are interested in is LOVE.73 4.74 4. .00 3. Women are more likely than men to report emotional symptoms of love.96 4.76 3.64 .42 4.39 .22 4. Results and Discussion Mean prototypicality ratings for the 20 target subtypes of love appear in Table 2. has nothing to do with how much you like the thing. and that's why they fail to understand one another.35. Reaction times are reported in seconds.53 .42 No.21 3." perhaps not even red at all.98. Orange and purple are even poorer examples of "red. Items listed by only one subject are omitted. You might prefer a purple-red or purple to a true red. On the following page is a list of different kinds of love. women are more likely to endorse pragmatic beliefs about love. Although you might still name the orange-red or the purple-red with the term "red.74 4.62 .63 Type of love Maternal love Parental love Friendship Sisterly love Romantic love Brotherly love Familial love Sibling love Affection Committed love Love for humanity Spiritual love Passionate love Platonic love Self-love Sexual love Patriotic love Love of work Puppy love Infatuation Prototypicality 5. times mentioned 4 4 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 Subjects rated each of the 20 target subtypes of love (listed in Table 2) on a scale ranging from 1 (extremely poor example of love) to 6 (extremely good example of love). Don't Note.81 .49 . The maximum possible number of errors was 20.39 5.27 4.

you would press the "1" key for TRUE. such that the category membership of prototypical instances was verified more quickly than that of peripheral instances (1. Reaction times to the practice items were not recorded." The 80 statements were presented on a computer screen in a different random order for each subject. Eighty sentences were needed so that the proportion of true and false statements was equal. Results and Discussion Reaction times were analyzed only for correct responses. The pool of statements consisted of (a) 10 true statements concerning prototypical types of love. a decision was required as to which subtypes of love should be considered central and which should be considered peripheral. 1976. p < . However. such as "Patriotic love is a type of love". or feeling as though floating on a cloud (Dion & Dion. Over half of all errors occurred in response to the four most peripheral subtypes. If the statement "Car is a type of vegetable" appeared. A direct challenge to the classical view is therefore the demonstration that membership is probabi- . whereas emotion. Subjects were predicted to verify a statement like "Maternal love is a type of love" more quickly than a statement like "Patriotic love is a type of love. have an internal structure.64 vs. chickens.38 vs. Over 90% of the subjects responded correctly to each statement. & Slapion-Foote. and penguins. they seem to have the same concept of love. The statements of primary interest were the 20 true statements concerning love.10 (1. Mean reaction time and number of errors for each subtype appear in Table 2. r .60 s) was not significant. Thus.57 s). but that these prototypicality ratings predicted how long subjects took to verify that they were birds. A response was considered correct if the subject responded "true" to a true statement or "false" to a false statement. (1983) showed that paradigm cases of classically definable concepts. (g) 10 false statements concerning prototypical fruit. A median split of the prototypicality ratings of Study 2 resulted in 10 subtypes (e. Similarly. The difference in reaction time between prototypical and peripheral false love statements was not significant. The cutoff we had used to divide central from peripheral emotions was similar. Classically definable concepts have precise borders.51 vs. /(26) = 4.. Fehr and Russell (1984) proposed a method.CONCEPT OF LOVE feeling euphoric. although for the more peripheral subtypes of love the beginnings of disagreement can be seen.05. So. Subjects spent approximately 10 min on this task. 1982) appeared on the screen: This is a simple study of the "belongingness" of items into categories. For the remaining studies." Method Subjects (N= 27) were asked to respond "true" or "false" to 80 statements in the form of "AT is a type of Y" The general categories (Y) of interest were love and fruit. such as "Sexual love is a type of cheese"..62 s). 1.52 vs. Try to respond as quickly and as accurately as you can. such as "Tomato is a type of fruit". such as "Maternal love is a type of love". (e) 10 true statements concerning prototypical fruit.001. such as odd number and plane geometric figure. Those that received ratings lower than 4. Study 3: Reaction Time to Verify Category Membership Rosch (1973) found not only that subjects could reliably rate how good an example of the concept bird were robins. (f) 10 true statements concerning peripheral fruit. (c) 10 false statements concerning prototypical types of love. maternal love) with ratings higher than 4. in our data.65 s). fruit. Dependent sample t tests showed a significant difference for reaction times to true love statements.94.19. In a related finding.30. The following instructions (adapted from Fehr et al.55 on the same 6-point scale (Fehr & Russell. Subjects were first presented with four practice statements (two true and two false) to familiarize them with the task (e.94. men's and women's ratings of prototypicality of various features of love were also highly correlated (Fehr. (d) 10 false statements concerning peripheral types of love. 1. if the statement "Car is a type of vehicle" appeared on the screen. such that the category membership of prototypical instances was verified more quickly than that of peripheral instances (1. p> . There was also a significant difference for reaction times to true fruit statements.68. vehicle. or in how they experience love. Fehr et al. And women love their partners more companionately.. Armstrong et al. although men and women may differ in their beliefs about love. Central members were verified faster than peripheral members. and the like do not. men's and women's prototypicality ratings were highly correlated. you would press the "2" key for FALSE. 4.07. whether prototypical or peripheral. r(26) = 1. eagles. whereas men love theirs more passionately (Traupmann & Hatfield. You will be presented with a series of statements of the form "X is a If" \bur task is to respond True or False to each statement as fast as you can. Hendrick. The types of love were the 10 central and 10 peripheral types as determined in Study 2. 1. having trouble concentrating. p < . Next followed the 80 statements. 1981). This demonstration seems to leave no method of distinguishing concepts that are classically definable from those that are not. The purpose of the present study was to test this prediction for the relationship between the basic and the subordinate level of categorization. "Hockey is a type of game" and "Soccer is a type of book"). such as "Olive is a type of dance.p<.g. such as "Pear is a type of vehicle". such as "Apple is a type of fruit". patriotic love) were taken to be peripheral.45 on the 6-point scale. 1988). 1984). these were taken to be central. men and women did not differ significantly far = .g. Study 4: Fuzzy Borders Internal structure and a classical definition are not mutually exclusive. The 20 types of fruit were taken from Rosch (1975) and were included to replicate her reaction time findings and to provide filler items for the task. for example.05) in which subtypes of love they rated as better or poorer examples of love. In fact. (b) 10 true statements concerning peripheral types of love. Foote. t(26) = . Hendrick. 1984). 429 and (h) 10 false statements concerning peripheral fruit. (1982) found that the prototypicality ratings of various emotions predicted how long subjects took to verify that they were emotions.45 (e. t(26) = 2. The difference in reaction time between prototypical and peripheral false fruit statements (1. such as "Romantic love is a type of boat".g. 1.

96 with its mean prototypicality rating. metaphors aren't to be taken literally.33 4.29 3.43 4. The present study addressed the question of whether love has sharp or fuzzy borders. Something can be a genuine case.38 2. . But most cases fell between these ex- Note. Prototypicality ratings. fruit. we don't mean it really. This finding implies that content-based representations rather than logically structured categories mediated their judgments.43 4.15 Results and Discussion The majority of subjects judged that romantic love is a type of love—only 2% disagreed with the majority decision. Road apples are not genuine apples. baby tiger is nonetheless a genuine tiger." However. We know that koala bears are not genuine bears.86 4. there was the expected disagreement found in the peripheral cases.'s (1987) cluster of subtypes of love. or vehicle. Just give us your opinion. A separate group of subjects provided prototypicality ratings for these 30. The probability of a candidate's being judged a genuine case of love correlated . Subjects' judgments were not random. Yet a penguin. The median percentage was 55. On one hand. at least 90% of the subjects agreed that each of the 20 target subtypes was a type of love. but the sample did not overrepresent prototypical cases.90 2. To take an extreme example. blind. Subjects (N = 21) not included in the study just described rated all 30 items in the same random order. nor Tom a genuine volcano. Table 3 Percentage of Subjects Denying Membership in the Category Love Item Romantic love A mother's love Caring A brother's love Friendship Tenderness Compassion Affection Love of humanity Self-love Passion Longing Fondness Liking Love of country Desire Love at first sight Sentimentality Attraction Adoration Puppy love Arousal Love of work Admiration Love of sports Love of books Love of food Love of money Infatuation Lust 2 6 8 11 14 24 24 27 35 36 36 44 52 53 58 59 60 63 63 65 66 68 68 75 75 76 77 80 82 87 Prototypicality 4. Don't worry about why you feel one way or the other.95 4. When the word love appeared in the item (e.05 3. many subjects still denied that it was a type of love. four-legged. following a procedure identical to that of Study 2.52 3. RUSSELL tremes. Study 3 had provided data on the question of whether the boundaries of love are sharp or fuzzy. we do.57 3. 1984)—for example. N-21 for prototypicality ratings. On the other hand.86 4.86 4. Consider each in turn.14 3. however. three-legged. Subjects (N = 118) read the following instructions: Sometimes when we use a word. Yet we know that an albino. not a bird. puppy love).43 3. each of the 30 items had some claim to being a genuine case of love. Other times. Most judged that lust is not a type of love—only 13% disagreed with the majority decision. Fifteen were taken from Shaver et al.62 2. As in: "Sally is an iceberg" or "Tom is a volcano. Study 5: Substitutability for the Category Name Asked to give us a sentence about emotion.86 3. one subject wrote: "Sometimes emotion is hard to control.48 5.430 BEVERLEY FEHR AND JAMES A. For 15 of the 30 candidates presented. We thus offered evidence to show that subjects can agree on which cases are and which are not odd numbers but cannot agree for emotion. You may decide that all are genuine.86 5. sharp-eyed adult tiger.90 4.57 3. even though it can't fly. Or you may decide that none are. listic rather than all or none. The results of Table 3 show that a sample of college-educated native speakers do not agree among themselves on what is and what is not included in the category of love. striped. No absurd filler items were presented. but the results were not clear. "Sometimes respect is hard to control. This degree of consensus might have occurred because subjects were asked to make quick judgments in a context where no responses went largely to absurd filler items such as "Is infatuation a type of fruit?" In this context.48 3. The percentage of subjects who denied that each item is a genuine case of love is shown in Table 3.29 3.81 2. A bird-of-paradise is a flower.76 2. N = 118 for percentages. Picture a tiger: a yellow. In this study. "Sometimes anger is hard to control. Subjects were given 30 subtypes and as much time as they wanted to answer each question. The remaining 15 were selected from responses obtained in Study 1 in such a way as to include a range of availability but to place an emphasis on the less available items in order to have a good sample of potential borderline cases." Such effects had been expected because the Method Membership ratings. however unusual or atypical. we're interested in your opinion about what are genuine cases of love and what are not. and please judge whether it is or is not a genuine case of love.g. as had been done in the target sample used in Study 3." Sally is not a genuine iceberg. Thus.33 3. The present study examined a broader range of potential subtypes. "Is infatuation a type of love?" may have seemed more reasonable than it otherwise would.95 2. is a real bird.62 3. On the next page is a list of possible cases. substituting a peripheral emotion produced a sentence that often sounded rather peculiar—for example." Substituting for the word emotion a prototypical type of emotion produced a sentence that sounded quite natural (Fehr & Russell. Next followed a list of 30 possible types of love. at least one third dissented from the majority.14 3.

**p<. Don't spend time wondering if your answer is right or wrong. In the present study. you are constantly exposed to people's views of love. The phenomena of love are not easy to pin down scientifically. Love has to be worked at and strived for to be truly achieved. On the other hand. Subjects(#= 37) were asked to generate sentences containing the word love. Instead. The first 10 sentences were written by undergraduate students.48* . There are no right or wrong answers. The naturalness of the resulting sentence was expected to depend on how prototypical was the subtype of love substituted. Subjects (N= 400) read the following instructions (adapted from Fehr & Russell.45* Note.001. Most of the time love does not involve intense physical reactions. 431 culiar. Your task is to read each sentence to yourself and rate how natural or peculiar it sounds to you." Substitution produced some sentences that sounded natural and others that sounded peculiar (e.81*** . the sentence "A bowl of apples makes a nice centerpiece for the table" probably sounds quite natural to most people.70*** . understanding the other. The second 10 sentences were taken from psychology textbooks. it can be viewed as a particular sort of attitude one person has toward another person.61** . we asked whether something similar would happen with sentences about basic-level categories. We're interested in sentences about LOVE. Love is tender and vulnerable no matter who shares it. good and bad. Love is something shared intensely by two people. For example. 17 . *p<. 19.01 ***p<. Ten of these sentences were selected for use in this study (Sentences 1 to 10 in Table 4). Below is a series of sentences—some make a lot of sense.05 . An additional 10 sentences were taken from recently published textbooks for courses in introductory or social psychology (Sentences 11 to 20 in Table 4). 8. 17.60** .33 . substitution of a prototypical subtype. Love is a very happy state—most of the time. yielded a sentence that subjects thought Method Phase 1. we're interested in how peculiar or how natural certain sentences sound. 12. Twenty sentences were drawn for each subject in such a way that each subtype of love and each sentence frame appeared only once. 1984. Love is merely an intense form of liking.05. and other superordinate categories. 18. Love is one of the most important human emotions. The emotions that accompany love are complex and far from being clearly understood. 16.17 —. Given the importance of love in promoting happiness and making the world go 'round. Love is a giving process. 3.14 . Correlation . 2. 20. "Friendship has to be worked at and strived for to be truly achieved" and "Infatuation has to be worked at and strived for to be truly achieved"). like the other emotions. Love may be both happy and sad. listen to records. 7. we're not interested in sentences about fruit. In our culture. In the example above. vehicles. 10.13 -. The 20 sentences were drawn from a pool of 400 sentences formed by substituting the 20 subtypes of love in place of the word love in the 20 sentences listed in Table 4. it may be surprising that psychologists have only recently begun systematic research on this topic.. Love is an emotion and. we learn about love from childhood on. . 4. just base your answer on your first impression from reading the sentence. may be seen as having both physiological and psychological components. yet it is poorly understood. and we took another 10 from psychology textbooks.31 . and some may make no sense at all.42 . Subjects then rated each of 20 sentences on a scale ranging from 1 (very peculiar) to 6 (very natural sounding).04 . 15. So. Love varies with the person and with the relationship.45* . 14. Rosch (1977) had earlier found similar results for sentences about fruit. 6.36 -. Another group of subjects was then asked to judge how peculiar or natural they found the sentences that resulted when the 20 target terms were substituted for the word love.CONCEPT OF LOVE more prototypical members are more representative of. friendship. others make some sense. 11. If you go to movies. Love is a feeling one gets in a special relationship with another person. or read novels.79*** . 13. and realizing the other's faults. Love is very painful if not reciprocated. 5. 13 -.g. In this study. Phase 2. Study 4): This questionnaire is part of a larger project on the sorts of things people have in mind when they hear and use words. Results and Discussion Consider the following sentence: "Love has to be worked at and strived for to be truly achieved. and closer to the meaning of. a sentence like "A bowl of watermelons makes a nice centerpiece for the table" sounds rather pe- Table 4 Correlations Between Rated Naturalness and Prototypicality Sentence 1. One group of subjects first gave us 10 sentences.03 -. Commitment and caring are important components of love. the superordinate category name. One of the difficulties in discussing love is that it means so many things to so many people. In this study. specifically with sentences about love. 9.

To derive an index of internal structure from these data. 1975). ranging from .08 3. Still. if you were asked to list the characteristics of a person experiencing a kind of fear. then statements about love that students find in their textbooks can be far removed from their own conceptualizations of love.61. If substitutability can be taken as an indication of the core meaning of the concept. and Medin and Smith (1984) cautioned that "comparable typicality effects in two domains do not imply common determinants of typicality" (p. Wittgenstein used the term family resemblance to refer to the degree to which a member resembles this overall pattern. These scores are given in Table 5.94 3.34 to . This idea has been discussed recently by Mascolo and Mancuso (1990). Family resemblance is a weighted average of features shared with other subcategories.53 3. Substitutability ratings were made on a scale ranging from 1 fyery peculiar) to 6 (very natural sounding).36 3. Barsalou (1987) found that family resemblance did not predict typicality for goal-derived concepts. the least prototypical type of love we included.90 4.84 to . prototypicality of an emotion may be the salience of associated bodily reaction. . the scores are means computed across 20 sentences.99 3.63 3. Fehr & Russell. Correlations between mean naturalness ratings for the subtypes of love and their prototypicality ratings obtained in Study 2 are shown for each of the 20 sentence frames in Table 4.94 for various measures of internal structure and for various natural object categories.69 4.13 3. features of love were taken from Fehr's (1988) list. such as terror. Rosch and Mervis (1975) found that more prototypical members shared more features with each other and had fewer features in common with members of neighboring categories. the most prototypical type of love. four of the textbook sentences yielded correlations that were negative. with the weight taken as the number of members sharing that attribute. yielded a sentence that sounded peculiar (M = \ . Family Resemblance.95 4.62 4. 1984).10. prototypicality of an emotion may be simply its intensity. Second. attributes that may be more difficult to articulate than the attributes of everyday objects. The range of correlations was similar to that found with the same task for other concepts (e.20).432 BEVERLEY FEHR AND JAMES A. Third. with correlations ranging from .40 4. received the highest mean naturalness rating (5. In contrast. The resulting correlations were generally in the expected positive direction. Study 6): This is a study on the characteristics and attributes that people associate with different kinds of emotion.89 4. This index of family resemblance was highly correlated with other measures of internal structure (Rosch & Mervis. the possibility remains that the internal structure of emotion depends on factors other than family resemblance.10). prototypicality may be related to valence. but were stronger for some sentences than for others. a "substitutability" score was computed by averaging across the 20 sentences the naturalness ratings that each subtype of love received.94 4. Consider "Love is merely an intense form of liking. facial expression.45). 130). and Common Features Scores for 20 Types of Love Type of love Maternal love Parental love Friendship Sisterly love Romantic love Brotherly love Familial love Sibling love Affection love Committed love Love for humanity Spiritual love Passionate love Platonic love Self-love Sexual love Patriotic love Love of work Puppy love Infatuation Substitutability 3.48 4.85 for various measures of internal structure.95). you might write: —possible danger occurs (may be real. received the lowest (2. the correlations were lower. or posture. Study 6 was designed to gather further evidence on the relationship of prototypicality to family resemblance. RUSSELL sounded natural (M = 4. 1984. First. For example.. sentences taken from psychology textbooks showed little if any relation between naturalness and prototypicality." Substituting a peripheral type of love yielded a more natural sounding sentence than did substituting a prototypical type: Puppy love.35 3. like a bear. In fact. This pattern was captured by calculating an index of famiry resemblance as a weighted count of shared features.06 3. 1984). The sentences constructed by students yielded correlations between naturalness and prototypicality that were generally high. like a ghost) —heart beats quickly —eyes open wider —eyebrows lift —palms and soles sweat —thoughts race through the person's mind —unpleasant sensations are experienced —the person runs as fast as he or she can —hands tremble Note. Family resemblance 135 144 145 122 109 143 121 129 101 124 61 67 85 100 49 65 74 61 41 41 Common features 13 17 17 12 12 18 12 13 10 17 6 7 10 11 7 10 7 8 3 5 Method Subjects (N = 40) read the following instructions (adapted from Fehr & Russell. Table 5 Substitutability. members of a category are linked through a complex pattern of criss-crossing and overlapping attributes. infatuation. Although a similar result was found in the domain of emotion (Fehr & Russell.g. whereas substitution of a peripheral subtype. whereas maternal love.91 3. may be imaginary. The reason for lower correlations in the domain of emotion is not known. Common features refers to the number of features the subcategory of love shared with love itself. average r = . In their research on the relationship of basic-level to superordinate-level categories. Although no single attribute may be shared by all. Family resemblance scores for emotions were based on what subjects listed as the attributes of emotional states. average r = . Tiller and Harris (1984) suggested three alternative determinants of typicality in the domain of emotion.68 Study 6: Family Resemblances Why are some members in a category considered more prototypical than others? Rosch and Mervis (1975) argued that Wittgenstein's (1953) idea of famiry resemblance provides an answer.53 3.

More generally. Family resemblance scores for prototypical subtypes were higher for love than for peripheral subtypes." listed only for brotherly love. how readily it can be substituted in sentences about love without their sounding unnatural. which thus had been listed by two or more subjects for at least one subtype. and . a weight of 14 was given to "caring" because it was listed as an attribute for 14 of the 20 subtypes. Convergence of Measures of Internal Structure This series of studies produced eight operations thought to correlate with internal structure: (a) frequency in a free listing . (f) substitutability into sentences about love (Study 5). and closeness (7). Each of the 205 attributes was then weighted by the number of subtypes for which it had been listed. respect (8). /(18) = 5.35 for emotions (Fehr & Russell. But emphasize a description of how one feels and acts. As expected. From a prototype perspective. in fact. The next most common attributes and their weights were helping (12). In other words. (e) probability of membership (Study 4). Subjects were given as much time as they wished to list the attributes for each of 10 subtypes of love presented in random order. Finally. although low. When thinking about each type of love. 60). Fehr and Russell (1984) had found family resemblance scores to be the weakest correlate of internal structure for emotion. feel free to talk about anything (9). attributes mentioned by only one subject were eliminated. a subtype's rank in this internal structure predicts its availability in a free listing task. you might ask yourself: What manifestations are there of it? What thoughts do you have about it? How do you show it? In what circumstances are you apt to be aware of it? It might help to imagine that you're explaining the meaning of each type of love to a foreigner or to someone who has never experienced it. Psychologists have attempted to specify one or two necessary and sufficient features for love. Try not just to free associate. but in different kinds of love. We first outline the differences and then. in which only identical or synonymous responses are combined. We also expected that more prototypical subtypes would bear a greater family resemblance to the general category love. The basic-level concept of love has an internal structure: Subtypes of love can be reliably ordered from better to poorer examples of love. In fact. for each target subtype. Judges' coding of the remaining attributes yielded a final list of 205 distinct attributes. This. who may also be the object of sexual desire" (p. All values were in the predicted direction. prototypical subtypes had more features in common with love than did peripheral subtypes. To consider just one example. "playfighting. Results and Discussion The total number of responses generated was 2. and those features had earlier been coded with the same procedure used here. In turn. If a particular kind of love makes you think of someone you know. The last finding is especially important in establishing that the concept of love lacks precise boundaries and therefore speaks strongly against the hypothesis that love is classically defined. Creating family resemblance scores required a decision as to which of these responses represented the same attribute and which represented different attributes. the speed with which it is verified as a type of love.37 for features of wisdom (Holliday & Chandler. for each type of love a family resemblance score was calculated as the sum of the weighted attribute scores listed for that type of love. Family resemblance cannot be put to rest as a possible mediator of typicality effects in the domain of emotion. Each of these variables is influenced by factors other than internal structure. and (h) number of features shared with love (Study 6). The correlations involving frequency of free listing. bond (11). A weight of 1 was given to attributes that were listed (by two or more subjects) for only one subtype of love: "act immaturely. The results are also shown in Table 5. . and therefore internal structure is best demonstrated by a convergence of these operations. (g) family resemblance score (Study 6).63. We tabulated the number of features each of the 20 types of love had in common with the features that had been listed for love.001.001. all operations except frequency of free listing were highly intercorrelated.35 for features of love (Fehr. (b) direct ratings of prototypicality (Study 2). (d) probability of membership (Study 3). For example. and "worrying. and the probability of its being considered a genuine type of love. family resemblance was one of the best. Scientific Definitions and Typologies of Love Contrasts of Folk and Scientific Concepts Psychologists have offered various definitions and typologies that attempt to specify what is to be included under the term love." listed only for puppy love. Which 10 of the 20 target subtypes were presented varied from one subject to the next such that each subtype of love was described by 20 subjects. Tell how it comes about and what happens after. We're interested in what is common to instances of love. 1985).31. These scores appear in Table 5.CONCEPT OF LOVE —relief is experienced after a few minutes 433 Here we are not interested in different kinds of fear. in the final section. The criterion adopted was the conservative procedure used by Fehr (1988). maternal love would have more features in common with love than would puppy love. Correlations between the eight operations. Next. we discuss what should be the relationship between folk and scientific concepts. sharing (10). calculated across types of love. Psychologists' offerings appear to conflict with the present results.333. 1986). The pattern of results obtained here was the one predicted from prototype theory. /(18) = 8. task (Study 1). p < . p< . Three judges coded the responses. In contrast.19 for features of persons in different occupations (Dahlgren. the extent to which it resembles other types of love. rated prototypicality and frequency of free listing correlate ." listed only for maternal love. most reliably so. 1988). Our results therefore failed to support the suggestion that basic-level categories are classically defined. these results provide an alternative account of how a person can use and understand the concept of love without knowing necessary and sufficient features for it. It was possible to test this prediction because Fehr (1988) had asked subjects to list the features of love. So include the obvious. or person. understanding (9). Johnson-Laird and Oatiey (1989) recently argued that love can be defined as follows: to love is "to experience internal happiness in relation to an object. in the present study. For example. . are shown in Table 6. don't write down the person's name. was the highest weight obtained. 1984). were similar to those obtained in other studies. (c) reaction time to verify membership (hypothesized to correlate negatively with internal structure) (Study 3).

Fehr. desire. were hypothesized to be positive.76 -. and commitment. Kirson.79 . The characteristics shared by most but not all types of love were caring. five. three. consummate (intimacy plus passion plus commitment). attraction. Thus. 7. Similarly. understanding. 6. helping. fondness. Schwartz. passionate love. 8. But doing so might impose order and boundaries where they did not before exist. 5. empty (commitment present). say. companionate (intimacy plus commitment). liking. Correlations were calculated across the 20 target types of love listed in Table 2. two. took this result as a model of the lay typology of love. by definition. Three types of love were described by Kelley (1983): passionate. • Not significant at a = . establishing a bond. would be a contradiction in terms. brotherly love. one. based on whether each person is high or low in power and in status: romantic love. Of course. then. indeed. and therefore several clusters could be created within this set. One could argue. Shaver et al. based on the pres- ence or absence of each of three components: intimacy. Our subjects did not list happiness as one of the features of most types of love. manic (possessive. The longing cluster consists only of the word longing. passion. This cluster resembles Hatfield and Walster's view of companionate love. lust. ludic (game-playing love).41 — . puppy love. and therefore they could all be included in one cluster. in fact. Two types were also described by Maslow(1955): D-love (deficiency love) and B-love (being love). . and sentimentality. infidelity. different types of love. 3).434 BEVERLEY FEHR AND JAMES A. and longing. and agapic (altruistic love). In contrast.47 . are recognized by many social psychologists (Hatfield & Walster.45 . fatuous (passion plus commitment). they differ in number. In other words. and therefore six types could be created. of the overlapping and cross-cutting features of similarity in the domain of love. affection. storgic (companionate love). respect. our subjects listed a great many different types of love (just as laypeople listed many more features of love than have scientists.63 — . Three types were listed by Shaver et al. and closeness. the folk typology of love is more organized 2 Shaver.48 .97 -. but no single clustering would capture all. pragmatic. and parent-infant love. 2. lust. happiness is likely to be only one of many features that characterize love—and perhaps not the most characteristic feature. tenderness. Johnson-Laird and Oatley's definition also implies that if love is experienced. and O'Connor (1987) reported a hierarchical cluster analysis of 135 emotion terms that yielded three types of love: affection. either. love is experienced. erotic love. motherly love. and love of God. romantic (intimacy plus passion).86 . four. Consider just six items from our list: infatuation. sharing.35" — -. lust.05.48 — . the types of love. 3. except for those involving consensus of membership from Study 4.50 . the typologies of love advocated by psychologists appear to conflict with the present results. lust. Undoubtedly. dependent love). companionate and passionate.2 Five types of love were described by Fromm (1956): brotherly love. would be a contradiction in terms. passion. Scientists have typically listed a small number of types of love. All correlations. 4. Eight types of love were described by Sternberg (1986). Each such clustering would capture some. infatuation. a loveless but happy relationship.70 -.42' .24" -.44 — . rather than allowed to generate. (1987): affection. that the list in Table 1 could be subjected to cluster analysis and thereby organized into a much smaller number of clusters. The resulting types of love are liking (intimacy present). Two types of love. compassion. as in. for instance. Seven types of love were described by Kemper (1978). The number of clusters appears to be arbitrary (a notion consistent with the varying numbers of types of love plausibly delineated by psychologists). self-love. it could." Johnson-Laird and Oatley's definition implies that if internal happiness in relation to an object or person is experienced. but this may not be so because their subjects were provided with. Johnson-Laird and Oatley's (1989) definition fails to capture the ordinary language concept of love. How might these be clustered? No two are identical. adulation by fans. pragmatic (logical. 1978).93 — Note. sentences about love sound peculiar when JohnsonLaird and Oatley's definition is substituted: The command "love thy neighbor" would thus translate as "experience internal happiness in relation to thy neighbor. except for those involving reaction time (No.64 .49 -. love at first sight. All contrast with companionate love. as in. then happiness must be experienced.49 . practical love). and altruistic. Thus. whereas B-love is based on autonomy and giving of oneself to the other. The affection cluster consisted of adoration. infatuation (passion present). and romantic love.56 .47 . love.63 . feeling free to talk. The lust cluster consisted of arousal. one could question whether many of the types of love listed in Table 1 are. But some are more similar than others. and longing. and nonlove (the absence of the components). unrequited love. Six styles of love were described by Lee (1977): erotic (passionate love). In short. a pleasant business association. RUSSELL Table 6 Correlations Between Measures of Internal Structure Measure 1. an unhappy love relationship. D-love is based on need and dependence. or six clusters could reasonably be formed.50 .55 .85 — . 1988). In other words. charismatic or discipleship love. and infatuation and resembles passionate love. And. caring. First. which were calculated across 11 types.86 -.32' . say. Frequency of free listing Prototypicality rating Reaction time Agreement on membership Probability of membership in love Substitutability Family resemblance Shared features with love 1 — .

sisterly. discrepancies between scientific and lay conceptions demonstrate serious shortcomings in the scientific account) A middle position between these extremes might be that scientific analysis correctly depends on everyday concepts. where five love objects are specified: brother. infidelity. fatuous love. and nonlove. We would suggest that love is organized around several prototypes: love of a parent for a child. (From this viewpoint. love between old friends. but other responses might more accurately be described as features of love. some types of love described by social scientists were not mentioned by our subjects: Maslow's (1955) deficiency love and being love. For example. love for humanity. and adulation by fans. But prototypes of love are not equivalent to a typology with a fixed number of types. Although some experts have mentioned types of love that were not mentioned by laypersons. and trust. and love for humanity in terms of love for one's brother. 1988)." Subjects found that this sentence makes sense for romantic love but little sense for other. (From this viewpoint. whereas psychologists have looked for one or two defining features of love.) In this section. self.CONCEPT OF LOVE than is implied by a list. Psychology of Love What should be the relation between the common concept of love and scientists' definitions and typologies of love? Different answers to this question could be proposed. cases of love: parental. In much of the research on love. Our subjects. the emphasis is different. 1988]). they do suggest that psychologists. and Sternberg's (1986) empty love. included selflove and love of God—most scientific typologies exclude them. Although these 10 are not a systematic sample. love between siblings. and decision/commitment. and so on. sex partner. therefore. For example. Subjects listed maternal. have drifted far from ordinary usage. Whereas experts focus on love between heterosexual adults. mother. For example. discrepancies between scientific and lay conceptions are irrelevant to science. sharing. Our subjects included many types of love that are not shared between two human beings: love of pets. (Lee [1988] saw his styles of love as applicable to homosexual relationships as well) Kelley's (1983) descriptions of passionate. in their use of the word love. although not so much in the familial sense. giving. the scientific study of love is the study of the concept of love. Hendrick and Hendrick (1986) designed scales to measure Lee's (1977) various love styles in which subjects are instructed to complete the scale with their current spouse or dating partner in mind (a past partner if they are not currently dating). but scientists cannot stray far. Other evidence was consistent with this hypothesis. were also given as features of love (Fehr. honesty. Psychologists' typologies can be divided into two types: those in which the divisions are specified by abstract features and those in which the divisions are specified by the type of love object. and God. we attempt to integrate them from a broader perspective. for example. An example of the second type is Fromm's. love between romantic partners. but meaning love for one's fellow human beings. 1956) and parent-infant love 435 (Kemper. related experiences. most are more likely peripheral types of love that do not come readily to mind. 1978) in their typologies. which are precipitates of accumulated wisdom but can tidy up. mentioned various forms of romantic love. Although some psychologists have written about a few of these types—Fromm. For example. the folk definition of love is complex and provides no sharp boundary between love and other. discrepancies can be tolerated. laypeople focus on less romantic forms. Further. We believe that none of these positions is adequate. but overall put less emphasis on it. (Features associated with companionate love were rated as more prototypical of love than were features associated with passionate love [Fehr. Many of these were listed frequently and also received among the highest prototypicality ratings. love of food. the folk typology seems to freely mix these two forms. Our subjects also put more emphasis on familial love than on romantic love. and altruistic love are framed largely in terms of adult heterosexual relationships.) At the other extreme is the position that the everyday concept of love is essential to the experience of love. Recall the 10 sentences about love (Table 4) taken from psychology textbooks. (From this viewpoint. spiritual love. love of sports. filial. love of God is often thought of in terms of love of a parent. In contrast. this was a much rarer occurrence. The folk concept of love also differs from psychologists' concept in focus. parental. Kemper's (1978) discipleship love. Scientists doing research on love are probably subscribing to a much narrower concept of love than are their subjects. Prototypicality is graded. At one extreme would be the position that science should ignore completely ordinary language concepts. especially love of the boy-meets-girl variety. love of money. intimacy. and brotherly love and love of grandparents. passion. pragmatic. of course. Lee's (1977) ludic (game-playing) love and pragmatic love. organize. we hope to contribute to an understanding of these issues. love of God. love of life. To summarize. consummate love. Conversely. all given here as types of love. subjects are asked to respond with reference to a romantic partner. even in love between two human beings. Other subtypes of love then resemble the prototypes to varying degrees. Scientific definitions and typologies of love have been generally aimed simultaneously at two goals: (a) to capture the mean- . Companionate love (friendship) was the more frequently mentioned type and was rated as more prototypical than was passionate love. love of art. The folk typology of love is more encompassing than most experts' typologies. Rather than viewing them as competing alternatives. even prototypical. and sibling love. love of beauty. Sentence 16 asserted that love has "both physiological and psychological components. as are many of Sternberg's (1986) types of love. including love between family members and many kinds of noninterpersonal types of love. and more prototypes would have to be added as more cases are considered. and improve these concepts. An example of the former is Steinberg's. Psychologists generally only consider love between two human beings. Some of the subjects' responses were objects of love. love of nature. Fromm and Kemper also considered brotherly love as a type of love. and. where types of love are based on the presence or absence of three components of his triangular theory: intimacy. although several mentioned maternal love (Fromm. Social psychologists have neglected familial love. Although these may represent cases laypersons would consider outside the domain. patriotic love.

As Medin (1989) stated. . Because prototype theory concerns concepts. Gleitman. First. it has no direct bearing on a prescriptive analysis. 11). So interpreted. Love. one must consider carefully the observation of Solomon (1977) on the cultural and historical specificity of the concept of love: "The French. The analogy here is to the concepts of witch and devil. Those phenomena are a worthy and legitimate topic in the psychology of emotion. Similarly. inspired by it.. analysis of the concept of love can help free scientists from hidden assumptions and confusion. R. L. set aside the question of whether a particular definition or typology coincides with the everyday concept of love. L. a man questions whether his spouse loves him. vehicle. Our suggestion therefore is to view the classical definitions and typologies offered by various theorists as prescriptive. If so. The prototype analysis offered in this article is a descriptive analysis of the everyday concept of love. a descriptive analysis of love and emotion would aim to clarify their role in today's thought. ing of love as everyone understands it and (b) to provide a conceptual framework for scientific study of the phenomena referred to by the word love. the hypothesis that things that look alike tend to share deeper properties" (pp. a husband loves his wife. Thus. Scientific psychology cannot rely on folk psychology any more than scientific physics or chemistry relies on folk physics and folk chemistry. & Gleitman. but this assumption may be just what is hindering scientific progress. This particular set of diverse experiences is called love by an historical process. In this case. So. "Despite the overwhelming evidence against the classical view." For purposes of a prescriptive analysis. in stating the topic of the prescriptive analysis we assumed that the phenomena to be analyzed were only those referred to by the word love. According to the classical theory they do. A prescriptive analysis concerns those events referred to by the word love: A mother loves her child. Prototype theory is as applicable to concepts such as witch or devil and unicorn or centaur as it is to emotion. 281). The historical evolution of the concept of love may be such a chain. love. or whether they coincide with the cases selected by Johnson-Laird and Oatley. is a worthy and legitimate topic in the psychology of social cognition.. Third. 263-308. From the classical view. not the event. The two goals may require distinct analyses. a teenager wonders whether what she is feeling is true love or just a passing fancy. of love: A mother worries whether she loves her children enough. The second analysis. Nevertheless. the distinction between concepts and events must not get lost. these two goals could be achieved simultaneously: The everyday concept of love already has scientifically desirable features of precise boundaries and common features. 13. (1983). or love. a prescriptive one. the scientist would assume neither the truth nor the falsity of folk conceptualizations. S. aims to describe the everyday conception of love. for these are all equal in their status as concepts. References Armstrong. a descriptive analysis. Scientific analysis could decide to draw different boundaries. it is irrelevant that this definition fails to capture the everyday concept of love. folk concepts and beliefs can provide hypotheses to be tested. and a miser loves his money. That is. therefore. Why are these experiences all called by the same name— love? Prototype theory asks one to consider an alternative to the classical answer that they all share a common essence. 1476-1477). Attempts to tidy up or improve these concepts would be inappropriate. one possibility is that the ordinary concept of love plays a causal role in events of love. but focus on this set of experiences. Many believe that an unloved child is doomed—courts have thus declared that "to be unloved is so debilitating a disease no person subject to this experience should be required to use the universal knowledge that maliciousness is wrong and the human ability to control malevolence" (Kagan. . take the definition as delimiting precisely a range of phenomena that may be explained by one set of scientific hypotheses. set aside the question of whether all and only such experiences are properly labeled love. as a concept. To tidy up or improve the concept of love would defeat the purpose of a descriptive analysis. prototype theory suggests that these cases need share no more than a resemblance to one another. To illustrate. RUSSELL Second. which played significant roles in the lives and deaths of people of the 17th century. There is a tendency to assume that all cases called by the same name. Set aside the question of whether all and only these experiences are properly labeled love. These worries and wonders and beliefs are tied to the everyday concept of love. One. Next. Then test those hypotheses through empirical means. let us return to Johnson-Laird and Oatley's definition—"to experience internal happiness in relation to an object or person. If the concept of love as commonly understood lacks precise boundaries. share certain essential features. . where each element shares some features with its neighbor. prototype theory can play an indirect role. The natural language concept of love guides people's official and unofficial interpretation of some of life's major and minor events. namely. but no common principle characterizes the entire set.436 BEVERLEY FEHR AND JAMES A. for example.. A descriptive analysis concerns the concept. people adopt an essentialist heuristic. What some concepts might not be. not events. and of course they might. each definition might be a first step in a scientifically useful analysis of a set of phenomena that do share enough features in common that they can be analyzed together. 1989. Vygotsky (1962) described some concepts as chains. For example. have many distinctions between the various forms of intimacy that we more cold-blooded Anglicans clumsily summarize with the clearly inadequate concepts of'love'and'like'" (p. p. Instead. Nevertheless. aims to prescribe a conceptualization of those phenomena referred to by the word love. and that concept must be understood —as it is. A descriptive analysis of witch and devil would clarify their role in the thought of the time. This assumption is challenged by prototype theory and by the empirical findings. H. consider Freud's definition of love as frustrated desire. then its scientific status must be reexamined. for the concept would be a causal antecedent to the event. such as those reported in this study. whatever their merits or demerits from a scientific perspective. Cognition. Nevertheless. then a prescriptive analysis requires an adequate descriptive analysis. A prescriptive analysis seeks to understand such phenomena. scientists and nonscientists alike have taken the classical view for granted. No classical definition or rigid taxonomy will provide an accurate descriptive analysis.

Davis. B.). M. M. Fehr. (1977).. R. DC: Georgetown University Press. On falling in love in conformance with the romantic ideal. R. In J. Assessment and modification of emotional behavior (pp. & J. Rule. Freud. Harnad (Ed.. C. 113. Annual Review of Psychology. & D. 12. R (1972). 11. Annual Review of Psychology. J. Lloyd (Eds. University of British Columbia. 15. In U. C. S.). M. 44. In S. Wisdom: Explorations in adult competence (Human Development Monograph No. The concept of intelligence. Izard.. 101-140). D. Journal of Personality and Social Psychology.105-126. J. A. K.181-193. Levinger. In R. (1980). (1989). A. 17-38). P. S. 1.). Kahneman.. T. & Russell. Barsalou. & Bekerian. 379-398. L. Cognitive Science. Rosch & B. Federal Republic of Germany: Springer-Verlag. J. S. 535-565). Prototype analysis of the concepts of love and commitment. H. 17). Norm theory: Comparing reality to its alternatives. Journal of Abnormal Psychology. fire. D. & Mezzich. Journal of Personality and Social Psychology. Pliner. W (1929). J. & Slapion-Foote. Cambridge. E. Maslow. In C. Personality and Social Psychology Bulletin. S. Jones (Ed. M. (1985). New York: Dover. Anger and aggression: An essay on emotion. E. Harvey. & Ward. (1986). Advances in descriptive psychology (Vol. In J. 9. J. W (1979). 392-402. Lee. A. Chicago: University of Chicago Press. Ortony. (1989). and trust in heterosexual relationships. 81-123. paradigm cases and relationship description. 93.13341346. F. Close relationships (pp. (1984). Concepts and concept formation. 341-364.. Conway.). Switzerland: Carger Publishing. Springfield. R. Hendrick. 45-77. Friendship and love relationships. 437 Hatneld.136-153. N. & Smith. M. Cambridge. Prototypes in person perception. McClintock. Kagan. L. New York: Random House. M. Love and commitment. Psychological Review. In K. A. W (1987). (1987). Category induction and representation. 79-122). (1980). Nebraska Symposium on Motivation (pp. 3. (1987).217-227.. K. Deficiency motivation and growth motivation. G. Mervis. W (1978). Psychiatric diagnosis as prototype categorization. Averill. Concepts and conceptual structure. E.). Universal and cultural differences in facial expression of emotion. Theory and method of love. J. and dangerous things: What categories reveal about the mind. Blankstein. H. The varieties of religious experience: A study in human nature. The act frequency approach to personality. Cole (Ed. Human emotions. 464-486. E. C. 557-579. 89-115. Cognition and Emotion. G. Cantor. (Original work published 1902) James. (1983). A. Cognitive Psychology. R. 49-71). (1984). Concepts and conceptual development (pp. 55. Newport. & Miller. 1.145-191. Intelligence. R. Freeman. J. Neisser. Peterson (Eds. Mascolo. Mischel. Bailey & R. Kelley.1469-1481. M. & Todd. Cognitive Science. Ed. Love. The forbidden colors of love: Patterns of gay love and gay liberation. DeCecco (Ed. CT: JAI Press. Buss. Unstable ideas. Journal of Experimental Psychology: General. F. & Bellugi. Fehr. K. Cambridge.173-182.. Centers. & Boothroyd. James. (1978). B. French. Cambridge. The art of loving. (1977). R. Beverly Hills. (Original work published 1890) Johnson-Laird. (1985). D. E. W (1950). D. CT: Yale University Press. L. Understanding personal relationships: An interdisciplinary approach (pp. 233-243. 1-39). & Schwartz. American Psychologist. A. 1971 (pp. Basel. N. R.). In H.187190. (1979). E. K. A social interactional theory of emotions (pp. A. 3-52. Kemper. Medin. B. Neisser. Lincoln: University of Nebraska Press. In E. A prototype analysis of psychological situations. Clore.). & Dion. MA: Addison-Wesley. Concepts and conceptual development. Lakoff. New \brk: Plenum Press. New \fork: Plenum Press. (1982).). Sternberg & M. liking. E. (1982). D. Ecological and intellectual factors in categorization. Motivation and Emotion. G. Manuscript submitted for publication. (1956). Fehr. (1984).). 283-309). pp. & Rosch. E. W Shuy (Eds. P. (1982). N. Polivy (Eds. Women. U. Green & Co. U.. Kelley. C. T. Greenwich. H.. A. New York: Liveright. G. MA: Harvard University Press. Russell. A. G. E. (1982). A. (1989). 340-373).. The structure and content of emotion knowledge: A prototype approach. J. Holliday. New York: Harrington Park Press. and Trans. 265-314). (1975).. (1988). & Foss. The principles of psychology. Barnes (Eds. Categorical perception (pp. (1983). (1951). 89. (1988). Foote. . Berlin. Washington. Berscheid. R. S. 20. Harnad.). & Chandler. Assessing friendship: Prototypes. D. K. (1976). Cantor. CA: Sage. Dahlgren. Concept of emotion viewed from a prototype perspective. Canada. Broughton. Davis & T.113-138. 3. Lee. Davis.. & Mischel. IL: Charles C Thomas. C. G. (1989). A. 235-247. Prototype categorization of emotion. B. J. (1984). (1987). Psychological Review. L. Vancouver. 1. A. Averill. M.. & Oatley. E. H.CONCEPT OF LOVE Averill. Cantor. (Original work published 1922) Fromm. (1981).. New ways of analyzing variation in English (pp. 7-45). (1987). 90. Perlman (Eds. L. Lysak. Labov.. N. K. Fehr. S. 253-254. Brehm. Hendrick. Hillsdale. B. E. 207-283) Lincoln: University of Nebraska Press. K. Intimate relationships. J. C. L. 14. & Hendrick. In S. 2. In K. 2. NJ: Erlbaum. J..). E. (1985). (1987). The psychology of love (pp. L.. & Todd. (1986). (1990). The referential structure ofthe affective lexicon. Smith. A prototype strategy for construction of personality scales. (1982). The instability of graded structure: Implications for the nature of concepts. Ekman. L. Cognition and categorization (pp. (1955). D. New York: W H. Bulletin of the Psychonomic Society. Linguistic expression of category levels in a visual-gestural language. Christensen. J. H.177-195. 11-32). 47. The boundaries of words and their meanings. D. Cognition and Emotion. E. B. W. A new look at love.). R. Buss. H. On the paucity of positive emotions. Peplau.. Advances in Experimental Social Psychology. J. Unpublished master's thesis. (1978). (1977). & Dobbs. Neisser (Ed.). Journal of Personality and Social Psychology. H. (1986). R. 35. Personality and Social Psychology Bulletin. Prototypicality of emotions: A reaction time study. R. New York: Wiley. & Mancuso. New Haven. 3. J. In M. M. The evolutionary biology of love. Situational knowledge and emotions. England: Cambridge University Press. U. Categorization of natural objects. Dion. Gay relationships (pp. P. L. Medin. Huston. W (1973). 32. M. The language of emotions: An analysis of a semantic field. Reading. 100-118). England: Cambridge University Press. S.. Hendrick. (1988). New York: Longmans. J. Conceptions of aggression: Prototype or defining features? Personality and Social Psychology Bulletin. Strachey. H. Nebraska Symposium on Motivation. L. N. New York: Harper & Row. 50. J... & Walster. England: Cambridge University Press. Duck & D. B. Sexual attraction and love. D. Mitchell (Eds. T. E. A typology of styles of loving. & Craik. Do men and women love differently? Journal of Social and Personal Relationships. P. J. Group psychology and the analysis ofthe ego (J. K. The cognitive structure of social categories..

16. A. Watson... (1953). In Psychology from the standpoint of a behaviorist (2nd ed. RUSSELL Skolnick. D. 111-144). Affect theory. D. Categories of environmental scenes. (1975). J.). Oxford University. R. 8. L. E. Christensen. W Fogel. & Bullock.. In E. B. & E. E. & O'Connor. Aging: Stability and change in the family (pp. In defense of a prototype approach to emotion concepts. (1988).1061-1086. A. (1970). MA: Harvard University Press. T. The behaviors of love. Peterson (Eds. England. J. J.). & Smith. P. J. (1987). The passions. L.. Shanas (Eds. Journal of Personality and Social Psychology. (1972). Vygotsky. P. E. New York: Simon & Schuster. & Medin. Cognitive development and the acquisition oj"language (pp. E. Swensen. 7.. D. H. MA: MIT Press. Kelley. K. B. Unpublished manuscript. Cognitive Psychology. W D. In T. pp. In H. (1975). 194-230). Emotion knowledge: Further explorations of a prototype approach.. E.. Rosch. (1981). 265-273.). About love. Prototypicality of emotion concepts: A discussion of normative data. E.192-233. In K. Love today (pp. In R. B. CA: Academic Press. Journal of Experimental Psychology: General.). L. (1924). & D. (1991). Tomkins. Hereditary modes of response: Emotion. Warren (Ed. Human categorization. 104. On the adequacy of prototype theory as a theory of concepts. Hatfield. (1978). H. NJ: Erlbaum. Philosophical investigations. Garden City. A. Huston.. E. (1983). Journal of Personality and Social Psychology. M. CA: Academic Press. B. Otto (Ed. R. Oxford. San Diego. Lloyd (Eds. Cognition. E. S. Wittgenstein. Hillsdale. Kirson. On the internal structure of perceptual and semantic categories. New York: Association Press. (1986). Cognitive Psychology. Social Cognition. Schwartz. Harvey. 1. 15. Ekman (Eds. In N. Hillsdale. C. H. & Boyes-Braem. 1-49). K. Boston: Little. Rosch.. (1977). (1983). (1978). P. Russell. E. Cambridge. Family resemblances in the internal structure of categories. New York: W H. (1962). 60. Smith. E. 35-58. L. Scherer & P. Cognitive representations of semantic categories. L. Close relationships (pp. C. C. Solomon. Moore (Ed. CA: Academic Press.1990 Accepted June 26. L. Tiller. S. (1976). L. Peplau. The intimate environment: Exploring marriage and the family (2nd ed. 37-47. & Mervis. NJ: Erlbaum. & Hemenway. Principles of categorization. R. R. McClintock. B. 52. M. 573-605. 27-71). 382-439. Roles and gender.119-135. D. E. Rosch. Thought and language. 309-341. Received July 31.). (1981). Basic objects in natural categories.121-149.). Gray. & Hatfield. Freeman and Company. Psychological Review. EX N. 253-274). Love and its effect on mental and physical health. J. NY: Anchor Press. Russell. Brown. & Harris. Approaches to emotion. A.. A. E. Jr. Sternberg. Osherson. (1973). San Diego.. Rosch. (1986).438 BEVERLEY FEHR AND JAMES A. A triangular theory of love.1990 • . Z. Rosch & B. Traupmann.). New York: Macmillan. Tversky. Solomon. 9. Philadelphia: Lippincott.1989 Revision received March 30. Rosch.. J. 220-264). R. Shaver.. Mervis. Rosch. E. pp. S. C. Cognitive Psychology. E. In A. San Diego. 86-101).). (1984). 4. Measurement of romantic love. (1977). C. C. E. (1984). Johnson. A. Studies in cross-cultural psychology (Vol. Levinger. Fuzzy concepts and the perception of emotion in facial expressions. G. Journal of Personality and Social Psychology. Berscheid. J. 93. Cambridge. Peplau. Categories and concepts. Rubin. Cognition and categorization (pp. H. (1981).