Learning Center
Plans & pricing Sign in
Sign Out
Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>



									Journal of Personality and Social Psychology                                                              Copyright 1991 by the American Psychological Association, Inc.
1991, Vol. 60, No. 3, 425-438                                                                                                                      0022-3514/91/$3.00

                                                      The Concept of Love
                                               Viewed From a Prototype Perspective

                                 Beverley Fehr                                                               James A. Russell
                         University of Winnipeg                                                       University of British Columbia
                       Winnipeg, Manitoba, Canada                                                  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 neces-
                              sary 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 process-
                              ing. Implications for a scientific definition and typology of love are discussed.

   Prototype theory has provided insights into concepts central                                             Prototype Approach
to psychology, including behavioral act (Buss & Craik, 1983),
personality trait (Cantor & Mischel, 1979), intelligence (Neisser,                        Traditionally, the general terms of a language were thought to
1979), social situation (Cantor, Mischel, & Schwartz, 1982), and                      denote categories of objects or events, each member of which
environmental setting (Tversky & Hemenway, 1983). The theory                          possessed features that were each necessary and together suffi-
has inspired important new approaches to psychiatric diagno-                          cient to define membership in that category. According to this
sis (Cantor, Smith, French, & Mezzich, 1980) and personality                          classical view, to know the meaning (sense) of a general term
assessment (Broughton, 1984). The studies reported in this arti-                      (i.e., to have the concept associated with it) was to know at least
cle are part of a larger project designed to explore the applicabil-                  implicitly these necessary and sufficient features. A definition
ity of prototype theory to the domain of emotion concepts.                            could therefore be formulated by philosophical discussion or—
   We had two related purposes. The first was to use the tools of                     because the defining features were also in the objects or events
prototype theory to throw light on the elusive concept of love.                       —by empirical investigation.
We shall have something to say about both the definition of and                           Although some writers continue to defend the classical view
typologies of love. The second purpose concerned prototype                            (Armstrong, Gleitman, & Gleitman, 1983; Harnad, 1987;
theory itself. Prior demonstrations of a prototype structure in                       Osherson & Smith, 1981), considerable psychological research
concepts focused on superordinate concepts, such as fruit, vehi-                      reinforces a growing skepticism over its plausibility as an ade-
cle, or emotion. Would the same results occur, we asked, for                          quate account of most concepts used in everyday speech and
basic-level concepts, such as apple, truck, or love? We shall                         thought (Medin, 1989; Mervis & Rosch, 1981; Smith & Medin,
begin with the second issue.                                                          1981). Rosch's (1975,1977,1978) proposal of a prototype ac-
                                                                                      count as an alternative to the classical view was followed by
                                                                                      various nonclassical accounts (Kahneman & Miller, 1986; La-
                                                                                      koff, 1987; Medin, 1989; Neisser, 1987; Smith & Medin, 1981).
  Portions of this research were presented at the Annual Convention                   Indeed, in one account, concepts are thought of not only as
of the Canadian Psychological Association, Montreal, Quebec, Can-                     ill-defined but as varying from one person to the next, and, for
ada, June 1988.                                                                       the same person, from one time to the next (Barsalou, 1987).
  This research was supported by a University of Winnipeg Group II                    We make no attempt in this article to differentiate within this
grant (4100-000) awarded to Beverley Fehr and by a grant from the                     family of nonclassical accounts. Rather, our purpose is to con-
Social Sciences and Humanities Research Council of Canada to James                    tinue to specify through empirical means the properties of
A. Russell.                                                                           emotion concepts.
  We thank Pat Keelan and Sandra Wolfram for their assistance in                          In previous research, we have compared the classical with the
data coding and Jim Averill, Michael Mascolo, and Phil Shaver for
                                                                                      prototype view in the domain of emotion (Fehr, 1982,1988;
helpful comments on a draft of this article.
  Correspondence concerning this article should be addressed to                       Fehr & Russell, 1984; Fehr, Russell, & Ward, 1982; Russell,
Beverley Fehr, Department of Psychology, University of Winnipeg,                      1991; Russell & Bullock, 1986), but we are not alone in this
Winnipeg, Manitoba, Canada, R3B 2E9, or James A. Russell, Depart-                     interest. Indeed, writers as far back as William James (1890/
ment of Psychology, University of British Columbia, 2136 West Mall,                   1950,1902/1929) have viewed emotion concepts in a way more
Vancouver, British Columbia, Canada V6T1Y7.                                           compatible with a prototype than a classical perspective. Averill

426                                          BEVERLEY FEHR AND JAMES A. RUSSELL

(1980,1982) has argued forcefully for viewing emotions as syn-         whether basic-level concepts are themselves prototypically or
dromes, no one feature of which is denning. Averill and                classically organized.
Boothroyd (1977) denned an ideal or paradigm case of roman-               Evidence of prototype structure found at one level of a hierar-
tic love and measured the degree to which actual cases resemble        chy does not show that concepts at other levels of even that same
it. Davis and Todd (1982,1985) drew explicitly on the idea of          hierarchy must be so structured. That fruit is organized around
prototypes when developing paradigm cases of friendship and            prototypes and has fuzzy borders does not show that apples and
love relationships. Other investigators of emotion have also           oranges are so organized. Newport and Bellugi (1978) analyzed
drawn on prototype theory (Conway & Bekerian, 1987; Mas-               American Sign Language, which sharply distinguishes superor-
colo & Mancuso, 1990; Shaver, Schwartz, Kirson, & O'Connor,            dinate-, basic-, and subordinate-level categories. Their analysis
1987; Tiller & Harris, 1984). The prototype analysis to be pre-        suggested that only superordinate concepts lack common fea-
sented here thus bears a family resemblance to other analyses of       tures shared by all members. At the basic level, features are
the emotion domain.                                                    shared by all or most members. At the subordinate level, fea-
   In applying the prototype perspective to the domain of emo-         tures are likewise shared by all or most members, but these
tion categories, it is helpful to begin by picturing that domain       same features are also shared with members of contrasting cate-
as an inclusion hierarchy. At the topmost, or superordinate,           gories.
level is the word emotion. At the middle level, emotion is di-            Lysak, Rule, and Dobbs (1989) offered another hypothesis in
vided into fear, love, happiness, and other prototypical and less      which the nature of concepts varies with level in the hierarchy.
prototypical emotions from anger to zest. This level is probably       They argued that the basic-level concept of aggression is more
basic in Rosch's sense. The number of categories at this level is      consistent with a classical than a prototype conceptualization.
indefinite because prototypes shade gradually into nonproto-           At the subordinate level, on the other hand, types of aggression
types, which shade gradually into nonemotions with no sharp            may be represented as prototypic. For example, having a fight
boundaries to be found. Many of the middle-level categories            with your wife or fighting in a hockey game may each show
may be further divisible, forming a subordinate level. Thus,           internal structure and fuzzy borders.1
anger is divided into wrath, rage, annoyance, and so on. Love is          Therefore, one purpose of the present research was to begin
divided into romantic love, filial love, and so on. Again, the         to examine empirically whether basic-level categories have in-
number of subcategories of love is indefinite.                         ternal structure and fuzzy boundaries. Earlier, we had hypothe-
   A middle-level category, or a particular event, is a member of      sized that they do—such is a corollary of viewing anger, fear,
the superordinate concept, emotion, not by virtue of possessing        and love as scripts (Fehr & Russell, 1984), for the notion of
all of a set of denning features, but by sufficient resemblance to     script extends to episodes the notion of prototype. More gener-
the prototypical, or exemplary, cases of emotion. On some ac-          ally, we hope to offer evidence that should address the concerns
counts, these cases are mentally represented as individual, ac-        of those who continue to defend a classical view in the domain
tual experiences; on other accounts, they are represented as a         of emotion (Johnson-Laird & Oatley, 1989; Ortony, Clore, &
more generalized schema. What is at the heart of the prototype         Foss, 1987).
perspective, however, is the notion of resemblance. Resem-
blance being a matter of degree, the number of categories at the
middle level is indeterminate: Events vary in the extent to                                      Concept of Love
which they are emotions, and no sharp boundary separates
emotions from nonemotions. Members resemble each other in                Love can be studied as a relationship, as an attitude, as an
overlapping and criss-crossing ways that vary in kind and              experience, and so on. In this article, we focus on love as an
number.                                                                emotion. Indeed, love is a prototypical emotion—it received
   Available evidence supports this line of thinking. As hypoth-       the highest prototypicality rating of any emotion in the Fehr
esized, the superordinate category of emotion was found to             and Russell (1984) study. Love is an important emotion, al-
have what Rosch called an internal structure—gradedness in             though sometimes forgotten—it has even been omitted from
membership—that predicted various indices of cognitive pro-            some psychologists' lists of the emotions (Ekman, 1972; Izard,
cessing involving emotion concepts. Moreover, the borders of           1977; Tomkins, 1984).
the concept of emotion were found to be fuzzy: Subjects could            In this article, we examine not cases of love, but the natural
not decide whether or not certain peripheral members, such as          language concept of love. Love has been defined in a variety of
pride and respect, were genuine examples of emotion.                   ways. As Brehm (1985, p. 90) commented:
   The evidence gathered so far pertains to the superordinate
concept, emotion. Indeed, as far as we know, almost all of the
studies to date offered as support for prototype theory involve          1
concepts at a superordinate level. The important exceptions are            Lysak, Rule, and Dobbs (1989) considered aggression a superordin-
Labov's (1973) study of cups and Rosch's (e.g., 1973) early work       ate concept, having a fight with your wife a basic-level concept, and
                                                                       throwing down your gloves a subordinate-level concept. We suspect that
on focal points of color categories. Rosch's (e.g., 1975) later work
                                                                       because of its relatively long name, having a fight with your wife is more
pertained to fruit, vehicle, furniture, vegetable, and other highly    likely a subordinate-level category, and that throwing down your gloves
general terms. Ironically, it was Rosch and her associates who         is more likely a feature than another level of categorization. In any case,
drew attention to the psychological significance both of proto-        what is important here is not so much the correct particular level in the
types and of the basic level (Rosch, Mervis, Gray, Johnson, &          hierarchy but the idea that concepts at different levels of the hierarchy
Boyes-Braem, 1976), and yet there is insufficient evidence on          may not all be structured in a prototypic way.
                                                              CONCEPT OF LOVE                                                                  427

    Social scientists have had as much trouble defining love as philoso-   Winnipeg who volunteered their participation. From one sam-
    phers and poets. We have books on love, theories on love, and          ple, we estimated their mean age as 20.6 years, with a range of
    research on love. Yet no one has a single, simple definition that is   17 to 47. In each study, roughly half were men, half women.
    widely accepted by other social scientists.

   Different writers have pointed to different features of love as                  Study 1: Free Listing of Subtypes of Love
denning: frustrated desire (Freud, 1922/1951), erogenous stim-
ulation (Watson, 1924), rewarding interactions (Centers, 1975),               Subjects were presented with the concept of love and asked to
and acts that achieve reproductive success (Buss, 1988). Rubin             list as many subtypes as came to mind.
(1970) defined love as an attitude held by one person toward
another, involving a predisposition to think, feel, and behave in          Method
certain ways toward that person. Skolnick (1978) defined love
                                                                             Subjects (N= 84) read the following instructions (adapted from Fehr
as "a constructed experience built with feelings, ideas, and cul-          & Russell, 1984, Study 1):
tural symbols" (p. 104). Swensen's (1972) definition of love fo-
cused on behaviors such as shared activities, disclosing intima-                  This questionnaire is part of a larger project on the sorts of
cies, and so on.                                                               things we have in mind when we hear and use words. In this study,
                                                                               we're concerned with types within a general category. For example,
   Similarly, psychologists have yet to agree on a typology of                 if asked to list types of the category CHAIR, you might write: rock-
love. Love has been said to consist of two (Hatfield & Walster,                ing chair, recliner, lawn chair, kitchen chair, stool, bean bag chair,
1978; Maslow, 1955), three (Kelley, 1983), seven (Kemper,                      and so on.
1978), or eight (Sternberg, 1986) subtypes.                                       The category we're interested in is LOVE. Please list as many
   Failure to agree on a definition or a typology of love suggests             types of LOVE as come to mind. Stop after a few minutes or 20
                                                                               items. Remember that there are no right or wrong answers—just
that psychologists are unsure what to include under the heading                give us your opinion.
of love. From a prototype perspective, the inability to achieve
consensus is understandable: There may be no small set of cri-
terial features common to all and only instances of love. There
                                                                           Results and Discussion
may be no fixed number of subtypes into which love can be                     Subjects generated an average of 8.69 responses each. In sum-
divided. One purpose of our research was to explore empiri-                marizing the responses, we treated syntactic variants (e.g., famil-
cally just this line of thinking. After describing that research,          ial love, family love, and love of family) as identical responses.
we shall then discuss the relationship of our findings to a scien-         This grouping left a total of 216 different items. Of these, 123
tific definition and typology of love.                                     were mentioned by only one subject and 93 by more than one.
                                                                           These 93 items appear in Table 1, along with the frequency with
                             Overview                                      which they were mentioned. As can be seen, the subtypes of
                                                                           love so generated varied widely in how readily they came to
   Six studies were conducted, aimed at establishing that subca-           mind. At one extreme was friendship, listed by over 60% of the
tegories of love vary in their degree of membership in love, that          subjects. At the other extreme were the 123 items each listed by
degree of membership predicts important indices of cognitive               only one subject. Between these extremes was no clear break
processing of members, that no sharp boundary separates                    between those available and those unavailable from memory.
members from nonmembers, and that members share a family                   The gradual change in availability is consistent with the idea of
resemblance to one another.                                                internal structure—that prototypical members shade gradually
   In the first study, subjects were asked to list whatever sub-           into nonprototypical members and then into nonmembers
types of love they could think of. In this way, subjects generated         with no sharp boundary to be found.
a portion of the next lower, or subordinate, level in the hierar-             Twenty target subtypes of love were selected from Table 1 for
chy of emotion. The ease with which subcategories of love came             further analysis in the subsequent studies. These were the 10
to mind also provided one measure of their degree of member-               terms listed most frequently, along with another 10 chosen from
ship within the category of love. Twenty of these subcategories            across the entire range of frequency; the 20 items are listed in
were then selected for analysis in the remaining studies. These            Table 2.
studies explored other predicted consequences of graded mem-
bership: direct ratings of the degree to which each subcategory
is an example of love (Study 2), reaction time to verify sentences
                                                                                          Study 2: Prototypicality Ratings
concerning its membership in the category love (Study 3), the                 In this study, subjects directly rated the goodness-of-example
probability of its inclusion in the category love (Study 4), sub-          (prototypicality) of each of the 20 target subtypes of love chosen
jective ratings of how natural or peculiar it sounds in sentences          in Study 1.
about love (Study 5), and the family resemblance pattern of
features listed for the types of love (Study 6). All together, these       Method
six studies yielded eight separate ways of mapping the internal
structure of the concept of love. Because internal structure is              Subjects (N = 92) read the following instructions (adapted from Fehr
best demonstrated through a convergence of these operations,               & Russell, 1984, Study 3):
comparison of one with another is postponed until a final sec-                   This study has to do with what we have in mind when we hear
tion.                                                                          and use words. Consider the word "red." Close your eyes and
   Subjects in these studies were students at the University of                imagine a true red. Now imagine an orangish red. Imagine a pur-
428                                               BEVERLEY FEHR AND JAMES A. RUSSELL

Table 1                                                                             worry about why you think something is or isn't a good example.
Free Listing of Types of Love                                                       Just give us your opinion.

                       No. times                               No. times          Subjects rated each of the 20 target subtypes of love (listed in Table 2)
       Item            mentioned               Item            mentioned       on a scale ranging from 1 (extremely poor example of love) to 6 (ex-
                                                                               tremely good example of love). For half the subjects, the 20 terms were
Friendship                  51         Mutual                        4         presented in reverse order.
Sexual                      25         Of sports                     4
Parental                    23         Agape                         3
Brotherly                   22         Ofart                         3         Results and Discussion
Sibling                     22         Casual                        3
Maternal                    20         Compassionate                 3            Mean prototypicality ratings for the 20 target subtypes of
Passionate                  19         Dependent                     3         love appear in Table 2. Two indices showed the reliability of
Romantic                    19         Offood                        3         these means: (a) Separate means calculated for the two orders of
Familial                    18         Of girlfriend                 3
Puppy                       18         Of grandparents               3         presentation were highly correlated, r= .95; and (b) the intra-
Paternal                    17         Heterosexual                  3         class correlation coefficient (which is equivalent to the average
Lust                        14         Of humanity                   3         of all possible split-half reliability coefficients) was .98. Of
Infatuation                 13         Long term                     3         course, individual prototypicality ratings were not as reliable as
Of pets                     10         Of money                      3         these aggregates (Barsalou, 1987). The mean correlation be-
Physical                    10         Occupational                  3
True                        10         Of self                       3         tween two raters across the 20 terms was .35.
Emotional                    9         Surface                       3            "Simone de Beauvoir once wrote that men and women have
Platonic                     9         Trust                         3         different concepts of love, and that's why they fail to understand
Respectful                   9         Unselfish                     3         one another. I don't doubt that this is true" (Solomon, 1988, p.
Sisterly                     9         Aesthetic                     2
Everlasting                  8         Ambition                      2         69). Consistent with this idea, previous studies have found sex
Of life                      8         Of beauty                     2         differences in beliefs and feelings about love. For example,
Marital                      7         Of books                      2         women are more likely to endorse pragmatic beliefs about love,
Of nature                    7         Brotherhood                   2         men more romantic ones (Peplau, 1983). Women are more
Obsessive                    7         Compulsive                    2         likely than men to report emotional symptoms of love, such as
Patriotic                    7         Fantasy                       2
Unrequited                   7         Free                          2
Affection                    6         Between friends of
Of animals                   6           the same sex                2
Erotic                       6         Helpful                       2         Table 2
Intimacy                     6         Hopeless                      2         Prototypicality Ratings, Reaction Times, and Errors in
Selfish                      6         Hurtful                       2         Verification for 20 Types of Love
Conditional                  5         Incestual                     2
First                        5         Inner                         2                                                           Nfean
Intense                      5         Kinship                       2                                                          reaction
Possessive                   5         Love-hate                     2            Type of love            Prototypicality         time         No. errors
Spiritual                    5         Love-30 (tennis)              2
Unconditional                5         Natural                       2         Maternal love                   5.39                 .53             0
Admiration                   4         Requited                      2         Parental love                   5.22                 .55             0
Blind                        4         Serious                       2         Friendship                      4.96                 .55             0
Committed                    4         Shallow                       2         Sisterly love                   4.84                 .46             1
Deep                         4         Sharing                       2         Romantic love                   4.76                 .53             0
Giving                       4         Spontaneous                   2         Brotherly love                  4.74                 .49             0
Of God                       4         Transitory                    2         Familial love                   4.74                 .51             1
Homosexual                   4         Understanding                 2         Sibling love                    4.73                 .57             0
Honesty                      4         Of work                       2         Affection                       4.60                 .47             1
Materialistic                4         Young                         2         Committed love                  4.47                 .39             0
                                                                               Love for humanity               4.42                 .61             0
Note. N = 84. Items listed by only one subject are omitted.                    Spiritual love                  4.27                 .47             0
                                                                               Passionate love                 4.00                 .62             0
                                                                               Platonic love                   3.98                 .57             1
                                                                               Self-love                       3.79                 .81             1
    plish red. Although you might still name the orange-red or the             Sexual love                     3.76                 .34             0
    purple-red with the term "red," they are not as good examples of           Patriotic love                  3.21                 .64             1
    "red" (not as clear cases of what red refers to) as the clear, true red.   Love of work                    3.14                 .78             1
    Orange and purple are even poorer examples of "red," perhaps not           Puppy love                      2.98                 .65             2
    even red at all.                                                           Infatuation                     2.42                 .63             2
       Notice that to judge how good an example something is, has
    nothing to do with how much you like the thing. You might prefer           Note. Prototypicality ratings were made on a scale ranging from 1 (ex-
    a purple-red or purple to a true red, but still recognize which is the     tremely poor example of love) to 6 (extremely good example of love).
    better example of "red."                                                   Reaction times are reported in seconds. Number of errors refers to the
       The word we are interested in is LOVE. On the following page is a       number of subjects reponding "false" when presented with the state-
    list of different kinds of love. You will be asked how good an             ment "X is a type of love," when X is the type listed. The maximum
    example of love the various types or instances of love are. Don't          possible number of errors was 20.
                                                                CONCEPT OF LOVE                                                                  429

feeling euphoric, having trouble concentrating, or feeling as                 and (h) 10 false statements concerning peripheral fruit, such as "Olive
though floating on a cloud (Dion & Dion, 1976; Hendrick,                      is a type of dance."
Hendrick, Foote, & Slapion-Foote, 1984). And women love                          The 80 statements were presented on a computer screen in a differ-
their partners more companionately, whereas men love theirs                   ent random order for each subject. The following instructions (adapted
more passionately (Traupmann & Hatfield, 1981). However, in                   from Fehr et al, 1982) appeared on the screen:
our data, men and women did not differ significantly far = .05)                      This is a simple study of the "belongingness" of items into cate-
in which subtypes of love they rated as better or poorer exam-                    gories. You will be presented with a series of statements of the
ples of love. In fact, men's and women's prototypicality ratings                  form "X is a If" \bur task is to respond True or False to each
were highly correlated, r - .94. In a related finding, men's and                  statement as fast as you can. So, for example, if the statement "Car
                                                                                  is a type of vehicle" appeared on the screen, you would press the
women's ratings of prototypicality of various features of love                    "1" key for TRUE. If the statement "Car is a type of vegetable"
were also highly correlated (Fehr, 1988). Thus, although men                      appeared, you would press the "2" key for FALSE. Try to respond as
and women may differ in their beliefs about love, or in how they                  quickly and as accurately as you can.
experience love, they seem to have the same concept of love.
   For the remaining studies, a decision was required as to                      Subjects were first presented with four practice statements (two true
                                                                              and two false) to familiarize them with the task (e.g., "Hockey is a type
which subtypes of love should be considered central and which
                                                                              of game" and "Soccer is a type of book"). Reaction times to the prac-
should be considered peripheral. A median split of the proto-                 tice items were not recorded. Next followed the 80 statements. Subjects
typicality ratings of Study 2 resulted in 10 subtypes (e.g, mater-            spent approximately 10 min on this task.
nal love) with ratings higher than 4.45 on the 6-point scale;
these were taken to be central. Those that received ratings lower
than 4.45 (e.g., patriotic love) were taken to be peripheral. The             Results and Discussion
cutoff we had used to divide central from peripheral emotions                    Reaction times were analyzed only for correct responses. A
was similar, 4.55 on the same 6-point scale (Fehr & Russell,                  response was considered correct if the subject responded "true"
1984).                                                                        to a true statement or "false" to a false statement, whether pro-
                                                                              totypical or peripheral.
Study 3: Reaction Time to Verify Category Membership                             The statements of primary interest were the 20 true state-
                                                                              ments concerning love. Mean reaction time and number of
   Rosch (1973) found not only that subjects could reliably rate              errors for each subtype appear in Table 2. Over 90% of the
how good an example of the concept bird were robins, eagles,                  subjects responded correctly to each statement, although for the
chickens, and penguins, but that these prototypicality ratings                more peripheral subtypes of love the beginnings of disagree-
predicted how long subjects took to verify that they were birds.              ment can be seen. Over half of all errors occurred in response to
Central members were verified faster than peripheral                          the four most peripheral subtypes.
members. Similarly, Fehr et al. (1982) found that the prototypi-                 Dependent sample t tests showed a significant difference for
cality ratings of various emotions predicted how long subjects                reaction times to true love statements, t(26) = 2.30, p < .05,
took to verify that they were emotions.                                       such that the category membership of prototypical instances
   The purpose of the present study was to test this prediction               was verified more quickly than that of peripheral instances
for the relationship between the basic and the subordinate level              (1.51 vs. 1.62 s). The difference in reaction time between proto-
of categorization. Subjects were predicted to verify a statement              typical and peripheral false love statements was not significant,
like "Maternal love is a type of love" more quickly than a state-             t(26) = .19, p> .10 (1.64 vs. 1.65 s).
ment like "Patriotic love is a type of love."                                    There was also a significant difference for reaction times to
                                                                              true fruit statements, /(26) = 4.68, p < .001, such that the cate-
Method                                                                        gory membership of prototypical instances was verified more
   Subjects (N= 27) were asked to respond "true" or "false" to 80 state-      quickly than that of peripheral instances (1.38 vs. 1.57 s). The
ments in the form of "AT is a type of Y" The general categories (Y) of        difference in reaction time between prototypical and periph-
interest were love and fruit. The types of love were the 10 central and 10    eral false fruit statements (1.52 vs. 1.60 s) was not significant,
peripheral types as determined in Study 2. The 20 types of fruit were         r(26) = 1.94,p<.07.
taken from Rosch (1975) and were included to replicate her reaction
time findings and to provide filler items for the task. Eighty sentences                           Study 4: Fuzzy Borders
were needed so that the proportion of true and false statements was
equal.                                                                           Internal structure and a classical definition are not mutually
   The pool of statements consisted of (a) 10 true statements concerning      exclusive. Armstrong et al. (1983) showed that paradigm cases
prototypical types of love, such as "Maternal love is a type of love"; (b)    of classically definable concepts, such as odd number and plane
10 true statements concerning peripheral types of love, such as "Patri-       geometric figure, have an internal structure. This demonstra-
otic love is a type of love"; (c) 10 false statements concerning prototypi-
                                                                              tion seems to leave no method of distinguishing concepts that
cal types of love, such as "Romantic love is a type of boat"; (d) 10 false
statements concerning peripheral types of love, such as "Sexual love is       are classically definable from those that are not.
a type of cheese"; (e) 10 true statements concerning prototypical fruit,         Fehr and Russell (1984) proposed a method. Classically de-
such as "Apple is a type of fruit"; (f) 10 true statements concerning         finable concepts have precise borders, whereas emotion, fruit,
peripheral fruit, such as "Tomato is a type of fruit"; (g) 10 false state-    vehicle, and the like do not. A direct challenge to the classical
ments concerning prototypical fruit, such as "Pear is a type of vehicle";     view is therefore the demonstration that membership is probabi-
430                                             BEVERLEY FEHR AND JAMES A. RUSSELL

listic rather than all or none. We thus offered evidence to show            tremes. The percentage of subjects who denied that each item is
that subjects can agree on which cases are and which are not                a genuine case of love is shown in Table 3. The median percent-
odd numbers but cannot agree for emotion, fruit, or vehicle.                age was 55. For 15 of the 30 candidates presented, at least one
The present study addressed the question of whether love has                third dissented from the majority. The results of Table 3 show
sharp or fuzzy borders.                                                     that a sample of college-educated native speakers do not agree
   Study 3 had provided data on the question of whether the                 among themselves on what is and what is not included in the
boundaries of love are sharp or fuzzy, but the results were not             category of love.
clear. On one hand, there was the expected disagreement found                  Subjects' judgments were not random, however. The proba-
in the peripheral cases. On the other hand, at least 90% of the             bility of a candidate's being judged a genuine case of love corre-
subjects agreed that each of the 20 target subtypes was a type of           lated .96 with its mean prototypicality rating. When the word
love. This degree of consensus might have occurred because                  love appeared in the item (e.g, puppy love), many subjects still
subjects were asked to make quick judgments in a context                    denied that it was a type of love. This finding implies that con-
where no responses went largely to absurd filler items such as              tent-based representations rather than logically structured cate-
"Is infatuation a type of fruit?" In this context, "Is infatuation a        gories mediated their judgments.
type of love?" may have seemed more reasonable than it other-
wise would.                                                                      Study 5: Substitutability for the Category Name
   The present study examined a broader range of potential
subtypes. Subjects were given 30 subtypes and as much time as                  Asked to give us a sentence about emotion, one subject wrote:
they wanted to answer each question. A separate group of sub-               "Sometimes emotion is hard to control." Substituting for the
jects provided prototypicality ratings for these 30. No absurd              word emotion a prototypical type of emotion produced a sen-
filler items were presented.                                                tence that sounded quite natural (Fehr & Russell, 1984)—for
                                                                            example, "Sometimes anger is hard to control." However, sub-
                                                                            stituting a peripheral emotion produced a sentence that often
Method                                                                      sounded rather peculiar—for example, "Sometimes respect is
   Membership ratings. Subjects (N = 118) read the following instruc-       hard to control." Such effects had been expected because the
       Sometimes when we use a word, we don't mean it really. Other
    times, we do. To take an extreme example, metaphors aren't to be        Table 3
    taken literally. As in: "Sally is an iceberg" or "Tom is a volcano."    Percentage of Subjects Denying Membership
    Sally is not a genuine iceberg, nor Tom a genuine volcano. We
    know that koala bears are not genuine bears. Road apples are not        in the Category Love
    genuine apples. A bird-of-paradise is a flower, not a bird. Yet a
    penguin, even though it can't fly, is a real bird. Something can be a          Item                                          Prototypicality
    genuine case, however unusual or atypical. Picture a tiger: a yel-
    low, striped, four-legged, sharp-eyed adult tiger. Yet we know that     Romantic love                      2                      4.86
    an albino, three-legged, blind, baby tiger is nonetheless a genuine     A mother's love                    6                      5.48
    tiger.                                                                  Caring                             8                      5.33
       In this study, we're interested in your opinion about what are       A brother's love                  11                      4.86
    genuine cases of love and what are not. On the next page is a list of   Friendship                        14                      4.95
    possible cases. Consider each in turn, and please judge whether it      Tenderness                        24                      4.90
    is or is not a genuine case of love.                                    Compassion                        24                      4.43
       Don't worry about why you feel one way or the other. You may         Affection                         27                      4.86
    decide that all are genuine. Or you may decide that none are. Just      Love of humanity                  35                      4.43
    give us your opinion.                                                   Self-love                         36                      4.43
                                                                            Passion                           36                      3.86
   Next followed a list of 30 possible types of love. Fifteen were taken    Longing                           44                      4.05
                                                                            Fondness                          52                      3.86
from Shaver et al.'s (1987) cluster of subtypes of love. The remaining 15   Liking                            53                      3.48
were selected from responses obtained in Study 1 in such a way as to        Love of country                   58                      3.57
include a range of availability but to place an emphasis on the less        Desire                            59                      3.29
available items in order to have a good sample of potential borderline      Love at first sight               60                      3.14
cases. Thus, each of the 30 items had some claim to being a genuine         Sentimentality                    63                      3.33
case of love, but the sample did not overrepresent prototypical cases, as   Attraction                        63                      3.62
had been done in the target sample used in Study 3.                         Adoration                         65                      3.52
   Prototypicality ratings. Subjects (N = 21) not included in the study     Puppy love                        66                      3.57
just described rated all 30 items in the same random order, following a     Arousal                           68                      3.14
                                                                            Love of work                      68                      3.29
procedure identical to that of Study 2.                                     Admiration                        75                      3.38
                                                                            Love of sports                    75                      2.95
Results and Discussion                                                      Love of books                     76                      2.90
                                                                            Love of food                      77                      2.76
  The majority of subjects judged that romantic love is a type of           Love of money                     80                      2.62
                                                                            Infatuation                       82                      2.81
love—only 2% disagreed with the majority decision. Most                     Lust                              87                      2.15
judged that lust is not a type of love—only 13% disagreed with
the majority decision. But most cases fell between these ex-                Note. N = 118 for percentages; N-21 for prototypicality ratings.
                                                               CONCEPT OF LOVE                                                                   431

more prototypical members are more representative of, and                          culiar. In this study, we're not interested in sentences about fruit.
closer to the meaning of, the superordinate category name.                         We're interested in sentences about LOVE.
Rosch (1977) had earlier found similar results for sentences                          Below is a series of sentences—some make a lot of sense, others
                                                                                   make some sense, and some may make no sense at all. Your task is
about fruit, vehicles, and other superordinate categories.                         to read each sentence to yourself and rate how natural or peculiar
   In the present study, we asked whether something similar                        it sounds to you. Don't spend time wondering if your answer is
would happen with sentences about basic-level categories, spe-                     right or wrong. There are no right or wrong answers. So, just base
cifically with sentences about love. One group of subjects first                   your answer on your first impression from reading the sentence.
gave us 10 sentences, and we took another 10 from psychology
                                                                                 Subjects then rated each of 20 sentences on a scale ranging from 1
textbooks. Another group of subjects was then asked to judge                  (very peculiar) to 6 (very natural sounding). The 20 sentences were
how peculiar or natural they found the sentences that resulted                drawn from a pool of 400 sentences formed by substituting the 20
when the 20 target terms were substituted for the word love.                  subtypes of love in place of the word love in the 20 sentences listed in
                                                                              Table 4. Twenty sentences were drawn for each subject in such a way
Method                                                                        that each subtype of love and each sentence frame appeared only once.
   Phase 1. Subjects(#= 37) were asked to generate sentences contain-
ing the word love. Ten of these sentences were selected for use in this       Results and Discussion
study (Sentences 1 to 10 in Table 4). An additional 10 sentences were
taken from recently published textbooks for courses in introductory or           Consider the following sentence: "Love has to be worked at
social psychology (Sentences 11 to 20 in Table 4).                            and strived for to be truly achieved." Substitution produced
   Phase 2. Subjects (N= 400) read the following instructions (adapted        some sentences that sounded natural and others that sounded
from Fehr & Russell, 1984, Study 4):                                          peculiar (e.g., "Friendship has to be worked at and strived for to
       This questionnaire is part of a larger project on the sorts of         be truly achieved" and "Infatuation has to be worked at and
    things people have in mind when they hear and use words. In this          strived for to be truly achieved").
    study, we're interested in how peculiar or how natural certain sen-          The naturalness of the resulting sentence was expected to
    tences sound. For example, the sentence "A bowl of apples makes a
    nice centerpiece for the table" probably sounds quite natural to          depend on how prototypical was the subtype of love substi-
    most people. On the other hand, a sentence like "A bowl of water-         tuted. In the example above, substitution of a prototypical sub-
    melons makes a nice centerpiece for the table" sounds rather pe-          type, friendship, yielded a sentence that subjects thought

                    Table 4
                    Correlations Between Rated Naturalness and Prototypicality
                                                             Sentence                                                 Correlation
                     1.   Love is a giving process, understanding the other, and realizing the other's faults.            .81***
                     2.   Commitment and caring are important components of love.                                         .79***
                     3.   Love has to be worked at and strived for to be truly achieved.                                  .61**
                     4.   Love is a very happy state—most of the                            time.                        .48*
                     5.   Love varies with the person and with the                       relationship.                   .45*
                     6.   Love is tender and vulnerable no matter who shares it.                                          .42
                     7.   Love is something shared intensely by two people.                                               .31
                     8.   Love is a feeling one gets in a special relationship with another person.                       . 17
                     9.   Love is very painful if not                            reciprocated.                            . 13
                    10.   Love may be both happy and sad, good and bad.                                                 -.04
                    11.   Love is one of the most important human emotions, yet it is poorly understood.                  .70***
                    12.   In our culture, we learn about love from childhood on.                                          .60**
                    13.   Given the importance of love in promoting happiness and making the world go
                          'round, it may be surprising that psychologists have only recently begun
                          systematic research on this topic.                                                             .33
                    14.   The phenomena of love are not easy to pin down scientifically.                                 .14
                    15.   Most of the time love does not involve intense physical reactions. Instead, it can
                          be viewed as a particular sort of attitude one person has toward another person.               .05
                    16.   Love is an emotion and, like the other emotions, may be seen as having both
                          physiological and psychological components.                                                    .03
                    17.   The emotions that accompany love are complex and far from being clearly
                          understood.                                                                                   -.13
                    18.   One of the difficulties in discussing love is that it means so many things to so
                          many people.                                                                                  -.17
                    19.   If you go to movies, listen to records, or read novels, you are constantly exposed
                          to people's views of love.                                                                    —.36
                    20.   Love is merely an intense form of liking.                                                     -.45*
                    Note. The first 10 sentences were written by undergraduate students. The second 10 sentences were taken
                    from psychology textbooks.
                    *p<.05. **p<.01 ***p<.001.
432                                            BEVERLEY FEHR AND JAMES A. RUSSELL

sounded natural (M = 4.45), whereas substitution of a periph-                              Study 6: Family Resemblances
eral subtype, infatuation, yielded a sentence that sounded pecu-
liar (M = \ .95). Correlations between mean naturalness ratings               Why are some members in a category considered more pro-
for the subtypes of love and their prototypicality ratings ob-             totypical than others? Rosch and Mervis (1975) argued that
tained in Study 2 are shown for each of the 20 sentence frames             Wittgenstein's (1953) idea of famiry resemblance provides an
in Table 4.                                                                answer. Although no single attribute may be shared by all,
   The resulting correlations were generally in the expected posi-         members of a category are linked through a complex pattern of
tive direction, but were stronger for some sentences than for              criss-crossing and overlapping attributes. Wittgenstein used
others. The sentences constructed by students yielded correla-             the term family resemblance to refer to the degree to which a
tions between naturalness and prototypicality that were gener-             member resembles this overall pattern. In their research on the
ally high, average r = .61. The range of correlations was similar          relationship of basic-level to superordinate-level categories,
to that found with the same task for other concepts (e.g., Fehr &          Rosch and Mervis (1975) found that more prototypical
Russell, 1984). In contrast, sentences taken from psychology               members shared more features with each other and had fewer
textbooks showed little if any relation between naturalness and            features in common with members of neighboring categories.
prototypicality, average r = .10. In fact, four of the textbook            This pattern was captured by calculating an index of famiry
sentences yielded correlations that were negative. Consider                resemblance as a weighted count of shared features, with the
"Love is merely an intense form of liking." Substituting a periph-         weight taken as the number of members sharing that attribute.
eral type of love yielded a more natural sounding sentence than               This index of family resemblance was highly correlated with
did substituting a prototypical type: Puppy love, the least pro-           other measures of internal structure (Rosch & Mervis, 1975),
totypical type of love we included, received the highest mean              with correlations ranging from .84 to .94 for various measures
naturalness rating (5.10), whereas maternal love, the most pro-            of internal structure and for various natural object categories.
totypical type of love, received the lowest (2.20). If substitut-          Although a similar result was found in the domain of emotion
ability can be taken as an indication of the core meaning of the           (Fehr & Russell, 1984), the correlations were lower, ranging
concept, then statements about love that students find in their            from .34 to .85 for various measures of internal structure.
textbooks can be far removed from their own conceptualiza-                    The reason for lower correlations in the domain of emotion is
tions of love.                                                             not known. Family resemblance scores for emotions were based
   To derive an index of internal structure from these data, a             on what subjects listed as the attributes of emotional states,
"substitutability" score was computed by averaging across the              attributes that may be more difficult to articulate than the at-
20 sentences the naturalness ratings that each subtype of love             tributes of everyday objects. Still, the possibility remains that
received. These scores are given in Table 5.                               the internal structure of emotion depends on factors other than
                                                                           family resemblance. Barsalou (1987) found that family resem-
                                                                           blance did not predict typicality for goal-derived concepts, and
Table 5                                                                    Medin and Smith (1984) cautioned that "comparable typicality
Substitutability, Family Resemblance, and Common Features                  effects in two domains do not imply common determinants of
Scores for 20 Types of Love                                                typicality" (p. 130). Tiller and Harris (1984) suggested three
                                                                           alternative determinants of typicality in the domain of emo-
                                              Family        Common         tion. First, prototypicality of an emotion may be simply its
   Type of love        Substitutability    resemblance      features
                                                                           intensity. This idea has been discussed recently by Mascolo and
Maternal love                3.94               135             13         Mancuso (1990). Second, prototypicality of an emotion may be
Parental love                3.94               144             17         the salience of associated bodily reaction, facial expression, or
Friendship                   4.63               145             17         posture. Third, prototypicality may be related to valence.
Sisterly love                3.69               122             12
Romantic love                4.53               109             12            Study 6 was designed to gather further evidence on the rela-
Brotherly love               3.90               143             18         tionship of prototypicality to family resemblance.
Familial love                4.08               121             12
Sibling love                 3.95               129             13
Affection love               4.48               101             10         Method
Committed love               4.40               124             17
Love for humanity            4.06                61              6           Subjects (N = 40) read the following instructions (adapted from Fehr
Spiritual love               3.89                67              7         & Russell, 1984, Study 6):
Passionate love              4.35                85             10
Platonic love                3.99               100             11               This is a study on the characteristics and attributes that people
Self-love                    3.62                49              7             associate with different kinds of emotion. For example, if you
Sexual love                  4.13                65             10             were asked to list the characteristics of a person experiencing a
Patriotic love               3.36                74              7             kind of fear, such as terror, you might write:
Love of work                 3.53                61              8               —possible danger occurs (may be real, like a bear; may be imagi-
Puppy love                   3.91                41              3             nary, like a ghost)
Infatuation                  3.68                41              5               —heart beats quickly
                                                                                 —eyes open wider
Note. Substitutability ratings were made on a scale ranging from 1 fyery         —eyebrows lift
peculiar) to 6 (very natural sounding); the scores are means computed            —palms and soles sweat
across 20 sentences. Common features refers to the number of features            —thoughts race through the person's mind
the subcategory of love shared with love itself; features of love were           —unpleasant sensations are experienced
taken from Fehr's (1988) list. Family resemblance is a weighted average          —the person runs as fast as he or she can
of features shared with other subcategories.                                     —hands tremble
                                                               CONCEPT OF LOVE                                                              433

       —relief is experienced after a few minutes                           task (Study 1), (b) direct ratings of prototypicality (Study 2), (c)
       Here we are not interested in different kinds of fear, but in        reaction time to verify membership (hypothesized to correlate
    different kinds of love. When thinking about each type of love,         negatively with internal structure) (Study 3), (d) probability of
    you might ask yourself: What manifestations are there of it? What
    thoughts do you have about it? How do you show it? In what cir-         membership (Study 3), (e) probability of membership (Study 4),
    cumstances are you apt to be aware of it? It might help to imagine      (f) substitutability into sentences about love (Study 5), (g) family
    that you're explaining the meaning of each type of love to a for-       resemblance score (Study 6), and (h) number of features shared
    eigner or to someone who has never experienced it. So include the       with love (Study 6).
    obvious. Tell how it comes about and what happens after. But
    emphasize a description of how one feels and acts.                         Each of these variables is influenced by factors other than
      Try not just to free associate. If a particular kind of love makes    internal structure, and therefore internal structure is best dem-
    you think of someone you know, don't write down the person's            onstrated by a convergence of these operations. Correlations
    name. We're interested in what is common to instances of love.          between the eight operations, calculated across types of love,
                                                                            are shown in Table 6. All values were in the predicted direction,
   Subjects were given as much time as they wished to list the attributes
for each of 10 subtypes of love presented in random order. Which 10 of      most reliably so. In fact, all operations except frequency of free
the 20 target subtypes were presented varied from one subject to the        listing were highly intercorrelated.
next such that each subtype of love was described by 20 subjects.              The correlations involving frequency of free listing, although
                                                                            low, were similar to those obtained in other studies. For exam-
Results and Discussion                                                      ple, rated prototypicality and frequency of free listing correlate
   The total number of responses generated was 2,333. Creating              .35 for emotions (Fehr & Russell, 1984), .35 for features of love
family resemblance scores required a decision as to which of                (Fehr, 1988), .37 for features of wisdom (Holliday & Chandler,
these responses represented the same attribute and which repre-             1986), and .19 for features of persons in different occupations
sented different attributes. Three judges coded the responses.              (Dahlgren, 1985).
The criterion adopted was the conservative procedure used by                   Fehr and Russell (1984) had found family resemblance scores
Fehr (1988), in which only identical or synonymous responses                to be the weakest correlate of internal structure for emotion. In
are combined. Next, for each target subtype, attributes men-                contrast, in the present study, family resemblance was one of
tioned by only one subject were eliminated. Judges' coding of               the best. Family resemblance cannot be put to rest as a possible
the remaining attributes yielded a final list of 205 distinct               mediator of typicality effects in the domain of emotion.
attributes, which thus had been listed by two or more subjects                 The pattern of results obtained here was the one predicted
for at least one subtype. Each of the 205 attributes was then               from prototype theory. The basic-level concept of love has an
weighted by the number of subtypes for which it had been                    internal structure: Subtypes of love can be reliably ordered
listed. For example, a weight of 14 was given to "caring" be-               from better to poorer examples of love. In turn, a subtype's rank
cause it was listed as an attribute for 14 of the 20 subtypes. This,        in this internal structure predicts its availability in a free listing
in fact, was the highest weight obtained. The next most com-                task, the speed with which it is verified as a type of love, how
mon attributes and their weights were helping (12), bond (11),              readily it can be substituted in sentences about love without
sharing (10), feel free to talk about anything (9), understanding           their sounding unnatural, the extent to which it resembles other
(9), respect (8), and closeness (7). A weight of 1 was given to             types of love, and the probability of its being considered a genu-
attributes that were listed (by two or more subjects) for only one          ine type of love. The last finding is especially important in
subtype of love: "act immaturely," listed only for puppy love;              establishing that the concept of love lacks precise boundaries
"playfighting," listed only for brotherly love; and "worrying,"             and therefore speaks strongly against the hypothesis that love is
listed only for maternal love.                                              classically defined. More generally, these results provide an al-
   Finally, for each type of love a family resemblance score was            ternative account of how a person can use and understand the
calculated as the sum of the weighted attribute scores listed for           concept of love without knowing necessary and sufficient fea-
that type of love. These scores appear in Table 5. Family resem-            tures for it. Our results therefore failed to support the sugges-
blance scores for prototypical subtypes were higher for love                tion that basic-level categories are classically defined.
than for peripheral subtypes, /(18) = 8.31, p < .001.
   We also expected that more prototypical subtypes would
bear a greater family resemblance to the general category love.                   Scientific Definitions and Typologies of Love
In other words, maternal love would have more features in                   Contrasts of Folk and Scientific Concepts
common with love than would puppy love. It was possible to
test this prediction because Fehr (1988) had asked subjects to                 Psychologists have offered various definitions and typologies
list the features of love, and those features had earlier been              that attempt to specify what is to be included under the term
coded with the same procedure used here. We tabulated the                   love. Psychologists' offerings appear to conflict with the present
number of features each of the 20 types of love had in common               results. We first outline the differences and then, in the final
with the features that had been listed for love. The results are            section, we discuss what should be the relationship between
also shown in Table 5. As expected, prototypical subtypes had               folk and scientific concepts.
more features in common with love than did peripheral sub-                     Psychologists have attempted to specify one or two necessary
types, /(18) = 5.63, p< .001.                                               and sufficient features for love. To consider just one example,
                                                                            Johnson-Laird and Oatiey (1989) recently argued that love can
     Convergence of Measures of Internal Structure                          be defined as follows: to love is "to experience internal happi-
  This series of studies produced eight operations thought to               ness in relation to an object, or person, who may also be the
correlate with internal structure: (a) frequency in a free listing          object of sexual desire" (p. 60). From a prototype perspective,
434                                           BEVERLEY FEHR AND JAMES A. RUSSELL

                   Table 6
                   Correlations Between Measures of Internal Structure
                                  Measure                     1

                   1.   Frequency of free listing            —      .35"    -.24"        .32'     .42'     .47       .50      .55
                   2.   Prototypicality rating                       —      -.48        .70       .97      .49       .86      .76
                   3.   Reaction time                                         —        -.44     -.64     -.50      -.48     -.49
                   4.   Agreement on membership                                         —         .85      .47       .56      .63
                   5.   Probability of membership in love                                                  .63       .86      .79
                   6.   Substitutability                                                         —                   .41      .45
                   7.   Family resemblance                                                                 —                  .93
                   8.   Shared features with love                                                                   —        —
                   Note. All correlations, except for those involving reaction time (No. 3), were hypothesized to be positive.
                   Correlations were calculated across the 20 target types of love listed in Table 2, except for those involving
                   consensus of membership from Study 4, which were calculated across 11 types.
                   • Not significant at a = .05.

happiness is likely to be only one of many features that charac-            ence or absence of each of three components: intimacy, passion,
terize love—and perhaps not the most characteristic feature,                and commitment. The resulting types of love are liking (inti-
either. Our subjects did not list happiness as one of the features          macy present), infatuation (passion present), empty (commit-
of most types of love. The characteristics shared by most but               ment present), romantic (intimacy plus passion), companionate
not all types of love were caring, helping, establishing a bond,            (intimacy plus commitment), fatuous (passion plus commit-
sharing, feeling free to talk, understanding, respect, and close-           ment), consummate (intimacy plus passion plus commitment),
ness. Thus, sentences about love sound peculiar when Johnson-               and nonlove (the absence of the components).
Laird and Oatley's definition is substituted: The command                      In contrast, our subjects listed a great many different types of
"love thy neighbor" would thus translate as "experience internal            love (just as laypeople listed many more features of love than
happiness in relation to thy neighbor." Johnson-Laird and Oat-              have scientists, Fehr, 1988). Of course, one could question
ley's definition implies that if internal happiness in relation to          whether many of the types of love listed in Table 1 are, in fact,
an object or person is experienced, then, by definition, love is            different types of love. One could argue, for instance, that the
experienced. In other words, a loveless but happy relationship,             list in Table 1 could be subjected to cluster analysis and thereby
as in, say, a pleasant business association, would be a contradic-          organized into a much smaller number of clusters. And, indeed,
tion in terms. Johnson-Laird and Oatley's definition also im-               it could. But doing so might impose order and boundaries
plies that if love is experienced, then happiness must be experi-           where they did not before exist. The number of clusters appears
enced. In other words, an unhappy love relationship, as in, say,            to be arbitrary (a notion consistent with the varying numbers of
unrequited love, would be a contradiction in terms. In short,               types of love plausibly delineated by psychologists). Consider
Johnson-Laird and Oatley's (1989) definition fails to capture the           just six items from our list: infatuation, puppy love, love at first
ordinary language concept of love.                                          sight, lust, passionate love, and romantic love. How might these
   Similarly, the typologies of love advocated by psychologists             be clustered? No two are identical, and therefore six types could
appear to conflict with the present results. First, they differ in          be created. But some are more similar than others, and there-
number. Scientists have typically listed a small number of types            fore several clusters could be created within this set. All contrast
of love. Two types of love, companionate and passionate, are                with companionate love, and therefore they could all be in-
recognized by many social psychologists (Hatfield & Walster,                cluded in one cluster. Thus, one, two, three, four, five, or six
1978). Two types were also described by Maslow(1955): D-love                clusters could reasonably be formed. Each such clustering
(deficiency love) and B-love (being love). D-love is based on               would capture some, but no single clustering would capture all,
need and dependence, whereas B-love is based on autonomy                    of the overlapping and cross-cutting features of similarity in the
and giving of oneself to the other. Three types of love were                domain of love.
described by Kelley (1983): passionate, pragmatic, and altru-                  Undoubtedly, the folk typology of love is more organized
istic. Three types were listed by Shaver et al. (1987): affection,
lust, and longing.2 Five types of love were described by Fromm
(1956): brotherly love, motherly love, erotic love, self-love, and              Shaver, Schwartz, Kirson, and O'Connor (1987) reported a hierar-
love of God. Six styles of love were described by Lee (1977):               chical cluster analysis of 135 emotion terms that yielded three types of
erotic (passionate love), ludic (game-playing love), storgic (com-          love: affection, lust, and longing. The affection cluster consisted of ador-
panionate love), manic (possessive, dependent love), pragmatic              ation, affection, love, fondness, liking, attraction, caring, tenderness,
                                                                            compassion, and sentimentality. This cluster resembles Hatfield and
(logical, practical love), and agapic (altruistic love). Seven types        Walster's view of companionate love. The lust cluster consisted of
of love were described by Kemper (1978), based on whether                   arousal, desire, lust, passion, and infatuation and resembles passionate
each person is high or low in power and in status: romantic love,           love. The longing cluster consists only of the word longing. Shaver et al.
brotherly love, charismatic or discipleship love, infidelity, infa-         took this result as a model of the lay typology of love, but this may not
tuation, adulation by fans, and parent-infant love. Eight types             be so because their subjects were provided with, rather than allowed to
of love were described by Sternberg (1986), based on the pres-              generate, the types of love.
                                                         CONCEPT OF LOVE                                                             435

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

ing of love as everyone understands it and (b) to provide a con-          Second, folk concepts and beliefs can provide hypotheses to
ceptual framework for scientific study of the phenomena re-            be tested. In this case, the scientist would assume neither the
ferred to by the word love. From the classical view, these two         truth nor the falsity of folk conceptualizations.
goals could be achieved simultaneously: The everyday concept              Third, analysis of the concept of love can help free scientists
of love already has scientifically desirable features of precise       from hidden assumptions and confusion. For example, in stat-
boundaries and common features. This assumption is chal-               ing the topic of the prescriptive analysis we assumed that the
lenged by prototype theory and by the empirical findings, such         phenomena to be analyzed were only those referred to by the
as those reported in this study, inspired by it. If the concept of     word love, but this assumption may be just what is hindering
love as commonly understood lacks precise boundaries, then             scientific progress. Scientific analysis could decide to draw dif-
its scientific status must be reexamined.                              ferent boundaries.
   The two goals may require distinct analyses. One, a descrip-           There is a tendency to assume that all cases called by the
tive analysis, aims to describe the everyday conception of love.       same name, love, share certain essential features. According to
Love, as a concept, is a worthy and legitimate topic in the psy-       the classical theory they do, and of course they might. Neverthe-
chology of social cognition. The prototype analysis offered in         less, prototype theory suggests that these cases need share no
this article is a descriptive analysis of the everyday concept of      more than a resemblance to one another. As Medin (1989)
love. The second analysis, a prescriptive one, aims to prescribe a     stated, "Despite the overwhelming evidence against the classi-
conceptualization of those phenomena referred to by the word           cal view,. . . people adopt an essentialist heuristic, namely, the
love. Those phenomena are a worthy and legitimate topic in the         hypothesis that things that look alike tend to share deeper prop-
psychology of emotion.                                                 erties" (pp. 1476-1477). That is, scientists and nonscientists
   A descriptive analysis concerns the concept, not the event, of      alike have taken the classical view for granted.
love: A mother worries whether she loves her children enough, a           Why are these experiences all called by the same name—
man questions whether his spouse loves him, a teenager                 love? Prototype theory asks one to consider an alternative to the
wonders whether what she is feeling is true love or just a passing     classical answer that they all share a common essence. This
fancy. Many believe that an unloved child is doomed—courts             particular set of diverse experiences is called love by an histori-
have thus declared that "to be unloved is so debilitating a dis-       cal process. Vygotsky (1962) described some concepts as chains,
ease no person subject to this experience should be required to        where each element shares some features with its neighbor, but
use the universal knowledge that maliciousness is wrong and            no common principle characterizes the entire set. The histori-
the human ability to control malevolence" (Kagan, 1989, p. 11).        cal evolution of the concept of love may be such a chain. Thus,
These worries and wonders and beliefs are tied to the everyday         one must consider carefully the observation of Solomon (1977)
concept of love. The natural language concept of love guides           on the cultural and historical specificity of the concept of love:
people's official and unofficial interpretation of some of life's      "The French, for example, have many distinctions between the
major and minor events, and that concept must be understood            various forms of intimacy that we more cold-blooded Angli-
—as it is.                                                             cans clumsily summarize with the clearly inadequate concepts
   The analogy here is to the concepts of witch and devil, which       of'love'and'like'" (p. 281).
played significant roles in the lives and deaths of people of the         Our suggestion therefore is to view the classical definitions
17th century. A descriptive analysis of witch and devil would          and typologies offered by various theorists as prescriptive. No
clarify their role in the thought of the time. Attempts to tidy up     classical definition or rigid taxonomy will provide an accurate
                                                                       descriptive analysis; therefore, set aside the question of whether
or improve these concepts would be inappropriate. Similarly, a
                                                                       a particular definition or typology coincides with the everyday
descriptive analysis of love and emotion would aim to clarify
                                                                       concept of love. Instead, take the definition as delimiting pre-
their role in today's thought. To tidy up or improve the concept
                                                                       cisely a range of phenomena that may be explained by one set of
of love would defeat the purpose of a descriptive analysis. Pro-
                                                                       scientific hypotheses. Then test those hypotheses through em-
totype theory is as applicable to concepts such as witch or devil
                                                                       pirical means.
and unicorn or centaur as it is to emotion, vehicle, or love, for
                                                                          To illustrate, let us return to Johnson-Laird and Oatley's defi-
these are all equal in their status as concepts, whatever their
                                                                       nition—"to experience internal happiness in relation to an ob-
merits or demerits from a scientific perspective.                      ject or person." For purposes of a prescriptive analysis, it is
   A prescriptive analysis concerns those events referred to by        irrelevant that this definition fails to capture the everyday con-
the word love: A mother loves her child, a husband loves his           cept of love. So, set aside the question of whether all and only
wife, and a miser loves his money. A prescriptive analysis seeks       such experiences are properly labeled love, but focus on this set
to understand such phenomena. Because prototype theory                 of experiences. Next, consider Freud's definition of love as frus-
concerns concepts, not events, it has no direct bearing on a           trated desire. Set aside the question of whether all and only
prescriptive analysis. Scientific psychology cannot rely on folk       these experiences are properly labeled love, or whether they
psychology any more than scientific physics or chemistry relies        coincide with the cases selected by Johnson-Laird and Oatley.
on folk physics and folk chemistry. Nevertheless, prototype            So interpreted, each definition might be a first step in a scientifi-
theory can play an indirect role.                                      cally useful analysis of a set of phenomena that do share enough
   First, one possibility is that the ordinary concept of love plays   features in common that they can be analyzed together.
a causal role in events of love. If so, then a prescriptive analysis
requires an adequate descriptive analysis. Nevertheless, the dis-                                 References
tinction between concepts and events must not get lost, for the        Armstrong, S. L., Gleitman, H., & Gleitman, L. R. (1983). What some
concept would be a causal antecedent to the event.                       concepts might not be. Cognition, 13, 263-308.
                                                              CONCEPT OF LOVE                                                                   437
Averill, J. R. (1980). On the paucity of positive emotions. In K. R.        Hatneld, E., & Walster, G. W (1978). A new look at love. Reading, MA:
  Blankstein, P. Pliner, & J. Polivy (Eds.), Assessment and modification      Addison-Wesley.
  of emotional behavior (pp. 7-45). New \fork: Plenum Press.                Hendrick, C, & Hendrick, S. (1986). Theory and method of love. Jour-
Averill, J. R. (1982). Anger and aggression: An essay on emotion. Berlin,     nal of Personality and Social Psychology, 50, 392-402.
  Federal Republic of Germany: Springer-Verlag.                             Hendrick, C, Hendrick, S., Foote, F. H., & Slapion-Foote, M. J. (1984).
Averill, J. R., & Boothroyd, P. (1977). On falling in love in conformance     Do men and women love differently? Journal of Social and Personal
  with the romantic ideal. Motivation and Emotion, 1, 235-247.                Relationships, 1,177-195.
Barsalou, L. W (1987). The instability of graded structure: Implica-        Holliday, S. G., & Chandler, M. (1986). Wisdom: Explorations in adult
  tions for the nature of concepts. In U. Neisser (Ed.), Concepts and         competence (Human Development Monograph No. 17). Basel, Swit-
  conceptual development (pp. 101-140). Cambridge, England: Cam-              zerland: Carger Publishing.
  bridge University Press.                                                  Izard, C. E. (1977). Human emotions. New \brk: Plenum Press.
Brehm, S. S. (1985). Intimate relationships. New York: Random House.        James, W (1929). The varieties of religious experience: A study in human
Broughton, R. (1984). A prototype strategy for construction of person-        nature. New York: Longmans, Green & Co. (Original work published
  ality scales. Journal of Personality and Social Psychology, 47,1334-        1902)
  1346.                                                                     James, W (1950). The principles of psychology. New York: Dover. (Origi-
Buss, D. M. (1988). The evolutionary biology of love. In R. J. Sternberg      nal work published 1890)
  & M. L. Barnes (Eds.), The psychology of love (pp. 100-118). New          Johnson-Laird, P. N, & Oatley, K. (1989). The language of emotions:
  Haven, CT: Yale University Press.                                           An analysis of a semantic field. Cognition and Emotion, 3, 81-123.
Buss, D. M., & Craik, K. H. (1983). The act frequency approach to           Kagan, J. (1989). Unstable ideas. Cambridge, MA: Harvard University
  personality. Psychological Review, 90,105-126.                              Press.
Cantor, N, & Mischel, W (1979). Prototypes in person perception.            Kahneman, D, & Miller, D. T. (1986). Norm theory: Comparing reality
  Advances in Experimental Social Psychology, 12, 3-52.                       to its alternatives. Psychological Review, 93,136-153.
Cantor, N, Mischel, W, & Schwartz, J. C. (1982). A prototype analysis       Kelley, H. H. (1983). Love and commitment. In H. H. Kelley, E. Bers-
  of psychological situations. Cognitive Psychology, 14, 45-77.               cheid, A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger, E.
Cantor, N, Smith, E. E., French, R. D., & Mezzich, J. (1980). Psychiatric     McClintock, L. A. Peplau, & D. R. Peterson (Eds.), Close relation-
  diagnosis as prototype categorization. Journal of Abnormal Psychol-         ships (pp. 265-314). New York: W H. Freeman.
  ogy, 89,181-193.                                                          Kemper, T. D. (1978). A social interactional theory of emotions (pp.
Centers, R. (1975). Sexual attraction and love. Springfield, IL: Charles      283-309). New York: Wiley.
  C Thomas.                                                                 Labov, W (1973). The boundaries of words and their meanings. In
Conway, M. A., & Bekerian, D. A. (1987). Situational knowledge and            C. J. N. Bailey & R. W Shuy (Eds.), New ways of analyzing variation in
  emotions. Cognition and Emotion, 1,145-191.                                 English (pp. 340-373). Washington, DC: Georgetown University
Dahlgren, K. (1985). The cognitive structure of social categories. Cog-       Press.
  nitive Science, 9, 379-398.                                               Lakoff, G. (1987). Women, fire, and dangerous things: What categories
Davis, K. E., & Todd, M. J. (1982). Friendship and love relationships. In     reveal about the mind. Chicago: University of Chicago Press.
  K. E. Davis & T. Mitchell (Eds.), Advances in descriptive psychology      Lee, J. A. (1977). A typology of styles of loving. Personality and Social
  (Vol. 2, pp. 79-122). Greenwich, CT: JAI Press.                             Psychology Bulletin, 3,173-182.
Davis, K. E., & Todd, M. J. (1985). Assessing friendship: Prototypes,       Lee, J. A. (1988). The forbidden colors of love: Patterns of gay love and
  paradigm cases and relationship description. In S. Duck & D. Perl-          gay liberation. In J. P. DeCecco (Ed.), Gay relationships (pp. 11-32).
  man (Eds.), Understanding personal relationships: An interdisciplin-        New York: Harrington Park Press.
  ary approach (pp. 17-38). Beverly Hills, CA: Sage.                        Lysak, H., Rule, B. G., & Dobbs, A. R. (1989). Conceptions of aggres-
Dion, K. L., & Dion, K. K. (1976). Love, liking, and trust in heterosex-      sion: Prototype or defining features? Personality and Social Psychol-
  ual relationships. Personality and Social Psychology Bulletin, 2,187-       ogy Bulletin, 15, 233-243.
  190.                                                                      Mascolo, M. F, & Mancuso, J. C. (1990). The structure and content of
Ekman, R (1972). Universal and cultural differences in facial expres-         emotion knowledge: A prototype approach. Manuscript submitted for
  sion of emotion. In J. R. Cole (Ed.), Nebraska Symposium on Motiva-         publication.
  tion, 1971 (pp. 207-283) Lincoln: University of Nebraska Press.           Maslow, A. H. (1955). Deficiency motivation and growth motivation.
Fehr, B. (1982). Prototype categorization of emotion. Unpublished mas-        In M. R. Jones (Ed.). Nebraska Symposium on Motivation (pp. 1-39).
  ter's thesis, University of British Columbia, Vancouver, Canada.            Lincoln: University of Nebraska Press.
Fehr, B. (1988). Prototype analysis of the concepts of love and commit-     Medin, D. L. (1989). Concepts and conceptual structure. American
  ment. Journal of Personality and Social Psychology, 55, 557-579.            Psychologist, 44,1469-1481.
Fehr, B., & Russell, J. A. (1984). Concept of emotion viewed from a         Medin, D. L., & Smith, E. E. (1984). Concepts and concept formation.
   prototype perspective. Journal of Experimental Psychology: General,        Annual Review of Psychology, 35,113-138.
   113, 464-486.                                                            Mervis, C. B., & Rosch, E. (1981). Categorization of natural objects.
Fehr, B., Russell, J. A., & Ward, L. M. (1982). Prototypicality of emo-       Annual Review of Psychology, 32, 89-115.
  tions: A reaction time study. Bulletin of the Psychonomic Society, 20,    Neisser, U. (1979). The concept of intelligence. Intelligence, 3,217-227.
  253-254.                                                                  Neisser, U. (1987). Concepts and conceptual development. Ecological
Freud, S. (1951). Group psychology and the analysis ofthe ego (J. Stra-       and intellectual factors in categorization. Cambridge, England: Cam-
  chey, Ed. and Trans.). New York: Liveright. (Original work published        bridge University Press.
   1922)                                                                    Newport, E. L., & Bellugi, U. (1978). Linguistic expression of category
Fromm, E. (1956). The art of loving. New York: Harper & Row.                   levels in a visual-gestural language. In E. Rosch & B. B. Lloyd (Eds.),
Harnad, S. (1987). Category induction and representation. In S. Har-          Cognition and categorization (pp. 49-71). Hillsdale, NJ: Erlbaum.
   nad (Ed.), Categorical perception (pp. 535-565). Cambridge, En-          Ortony, A., Clore, G. L., & Foss, M. A. (1987). The referential structure
   gland: Cambridge University Press.                                         ofthe affective lexicon. Cognitive Science, 11, 341-364.
438                                            BEVERLEY FEHR AND JAMES A. RUSSELL

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

To top