f - , , p"j
;2 - i
ELSEVIER Cognition 60 (1996) 173-204
COGNIT ION
On metaphoric representation
Gregory L. Murphy*
University of Illinois, 405 North Mathews Avenue, Urbana, IL 61801, USA
Received 9 January 1995, final version accepted 17 January 1996
Abstract
The article discusses claims that conceptual structure is in some part metaphorical, as
identified by verbal metaphors like LOVE IS A JOURNEY. Two main interpretations of
this view are discussed. In the first, a target domain is not explicitly represented but is
instead understood through reference to a different domain. For example, rather than a
detailed concept of love per se, one could make reference to the concept of a journey. In the
second interpretation, there is a separate representation of love, but the content of that
representation is influenced by the metaphor such that the love concept takes on the same
structure as the journey concept. It is argued that the first interpretation is not fully coherent.
The second interpretation is a possible theory of mental representation, but the article raises
a number of empirical and theoretical problems for it. It is concluded that many of the data
cited as evidence for metaphoric representations can be accounted for by structural
similarity between domains.
1. Introduction
An intriguing view has recently arisen in cognitive science concerning the
nature of mental representation. This view is exemplified by the following
passages:
. . . most people think they can get along perfectly well without metaphor.
We have found, on the contrary, that metaphor is pervasive in everyday life,
not just in language but in thought and action. Our ordinary conceptual
system, in terms of which we both think and act, is fundamentally
metaphorical in nature. (Lakoff and Johnson, 1980, p. 3)
* E-mail: gmurphy@s.psych.uiuc.edu.
0010-0277/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved
PII S0010-0277(96)007! ! -1
174 G.L. Murphy / Cognition 60 (1996) 173-204
The basic-level metaphors allow us to comprehend and draw inferences
about these [emotion] concepts, using our knowledge of familiar, well-
structured domains. (Kovecses, 1986, p. 117)
In short, the locus of metaphor is not in language at all, but in the way
we conceptualize one mental domain in terms of another. (Lakoff, 1993,
p. 203)
. . . human cognition is fundamentally shaped by various poetic or figurative
processes. (Gibbs, 1994, p. 1)
As these quotations indicate, some researchers in cognition and language have
argued that mental representation is at least in part metaphoric. Rather than seeing
metaphors as being solely or even primarily a linguistic phenomenon, they have
proposed it as a mode of representation and thought. The reasoning behind this is
that certain aspects of our knowledge are difficult for people to represent: They are
overly abstract and complex, and therefore they are represented in terms of
easier-to-understand domains, that is, metaphorically (Kovecses, 1986, p. 6;
Lakoff, 1987, pp. 84-85; Lakoff and Johnson, 1980, p. 115). Thus, when we think
about abstract ideas such as inflation, the mind, or anger, we use more concrete
concepts, a process which "allows us to refer to it [an abstract concept], quantify
it, identify a particular aspect of it . . . and perhaps even believe that we
understand it" (Lakoff and Johnson, 1980, p. 26).
The argument for metaphoric representation is often made as part of an
argument for Cognitive Linguistics, as championed by Lakoff and his colleagues.
However, I believe that the issue of metaphoric representation is an interesting and
radical idea which deserves attention in its own right. Lakoff (1987) and Lakoff
and Johnson (1980) present the use of metaphor in thought as just one part of an
overarching theory of the nature of the mind. But their arguments about
"objectivist metaphysics," generative approaches to linguistics and other contro-
versial ideas may have drawn attention away from this specific claim. Therefore,
in this article I will examine metaphoric representation as a theory of conceptual
structure. I will not be addressing most of the other views of its proponents. Of
those views, metaphoric representation has probably had the most direct influence
on psychological research, through theories of idiom and metaphor comprehension
(see Gibbs, 1992, Gibbs, 1993; Gibbs and O'Brien, 1990; Glucksberg, 1993;
Keysar and Bly, in press). That said, however, it will be impossible to avoid
mentioning other views expressed by Lakoff and Johnson (1980) or Lakoff
(1987), because some of them are used to provide support for the notion of
metaphoric representation.
The next section of this article will briefly review basic terminology and facts
about linguistic metaphor. Then two notions of metaphoric representation will be
described and criticized, in turn. Finally, I will draw some conclusions on how
proponents of this view might address the problems raised in this article.
G.L. Murphy / Cognition 60 (1996) 173-204 175
2. Metaphor
One disadvantage in discussing metaphoric representation is that it is necessary
to discuss verbal metaphor. And as is well known, metaphor is one of the most
difficult and intransigent problems in language. Indeed, even defining metaphor is
by no means an easy task. For the present purposes, then, a minimal description
(not really a definition) will have to suffice.
In verbal metaphor, there are usually two explicit parts: a topic, which is the
entity being talked about, and the vehicle, which is the metaphoric material being
predicated of the topic. (Some authors refer to these as the target and source,
respectively.) For example, in Lee is a block of ice, Lee is the topic, and block of
ice is the vehicle. The implicit connection between the two is often referred to as
the ground. In this case, the ground is that blocks of ice have a cold temperature,
and this is interpreted in terms of emotional unresponsiveness. More generally, the
relation between temperature and emotion provides the ground.
In order for a sentence to be perceived as metaphoric, the vehicle cannot apply
in a straightforward way to the topic. For example, Lee is an attorney does not (in
this context) require any special ground for its interpretation. The usual, familiar
meaning of attorney specifies a person with a particular profession, and since Lee
is the name of a person, there is no inconsistency in calling Lee an attorney.
However, block of ice literally means the solid, frozen state of the substance H20,
and since a person is typically neither H20 nor frozen, this predicate cannot be
applied to Lee in a straightforward way. There must be some kind of mapping
from the usual meaning of this phrase to the conveyed, nonliteral meaning.
It is not necessary that something have no literal interpretation in order to be
interpreted as a metaphor, as has often been pointed out (e.g., Keysar, t989;
Morgan, 1979). For example, a (bad) rock 'n' roll song included the lyric "A guy
could get pneumonia sitting next to you." Although this has a clear literal
interpretation (perhaps the addressee had viral pneumonia), in the context of the
song the metaphoric interpretation, again equating temperature with emotional
properties, was much more salient. In that interpretation there was a mapping
between the literal meaning (catching a disease) and the intended meaning
(emotional coldness).
Although this brief description has not gone very far in defining metaphor,
much tess explaining how people understand it, it includes two points that will be
important later. First, understanding a metaphor requires one to go beyond the
literal, straightforward interpretation of the vehicle. In some sense, Lee is clearly
not a block of ice, billboards are not really warts, encyclopedias are not really gold
mines, and so on. Note that this claim does not assume any particular psycho-
linguistic model of how metaphors are understood. That is, although metaphors
can be identified as different from literal sentences, this does not entail that
metaphors are more difficult to understand or that literal language has priority (see
Gibbs, 1984, Gibbs, 1994; Keysar, 1989; Ortony et al., 1978). My point is only
that Lee is a block of ice and Put a block of ice in the cooler involve rather
different interpretations of block of ice.
176 G.L. Murphy / Cognition 60 (1996) 173-204
Second, the interpretation of a metaphor requires constructing a relation (the
ground) or accessing a mapping between the topic domain and the vehicle domain
(Black, 1979). In order to understand Lee is a block of ice and Encyclopedias are
gold mines, coldness must be equated with emotional unresponsiveness, and the
financial value of a gold mine must be related to the intellectual value of an
encyclopedia. Unless this connection is made, the sentence appears anomalous or
just false.
Although Lakoff and Johnson (1980) do not provide a detailed theory of verbal
metaphor, their discussion seems to accept this kind of view. That is, in insisting
that representations are metaphoric or metonymic, they are contrasting them with a
more straightforward relation (called direct understanding by Lakoff and Johnson,
1980, pp. 176 ff.). (Henceforth I shall refer to Lakoff and Johnson, 1980, as
"L&J.") Since the metaphoric relation is not direct, some kind of mapping is
necessary. And in fact, much of L&J and Kovecses (1986) consists in spelling out
the mappings behind various conceptual metaphors.
3. Two interpretations of metaphoric representation
As linguist and philosopher, respectively, Lakoff and Johnson do not provide a
detailed psychological model of metaphoric representation, nor a process model
for how such representations would be used in understanding and thought - at
least, models of the sort expected in cognitive psychology. Nonetheless, they make
constant reference to conceptual structure, thought and understanding (e.g., see the
opening quotation). Most of the literature on metaphoric concepts focuses on
representation, with little to say about process. Cognitive psychology has long held
that both representation and process are necessary to define any psychological
model, because it is the processes that transform the representations into overt
behavior (Anderson, 1978). Thus, lacking an explicit psychological model of
metaphoric representation, it will be necessary for me to make certain assumptions
about how metaphoric representations are constructed and used. I have made what
I believe to be plausible assumptions about processing, but without detailed
discussion in most descriptions of metaphoric representation it is possible that I
have missed other possibilities. In short, this article is about the psychological
model I have derived from the writings of L&J (and others) rather than being
about their own explicit models. However, it is surely incumbent on the
proponents themselves to make such matters clear, and the lack of a detailed
processing model is a problem with the view as a psychological account of
conceptual structure.
I will describe two general interpretations of the claim for metaphoric
representation, which I call the strong and weak versions. In brief, the strong
version argues that some concepts are not understood via their own representations
but instead by (metaphoric) reference to a different domain. We don't really
understand emotions very well, for example, and so we think of them in terms of
temperatures, which we do understand. The weak version does not claim that we
G.L. Murphy / Cognition 60 (1996) 173-204 177
think of actual temperatures when considering emotions, but instead that we do
have a well-developed conceptual structure for emotions (or whatever). However,
the existence of systematic verbal metaphors in our culture has influenced the
structure of our concepts so that they are consistent with the metaphor. That is, the
structure of the domain of emotions has been influenced by the many metaphors
relating emotions to temperatures.
3.1. The strong version of metaphoric representation
In the strong version of metaphoric representation, some mental representation
is metaphorical in the same sense that Lee is a block of ice is metaphorical: One
thing is understood through its relation to another thing that is not truly (literally)
the same sort of entity. L&J (p. 176) acknowledge that there are some things that
are directly (nonmetaphorically) understood, via our immediate experience with
our environment. But this is not true for all concepts: "many aspects of our
experience cannot be clearly delineated in terms of the naturally emergent
dimensions of our experience. This is typically the case for human emotions,
abstract concepts, mental activity ... Though most of these can be experienced
directly, none of them can be fully comprehended on their own terms. Instead, we
must understand them in terms of other entities and experiences, typically other
kinds of entities and experiences" (p. 177). Thus, the strong version of metaphoric
representation takes the notion of understanding one thing in terms of another
seriously. One reason for these metaphoric mappings is that an abstract concept
becomes grounded through its relation to a more basic domain. Eventually, the
mappings may lead back to sensorimotor or other bodily bases for concepts (L&J,
Chs. 12-13).
For example, take the metaphor ARGUMENT IS WAR (L&J capitalize
conceptual metaphors). This metaphor is evidenced by a number of common
expressions (L&J, p. 4, italics in the original, indicating the metaphoric com-
ponent):
Your claims are indefensible.
He attacked every weak point in my argument.
His criticisms were right on target.
I demolished his argument.
In the strong view, our direct representation of arguments is a set of connections
to another domain, which provides an interpretation of the entities in the argument
concept. For example, the matter under dispute in the argument corresponds to the
object (land, power) being fought over in a war. Each person is interpreted as a
combatant; the arguments are weapons used to protect one's own view or to attack
the opponent's; arguments in favor of one's position are viewed as defenses;
arguments criticizing the other person's assumptions or position are viewed as
offensive maneuvers; and so on. On this view, when I think about arguments, I use
my knowledge of war to reason about and understand the argument.
178 G,L. Murphy / Cognition 60 (1996) 173-204
When stated explicitly, the strong view may seem too extreme to describe
anyone's view. Nonetheless, some of the strongest claims made for metaphoric
representation seem to require this sort of representation. And, as I will argue
below, some of the more recent discussions of metaphoric representation seem to
include the strong view as one part of a more complex theory. Thus, this view
merits further attention.
L&J make it clear on virtually every page of their book (and in the title) that
they do not view metaphors as simply ways of talking about things: Metaphors are
vehicles of thought. In fact, Lakoff (1993) argues that metaphor is really a
conceptual mapping of a certain kind, and verbal metaphors are simply expres-
sions of the underlying mapping - that is, the verbal metaphor is secondary. On
the strong interpretation of their view, there is little structure to the metaphorical
concept itself; instead, the structure comes from the "vehicle" concept. For
example, people don't have independent ideas about the parts and events involved
in an argument. Instead, what they have is a set of pointers to their concept of war.
On the strong interpretation, it really is our concept of war that we use when
thinking about arguments. Perhaps there are separate units for the components of
an argument, such as the opponents, the positions, or the possible resolutions of
the argument. However, if the concept is represented metaphorically, these
components cannot be independently defined and represented. Instead, they must
merely be linked to the appropriate element of the vehicle concept. In a real sense,
then, one does not really understand an argument - one only understands war, and
the understanding of arguments is parasitic on this concept. Similarly, Lakoff and
Turner (1989, p. 62) say that the LIFE IS A JOURNEY metaphor shows that the
"structuring of our understanding of life comes from the structure of our
knowledge about journeys." This claim seems consistent with the strong view, as
does the statement (L&J, p. 5), "The essence of metaphor is understanding and
experiencing one kind of thing in terms of another." (See also Lakoff, 1993, p.
206.)
3.2. The weak view of metaphoric representation
A different interpretation of metaphoric representation is that the metaphors
have some influence on the representation of the topic concepts, but these concepts
nonetheless have their own separate representations. That is, the metaphor may
influence the structure of the topic concept, but the representation itself is not
metaphoric.
For example, consider the metaphor ARGUMENTS ARE WAR. On the weak
view, the concept of argument has its own representation, separate from that of
war. The concept has some conceptual primitives and relations that are not the
same as that of war - its parts are defined in their own terms rather than in terms
of the metaphor. Nonetheless, one might argue that the prevalence of this
metaphor in language and our thinking has resulted in the structure of the
argument concept being similar to the structure of the war concept. Thus, the
G.L. Murphy / Cognition 60 (1996) 173-204 179
metaphor has had some causal effect on the concept of argument, but the
representation of the concept is not itself metaphorical. The critical difference
between the strong and weak versions, then, has to do with independence of
representation. In the strong version, I really do not have a well-developed concept
of argument separate from war - when I think about arguments, I use my
concepts of wars (I'm "understanding one kind of thing in terms of another"). In
the weak version, I have a complete, separate concept of argument which I use to
reason about arguments, but its content and structure have been influenced by my
knowledge of war.
3.3. An alternative view
One problem in thinking about metaphoric representation is that a plausible
alternative hypothesis is not readily found in L&J or Lakoff (1987). They
primarily contrast their views with a collection of claims about metaphysics,
language and the mind that they call "Objectivism." This monolithic view is one
that many psychologists would not want to commit themselves to. Nonetheless,
they may feel that L&J's strong views about metaphoric representation are not
correct. Even when discussing metaphoric representation more specifically, Lakoff
(1993) identifies its opponents (the "traditional" view) as having very extreme
views, such as "'All everyday conventional language is literal, and none is
metaphorical. All subject matter can be comprehended literally, without
metaphor" (p. 204). Although there may be people who hold these views, there
seems to be considerable ground between this extreme position and the metaphoric
representation view, and this middle ground is not addressed by proponents of
metaphoric representation.
I will, therefore, propose an alternative hypothesis that could serve as a basis of
comparison, the structural similarity view. On this view, there is no strong form of
metaphoric representation - all concepts are directly represented. Furthermore, the
prevalence of metaphoric ways of talki