George Lakoff - Cognitive Models

C E N T E R

F O R

http://www.hum.au.dk/semiotics ©1997-2002, Center for Semiotics & the author You are not allowed to distribute this paper without permission

S E M I OT I C S University of Aarhus

INTRODUCTION

TO

COGNITIVE MODELS

LARS ANDREASSEN [email protected]

1. Introduction The ambition of the following is to present an outline of the theory of cognitive models. To begin with I’ll try to give a rough idea of what a cognitive model is, what it consists of and where the elements come from, then it’s appropriate with some small examples of cognitive models (part 2). And since Lakoff asserts that we are to understand categorization and prototype effects as evoked by the organization of cognitive models, we’ll have to touch upon the theory of categorization and prototypes in order to see what is meant by that assertion (part 3). Finally it’s the intention to dwell upon some details in a more theoretically elaborated cognitive model, namely that of anger, a case study provided by Lakoff and Kövecses (part 4).

2. Models - in Short A cognitive Model is an organized whole of generalized experiences of individual, but comparable phenomena – objects, actions, and events. Our everyday encounters with different, but comparable phenomena gives rise to ideas of what is normal to come across and experience in our respective environments. These ideas of what is normal, or typical evoke prototype effects, which preliminary can be compared to stereotypes, role models (paragons), typical events, and so on. We create notions of persons, of behavior, of objects, of future events, of how things must have happened in the past. These experiences sort of constitute an experiential horizon which in turn constitute our everyday reasoning. The overall idea of cognitive models is then, that they structure our experiences into complex structured wholes - a sort of (unconscious) conceptual gestalts: an idea stemming from Metaphors We Live By, where it is said that “[…] multidimensional structures characterize experiential gestalts, which are ways of organizing experi-

Lard Andreassen: Introduction to Cognitive Models

2

ences into structured wholes.” (Lakoff & Johnson 1980 p. 81, Lakoff 1987 p. 68). Thus, cognitive models constitute a large part of the background knowledge on behalf of which we reason. Cognitive models define what we take to be the (overall) representative case of comparable cases, and “provides a conventionalized way of comprehending experience in an oversimplified manner. It may fit real experience well or it may not.” (Lakoff 1987 p. 126). Most of the theoretical assumptions stems from within the cognitive linguistics and semantics and tries to capture and put together and develop the notions of Fillmore’s frame semantics, his own and Mark Johnson’s theory of conceptual metaphors (an elaboration of the notion of structural metaphors) and metonymy, Langacker’s cognitive grammar, and Fauconnier’s theory of mental spaces. Lakoff claims that a mental space is structured as an ICMi in the way it is used in a given case. He tries to give a theoretical account of how we structure and organize all of our experiences and our knowledge. As it is suggested by the use of the insights from Lakoff & Johnson 1980 the whole theoretical building of ICMs should be understood in the light of the experiential point of view, the experientialism, or the embodiment hypothesis as I believe it is properly called. An ICM makes use of (at least) the following four structuring principles which are (and I quote from Lakoff 1987 p. 68): -

propositional structure, as in Fillmore’s frames image schematic structure, as in Langacker’s cognitive grammar metaphoric mappings, as described by Lakoff and Johnson metonymic mappings, as described by Lakoff and Johnson

2.1. Bachelor vs. Pope When the lexicalized concept of »bachelor« meets the concept of »pope« we get a clear cut example of reasoning according to the theory of ICMs. Both concepts is made up from a host of background assumptions (what I have called a horizon of expectations). The members of the category “bachelor”, as we know, has the features human, male, adult, and not married. But as Lakoff, with reference to Charles Fillmore, puts it, the word “bachelor is defined with respect to an ICM”: i.e. a context of a human society were people are met with certain expectations about marriage and marriageable age. But the ICM of »bachelor« “is [still] simply an unmarried adult man.” But there’s a lot of instances of unmarried adult men, that do not fit this rigid and non-graded category of bachelors very well. The bachelor-ICM tells us nothing about couples, who has lived together for a long time without getting married, eunuchs, priests, that are not aloud to marry, homosexuals, Tarzan, Mowgli (though he hasn’t got the age yet), or the pope.

Lard Andreassen: Introduction to Cognitive Models

3

The ICM of the pope do in some respects fit the “feature-definition” of bachelor – male, adult, unmarried - but it certainly differs in a lot of very important aspects. Thus, one would make a very poor impression if one asked the pope how many children he has, or of his favorite brothel - that is, meeting the pope with our common knowledge of bachelors. We know that the pope is head (metonymy) of the romancatholic church, that he resides in the Vatican State, and that he is not supposed to marry. This knowledge is also part of our pope-ICM and does not fit the category of bachelors. What does this tells us about reasoning? According to Lakoff this tells us that reasoning implies our ability to adjust an ICM to a given situation, or to adjust and compare different ICMs “noting the ways in which they overlap and the ways in which they differ. One needs the concept of “fitting” one’s ICMs to one’s understanding of a given situation and keeping track of the respects in which the fit is imperfect.” (Lakoff 1987 p. 71) The example tells us about a very important assumption in the theory of experientialism, namely that our knowledge is not truth conditional (that we cannot define every adult unmarried man in accordance to the definition of a bachelor). Meaning is embodied, as they say, which means that human knowledge is based on human perception and interaction with one’s respective environment. The example of bachelor versus pope is not much more than a simple comparison. One does not even have to fully consider the other part, say the pope, in such a clash. Therefore, let us proceed with a more complex example of a cognitive model. 2.2. When is a lie really a lie? A lie is a false statement. This must be clear to everyone and it is also what people consistently states, when they are asked to give a definition of the concept of lying. According to this there should be no problems determining whether something is a lie or not. But the definition doesn’t seem to fit the actual cases of lies (relative to contexts), and the understanding of these very well. As Coleman and Kay wrote in 1981 on the basis of empirical research: “Informants fairly easily and reliably assign the word lie to reported speech acts in a more-or-less, rather than all-or-none, fashion”. (Quoted from Lakoff 1987 p. 71 - my italics). As the matter of fact “actual falsity” turned out to be the least important condition. So how should we account for this more-or-less notion of lies? Eve Sweetser (1984) has discussed the findings of Coleman and Kay. A study that Lakoff refers to and it is his version that form the basis for the following resume. The ICM of lie (from Lakoff 1987 p. 72: partly modified)

Lard Andreassen: Introduction to Cognitive Models

4

THE MAXIM OF HELPFULNESS

People intend to help one another. THE ICM OF ORDINARY COMMUNICATION.

(a) (b)

If people say something, they’re intending to help if and only if they believe what they say. People intend to deceive if and only if they don’t intend to help.

THE ICM OF JUSTIFIED BELIEF

(c) (d)

People have adequate reasons for their beliefs. What people have adequate reason to believe is true.

An example of a prototypical lie is if you’re accused for stealing something and you know (by your full senses) that you did it but deny it. A less representative lie is if you say to your mother that it was a great meal she did prepare, even though you had huge difficulties in swallowing it. In such a situation you’re actually trying to be nice and helpful (helping to make her glad). This example of a lie violates (a) and (b): You don’t mean/believe what you say, but are never the less intending to help. That is a called a social lie, and can be seen as a systematic deviation from the ICM OF ORDINARY COMMUNICATION, where deceit is actually rendered as helpful. Another kind of less representative lies is white lies. They also occur when (a) and (b) is deviated from, but white lies are not helpful, instead they are believed not to harm anyone - e.g. giving someone irrelevant information is not a kind of deceit that is necessarily harmful. I guess teenagers are experts in using white lies: an example could be that you tell your parents that you will be sleeping at a friends place, while the real purpose is to be able to go to a party and you know that your parents would have prohibited it if you asked them for permission. 2.3. Summary The conditions (a), (b), (c), and (d) provides a way to understand the more-or-less flexibility in the graded category of lies. That is, the systematic ways in which we understand different kinds of lies according to context. The conditions also help us to see that these categorizations are based on an internal logic and thus, that our behavior is not totally arbitrary, even though it is not predictable in a strict sense. A very important deviation from the classical point of view to this experiential point of view could be illustrated by the conditions (c) and (d). From these conditions it follows that truth is related to peoples beliefs, rather than to objective “real-world” attributes. As we have just read, it’s not the “actual falsity” that is important (though falsity necessary is involved in lying). According to THE MAXIM OF HELPFULNESS and

Lard Andreassen: Introduction to Cognitive Models

5

the two ICMs it is peoples intentions as related to their beliefs, that plays the crucial role in the categorization of lies. Another important thing concerning the classification of lies is to whether it is harmful or not, or if it’s even considered helpful. Once again, the ICMs gives an account of the flexible reasoning and behavior that governs the understanding of typical communicative situations, rather than a strict truth conditional definition of lies. “The ICMs used are not made up just to account for lie. Rather they govern our everyday common sense reasoning.” (Lakoff 1987 p. 73). But, as the quote indicates (by the word ‘just’), the ICMs does fit the classical view of lies, of course one cannot lie by telling the truth. Thus, the most important characteristic of a lie is still falsity, but falsity is not by necessity rated a bad thing (a sinn as in most theologies, or as in Kant’s moral philosophy a violation of the categorical imperative). What the lying example stresses is that our understanding of situations is by far more complex and flexible than a binary code and requires much more than a certain and rigid set of attributes to guide us. What we also learn from the above examples is some of those features that are recognized as very important to the validity of the theory: “The theory of ICMs allows one to recognize unrepresentative cases and to say precisely what is unrepresentative about them. It also permits the statement of general principles [e.g. conditions (a) to (d) of the two above ICMs] […] but accords them […] cognitive rather than logical.” (Lakoff 1987 p. 130)

3. Categorization, Prototypes and Cognitive Models As I mentioned in the introduction Lakoff asserts that cognitive models to large extent is what structures categories. I have now come to the point where it is appropriate to investigate this assertion and look further on the organization and formation of categories. This could also yield an impression of how the “objective world” influences on the formation of our cognitive capabilities and thus to the formation of cognitive models. This last point has to mainly to do with the second basic principle that Rosch put forward in the following passage: “Two general and basic principles are proposed for the formation of categories: The first has to do with the function of category systems and asserts that the task of category systems is to provide maximum of information with the least cognitive effort; the second has to do with the structure of the information so provided and asserts that the perceived world comes as structured information rather than as arbitrary or unpredictable attributes. Thus maximum information with least cognitive effort is achieved if categories map the perceived world structure as closely as possible.” (Rosch 1978 p. 28).

Lard Andreassen: Introduction to Cognitive Models

6

When we talk about categories we’re talking about how cognitive systems organize perceptual experience, and cognitive models plays, as we will see, an important role in this organization. Categorization is about organisms abilities to identify entities in their environment, i.e. about their ability to perceive equivalencies between different stimuli from members in the same category but also these members differences from members of contrasting categories. A thing that is emphasized by both Rosch and Lakoff is that categories does not reflect a metaphysical world but a perceived world. Though, as Rosch said, it is not totally arbitrary how categories does form. The organization of categories rest on “high correlational structure” (as seen from a perceivers point of view): for example “it is an empirical fact provided by the perceived world that wings co-occur with feathers more than with fur.” (ibid. 29, my italics). There is no meaning to categorization if it doesn’t somehow consent with the surrounding world - the “out there”. On the other hand, if we were to consider every single detail in the phenomenological surroundings - if every thing had its own category - we would be mortally delayed in our abilities to take action. So what is perceived by an organism depend on, or is determined by, many different factors concerning the organisms different needs in order to function successfully in its environment. Therefore, the two above stated principles amounts to the abilities 1) the cognitive economical principle induce organisms to abstract differences among entities (species) as to avoid information overload - that is, not to differentiate when not necessary, 2) but at the same time organisms must have the ability to differentiate a potentially dangerous dog (a wolf, or a hyena) from a harmless one. From this it follows that there must be some sort of correlation between perceived (immanent) and objective (transcendent) structure - otherwise we would simply not be able to behave in an adequate manner. (And of course, it might be added, to see when something is not a dog but a tiger, or to make all dogs look like each other but not as much that they (in our cognition) become cats). 3.1. Basic Level “By category is meant a number of objects that are considered equivalent.” (Rosch 1978 p. 30). This equivalence is, as above suggested, organized as consequence of both organisms perceptual and cognitive abilities and their environments. Now the theory of cognitive categorization claims that there is a basic level of abstraction. And that the basic level can be formalized in terms of a probabilistic concept called cue validity. “The cue validity of an entire category may be defined as the summation of the cue validities for that category of each of the attributes of the category. “ (Ibid. p. 30 - 31). Further-more this basic level is the most inclusive level of a category (consisting of superordinate, basic, and subordinate levels), that is, the level where one find attributes in common of most of the members of the category. This

Lard Andreassen: Introduction to Cognitive Models

7

means that at the level where basic objects are categorized both the total cue validity and the resemblance between members are maximized. This implies that other levels of categorization can be determined by their degree of abstractness: more abstract categories, where members has less attributes in common, are then superordinate (furniture, vehicle), and more detailed levels with a lesser degree of abstraction than both superordinate and basic level, is then subordinate (coffee table, sports car). The objects included under the subordinate level have more attributes in common with objects from contrasting categories than is found among objects on the other levels. It is also the case that both the superordinate and the subordinate level have lower total cue validity than does the basic level. (Ibid. p. 31). In connection to these abstraction processes research has come up with four converging operational principles which has to do with our perception, cognition, and interaction with the entities that are classified as basic-level objects (which has a lot in common with gestalt principles), these principles are: “attributes in common, motor movements in common, objective similarity in shape, and identifiability of averaged shapes.” (Loc. Cit.). These determines the overall shape of things. And the overall shape of things constitute the ability to perceive and psychologically comprehend objects, actions, and events in a holistic way, i.e. to assemble them into coherent structured wholes that are easy to employ for a cognitive system. All of which depends on experiential aspects of human psychology. Thus ease of perception, memory, learning, naming, use, and social functions plays important roles in the constituting of basic level phenomena. (Lakoff 1987 p. 36f, Rosch 1978). A list of these characteristics could look something like the following. (From Lakoff 1987 p. 33): - People name things more readily at that level. - Languages have simpler names for things at that level [largely one-syllable words]. - Categories at that level have greater cultural significance. - Things are remembered more readily at that level. - At that level, things are perceived holistically, as a single gestalt, while for identification at a lower level, specific details (called distinctive features) have to be picked out to distinguish, for example, among the kinds of oak [the kinds of a species]. These assumptions can also account for cultural differences in categories. Lakoff says it in the following way: “Basicness of level has no objective status external to

Lard Andreassen: Introduction to Cognitive Models

8

human beings. It is constant only to the extent that relevant human capacities are utilized in the same way. Basicness varies when those capacities either are underutilized in a culture or are specially developed to a level of expertise.” (Lakoff 1987 p. 38). 3.1.1. Preconceptual Structure Another characteristic of basic level objects is that they seem to be mentally represented by image-schemas, abstract representations that doesn’t fit one specific individual member of a basic level category, but all (or most) of them. This pre-conceptual structure also includes general sensory-motor schemas (kinesthetic imageschemas), schemas for bodily movement and interaction with basic level objects. Image-schemas (of objects) are not rich on details and specific knowledge does not associated to them. They are alike in the sense that they are not defined by context and they are activated without conscious activity. “They are relatively abstract schemas that organize what can be perceived and visualized, but they themselves cannot be directly visualized the way a rich image can be.” (Lakoff 1987 p. 453 - my italics, see also Mandler 1992 p. 592 on this point). This could be exemplified by a very banal but short example: we have an imageschema for “chair” (basic level object) so we quickly categorize something as a chair (or chair-like), and a sensory-motor schema for ‘sitting’ (on/in chairs). Thus something just remotely chair-like, or as a matter of fact not even chair-like, e.g. a lying log, or a tree stump in a wood, could activate the sensory-motor schema for sitting. On the other hand, how to act toward a piece of furniture (superordinate) would depend on what kind of furniture it is – a bed, a table, a drawer, etc - i.e. what we find to be included at the basic level of the category. 3.2. Prototype Effects What specifically is of relevance to this discussion is the prototype effects, because these effects shows, in a very pedagogical way, how categories are formed on the basis of cognitive models. In categories we find asymmetries between the members, that are not supposed to be there according to the classical theory, in which categories were assumed to consist of equivalent members equally interchangeable. Though, Rosch found that in a given category there are members which have a privileged cognitive status. These privileged members are called “best examples” or (when they are relative to contexts) prototypes. “In short, prototypes appear to be just those members of a category that most reflect the redundancy structure of the category as a whole. That is, if categories form to maximize the information-rich cluster of attributes in the environment and, thus, the cue validity or category resemblance of the attributes of categories, prototypes of categories appear to form in such a manner as to maximize such clusters

Lard Andreassen: Introduction to Cognitive Models

9

and such cue validity still further within categories.” (Rosch 1978 p. 37 -my italics).

Prototypes, thus, do often form a reference point in reasoning, e.g. a word in a sentence designating a basic-level object, say a “dog”, does in a given context yield a prototype effect. Thus, as a reference point and on the basis of a cognitive model this designated basic level object gives rise to the recategorization within a category. This also implies that a shift in context often would lead to a new prototype effect. And that’s an important point to make, because it shows how categories reflects not only a perceived world (and not a metaphysical) but also how objects and organisms take on different value and relevance in our eyes as a consequence of our interacting with them. How this is supposed to go on, the following will try to account for. 3.2.1. The Word and the Context In a paper from 1977 Rosch showed that the meaning of words is intimately tied to their use in sentences. The substitutability of a word in a sentence shows the degree of prototypicality to a given category member. “Thus, in the sentence “Twenty or so birds often perch on the telephone wires outside my window and twitter in the morning,” the term “sparrow” may readily be substituted for “bird” but the result turn ludicrous by substitution of “turkey” […]” (Rosch 1978 p. 39). This is one way or the other the working principle in the following examples. The examples also show what propositional structure in a cognitive model might be and thus, how words take on meaning from the frame they are presented in (because the frame includes propositions) (Fillmore 1982). They also illustrate how sentences evoke different cognitive models and how cognitive models evoke a prototype effect (and a recategorization). Examples that illustrates the flexibility in reasoning (once again) and categorization on the basis of cognitive models and prototypes as reference points. (From Ungerer and Schmid 1996 p. 43) 1 The hunter took his gun, left the lodge and called his dog. 2 Right from the start of the race the dogs began chasing the rabbit. 3 She took her dog to the salon to have its curls reset. 4 The policemen lined up with the dogs to face the rioters. [5 The dogs were straying while the alcoholics sat on their bench drinking.]

Lard Andreassen: Introduction to Cognitive Models

10

I believe that all these principles are highly productive in literature. The next examples therefore are supposed to show how the above discussed principles of prototype effects play a crucial role in understanding. First, a ‘real life’ example of how context gives meaning to the word knife - i.e. the more or less precise but never totally arbitrary meaning of a word - which then come to function as the reference point for a recategorization. The example is from a short story of J.L. Borges. (A) ”… toward midnight, in a dive on the Paso del Molino, he is witness to an altercation between a number of cattle drovers. A knife flashes; Otálora does not know which side is in the right but he is attracted by the pure taste of danger, as others are attracted by playing cards or music.”

In the set up with the dive and the cattle drovers its is sufficient to denote “knife” as basic level category object (as opposed to superordinate category of weapon and subordinate as pocket knife) to vividly set the image of a lethal knife before our eyes. The frame (cattle drovers, dives, and fighting) immediately makes us understand what kind of knife we’re dealing with, and that it is not a butter knife or a paper knife (could it be the butchers knife flashing just before he cuts out a fine juicy steak?). As opposed to the above example consider this sentence (which is made up if anyone should doubt) - “The boys were fencing on the playground” - I guess nobody would really believe that the boys would really harm themselves with their weapons. (B) ”Otálora now begins a different life, a life of vast dawns and of days smelling of horses.”

It is clear, I believe, that Otálora is not an accountant and work in a big downtown office building - those kind of places doesn’t smell of horses. (also from J.L. Borges). Now, consider the sentences in (C) (from a novel by William Gibson.). (C) ”Rydell got the Brazilian glasses out, put them on, and scrolled for the number Yamazaki had given him in Tokyo. Someone picked up on the third ring, but the glasses didn’t map a location for the answering phone. Probably meant another mobile.”

(C) is a case were an unknown object is described by the way it is handled by the protagonist. It should be noted, that the reader at this point know that the glasses is equipped with the features of the Global Positioning System (GPS) - that’s what would have enabled the protagonist to “map a location” if it had worked. Though, the glasses has attributes from different “real space” objects that we all ready are familiar with. It could be speculated that the words “scrolled for the number” and “picked

Lard Andreassen: Introduction to Cognitive Models

11

up on the third ring” evoke experiences (cognitive models) respectively of computers (‘scrolling’) and phones. Together with the glasses we mingleii these things together and develop a blend-space with “computer-phone-glasses” in it. This example then show how cognitive models organizes mental spaces “as used” (Lakoff 1987 p. 68), and these mental spaces can integrate by projecting structure from one to another and thereby create emergent structure in the blended space. Never the less, it seems to me, that something more is going on in the “phoneglass” examples - some new meaning popped up from the context which was not to be found in any of the cognitive models of computers, phones, or glasses. This new meaning could be explained in terms of Conceptual Integration Networks and the formation of mental spaces. iii Like all of the above examples this shows a little of how cognitive models (that is, if Lakoff is right in including every aspect of cognition in his theory) structure and organize reasoning, i.e. how to integrate, adjust and make sense of (partly) unknown things by comparing the given situation (the ongoing on-line perception of things) to stored knowledge. I believe that the blend, the creation, or understanding of the glasses as a small “wearable” computer tells us something about how we “live ourselves into” a fictionalized universe, and that the way we do this, is no different from our everyday understanding and reasoning and therefore requires no such thing as a poetic language.iv The example illustrates how newly cognized items requires rearrangement of items already in memory. Later on in the story, as we see in (D) the author can elaborate on these hi-tech “phoneglasses” and/or add further information to the way they work or are handled without creating unnecessary confusion. (D) “Rydell was midway back, through the lower-level crunch, when his sunglasses rang. […] The glasses were acting up; weirdly elongated segments of Rio street maps were scrolling down his field of vision.”

What we were supposed to learn from the examples (C) and (D) was how cognitive models helps us to adjust elements of what we already know to new material. Though without thoroughly analyzing the example it seemed as if the theory of cognitive models ran into some problems but that these might be overcome with the blend theory. In the examples (A) and (B) the objects were denoted at the basic level (knife, dog). As we saw, it was not necessary to say hunting-dog, lap-dog, or racing dog (designating subordinate objects). Though, it could work just to say weapon or animal but it would properly be to unprecise in most cases (it is not easy to give the right impression of mood in a situation or type of person by designating all objects at the superordinate level). The context in which the word is presented evoke the relevant frame (ICM) and we automatically pick out a relevant “image” for the scenario.

Lard Andreassen: Introduction to Cognitive Models

12

The above examples of prototype shift evoked by different situations and the connection of cognitive models to such situations gives an impression of the flexibility in ICMs and the plasticity of our thought or conceptual systems, a flexibility that are necessary to react swift, firmly, and adequate in different situations (which are, as we know, never exactly similar).

4. Anger This final part has as it purpose to go further into details with a cognitive model and to try to explicate some of the structuring and organizing principles in cognitive models themselves as opposed to the former parts that tried to show how cognitive models models our reasoning. The question is, what is ‘inside’ cognitive models? The case study of anger in Women, Fire, and Dangerous Things, made by Zoltan Kövecses and Lakoff argues that there can be found conceptual structure underlying our understanding of emotions. And that this structure is to be found and studied by the theory of conceptual metaphor as devised by Lakoff and Mark Johnson (1980). The study of anger sets out to find the correlation’s and systematicity between lexicalized, but metaphorical understood expressions of anger. (The following exposition is a lightly edited and modified resume of the pages 380 to 400 of that case study.) -

Some lexicalized examples are:

-

He was foaming at the mouth.

-

He lost his cool.

-

She was looking daggers at me.

-

He was doing a slow burn.

-

Watch out! He’s on a short fuse.

-

You make my blood boil.

-

He channeled his anger into something constructive.

-

When I told my mother, she had a cow.

To give a hint of the systematicity in between the above expressions Lakoff writes: “We know, for example, that someone who is foaming at the mouth has lost his cool. We know that someone looking daggers a you is likely to be doing a slow burn or be on a short fuse. We know that someone who’s blood is boiling has not had his anger appeased. We know that someone who has channeled his anger into something constructive has not had a cow. How do we know these things?” (p. 381)

Lard Andreassen: Introduction to Cognitive Models

13

4.1. Folk Physiology, Metaphor, and Metonymy in the ICM of Anger According to a common folk theory anger has some physiological impacts.v Anger is thus understood as having the effect of increasing body heat, internal blood pressure, tightening of muscles (a kinesthetic effect), agitation, and interference with normal perception and functioning. These physiological effects is increased with the intensity of anger, but only to a certain degree. There’s a limit beyond which normal functioning is impaired. On the basis of this folk theory we make use of a general metonymic principle to tell whether people are angry or not, or to judge about the degree of anger. The metonymic principle is this: “The physiological effects of an emotion stands for the emotion.” (p. 382).

This principle, coupled with the folk theory of anger, yields a system of metonymies for anger, where the expressions indicate anger via its supposed physiological effects. As concerned to body heat we find expressions such as hothead, with internal pressure: I bursted a blood vessel. Heat and blood pressure, supposedly, is the cause to redness in the face and neck area and leads to expressions like scarlet, or flushed with anger. Further we find expressions of agitation like quivering with rage, excited/shaking with anger, and where perception and function is interfered we find expressions like blind with rage, seeing red, and I couldn’t see straight. The part of the folk theory of physiological effect that emphasize heat forms the most pervasive metaphor for anger: ANGER IS HEAT. This metaphorical construction applies heat to fluids and solids, and respectively leads to the two metaphorical constructions: ANGER IS THE HEAT OF A FLUID IN A CONTAINER and ANGER IS FIRE. The first of these are much more elaborated in language than is the latter, and the reason for this, Lakoff suspects, is that we in our overall conceptual system have a general metaphor for the understanding of emotions, THE BODY IS A CONTAINER FOR THE EMOTIONS (and I guess that emotions are best conceived of as fluids). THE BODY IS A CONTAINER FOR THE EMOTIONS

-

He was filled with anger/fear/joy/sorrow.

-

She couldn’t contain her joy/anger/fear/sorrow.

-

She was brimming with rage/joy/fear/sorrow.

-

Try to get the anger/fear/sorrow out of your system.

Thus, when the ANGER IS HEAT metaphor applies to fluids and links up with THE BODY IS A CONTAINER FOR THE EMOTIONS we get the central metaphor in the conceptual system (the ICM) of anger. The metaphor is this: ANGER IS THE HEAT OF A FLUID IN A CONTAINER and the following expressions are examples of it:

Lard Andreassen: Introduction to Cognitive Models

-

You make my blood boil.

-

Simmer down

-

I had reached the boiling point.

-

Let him stew.

14

When the heat is off the liquid is cool and calm and thus, coolness and calmness corresponds to lack of anger. 4.2. Elaborating the Central Metaphor There are two ways in which conceptual metaphors is rated productive. 1) The amount of lexical expressions it give rise to. In addition to this “the words and fixed expression of a language can elaborate the conceptual metaphor.” The second way in which metaphors can be productive is by “carryovers” of knowledge from one domain to another. Carryovers are referred to as “metaphorical entailments” and are a part of our conceptual system. In order to investigate this aspect of metaphors Lakoff reminds us of one of the findings from Metaphors We Live By, namely that abstract domains (such as emotions) often is structured by a more concrete domain (with image-schematic structure as vehicle (not to confuse with the old metaphor-pair of tenor and vehicle)). And that we usually have more knowledge of source domains than we have of target domains. This is the structural aspect of the conceptual metaphor theory applied to the theory of ICMs. In the central metaphor of the anger concept the target domain is ANGER and the source domain is HEAT OF A FLUID IN A CONTAINER. Now, lets see what we know of the source domain and how we can extent this knowledge to the target domain. What do we know of fluids? (A) We know that fluids when heated rise in the pot. This knowledge is projected onto the target ANGER in: -

His pent-up anger welled up inside him.

-

She could feel her gorge rising.

-

We got a rise out of him

-

My anger kept building up inside me.

-

Pretty soon I was in a towering rage.

(B) A heated fluid also produces steam and pressure if its in a closed container, so anger (as a target) produces steam and pressure in the body: -

She got all steamed up.

-

Billy’s just blowing off steam.

Lard Andreassen: Introduction to Cognitive Models

-

I was fuming.

-

He was bursting with anger.

-

I could barely contain my rage.

-

I could barely keep it in anymore.

15

(C) Thus the pressure sometimes must be withhold: -

I suppressed my anger.

-

He turned hi anger inward.

-

He managed to keep his anger bottled up inside him.

(D) We also know that things can explode under high pressure, so a person can explode if he’s really angry: -

When I told him, he just exploded.

-

She blew up at me.

-

We won’t tolerate anymore of your outbursts.

(E) In explosions things break and parts dissipate and what was inside the container wells out, this goes for angry persons as well: -

I blew my stack.

-

I blew my top.

-

She flipped her lid.

-

He hit the ceiling.

-

I went trough the roof.

-

His anger finally came out.

-

Smoke was pouring out of his ears.

4.3. Ontology, Epistemology and Structure in Metaphors of Anger When looking further at the central metaphor of anger and the way it is used to express and speak about anger it is, according to Lakoff, possible to say something about how we conceptualize the ontology of anger. He states that anger is conceptualized as a mass and thus, understood as a “mass entity”, which can be seen by the fact that it is the grammar of mass nouns as opposed to count nouns that apply to the word anger: You can ask ‘how much anger …’, but not ‘how many anger … did they evoke at the meeting today.’ Masses usually has scales from zero and up indicating their amount, and thus masses do not exist when the amount is zero. This goes for anger as well, but the mass of anger also has a limit. Fluid in a closed container can only surpress a certain amount of heat before it goes

Lard Andreassen: Introduction to Cognitive Models

16

off. This correlates with the folk theory of physiological effects of anger. A person can only be angry to a certain point, before he loses control and his normal functioning is seriously impaired. We also know that an outburst of anger could be dangerous to the people near the angry person. Lets look further upon the structural aspect of the ANGER IS THE HEAT OF A FLUID IN A CONTAINER metaphor. “The structural aspect of a conceptual metaphor consists of correspondences between a source domain and a target domain. These correspondences can be factored into two types: ontological and epistemic.” (pp. 386 – 387). The ontological factor concerns the metaphorical mapping between entities of source (FLUIDS) and target domains. This illustrate how the image-schemas for -CONTAINER, -UP-DOWN-, and -SPREADING- helps structure the target domain of anger. In this case the container in the source domain maps onto the body in the target domain (ANGER). So what we see here is how image-schematic structure is activated to transport structure from the well known domain of fluids to the complicated and abstract domain of anger via THE BODY IS A CONTAINER FOR EMOTIONS. Listed, the ontological correspondences looks something like the following: -

The container is the body.

-

The heat of fluid is the anger.

-

The heat scale is the anger scale, with end point zero and limit.

-

Container heat is body heat.

-

Pressure in container is internal pressure in the body.

-

Agitation of fluid and container is physical agitation.

-

The limit of the container’s capacity to withstand pressure caused by heat is the limit on the anger scale.

-

Explosion is loss of control.

-

Danger of explosion is danger of loss of control.

-

Coolness in the fluid is lacking of anger

-

Calmness of the fluid is lack of agitation.

The epistemic factor concerns knowledge correspondence between the two domains. In this case it concerns the relations between fluid and anger. Listed: Source: The effect of intense fluid heat is container heat, internal pressure, and agitation.

Target: The effect of intense anger is body heat, internal pressure, and agitation.

Lard Andreassen: Introduction to Cognitive Models

17

Source: When the fluid is heated past a certain limit, pressure increases to the point at which the container explodes. Target: When anger increases past a certain limit, pressure increases to the point at which the person loses control. Source: An explosion is damaging to the container and dangerous to bystanders. Target: A loss of control is damaging to an angry person and dangerous to other people. Source: An explosion may be prevented by the application of sufficient force and energy to keep the fluid in. Target: A loss of control might be prevented by the application of sufficient force and energy to keep the anger in. Source: It is sometimes possible to control the release of heated fluid for either destructive or constructive purposes, this has the effect of lowering the level of heat and pressure. Target: It is sometimes possible to control the release of anger for either destructive or constructive purposes, this has the effect of lowering the level of anger and pressure. What Lakoff has accounted for is how the central metaphor anger IS THE HEAT OF A FLUID IN A CONTAINER characterize correspondences between the source domain and the target domain and how detailed these correspondences are. So far the account enables him to give answers to the question he raised in the beginning of the case study (p. 381). “We can see why someone who is in a towering rage has not kept his cool, why someone who is stewing may have contained his anger but has not got it out of his system, why someone who has channeled his anger in to something constructive has not had a cow.” (p. 388). 4.4. Putting an End to this Never Ending Story As Ungerer and Schmid writes cognitive models are in principle open-ended. Some content of a cognitive model can be related to another, and Lakoff doesn’t end his study of anger at this point. He continues by analyzing the structure of the ANGER IS HEAT metaphor when applied to solids. Then he proceeds to elaborate on other prin-

Lard Andreassen: Introduction to Cognitive Models

18

cipal metaphors expressing different aspects of the understanding of anger. How anger and madness overlap so we get the construction ANGER IS INSANITY, and further to ANGER IS AN OPPONENT (in an internal psychological struggle). The latter shows the way we conceive of anger as a psychological opponent to the way we wish to behave and how we struggle with it in order to prevent us and our surroundings from its disabling or impairing consequences. He also gives an account of another pervasive conceptualization of anger, namely ANGER IS A DANGEROUS ANIMAL (the aggressive behavior of the dangerous animal is angry behavior) and a metonymic elaboration on this: aggressive verbal behavior stands for anger. And the CAUSE OF ANGER IS PHYSICAL ANNOYANCE, CAUSING ANGER IS TRESPASSING, a few more and, in addition, a few minor metaphors (minor to anger but, so it seems, more general to the conceptual system, the examples are EXISTENCE IS PRESENCE and EMOTIONS ARE BOUNDED SPACES. This is not all. Before he ends his study with a discussion concerned the links between sex, anger, and violence he hypothesis of the embodiment of anger. But before that he discusses the prototypical scenario of anger. He finds that this has 5 basic stages: 1) An offending event that evokes 2) anger, and the struggling with this 3) the attempt to control it, 4) the loss of control and 5) the act of retribution.

5. Conclusions What can be concluded from all this? We have seen how context-depended category structure determines the choice of prototype and how this leads to the formation of ad hoc categories that adjust to the concrete perceptual situation. A certain context could emphasize attributes that was not at all prominent in a goodness-of-example rating, or a context could yield what would normally be a very bad example of a category member. “We went ashore at the Antarctic peninsula and the first birds we saw were …” This was explained by the facts that different attributes have different value in different situations. Once again, this adds weight to arguments, that our cognition cannot be explained by classical or objectivist category theories, because classical categories, made up from truth conditionals, are not made to fit the flexibility in cognition and the adjustments adequate reasoning has to do, and which seems necessary when we interact with objects and organism in the world (on the other side of our skulls). The examples in part 2.1. very clearly showed the importance of being able to adjust ones clusters of experience and knowledge to concrete situations in order to function properly in social context. We have seen that even though our thought does not move on behalf of logically connected strings of symbols eating data from a digital worldtape we do not have to consent with pure relativism and constructivism. We have considered some of the notions of structuring and organizing principles with which the “embodiment hypothesis” tries to account for the systematicity in cognition. How

Lard Andreassen: Introduction to Cognitive Models

19

the principle of metonymy in a systematic way stands for a part of something larger and perhaps less comprehendible and how structure by metaphorical entailments in a systematic way is mapped from one domain to another. And that we should accord these general principles cognitive rather than logical. Finally, it seems as if a sort of cognitive hierarchy can be deduced from all of the above: A hierarchy spanning from the notions of image-schematic pre-conceptual structure which play an important part in perception and in organizing and keeping track of the experiential and conceptual content in cognitive models, which then constitute the formation of category structure and prototype effects. We have seen how it can be said that cognitive models, as Lakoff thinks of them, structures the content in mental spaces. But it also seems as if the theory of cognitive models cannot account for the creation of new meaning in on-line conceptual cognition. That the theory reaches an end-point in the meeting with emergent structure in blended spaces and that the theory of conceptual integration is the theory that present the best analyses of that phenomena.

Bibliography Jorge Luis Borges 1967 “The Dead Man” trans. by Anthony Kerrigan. In. A Personal Anthology edited by Anthony Kerrigan. Grove Press. NY. Charles Fillmore 1982 “Frame Semantics” The Linguistic Society of Korea (ed.) Linguistics in the Morning Calm, Seul: Hanshin Publishing Co. William Gibson 1999 All Tomorrows Parties. Penguin. GB. George Lakoff & Mark Johnson 1980 Metaphors We Live By. The University of Chicago Press. George Lakoff 1987 Women, Fire, and Dangerous Things. The University of Chicago Press. Jean M. Mandler 1992 “How to Build a Baby: II. Conceptual Primitives”. Psychological Review, vol. 99, No. 4, 587 - 604 Eleanor Rosch 1978 “Principles of Categorizations” in. Cognition and Categorization. (ed.)

Lard Andreassen: Introduction to Cognitive Models

20

Barbara B. Lloyd & Eleanor Rosch. Lawrence Erlbaum Associates, Publishers. New Jersey. Frederik Stjernfelt 1997 “Kategoriernes kategori” in. Rationalitetens himmel og andre essays. Gyldendal. Dk. Michael Tomasello 1998 “Cognitive Linguistics” in. A Companion to Cognitive Science, (ed.) William Bechtel & George Graham. Blackwell publisher Friedrich Ungerer & Hans-Jörg Schmid 1996 An Introduction to Cognitive Linguistics. Addison, Wesley & Longmann. Malaysia.

Notes i

Idealized Cognitive Model (ICM) is Lakoff’s term for Cognitive Models. In the following I will not put an effort into distinguishing between Idealized Cognitive Models and Cognitive Models. Mainly the difference is that Lakoff, as mentioned, tries to unite all of the theoretical assumptions made in cognitive linguistics and cognitive semantics up to 1987, and therefore, properly, finds it appropriate to invent a new term.

ii

This is not the place to consider the principles governing conceptual integration, so we’ll have to do with “mingle”. Readers interested in these issues could attend to Gilles Fauconnier & Mark Turner. “Conceptual Integration Networks, Cognitive Science.Vol 22 (2) 1998 pp. 133 - 187.

iii

It is not my intentions to go into details with these questions but I believe that the theory of cognitive models would run into the same problem s as the theory of conceptual metaphor does when it comes to accounting for the emergency of new meaning. For a discussion see Joseph Grady, Todd Oakley, and Seana Coulson; “Blending and Metaphor” in. Gibbs & Steen (ed.) Metaphor in Cognitive Linguistics. Benjamins. 1997.

iv

Of course it could be speculated, as Mark Turner does, that the mind is literary from the beginning. The Literary Mind - The Origins of Thought and Language. Oxford University Press. NY. Oxford. 1996.

v

It should be noted that this folk theory in some concerns is confirmed by scientific empirical investigation into emotions. Lakoff himself refers to such a study by Paul Ekman (pp. 38f), and again (pp. 406ff).

Updated Sep 16, 2000 by J-H Skovgaard Please, mail comments to the Webmaster: [email protected]