See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/229060050

Towards A Cognitive Theory of Emotions

Article  in  Cognition and Emotion · March 1987

DOI: 10.1080/02699938708408362

CITATIONS READS

1,111 1,413

2 authors, including:

Keith Oatley
University of Toronto
141 PUBLICATIONS   6,566 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

On Truth and Fiction: Dialogues between Literature and Cognition View project

All content following this page was uploaded by Keith Oatley on 22 February 2019.

The user has requested enhancement of the downloaded file.

This article was downloaded by: [University of Toronto]
On: 27 July 2009
Access details: Access Details: [subscription number 911810122]
Publisher Psychology Press
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Cognition & Emotion

Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713682755

Towards a Cognitive Theory of Emotions

Keith Oatley a; P. N. Johnson-laird b
a
University of Glasgow, Glasgow, UK b MRC Applied Psychology Unit, Cambridge, UK

Online Publication Date: 01 March 1987

To cite this Article Oatley, Keith and Johnson-laird, P. N.(1987)'Towards a Cognitive Theory of Emotions',Cognition & Emotion,1:1,29
— 50
To link to this Article: DOI: 10.1080/02699938708408362
URL: http://dx.doi.org/10.1080/02699938708408362

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
 COGNITION AND EMOTION, 1 9 8 7 , l (1) 29-50

Towards a Cognitive
Theory of Emotions

Keith Oatley
University of Glasgow, Glasgow, U.K.
P. N. Johnson-Laird
MRC Applied Psychology Unit, Cambridge, U.K .
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

A theory is proposed that emotions are cognitively based states which

co-ordinate quasi-autonomous processes in the nervous system. Emotions
provide a biological solution to certain problems of transition between plans,
in systems with multiple goals. Their function is to accomplish and maintain
these transitions, and to communicate them to ourselves and others. Transi-
tions occur at significant junctures of plans when the evaluation of success in
a plan changes. Complex emotions are derived from a small number of basic
emotions and arise at junctures of social plans.

INTRODUCTION
In the first study of emotions to be based on the theory of evolution,
Darwin (1 872) concluded that emotional expressions are a kind of neural
accident. They result from overflows of neural excitation which may serve
no function in the actions of adults. Had he lived later, Darwin might have
used as an example of this superfluity the facial expressions that people
make when speaking on the telephone. His theory was that emotional
expressions are vestiges of evolutionary history or of childhood habit.
This idea that major aspects of emotions are not functional has been
taken up by later writers, who include Dewey (1895) in his attempt to
reconcile Darwin’s work with James’ (1890). He proposed that emotions
are disturbances which occur when habits are no longer useful in present
contexts.

Requests for reprints should be sent to Professor K. Oatley, Department of Psychology,

University of Glasgow, Glasgow G12 8RT, U.K. This paper is based on a presentation to the
Summer Institute on Cognition-Emotion Interrelations held in Colorado, 18-23 August,
1985, and supported by NIMH. We very gratefully acknowledge helpful discussions with
Steven Draper, George Mandler and Richard Power, all of whom commented on a draft of
the paper.

@ 1987 Lawrence Erlbaum Associates Limited

 30 OATLEY AND JOHNSON-LAIRD

Dewey’s proposal was an example of what Mandler (1984) has called a

conflict theory of emotion. In this type of theory, and putting the problem
in cognitive terms, emotions arise as disturbances which accompany inter-
ruptions and discrepancies among multiple goals and representations. The
idea has been developed by cognitive and computational theorists (Mand-
ler, 1964; Simon, 1967; Sloman and Croucher, 1981; Sloman, in press).
Our theory is a conflict theory in Mandler’s sense. Deriving from the
version of conflict theory that Simon (1967) proposed, our theory also
proposes that emotions have important cognitive functions. Emotions may
be disturbing, but they are not just inco-ordinations or side effects.
In this paper we use the term “cognitive” to refer to psychological
explanations in terms of the representation and transformation of know-
ledge which may or may not be conscious. Tacit knowledge is used, for
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

instance, when a cuckoo, hatched in Britain, leaves at the end of the

summer to fly to Africa without benefit of parental training; or when a
human speaker of a language utters sentences grammatically although
unable to give an explicit account of that language’s syntax.
In this cognitive approach, we refer to goals as symbolic representations
of possible states of the environment that a system will try to achieve. Plans
are sequences of transformations between representations, that link a
current state of the environment to a goal. To make a plan is to assemble
such a sequence. To carry it out is to enact the sequence in the world. We
use the terminology of goals and plans to indicate aims and assembled
sequences of action in a way that is neutral as to whether they are
conscious. Thus a conscious plan might be an arrangement to catch a
particular train; plans which are at least partly unconscious include both
those that are instinctive and those that are highly practiced like tying a
shoelace.
We will also use the notion of evaluation of a plan in a neutral way. If an
evaluation is conscious we will use the word “conscious”, which will here
not be a synonym of “cognitive”.
The theory is not a contribution to the debate about whether emotions
require conscious thought, which has been stimulated by Zajonc’s (1980)
argument that preferences need no inferences. We do not claim that all
emotions derive from thinking. Some do and some do not.
The purpose of this approach and this terminology is to lay out a
groundwork of emotional theory that relates to the computationally based
theories of language and perception in cognitive science (see e.g.
Johnson-Laird, 1983a; Gardner, 1985) in which emotions have been very
neglected, although, as we argue, they are central to the organisation of
cognitive processing. The approach is intended t o be a step towards
theories of emotions that can be tested formally, and their implications
explored in computer programs,
 A COGNITIVE THEORY OF EMOTIONS 31

As part of the argument that emotions have important cognitive func-

tions, we propose that they are part of a management system to co-ordinate
each individual’s multiple plans and goals under constraints of time and
other limited resources. We go beyond Simon’s (1 967) suggestions along
these lines, to propose a specific system of internal communication to
achieve this management, and we also propose that emotions are further
evolved in social species such as ourselves to communicate junctures in
mutual plans among individuals in social groups. The human skin is, as it
were, permeable to emotions: an emotion such as anxiety is not only
communicated through the individual human body and brain to set them
into a particular state appropriate to danger, but it also propagates beyond
the individual to influence others and may set them into a similar or
complementary state.
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

What should a theory of the emotions explain? The basic observation is

that adult human beings report subjective experiences with a particular
phenomenological tone that they describe as emotional. These experiences
are consciously preoccupying. They are typically accompanied by certain
somatic events, revealed by facial and other expressions, and lead to
certain characteristic courses of action. A theory should account for these
components, for the diversity of emotions, for the variation in their quality,
and for their relations with other aspects of mental life. It should also
answer the question of whether emotions have a function.
In our theory we bring together and develop two threads of theorising:
one concerns the function of emotions in modular nervous systems, the
other describes the occurrence of emotions at significant junctures of plans
influenced by multiple goals.

SCHEDULING PROCESSES IN A MODULAR

COGNITIVE SYSTEM

Hierarchical Organisation of Processors

Basic to our theory is the proposal that the human cognitive system is
modular and asynchronous as described by Johnson-Laird (1983a, 1983b).
Johnson-Laird describes a module as an autonomous processor. It may be
thought of as a procedure that is relatively self-contained, and that attempts
to complete its computation once it is activated. Each module has an
associated goal, and can accomplish it given certain preconditions and
absence of interruptions. In such systems problems arise as to how to
co-ordinate processors that only compute when they receive the right
input; how, for instance, to avoid pathological situations in which two
processors are each awaiting an input from the other. One solution is to
build hierarchies in which processors at higher levels invoke lower proces-
 32 OATLEY AND JOHNSON-LAIRD

SOTS, in the way described by Miller, Galanter and Pribram (1960). This
method is familiar both from neurological theorising and from computer
programs with many levels of embedding in which one procedure can call
others as subroutines, these in turn can call sub-sub-routines, and so on.
At the top level of the hierarchy of modules in the human cognitive
system there is a processor corresponding to an operating system, capable
of invoking lower level processors in specific sequences o r according to
particular pattern matches. The operating system needs to include, as
Minsky (1968) implies, some model of the whole system. That is, the
human cognitive system needs to include a model of itself, though the
implications of this idea have as yet to be worked out in cognitive science.
The core of our-proposal is that emotions are based on one of two
specific kinds of communication in such systems. One kind of communi-
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

cation is propositional. Propositional signals are symbolic: i.e. they have

internal structure that plays a part in denotation within the system. These
signals correspond to calling patterns and procedure names that pass down
the hierarchy to invoke lower level functions, to representations of differ-
ent aspects of the world, to results and arguments of functions, and to
messages that can construct new procedures.
The other kind of communication is non-propositional. It is simpler,
cruder, and evolutionarily older. Non-propositional signals have no inter-
nal symbolic structure of significance to the system. They do not denote
anything. Like hormones, they function purely causally. They propagate
globally among the processors to set them into specific modes at particular
junctures of multi-goal planning sequences. Emotions are based on non-
propositional communications which we will call “emotion signals”. They
function both to set the whole system suddenly into a particular mode, and
to maintain it tonically in that mode. We will call these “emotion modes”.
Taking fear or anxiety as an example once more: anxiety may occur
when a background self-preservation goal is threatened in the course of
action directed towards a different goal. For instance, while watching
television alone in an empty house you may hear a door open. A fright
interrupts your activity. It sets the whole system into a mode of prepared-
ness for escape o r response, and maintains for a while a state of wary
vigilance (cf. Gray, 1982) with various physiological accompaniments. An
activity has been interrupted. As Mandler (1964) has shown, with such an
interruption an emotion is likely to occur. Some theorists, e.g. Simon
(1967) identify most emotion with the arousal that accompanies the
interrupt signals that are necessary when managing multiple goals. Emo-
tions, however, do not just occur with interruption. Emotionally toned
moods can maintain the system in specific states, and it is a common
observation that episodes of emotion can occur, and moods can persist,
long after the event that elicited them is past. We suggest that the functions
 A COGNITIVE THEORY OF EMOTIONS 33
of emotion modes are both to enable one priority to be exchanged for
another in the system of multiple goals, and to maintain this priority until it
is satisfied o r abandoned.
Emotion signals provide a specific communication system which can
invoke the actions of some processors and switch others off. It sets the
whole system into an organised emotion mode without propositional data
having to be evaluated by a high-level conscious operating system which
would have to reason about an appropriate action. The emotion signal
simply propagates globally through the system to set it into one of a small
number of emotion modes. We imagine that these global signals may be
based on non-specific neural pathways for quick responses, and perhaps on
non-specific peptide or other chefiiical transmission in sustained responses.
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

Five Basic Emotions

We postulate that there is a small number of basic emotion modes which
occur universally in the human species. Each has a characteristic
phenomenological tone-though no meaning as such, as each is based on a
non-propositional signal. On the basis of a variety of classificatory studies
reviewed by Ekman, Friesen and Ellsworth (1982) one may infer that there
are at least five basic emotion modes: they correspond to happiness,
sadness, anxiety (or fear), anger, and disgust. One important criterion for a
basic emotion is that the facial expression associated with it should be
recognised panculturally. On this basis Ekman (1973) has argued that
surprise may be a universal emotion, Izard (1971) that interest may be, and
Panksepp (1982) that there is a specific neural circuit that subserves
curiosity, and that this may be responsible for mediating surprise and
interest. We would argue that surprise and interest are not proper members
of the set because they are not single emotions, but may be aspects of many
emotions. Surprise is elicited by a sudden unexpected event, such as the
door opening in the example just given, and it can indicate an interruption
and an abrupt transition into one of the basic emotion modes. Interest
implies sustained attention to certain external events, and again may be a
feature of other emotion modes. Empirical criteria for universal emotions,
however, are only recently being agreed upon and hence the number of
basic emotions may be adjusted with further evidence. Surpriselinterest
may be a basic emotion in the sense that we are discussing, but with
properties that allow it to combine with other emotions. We concentrate on
just five basic emotions here because the evidence for their universality as
emotions is strongest, and because evidence on whether basic emotions can
co-exist is sketchy.
There are, in addition, several non-emotional modes of the cognitive
system. One such waking mode is conscious construction of a plan-a
CAE 1 I-c
 34 OATLEY AND JOHNSON-LAIRD

certain kind of goal-directed thinking in which the operating system

schedules lower level components. This mode, involving as it does, search,
inference, and evaluation, is slow and liable to mistakes. Some of these
mistakes might violate some of the system’s multiple goals, including
self-preservation. Sussman (1975)found he had to construct a specific and
slow “careful” mode for such reasoning. A second non-emotional mode is
free-association or daydreaming in which a person may be musing with
memories and associations coming to mind asynchronously from a variety
of sources without their being scheduled deliberately by a plan of the
operating system.
We postulate that each emotional mode tends to inhibit the others.
There may also be conflicts in which the system does not settle into one
mode. The system of quasi-autonomous processors will require simulation
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

on the basis of the properties of the emotion signals before its functioning
can be fully specified. We imagine it will have some of the properties of
parallel distributed processors (e.g. McClelland and Rumelhart, 1985).
For an emotion to occur, the cognitive system needs to be in one emotion
mode o r oscillating between two. The intensity of an emotion corresponds
to the amount of the system entrained in a particular mode and to the
consequent degree of locking into that mode.
An emotion mode is a necessary but not a sufficient condition for the full
experience and expression of an emotion-the distinctive phenomenologi-
cal tone, the somatic changes, the behavioural expressions, and courses of
action. By itself the emotion mode based on non-propositional signals only
prepares for action. In adults, the full emotion typically also includes a
conscious evaluation of the juncture in planning, based on propositional
signals reaching the operating system so that it is able to ascribe a meaning
to the emotion mode, and so that voluntary action can be scheduled.
Less developed states also occur. For example, a person may feel a
dysphoric phenomenological tone without being able to ascribe a meaning
to it, or a speaker may talk loudly while leaning forward, but not until
someone else asks why he or she is angry does the speaker experience the
phenomenological tone.
Distinctions must be drawn between emotions and related psychological
states. One distinction is with predispositions to emotion. Enduring predis-
positions are called temperaments.
Temporary predispositions are more difficult. The evidence of physiolog-
ical and facial changes indicates that emotions are brief transitional
phenomena, lasting usually for a few seconds o r at most for a few minutes
(Ekman, 1986). Phenomenologically, emotional states may last longer
than this. So it seems best to regard the term “mood” as ambiguous. It may
either refer to a temporary predisposition to emotion, o r it may refer to an
emotional state, perhaps of low intensity, capable of lasting for many
minutes or several-hours. Either of these meanings could imply a back-
 A COGNITIVE THEORY OF EMOTIONS 35
ground against which more prominent, short-lasting episodes of emotion
occur, and are expressed facially and physiologically.
Other distinctions are with instinctual action patterns such as a carnivore
killing its prey, with motivations like aquisitiveness or hunger, and bodily
states like pain. None of these states are in themselves emotional, although
some motivations or bodily phenomena may lead to a transition by way of
an emotional state.
We propose that a complex emotion, e.g. jealousy or remorse, is an
elaboration of one of the five distinctive modes by means of the proposi-
tional meanings that are ascribed to it, and we will discuss the nature of this
elaboration further.

THE JUNCTURES OF PLANS

Downloaded By: [University of Toronto] At: 18:48 27 July 2009

So far we have described one thread of the argument, that emotions have
evolved as a primitive means of co-ordinating a modular nervous system.
We will now draw out the other main thread, extending the intuition shared
by various theorists since Miller, Galanter, and Pribram (1960), Simon
(1967), and Mandler (1975), that many emotions occur when planned
behaviour is interrupted.
The cognitive system adopts an emotion mode at a significant juncture of
a plan, i.e. typically, as Draper (1985) has pointed out, when the evaluation
(conscious or unconscious) of the likely success of a plan changes. We
assume that these junctures are both distinctive and recurring, so that the
emotional system in mammals has evolved to recognise them and to
establish distinctive responses to them. Indeed the function of these modes
is to organise a transition to a new phase of planned activity directed to the
priorities of the mode with associated goals and certain stored plans for
dealing with what has happened. This mode is then sustained until another
transition occurs.
In some cases a transition is made by default to an instinctual action, e.g.
to freeze when frightened. Such default options have been wired into the
system in the course of evolution as the best general plans for certain kinds
of recurring juncture, where there is danger or when there is insufficient
time or other resources available for reasoning carefully (cf. Sussman,
1975) in the conscious planning mode.
Table 1 indicates five distinctive types of juncture that occur generally in
plans, and the transitions to new sets of goals that are typically accom-
plished by emotions occurring at these junctures.

The Function of Emotions in Planning

Emotions are part of the biological solution to the problem of how to plan
and to carry out action aimed at satisfying multiple goals in environments
which are not perfectly predictable. Examples of the multiple goals simul-
 36 OATLEY AND JOHNSON-LAIRD

TABLE 1
Five Basic Emotions Together With Their Elicitors (the Junctures at Which They
Occur) and Their Effects (the Transitions They Accomplish)

State to Which Transition

Emotion Juncture of Current Plan Occurs

Euphoric
Happiness Subgoals being achieved Continue with plan, modifying
as necessary
Dysphoric
Sadness Failure of major plan or loss D o nothingkcarch for new
of active goal plan
Anxiety Self-preservation goal Stop, attend vigilantly to
threatened environment andor escape
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

Anger Active plan frustrated Try harder, andor aggress

Disgust Gustatory goal violated Reject substance and/or
withdraw

taneously pursued by mammals include: to find supplies of food and water,

to hoard such supplies, to maintain oneself in proper climatic conditions, to
avoid predators, to maintain territory, to find and court mating partners, to
care for young, to guard one’s position in the dominance hierarchy.
Because the environmental niches of sub-human mammals are some-
what unpredictable, models of the environment, although useful (see e.g.
Oatley, 1974), can in principle be neither complete nor wholly accurate.
Likewise, uncertainty prevents any complete dependence o n predictive
models in human planning. Moreover, people typically think only a step o r
two ahead and they respond, moment-by-moment, t o the new arrange-
ments of the environment that their actions help to create. Human plans
are much more flexible than those so far explored in artificial intelligence
(AI).
Most current A1 programs for planning pursue a single main goal using
models of the world to predict future states. They then unreel long
sequences of steps ballistically, receptive only at certain moments to
specific cues anticipated from the environment.
Insects seem to behave somewhat like A1 programs: species-specific cues
act to trigger action patterns rather in the manner of production rules, in
which a simple cue elicits the performance of a specific action. The inherent
dangers of relying on simple cues are absorbed by a low priority given to
the survival of the individual. In mammals, the goals of individual self-
preservation have higher priority. Their actions have multiple goals and
new pieces of planned behaviour can be constructed according to
unforeseen environmental contingencies.
 A COGNITIVE THEORY OF EMOTIONS 37
Mammalian problems of scheduling behaviour are solved by devoting
cognitive resources dynamically in the course of action. This process may
be understood by contrasting it with the switching between fixed action
patterns observed in simpler animals (Tinbergen, 1951) or A1 systems
based on production rules. In such systems a releaser, cue, o r invoking
pattern occurs, and this triggers an action pattern. In mammals, by con-
trast, switching may occur, not between one fixed action pattern and
another, but between modes which are each associated with a small set of
goal priorities. Here an emotion mode activates a set of modules that are
compatible with each other. Hence an emotion mode invokes a limited
suite of goals, action possibilities, and skills. This is indicated in Table 1,
where each emotion calls a small range of alternative plans into consider-
ation. This is not to say that mammals do not exhibit stereotypic behaviour,
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

rather, that when they do, this behaviour is not an emotion.

This contrast between switching of action patterns in response to specific
environmental cues, and transitions by means of emotion modes is illus-
trated by imagining how an insect o r a car-assembly robot would act, as
compared with a mammal, in an environment to which it was not adapted.
An insect or robot would fail disastrously. A mammal would enter into an
emotional state. Although it too might be incapacitated, it might be able to
choose among some alternative actions. Likewise while one can imagine an
insect or a robot in a repetitive oscillation of maladaptive behaviour, it is
hard to imagine it in a conflict or a “difficulty” which seem to be the
prerogative of mammals.
The small range of options invoked by an emotion mode explains some
of the phenomenological quality of emotions: an emotion mode creates a
sense of compulsion though with some slight flexibility. Options in an
emotion mode are more narrowly focused than when in, say, the mode of
free association, but action is still not completely automatic as with a fixed
action pattern of an insect, or a reflex. Emotions function to focus attention
on the matters of the transition: as Tomkins (1979) has argued, they
amplify motivation. According to us, this occurs because an emotion mode
makes some goals into figure while others become ground. Changes in the
evaluation of plans’ success initiate transitions into emotional states in
which what has happened, and what should be done about it, remain
focussed but ambiguous, and hence open to some reasoning.

Emotions: Local Difficulties and Global Problems

Why should there be both euphoric and dysphoric emotions? We propose
that where ambiguity about what should be done at any given point in a
plan is low, and goals of self-preservation are not threatened, mistakes and
blocked paths in plans are assimilated smoothly. The system remains in the
 38 OATLEY AND JOHNSON-LAIRD

emotion mode of “happiness”. “Bugs” (mistakes in the plan) are treated as

local difficulties and “patches” (new pieces of program to repair the bugs)
are created for the current plan from available resources, perhaps in the
way described by Sussman (1975). But when a self-preservation goal is
threatened or when a plan is blocked and the problem cannot be solved
using current resources, or when something happens to reveal an incom-
patibility among the multiple goals, the difficulty ceases to be local. The
previous plan is interrupted, and there is a transition to- a dysphoric
emotion mode. Within this mode, ambiguities that have arisen in the
evaluation of the event have to be resolved. Decisions have to be made as
to whether the current plan should be abandoned altogether o r only
temporanly, what levels of change to it might be required, whether goals
should be changed, and whether current models of the world need to be or
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

can be revised.
Euphoric and dysphoric emotions are well illustrated by, respectively, an
experienced and a novice programmer. For an experienced person, compu-
ter programming can be euphorically fascinating (see e.g. Turkle, 1984),
because it is easy to attain a level of skill in which difficulties remain local.
Bugs occur in a form in which the programmer can see what to do about
them. Though some bugs are undeniably difficult to find, their virtue as a
species is that they unambiguously invite a particular kind of solution, a
patch to the existing program.
Novices, however, do not always cope with difficulties smoothly and
locally. They often suffer dysphoric emotions (e.g. anxiety, anger, disgust,
hopelessness). In a typical scenario the computer waits for some input
while the programmer does not know what to do. Two cognitive systems
are interacting-a technological system with some properties of an A1 plan
as discussed earlier, and a human planner used to flexibility and repair.
Suchman (1985) points out that as far as the technological system is
concerned nothing has gone wrong. In consequence there is no way of
letting it know that anything might be wrong. The system is simply in a
particular state waiting for a cue that it can recognise in that state. The
human novice is distressed and confused because he or she can not use
familiar means of repair, such as those used when a misunderstanding
occurs in a conversation. For the human the situation is deeply ambiguous.
Is one being stupid? Is one likely to be judged as such? If the wrong action
is taken will that further damage the situation? Will the effort of learning
the system be worth it? A potentially local difficulty has become global,
and a transition takes place into a dysphoric emotional state.

THE HUMAN SOCIAL WORLD

In considering emotions in the human social world, the two threads of our
argument, from co-ordination of modular systems, and from signalling the
 A COGNITIVE THEORY OF EMOTIONS 39
junctures of plans, come together and we incorporate evidence from the
evolution of species and from the development of the individual from
childhood to adulthood.
Evolution takes place, as Lorenz (1969) has pointed out, largely where
existing structures and aspects of behaviour are appropriated to new uses.
In computational terms, what starts off as a side effect is developed into
something functional. For example, the control structures of biochemical
systems are liable to oscillate as a side effect of delays in the system. They
have been pressed into service as representational oscillators capable of
entraining on environmental rhythms such as the day-night cycle, and
hence they provide the basis of biological rhythms (Oatley, 1978; 1985).
Emotions in adult humans seem to have resulted from lines of evolution-
ary and individual development in which new functions have developed
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

for existing structures. While many writers have assumed that emotions are
dependent on evolutionarily older parts of the brain, functional arguments
about this evolution are rarer.

Evolution of Emotions
Evolution must have solved a set of design problems in scheduling progres-
sively more complex nervous systems. Switching between action patterns,
characteristic of insect-like invertebrates, has evolved into transitions
between emotion modes characteristic of mammals. We speculate that this
step was taken by an elaboration of nervous systems in which parts became
progressively more specialised for specific goal-directed functions, such as
specific processors to control the escape apparatus of invertebrates like
squids and crayfish, activated by specific cues. Such specialisations form the
basis of nervous systems composed of quasi-autonomous processors, o r
agencies as Minsky (1979) calls them, each with a specific goal, and in
which overall organisation is hierarchical. The emotion of fear o r anxiety
evolved, according to this argument, in animals that had several means of
escape; the choice of freezing or fleeing, the choice of fighting or fleeing.
Darwin (1 872) observed that certain action patterns in lower mammals
have vestigial descendants in humans even though they seem to serve no
useful purpose, e.g. the sneer is a one-sided uncovering of a canine tooth.
Though in humans this expression might accompany a mordant remark, it
no longer prepares for a physical bite. But it may have become a communi-
cation important for the regulation of intra-specific aggression. This type of
progression from innate action pattern to social communication implies an
elaboration of the functions of the emotion of anger, and helps explain how
the sneer has a current human function. At the same time, some expres-
sions including those made by the mouth, which in humans is specialised
for deliberate movements in speech, may have become more subject to
voluntary control.
 40 OATLEY AND JOHNSON-LAIRD

Emotions in Individual Development

Human infants, in comparison with those of other species, are born very
immature. We all start life in a close relationship with another person, a
caregiver, on whom we depend, We are equipped with a repertoire of
expressions: crying, gurgling, and so on. Our caregivers’ responses at this
stage extend our immature behavioural repertoire.
To start with, as Emde argues (see e.g. Emde, 1983; Johnson, Emde,
Pannabecker, Stenberg, & Davis, 1982) a baby’s expressions are simply
biological, perhaps reflex. But caregivers interpret them as intentional, and
as emotional: the baby wants to be fed and is imtable, or is uncomfortable
and sad, or is calm and happy. Moreover, adults judge photographs of
infant facial expressions in the same discrete categories that are found
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

cross-culturally: namely happiness, sadness, fear, anger, and disgust-as

well as surprise and pain. Caregivers act as if such expressions are signals to
make transitions appropriately into specific modes of caregiving: to feed, to
comfort, to gaze lovingly. The expressions acquire significance for the child
only through interactions with the caregiver. Later in childhood as Emde
(1983) argues, they will be experienced as emotions by the child. The child
also becomes sensitive to emotions in others and uses these as signals of
“social referencing” to regulate his or her own behaviour.
Table 2 illustrates how emotions develop in the individual. The leftmost
column describes the set of characteristic junctures in human plans that
give rise to emotions, and the second column shows the corresponding
basic emotions. In early childhood the crucial junctures typically concern
the relations between caregiver and child (see Bowlby, 1969-80). The
early social emotions engendered in this way are shown in the third
column.
The early social emotions seem to be programmed innately to arise when
crucial junctures occur, and consequently, as Ainsworth (e.g. 1967) has
shown, they take similar forms in widely different cultures. As children
grow older and become socialised, their planned activities diversify. They
enter into mutual plans and arrangements that call for intentional commit-
ment to others. Mutual plans are partly under the control of both partici-
pants and partly governed by conventions of their society: Certain junctures
are still critical, and they still give rise to the same emotion modes. What
changes as a concomitant of mutuality is that the range of cognitive
interpretations of the emotion modes is extended to generate the adult
social emotions. The fourth column of Table 2 gives some examples of
these adult social emotions.
Moving rightwards along any line of Table 2 the reader encounters a
developmental sequence of increasingly more elaborate cognitive interpre-
tations that create emotion modes, and that become part of the actors’
 A COGNITIVE THEORY OF EMOTIONS 41

TABLE 2
Examples of Social Emotions Developed on the Original Basis of Biological
Emotions of a Single Actor

Juncrurcs Basic I nfanr Social Adult Social

of Plam Emotion Emotion Emotion

Sub-goals being Happiness Emotions of Sexual love,

achieved attachment Delight
Failure of major Sadness Emotions of Depression,
plan or loss of loss Disappointment
active goal
Self-preservation Anxiety Separation Embarrassment,
goal threatened anxiety Horror
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

Active plan Anger Rage Vengefulness,

frustrated Bitterness
Gustatory goal Disgust Disgust at Distaste,
violated faeces etc. Loathing

understanding of them. Only humans reach the last stage (in the last
column) in which instinctual structures of attachment provide foundations
for culture and language.

THE DEVELOPMENT OF MUTUALITY

With the exception of fear, which often occurs in modem life as the result
of such events as near traffic accidents (Scherer, 1984), most emotions of
interest to humans occur in the course of our relations with others. They
are social emotions, and any theory of emotions must take this social
dimension seriously, and not merely assume that our relations with other
people are an extension of the way we treat the physical environment. We
have argued that emotions are a biological solution to the problem of
co-ordinating planned action with multiple goals in a world that is only
partly predictable. The co-ordination of action among a social group
involves a type of cognitive processing different from that used to interact
with the physical world. For adults, to interact socially requires mental
processes that allow the construction and execution of mutual plans, in
which two cognitive systems co-operate. These processes depend crucially
on each actor having a “model of the self”.
There are two main questions about the nature of this model: How the
nervous system might have a structure to contain such a model, and what
the contents of the model might be.
As to structure, Johnson-Laird (1983a; 1983b) has argued that the
human cognitive system has evolved to contain a recursively defined model
 42 OATLEY AND JOHNSON-LAIRD

of itself. The mind is aware of itself, at least to some extent. Part of the
mind’s ability to construct models involves the ability to embed models of
our own mind in our mind. In deliberate action, for instance, we not only
act according to a goal, but we can know that we are able to do so. This
ability, is re-presented in a model of the self, and indeed only with such a
re-presentation would we be said to have an intention to act.
As to content, as argued by Oatley and Bolton (1985), the model of self
develops from culture and language. Selman (1980) and Damon and Hart
(1982) have traced the development of the sense of self. By adolescence
people talk of this self as able to monitor and control some thoughts and
emotions. Harris, Olthof, and Terwogt (1981) have shown that children
are articulate about a range of emotions as “inner” experiences by the age
of 11. We argue that only with the development of a reflective sense of self
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

can the full set of complex emotions occur. Some of these depend on a
person feeling the self to be enhanced (e.g. by falling in love), or damaged
(e.g. in betrayal by others, or, as we will describe in a later example, by
contradicting one’s own definition of self). Thus in early childhood an
individual might be anxious and clingy, but to talk of a lack of confidence
depends on an adult sense of self. Though sadness is common in childhood,
low self-esteem in depression is an adult experience.
It is a common assumption that representations of the self are inherently
social, and first become accessible in consciousness as a result of relation-
ships with others. Mead (1912, p. 141) wrote: “Inner consciousness is
socially organized by the importation of the social organization of the outer
world.” Like James (e.g. 1890) he proposed that there was an aspect of the
self that can be an object of thought. It corresponds to the model of self.
Mead (1913, p. 145) also argued that this aspect of self acts to monitor
ongoing activity, “criticising, approving and suggesting, and consciously
planning.” Though without the computational metaphor, Mead described
some of the functions and reasons for an operating system containing a
model of the whole system, including some of its goals and operations.
The content of the model of self includes an abstraction of what we have
experienced in others’ reactions to us. At first it is parents who hold up the
social mirror. Mead (1913, pp. 146-147) states that:

The child can think about his conduct as good or bad only as he reacts to his
own acts in the remembered words of his parents . . . and the self which is a
fusion of the remembered actor and this accompanying chorus is somewhat
loosely organized and very clearly social. Later the inner stage changes into
the forum and workshop of thought. The features and intonations of the
dramatis personae fade out and the emphasis falls upon the meaning of the
inner speech, the imagery becomes the barely necessary cues.

Mead thus claims, in essence, that an upbringing by adults programs new

processors in the child, and one of these is the model of self, with the
 A COGNITIVE THEORY OF EMOTIONS 43

consequences of being able in adulthood to talk about “looking after

oneself” o r of “not being able to control oneself”. As Mead makes clear,
the model of self contains some of the conventions of a community. It
provides the means by which values are maintained and propagated from
one generation to another. It is the social glue that holds a society together.
Mead (1913) went on to remark how- in normal activity, self-
consciousness is rare. People’s actions are in register with their monitoring
self, and correspond to habit, to character, to what they expect, and to what
others expsct of them. It is only when “an essential problem appears, there
is some disintegration in this organization, and different tendencies appear
in reflective thought as different voices in conflict with each other. In a
sense the old self has disintegrated, and out of the moral process a new self
appears (p. 147):’
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

This phenomenon of becoming self-conscious when a problem arises in a

social plan is a typical part of the experience of adult emotion. The emotion
mode generates pervasive signals which co-ordinate lower level modules
and perhaps initiate bodily changes; the emotion signals focus attention
and hence preoccupy the operating system. So for instance with a severe
loss, a person might experience grief, and the inner dialogue is devoted to
coming to terms with the loss. After an insult, a person feels angry and the
inner debate may concern the means and advisability of retaliation. In
general, conscious reflection arises from the critical juncture, and concerns
such matters as its cause and its consequences for goals and plans. Each of
these matters is usually highly ambiguous, and so the inner debate may be
about whether to adjust the model of the self or the model of the other,
about new plans, or new goals. Plans are evaluated in mental simulations in
which conflicts are often detected among the multiple goals of the goal
hierarchy. It is the propositional messages associated with conclusions from
such operations that rise to consciousness, as Mead (1913) described, as
voices in debate.

Mutual Plans
Many adult human plans are mutual: They are social, but unlike the
attachment activities of infant and caregiver, they depend partly on con-
scious negotiation and cultural conventions. Mutual plans cannot be
innately wired into the cognitive system; they must be created in the minds
of more than one individual by implicit or explicit agreement. Such plans
are among the most important that we make: in marriage, parenthood,
employment, friendships etc. Many of our more intense and problematic
emotions concern plans where mutuality has been sought for, set up, o r
assumed.
The creation of a mutual plan requires a more complex kind of operation
than one to schedule actions in the physical world. One cannot model other
people merely as complicated physical objects, o r treat them by simple
 44 OATLEY AND JOHNSON-LAIRD

strategies as in the theory of games. Plans become mutual when we

negotiate, exchange knowledge, correct misunderstandings, and enter into
shared intentions. Much of language is used in setting up, readjusting, and
commenting on mutual plans and the assumed o r established conventions
that underlie them.

How to do Things With Promises

We have explained some of the biological mechanism underlying basic
emotions, but some of the complex emotions that human beings experience
require an understanding of the setting up and maintenance of mutual
plans. Since the setting up of mutual plans is fundamental to social
emotions, and different from planning that involves only a single actor in
the physical world, we give here an example of setting up a mutual plan,
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

and the emotions elicited by its non-completion. One way to set up a

mutual plan is for one person to make a promise to another. A promise
creates an obligation in the speaker and a corresponding expectation in the
recipient of the promise.
Consider Searle’s (1969) analysis of promising. We follow Power (1984)
in naming the two actors Xavier (X), the promiser, and Yolande (Y), the
promisee. Searle argues that an utterance (U) of X is a validly performed
promise to perform the action (A) if, and only if, an extensive set of
extra-linguistic conditions are met. These include what Searle calls the
“essential” conditions for promising, as follows.
(a) X intends U to place him under an obligation to do A.
(b) X intends to produce in Y the knowledge that the utterance of U is
to count as placing X under the obligation to d o A.
Now imagine that Xavier has promised Yolande t o call at her house to
feed her cat while she departs for a week’s holiday. Yolande is happy to
have placed her cat in safe hands, and happy to have established a mutual
plan with her friend. Xavier is working on a paper for a conference and
forgets to feed the cat. A day before Yolande’s return he comes across a
reference to T. S. Eliot. He thinks of Eliot’s poems about cats, and
suddenly with a start, and a pounding of the heart, remembers his promise.
He interrupts what he is doing, finds the key she has left, and rushes to her
house. The cat is nowhere to be seen. The cat does not turn up. The
neighbours have not seen it for days. Xavier finds himself preoccupied with
thinking what to say to Yolande. He sadly contemplates the other implica-
tions of his omission, and experiences the emotion of remorse. When he
finally meets Yolande his apology is accompanied by bodily disturbances
and by gestures of agitated deference.
Xavier’s emotion occurs partly because there has been a mismatch
between his goal, the obligation to d o A, and his actual behaviour, non-A.
 A COGNITIVE THEORY OF EMOTIONS 45

One condition for the emotion of remorse is a sincere promise which has
been broken. A promise, in Searle’s sense, involves the act (A) being
anticipated. In remorse non-A is remembered. Xavier would not feel
remorse for the loss of Yolande’s cat if he had not made any promise, if his
promise had been insincere, or if he had amnesia for his omission.
To experience remorse, however, more is needed than the mismatch
between the obligation or goal of doing A and the non-performance of A:
it is a higher order cognitive appraisal with at least some conscious
components, and based on the model of the self that played an essential
part in mediating the mutually agreed plan. To experience remorse Xavier
must infer that Yolande will regard his broken promise as an instance of
untrustworthiness, and this perception of him by Yolande becomes dis-
crepant with Xavier’s own model of himself, that he is trustworthy. It is
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

Xavier’s understanding of Yolande’s perception of him which is discrepant

from his model of himself as a trustworthy person: the sort of person who
would kindly offer to look after a friend’s cat while she was away.
Moreover this understanding of Yolande’s perception of him has sufficient
weight for Xavier to suffer a loss in his conception of himself. His model of
himself suffers damage and he is no longer able to experience himself as a
trustworthy person, at least in his relations with Yolande.
Power (1984) has provided an analysis of mutual intention as follows. In
order to have a mutual intention both X and Y must intend some particular
goal (G). They must assume that the other intends G, and they must
assume also that the other assumes that they intend G. Assuming what the
other assumes is theoretically an infinite series, but rather than cutting it at
some arbitrary point, Power proposes a recursive formulation of the
concept. Each actor intends his or her action (A) to achieve the-goal. The
two cognitive systems then include the goals and the assumptions each has
of him or herself, and of the other.
Power’s formulation is based on e E h actor assuming that the other will
act in a specific plan, i.e. on inter-reliance. A slightly different formulation
would be that each actor has some knowledge or beliefs about the cognitive
states of the other. This modification, while losing something from Power’s
formulation as a competence model of mutual intention, allows a clearer
view of the performance issues; the requirement that actors convey rele-
vant knowledge about intentions on which reliance could then be based. A
revised formulation of Power’s analysis would then yield the following for
the relevant parts of the cognitive systems of two actors who have estab-
lished a mutual plan.

For X: For Y:
1. X intends G Y intends G
2. X knows (X intends G) Y knows (Y intends G)
3. X knows what Y knows Y knows what X knows
 46 OATLEY AND JOHNSON-LAIRD

Line 2 of this analysis is not in Power’s account but is important, and part
of the idea that social actors must have a model of the self. It follows from
this analysis that for two actors to make a mutual promise is to set up these
cognitive structures. Together they must establish the mutual goal G, each
person’s action A to fulfil it, and a joint knowledge of everything relevant
to that goal. A promise, therefore, is the paradigm of the type of speech act
by which a mutual understanding, alliance o r contract is established.
Mutuality is important for the theory of emotions, partly because achiev-
ing social co-operation itself creates an emotion mode (happiness), and
partly because failures to achieve o r sustain it have dysphoric emotional
consequences. The emotion mode arising from mutuality or its breach is
communicated to oneself and other persons involved. Moreover, euphoric
emotions themselves can become goals to be achieved by mutual action.
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

As well as Searle’s conditions for promising, an actor requires a model of

self to make a promise or undertake a mutual intention: line 2 was added to
Power’s analysis because X and Y do not just intend, they must also know
that they intend the mutual goal and their part in achieving it. As well as
representing goals in this way, the model of self also represents expectancies
about oneself, for instance that one will not break a promise. Keeping part
of a mutual agreement then constitutes a significant juncture, confirming
the model of self, and potentially giving rise to a euphoric emotion.
Moreover the social communication of emotions leads each actor to
become aware of the other’s euphoric feelings, and a euphoric mutual
emotion is created. Such emotions act to cement social relations.
Failure to keep an agreement, however, produces a discrepancy between
outcome and the expectation generated by the self-model, hence a dys-
phoric emotion, such as Xavier’s remorse.
Many mutual relations are established without explicit promises or
acknowledgments. They arise implicitly by precedent and custom, in
families, friendships, in larger communities, and even in nations. The fact
that we can feel pride, disappointment, or outrage as evaluations of what
happens to other people in our social groupings indicates that cognitive
structures similar to those entered into explicitly also underlie our partici-
pation in these groupings, and that these too can be analysed in the kind of
way that Power has shown. For instance Rawls (1972) has shown that the
principle of justice as fairness in society has a similar implicit contractual
basis.
THE COMPLEX EMOTIONS
Many adult emotions are complex in the sense that they are founded on a
basic, non-propositional, emotion mode, but have a propositional evalu-
ation which is social and includes reference to the model of the self.
In the foregoing section we analysed an example of a complex emotion,
 A COGNITIVE THEORY OF EMOTIONS 47
remorse. It is founded on a basic mode, with its underlying phenomenolog-
ical tone of sadness, physiological accompaniments, preoccupation,
behavioural expressions, and so on.Such emotions depend on an appraisal
in which performance is compared with that which is expected on the basis
of the model of self. In remorse, an aspect of the sense of self is lost: The
model of self suffers a decrement in its attributed positive characteristics.
The example of Xavier’s remorse also illustrates two further points.
First, as Katz (1980) has pointed out, an emotion may be part of a
sequence of emotional states in which one mode gives way to another as
events unfold or new evaluations occur. Phenomenologically this can appear
as emotions following one another in sequence. For Xavier, anxiety was
replaced by remorse as the threat that something bad might have happened
was replaced by the realisation that it had.
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

Secondly, a complex emotion may start by being quite inchoate: Only

with substantial reasoning about the situation and its implications may the
full complex emotion develop as it did for Xavier as he contemplated his
unreliability. A single emotion mode can thus give rise to a considerable
range of complex emotions depending on the details of propositional
evaluations that have been made. This phenomenon suggests an alternative
account of the psychoanalytic concept of displacement (e.g. Freud, 1901).
But it also implies that the range of possible interpretations of a basic
emotional state is narrower than that implied by Schachter and Singer
(1962). The hypothesis that autonomic arousal can be interpreted as any
emotion has been disconfirmed (see e.g. Manstead and Wagner, 1981;
Reisenzein, 1983). Our theory retains the idea that the experience of an
emotion can change as one mode gives way to another and as shifts in the
evaluation of a single mode occur. It does not predict, as did Schachter and
Singer’s theory, that bodily states are completely ambiguous as to the
emotion modes from which they have arisen. Instead it is consistent with
Ekman, Levenson and Friesen’s (1983) finding that several basic emotions
are physiologically distinguishable.
Basic emotions are developed from universal biological mechanisms.
Complex emotions are founded on these, but plans and their evaluations
vary from culture to culture and from person to person. Thus the complex
emotion of remorse that we have described might be appropriate to a
Western culture. By contrast Harrt, Clarke and De Carlo (1985) describe
how in Aquinas’ discussion of emotions “accidie” is accorded the lengthiest
treatment. In mediaeval times this emotion occurred with a failure of
religious duty. Although related to remorse in being based on sadness it
had cognitively the sense of loss of intimacy with God. In a cross-cultural
example, Morsbach and Tyler (1976) describe “arnae”, a Japanese emo-
tion evidently based on happiness, but with a sweetish quality of childlike
dependence that occurs between adults as lovers.
 48 OATLEY AND JOHNSON-LAIRD

CONCLUSION
We have proposed that emotions serve two principal functions. First they
are a form of internal communication that sets cognitive processors into
one of a small number of characteristic modes. This system of non-
propositional communication has evolved as a method of changing the
relative priorities of goals, and maintaining these priorities, within a
parallel system of planning. The cognitive evaluation of the situations that
create such junctures contributes to the phenomenology of emotional
experience. Second, emotions are a form of external communication,
important in the adjustment of social relations, Complex emotions charac-
teristic of adult life arise at junctures in social plans, which often concern
mutual goals and a model of self.
The theory resembles some existing theories of emotion at several
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

points, but the constellation of its proposals is novel. Amongst its major
implications are the following. There is a small number of innately deter-
mined emotion modes that underlie all emotional experiences, facial
expressions, and behaviours, based on a specific cognitive signalling sys-
tem. The indefinite number of complex emotions derives not from mixtures
of these primitive emotions, but on propositional evaluations that interpret
specific basic emotion modes. These evaluations of junctures of plans are
likely to vary from one culture to another. They cannot, however, modu-
late just any state of arousal to produce a specific emotion. The basic
emotion modes are physiologically as well as psychologically distinct.

Manuscript received 14 June 1986

Revised manuscript received 23 September 1986

REFERENCES
Ainsworth, M. D . S. (1967). lnfnncy in Uganda: Infant care and rhe growth of artachment.
Baltimore: Johns Hopkins Press.
Bowlby, J . (1969-1980). Afrachmenf and loss, Vo& I , 2 and 3 . London: Hogarth Press.
Damon, W. & Hart, D . (1982). The development of self understanding from infancy
through adolescence. Child Dcvefoprnenf,53, 841-864.
Darwin. C. (1872). The expression of rhe emotiom in man and rhe animalr. Reprinted,
Chicago: University of Chicago Press, 1965.
Dewey, J. (1895). The theory of emotions 11: The significance of emotions. Psychological
Review, 2 , 13-32.
Draper, S. W. (1985). A micro-survey of ten theories of emotion. Unpublished technical
memorandum, University of Sussex.
Ekman, P. (1973). Cross-cultural studies of facial expression. In P. Ekman (Ed.), Darwin
and fociul expression: A century of research in review. New York: Academic Press.
Ekman, P. (1986). Expression and the nature of emotion. In K. R. Scherer & P. Ekman
(Eds.), Approuches ro emotion. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc.
 A COGNITIVE THEORY OF EMOTIONS 49
Ekman, P., Friescn, W. V., & Ellsworth, P. (1982). What emotion categories or dimensions
can observers judge from facial behaviour. In P. Ekman (Ed.), Emotion in the humun face,
(2nd ed.) Cambridge: Cambridge University Press.
Ekman, P., Levenson, R. W., & Friesen. W. V. (1983). Autonomic nervous activity distin-
guishes among emotions. Science, 2 2 1 , 1208-1210.
Emde, R. N. (1983). The prerepresentational self and its affective core. The Psychoundytic
Study o f the Child, 38, 165-192.
Freud, S. (1901). The psychopathology of everyday life. English translati0n:'Standard edi-
tion of the complete psychological works of Sigmund Freud, Vol 6, London: Hogarth
Press, 1960.
Gardner, H. (1985). The mind's new science: A history of the cognitive revolution. New
York: Basic Books.
Gray, J. A. (1982). The neuropsychology of unriety: An enquiry into the functions o f the
septo-hippocampal system. Oxford: Oxford University Press.
HarrC, R., Clarke, D., & De Carlo, N. (1985). Motives und mechanisms: An introduction to
the psychology of action. London: Methuen.
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

Hams, P., Olthof, T., & Terwogt, M. M. (1981). Children's knowledge of emotion. Journal
o f Child Psychology and Psychinrry, 2 2 , 247-261.
Izard, C. E. (1971). The face of emotion. New York: Appleton Century Crofts.
James, W. (1890). The principles ofpsychology. New York: Holt.
Johnson, W. F., Emde. R. N., Pannabecker, B. J., Stenberg, C. R., & Davis,
M. H. (1982). Maternal perception of infant emotion from birth through I 8 months.
Infunt Behavior and Development, 5 , 313-322.
Johnson-Laird, P. N. (1983a). Mental models: Towards a cognitive science o f lunguuge,
inference and consciousness. Cambridge: Cambridge University Press.
Johnson-Laird, P. N. (1983b). A computational analysis of consciousness. Cognition und
Brain Theory, 6 . 499-508.
Katz, J. M. (1980). Discrepancy, arousal and labelling: Towards a psychosocial theory of
emotion. Sociological Inquiry. 50, 147-156.
Lorenz, K. (1969). Innate bases of learning. In K. H. Pribrarn (Ed.), On the biology of
learning. New York: Harcourt, Brace & World.
Mandler, G. (1964). The interruption of behavior, In D. Levine (Ed.), Nebraska symposium
on motivation, Vol. XII.Lincoln: University of Nebraska Press. pp. 163-219.
Mandler, G. (1975). Mind and emotion. New York: Wiley.
Mandler, G. (1984). Mind and body; Psychology ofemotion and stress. New York: Norton.
Manstead, A. S. R. & Wanger, H. L. (1981). Arousal, cognition and emotion: An appraisal
of two-factor theory. Current Psychologicul Reviews, 1 , 35-54.
McClelland, J. L. & Rumelhart, D. E. (1985). Distributed memory and the representation
of general and specific information. Journal of Experimental Psychology: Generul, 114,
159-188.
Mead, G. H. (1912). The mechanism of social consciousness. Reprinted in A. J. Reck (Ed.),
Selected writings of George Herbert Mead. Indianapolis: Bobbs-Memll. 1964.
Mead. G. H. (1913). The social self. Reprinted in A. J. Reck (Ed.), Selected writings o f
George Herbert Meud. Indianapolis: Bobbs-Merrill. 1964.
Miller, G. A., Galanter, E., & Pribrarn, K. (1960). Plans and rhe srrucfureof behavior. New
York: Holt, Rinehart & Winston.
Minsky, M. (1968). Matter, mind and models. In M. Minsky (Ed.), Semantic informution
processing, Cambridge, Mass: MIT Press.
Minsky, M. (1979). The society theory of thinking. In P. H. Winston & R. H. Brown (Eds.),
Artificial Intelligence: An MIT perspective, Vol. 1. Cambridge, Mass: MIT Press.
Morsbach, H . & Tyler, W. J. (1976). Some Japanese-Western linguistic differences concern-
 50 OATLEY AND JOHNSON-LAIRD

ing dependency needs: The case of “Amae”. In R. H a r d (Ed.), Life sentences: Aspecu of
the social role of language. Chichester: Wiley.
Oatley, K. (1974). Circadian rhythms and representations of the environment in motiva-
tional systems. In D. J. McFarland, (Ed.), Motivational control systems analysis. London:
Academic Press.
Oatley, K. (1978). Perceptions and representations: The theoretical bares of brain research.
London: Methuen.
Oatley, K. (1985). Representations of the physical and social world. In D. A. Oakley (Ed.),
Brain and mind. London: Methuen.
Oatley, K. & Bolton, W. (1985). A social-cognitive theory of depression in reaction to life
events. Psychological Review, 92, 372-388.
Panksepp, J. (1982). Towards a general psychobiological theory of emotions. Behavioral
and Brain Sciences, 5 , 407-468.
Power, R. (1984). Mutual intention. Journal for the Theory of Social Behaviour. 1 4 ,
85-102.
Rawls. J . (1972). A theory of justice. Oxford: Oxford University Press.
Downloaded By: [University of Toronto] At: 18:48 27 July 2009

Reisenzein, R. (1983). The Schachter theory of emotion: Two decades later. Psychological
Bulletin, 94, 239-264.
Scherer, K. R. (1984). Emotion as a multi-component process: A model and some cross-
cultural data. In P. Shaver (Ed.), Review of personalify and social psychology, 5:
Emolions, relationships, and health. Beverly Hills: Sage.
Schachter, S. & Singer, J. (1962). Cognitive, social and physiological determinants of
emotional state. Psychological Review, 69, 379-399.
Scarle, J. R. (1969). Speech a m : A n essay in the philosophy of language. Cambridge:
Cambridge University Press.
Sclman, R. (1980). Thrgrowrh ofinterpersonal understanding. New York: Academic Press.
Simon, H. A. (1967). Motivational and emotional controls of cognition. Psychological
Review, 74, 29-39.
Sloman, A. (In press). Motives, mechanisms and emotions. Cognition and Emofion.
Sloman, A. & Croucher, M. (1981). Why robots will have emotions. International joint
conference on arrificial intelligence, 7.
Suchman. L. A. (1985). Plum and siruated actions: The problem of human-machine com-
munication. Xerox PARC, ISL 6.
Sussman, G. J . (1975). A computer model of skill acquisition. New York: American
Elsevier.
Tinbergen, N. (1951). The study of instincr. Oxford: Oxford University Press.
Tomkins, S. S. (1979). Script theory: Differential magnification of affect. Nebraska Sym-
posium on Motivation, 26, 201-240.
Turkle, S. (1984). The second selt Computers and the humnn spirit. New York: Simon &
Schuster.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. Americnn
-
PSyChOlOgist, 35, 15 1 175.

View publication stats