New Scientist vol 123 issue 1678 - 19 August 89
The importance of being emotional: Recent theories in cognitive psychology allow us to understand that emotions are not especially irrational. Rather, they are important in the management of our goals and actions
WE ARE ambivalent about our emotions. Sometimes they seem to make us think in a distorted way. To say that someone is being emotional is to be insulting. But on the other hand, we regard emotions as important to our humanity. To be without them would be less than human.
This ambivalence is depicted in science fiction. Mr Spock of Star Trek is superintelligent and without emotion. But he is a lonely figure - not the person to identify with as one boldly goes across the universe. So the question is, do emotions impede rationality? If we were fully rational, would we need them? Would an intelligent being from another planet have emotions? Would a robot? Are emotions an important part of being human? And if so, how? Perhaps science can help to answer such questions. Most important here has been the work of Charles Darwin. His book published in 1872, The Expression of the Emotions in Man and the Animals, touches on a fundamental dilemma about the nature of emotions, and the way we view them.
Darwin thought of emotional expressions as vestiges of patterns of action that once were useful, but are so no longer. Earlier studies of anatomical vestiges had provided evidence for evolution. We have, for instance, a row of vertebrae at the base of our spine which indicate that our ancestors had tails. Now that we are tail-less, these bones have no function. It is the same with emotional expressions, Darwin argued. He wrote that their study 'confirms to a certain limited extent the conclusion that man is derived from some lower animal form'.
With this theory, Darwin gave support, perhaps unintentionally, to an intuition already strong in Western culture, that emotions are subverters of reason, matters for infants and beasts, but scarcely to be approved of in adult humans.
At the same time, he admitted that emotional expressions are important for human welfare. He stopped short of saying that they have functions - that is, that they have evolved because they are adaptive in some way. To do so would have contradicted his observation that many emotional expressions are not functional in many circumstances. He collected evidence of activity that was superfluous to efficient action: tears that do not serve to lubricate the eyes, hair that stands on end, adding nothing to the skill of an attack, laughter that seems not to improve the execution of any task. He would have been fascinated by the expressions of people talking on the telephone.
So we are left with a problem: how can emotions be important, when their expressions can happen whether or not they serve any purpose, and sometimes seem irrational? Darwin, of course, was primarily interested in expressions of emotion as evidence for evolution. But the issues he explored point to another direction which helps to resolve the paradox. It is this: mammals and birds often find themselves in situations where they lack appropriate patterns of behaviour - when they are not fully adapted to an environment that has changed or when no habit or instinct fits a situation. Could emotions be important as part of the solution to the problem of what to do at these junctures? Could they be useful because they prompt us towards certain types of action when perhaps we should do something, but lack a well-adapted way of acting? Over the past 20 years or so, cognitive psychologists have begun to answer 'Yes' to such questions. They have studied phenomena of the kind that Darwin described, but perhaps because they no longer need to argue as staunchly as he did for the theory of evolution, they have come to a conclusion more appropriate to understanding emotions themselves. It is that emotions are not just vestiges of an infantile and bestial history. They are important now in our lives, in the everyday management of action.
The issue turns out to be a very general one. It would apply not just to us, but also to Martians, or to general-purpose robots. It would apply to any intelligent being that makes new goals and plans as it goes along, if that being had only imperfect knowledge and other limitations of its resources, if it had a number of goals that were not always mutually consistent, if it needed to cooperate with others.
Ronald de Sousa, in The Rationality of Emotion, puts the problem like this: we are neither completely determinate machines, nor angels with pure and rational wills. We are somewhere in between. Let me enlarge on this idea.
We can think of insects as being equipped with patterns of action shaped by natural selection for their particular form of life, and perhaps for particular patterns of interaction with other members of their species. Insects have so-called fixed action patterns that are triggered by stimuli, and this arrangement works well for them. An insect can be thought of as a little automaton, programmed by genetics. These action patterns can be directed in various ways, but the basic patterns are wired in. Where the insect does learn something - for instance, where an object is - it is mainly by a process in which the relevant information is inserted in slots that accept these data. Similarly, although we may think a swarm of bees seems 'angry', this may be an inappropriate assumption. More probably, the members of the swarm are displaying action patterns that have been triggered off. In general, insects do not have much need for emotions.
By a comparable argument, there is even less question of whether present types of robot have emotions. The reason is that a robot is an engineering means of achieving just one goal at a time. Like many such solutions, it requires the world to be simplified somewhat. The wheel was certainly a good idea, but to work properly it has needed parts of the Earth's surface to be made hard and flat. Similarly, robots work well, but only in the simplified situation for which they are designed, perhaps to assemble a part of a car. In their world, nothing unexpected happens. They can be programmed fully and rationally, precisely because they must fulfil one single function at a time, in a known and simplified environment.
We are not insects or robots - but neither are we gods. If we were, we would be all knowing, all seeing, omnipotent. For such beings, nothing unexpected could happen. Everything would be subordinated to a grand design, and a rational will. Instead, we are somewhere in between, neither automata nor omnipotent beings. We act with a degree of voluntariness and rationality, but because we are not all knowing our actions often have consequences we do not anticipate. Moreover, we have not one grand design, but many rather smaller goals, which are not always clear cut or compatible with each other. Sometimes when we act in pursuit of one of them, something happens that is relevant to another. And, rather than being like ants whose interactions with other ants are programmed, people make arrangements in the tasks that they share with others, and these may turn our differently from any one person's plans.
To understand emotions, we need to know when and where we tend to experience them. In fact, we tend to experience them at just the points that happen frequently to higher animals including humans, but hardly ever to automata or gods. They arise when something unexpected happens, a situation to which we are not fully adapted, an event at which two different concerns clash, or when someone else does something more or less than we expected.
So we need some mechanism that can do three things. First, it must be able to handle interruptions and potential interruptions. It must signal when something urgent happens, or something that makes it necessary to abandon a plan, or when we must respond to some person with whom we have joint interests. At the same time, if we are doing something important to us, this same mechanism should screen out events of lower priority, and help us to continue with what we are doing.
Secondly, when a potential interruption does arise, the mechanism must be able to change priorities, and manage the problem of whether and how we should make a transition from one activity to another.
Thirdly, because some events are both important and unanticipated, we might want to reprogram ourselves in the light of whatever new knowledge we have acquired as a result of these events. This would be learning of a kind that is not just taking on specific data, but which involves making new plans or modifying existing goals. The mechanism must allow us to concentrate on reprogramming ourselves, even though the urgency of the moment has passed.
These functions correspond rather closely to what happens with emotions. Emotions happen when certain events affect our goals. Here we need to distinguish between emotions and moods. Emotions arise suddenly, and they last for seconds or minutes. Moods are emotional states that may be more vague, and they last for hours or days. The distinction between emotions and moods is like that between two types of muscular activity: contractions, which change the position of a limb, and muscle tone, which maintains posture. Discrete emotions are concerned with changing something, and moods with maintaining something.
Emotions have five salient characteristics: first, they usually include an involuntary urge to act; secondly, there is often some bodily perturbation; thirdly, there is usually distinctive conscious feeling; fourthly, recognisable expressions of emotion, such as smiling or frowning, occur; and fifthly, thoughts may come to mind involuntarily and may reverberate for some time. We can explain these five characteristics by the idea that emotions manage transitions, or potential transitions, between different goals and plans.
Focusing on the first of these characteristics, most theorists now agree that emotion must be understood in relation to action. Emotions involve readiness and involuntary predispositions to act. Philip Johnson-Laird, of the Medical Research Council's Applied Psychology Unit at Cambridge, and I have devised a cognitive theory to explain some of these issues. We propose that all our various emotions are based on just a few distinctive mental states that go with readiness for action, and that each is set off when we evaluate an event in relation to our goals.
We suggest that events are evaluated, consciously or unconsciously, in terms of the following categories: the achievement of minor goals in solving problems as they arise; the loss of a goal; the frustration of a plan or goal by some person or circumstance; a conflict of goals including conflict with a goal of self-preservation; and the perception that something or someone is noxious. Each evaluation produces a mental state that is a basic emotion. The nearest English terms are happiness, sadness, anger, fear and disgust.
If we are happy, we are ready to keep going with what we were doing, and we may adopt expressions such as smiling. If sad, we become ready to do nothing for a while, perhaps hoping to be rescued, or hoping to change our plans. Perhaps we cry. If angry, we prepare to make redress of some kind. If frightened, we may freeze, or prepare to flee, or perhaps even to fight. If disgusted, or experiencing the interpersonal form of disgust known as hatred, we withdraw and may sneer and belittle the person concerned.
The mechanism is not that of the fixed action pattern - the stereotyped behaviours performed by some animals in some situation. In a changeable environment, we could not be properly programmed in advance. But neither do we depend solely on ordinary thinking, which is slow and prone to error - in any case, we seldom have the necessary knowledge to come to a rational conclusion. Instead, evolution seems to have provided us with an intermediate mechanism that involves being able, at each juncture, to make ready one of several small repertoires of action.
The mechanism must enable us to evaluate a situation, to interrupt what we are doing, and to move towards actions that are appropriate to a recurring type of juncture. The mechanism simplifies the choices available to us. Then, once the moment of what to do immediately is passed, the same mechanism may prolong the state into a mood - a period in which we may think consciously about what has happened, and plan what to do next.
Although there are just five basic emotions, there is an indefinite number of specific emotions, each made up of the basic emotions plus information about what caused it, or to whom it is directed. So being in love is a kind of happiness directed towards another person, with sexual implications. Jealousy, on the other hand, is based either on hatred or perhaps on anger, and caused by a possibility of being displaced from a love relationship by a third person. Johnson-Laird and I believe that the semantics of nearly 600 English words for emotions can be understood in this way.
We propose that it has been cognitively efficient for people to be ready to evaluate events in these five basic ways, and that evolution has selected for these states. When events evoke any one of the five, appropriate habits and any genetically programmed instincts are made ready.
Take, for example, the initiation of fear: let us say that we use 'fear' for a discrete emotion, and 'anxiety' when it is prolonged into a mood. Some part of the whole cognitive system - everything that enables us to perceive and think - detects a danger. There may be no single thing to do that is best, no fixed action pattern. But on the other hand there is usually neither the time nor knowledge to think through the best course of action. So evolution has equipped us with an intermediate mechanism based on an emotion. The recognition of a danger, in relation to the concerns about ourselves that we have, triggers a state of fear which summons, as it were, a small suite of action patterns derived from genes and habit.
As Jeffrey Gray of the Institute of Psychiatry in London has proposed, the actions that are prompted include stopping what we were doing when we become fearful or anxious. We become ready to freeze, to flee or perhaps to fight. We check what we have done carefully. We monitor the environment assiduously, and we perhaps call on any special training we may have for the situation in question. One thing is certain: in an emergency or a fight, there is not much time to think and almost no chance that we would arrive at the best, 'god's eye', solution anyway. So instead we have this set of more general promptings that have apparently been successful in evolution.
What about the second characteristic - that emotions are often accompanied by bodily perturbations? We can probably understand this best by thinking that, just as action patterns are loaded in readiness, so too are the physiological mechanisms necessary for their support. So, if we may need to fight or run, the heart starts to pump faster, and the body shifts resources away from activities that can be suspended, such as digestion, towards the muscles. This kind of process is the best we can propose at present, but it is not fully satisfactory. It makes a poor showing, for instance, in explaining why we blush when we are embarrassed and find ourselves the focus of unwelcome attention.
The third characteristic is that each emotion also has a distinctive conscious feeling. Johnson-Laird and I think of it in terms of the emotion signal reaching the topmost module of the cognitive system, where the results of some of the processes become conscious. The system then interprets the signal consciously in a particular way - as a feeling of happiness, sadness or anger, say. We believe that the signal indicates that emotions are a kind of communication. In our conscious awareness of them, they are communications to ourselves. The distinctiveness of each emotion means that we can recognise it and then talk about it with other people, so allowing us to reach understandings that our animal ancestors could not. Hence our fascination with cultural elaborations of ideas about emotions in conversation, in novels and in plays.
According to this theory, emotions depend on processes that monitor our goals all the time, to assess whether they are contributing to an ongoing plan. From these monitoring processes, simple signals are sent out whenever progress toward any goal changes substantially, for better or worse. Like an alarm bell, the signal indicates that something has happened that demands attention. It also tells us what kind of thing the event is. But in itself, the simple signal does not say what caused it, nor exactly what to do about it.
One clue suggesting that there are several distinct signals of emotion is that emotions and moods sometimes happen without there being any meaning attached to them. In other words, they may occur simply as 'feelings' without necessarily having any particular content. Although this happens rarely, it is very significant. For instance, someone might feel - in Wordsworth's words - 'surprised by joy': the emotion of happiness may arise for reasons that no one can describe. In the same way, people sometimes suffer from so-called free-floating anxiety, a sense of utter dread that is nameless. They are unable to say what it is that they fear. Moreover, certain drugs, including popular illegal ones as well as those used as tranquillisers and antidepressants, can alter moods without changing events in the outside world. Such drugs act to induce a basic mood state, or alter its intensity. They further indicate that these states are probably based on distinctive physiological mechanisms involving hormones, peptides and neural pathways.
The signal of an emotion is simple and, in evolutionary terms, old. Its detailed structure is of no meaningful significance to the system, because the signal merely switches some cognitive functions on and others off, producing its distinctive conscious feeling. We call such signals nonpropositional. Each induces one of the five basic states of emotion, and each can happen in a free-floating way.
Usually, however, the nonpropositional emotion signal is accompanied by what we call propositional information, and the feeling is closely bound to information about what caused it, or to whom it is directed, or what we might do about it. Some emotions - for instance, those that involve some evaluation of the self, such as guilt, remorse and envy - cannot happen in a free-floating way because some aspect of the reason why they occur must be conscious. So, although we may suffer free-floating anxiety, a basic emotion of fear, without knowing why, we cannot feel embarrassed without knowing something about why. Embarrassment is fear or anxiety that includes the knowledge that we are the focus of attention.
As to the fourth salient characteristic, as well as being communications to ourselves, emotions also communicate to others. Many animals, including humans, communicate happiness, sadness, anger, fear and disgust to each other principally by expressions and gestures. Vocal communications of fear even have a special name in the study of animal behaviour - alarm calls. We know, too, that human facial expressions of these five basic emotions are recognisable across all the cultures that have been studied, and recent research also indicates that each emotion produces a different constellation of physiological effects.
The shape of my eyebrows when I frown does not say what I am angry about, nor does a look of sadness indicate what has been lost. But when communicated, such signals may have the effect of propagating an emotion within the cognitive system of the person receiving it. In empathy, the emotion is the same as that of the sender. When the emotion expressed is anger, it may have a complementary effect on someone else, inducing fear, or an escalatory effect, making the other person angry too. So emotion signals between people, like their internal counterparts, are simple. They are like a police or ambulance siren: the siren does not tell you what has happened, but if you are driving, it prompts you towards a simple action, to make way for the vehicle to pass.
The fifth salient characteristic of emotions is that we may find thoughts coming to mind that are difficult to stop, especially when emotions are prolonged into moods. So, when we are in love, we think obsessively about the loved one. If sad, we may not be able to get what happened out of mind. If anxious, we find it hard to do anything else except worry about what may happen and how we could make ourselves safe. If angry, we may plot revenge - indeed, some people may shape large parts of their lives around such plans.
Such driven patterns of thought, with their conscious feelings of emotion, indicate that we may enter into a maintained mood after the rapid interruption by the emotion. Moods of this kind seem to concentrate the attention, to ensure that we think about making sense of what has happened, construct new plans about what to do about it, and perhaps modify our goals in relation to the new events.
So emotions are not just useless vestiges of our evolutionary past. They remain important in human thinking about the world because action is typically influenced by many simultaneous goals, is directed without complete mental models of the world and involves coordination with other people. Because the world is not fully predictable, and because conflicting goals cannot always be reconciled either with ourselves or between people, human action can hardly ever be perfectly rational. So to argue that emotions are irrational, as Plato did, for instance, misses an important point: rational solutions to problems of human action are rarely possible. Emotions help us to act in a world that can be only imperfectly known. Then, where something has not gone as expected, a change in readiness to act may be followed by a longer lasting mental preoccupation in which we can concentrate on reprogramming the way we think about our lives.
These properties of emotions are a biological and cognitive solution to the problems of managing goals and plans. Some solution to these problems would be necessary for any intelligent being capable of planning actions in an imperfectly knowable world.
Keith Oatley is professor of psychology at the University of Glasgow and currently president of the psychology section of the British Association for the Advancement of Science. This article is adapted from a British Association Lecture he gave at the first Edinburgh Science Festival in April 1989.
Further reading Keith Oatley and Philip Johnson-Laird, 'Towards a cognitive theory of emotions', Cognition and Emotions, vol 1, p 29.