Much attention in the life sciences is now being directed to the emotions. For those who are fascinated by the fads and fashions of science, this is especially interesting because emotion had almost been written off as a legitimate subject for theory and research until it began to revive in the 1960s and 1970s. In much of my writing, including my own book on emotion ( Lazarus, 1991 ; see also Smith & Lazarus, 1990 ), I have been concerned with stress and emotion, which I view from a cognitive and phenomenological perspective. Now I would like to clarify and deepen my analysis of appraisal. The main concerns of this article are the role of cognition in emotion, the ways meaning is generated, unconscious appraising, and the implications of these issues for life-span developmental psychology.

The cognitive movement in social science, which got rolling in the 1960s and 1970s, brought with it a return to a position about emotion that seems to have originated with Aristotle: The ways people construe their social situations have causal significance for their emotional reactions. Aristotle suggested that people are made angry by the thought that they had been unfairly slighted. Lyons (1980) expressed the classical readers' surprise that Aristotle discussed the emotions in Rhetoric rather than in De Anima . This reflects the latter's special interest in how emotions could be manipulated by orators and politicians (see Aristotle, 1941 ). Aristotle wrote that "Anger may be defined as a belief that we, or our friends, have been unfairly slighted, which causes in us both painful feelings and a desire or impulse for revenge" (p. 1380). Notice that his analysis makes implicit use of the ideas of relationship, appraisal, and action tendency, which are central concepts in many recent emotion theories. Nevertheless, there is no mention of motivation, the other essential construct of a cognitive—motivational—relational theory of emotion.

Over centuries of psychological thought there have been diverse conceptualizations of the relations among cognition, motivation, and emotion–or what Hilgard (1980) referred to as the "trilogy of mind." A theory of emotion is, in effect, a theory of how motivation and cognition produce emotions in adaptationally relevant encounters. The basic theme that thought causes emotion is often referred to as cognitive mediation , but it is also a variant of neobehavioristic thought, is consistent with subjectivism, and has been called folk or naive psychology because it depends on how ordinary people construe and evaluate what is happening in their lives.

Two difficulties with concepts like cognition and motivation should be mentioned: First, there has been a tendency to overextend cognition and treat it as equivalent to mind; second, overlaps between cognition, motivation, and emotion make it difficult to separate and distinguish their respective territories. I believe that although they are interdependent and difficult to separate in nature, each contains distinctive contents that make them all indispensable for understanding human adaptation and emotion.

Historically, emotion has been regarded as dependent on both cognition and motivation, sometimes within a hedonistic framework and sometimes cast in terms of connation or will. Without some version of a motivational principle, emotion makes little sense, inasmuch as what is important or unimportant to us determines what we define as harmful or beneficial, hence emotional. The implications of this are profound; if we wish to study how emotions are caused, then we must also study and measure the similarities and variations in motivation among different persons (see Novacek & Lazarus, 1990 ). Because I want to dwell mainly on cognition, I cannot take on the task of providing further detail on this topic here.

Without cognitive activity to guide us, we could not grasp the significance of what is happening in our adaptational encounters with the environment, nor could we choose among alternative values and courses of action. Emotion without thought would be mere activation without the directionally distinctive impulses of attacking in anger or fleeing in fear. Motivation without cognition too would be merely a diffuse, undifferentiated state of activation, a tissue tension that does not specify the consumatory goal or means to attain it. Finally, integration of behavior would also be impossible without cognitive direction ( Miller, Galanter, & Pribram, 1960 ); there would be no possibility, for example, of feedback control of behavior if we did not have the ability to take cognizance of what is happening. This, of course, is old hat and has been said in many different ways.

As a cause or independent variable, emotion may impair or interfere with subsequent thought and also produce feedback about its consequences, which engender further thoughts that are emotional. The moment an emotion occurs it becomes food, so to speak, for the next appraisal and emotion. Thus, if we feel ashamed by having been made angry because we regard anger as an unwarranted personal lapse, the anger could be said to have generated the shame. In both of the aforementioned cases, an emotion has influenced subsequent thoughts and emotions. Given my early research on the effects of stress on cognitive activity, which was a prominent concern of the times, it would indeed be odd for me to argue that the flow of cognition and emotion is only in one direction. How we view the direction depends on where in the psychological process one chooses to stop the action to identify the variables that precede or follow each other (see also Lazarus, Coyne, & Folkman, 1982 ).

In the following section, I try to make the strongest case I can that cognition is both a necessary and sufficient condition of emotion. I begin with the decision to make the cause–appraisal–part of the effect, distinguish knowledge and appraisal as different kinds of cognition, address whether emotion can occur without appraisal, and speak about reductionism and the problems inherent in treating cognition and emotion as separate systems.

A logical complication in the analysis of the relations between cognition and emotion is the presumption I make that emotion includes cognition (and meaning). However, although I think emotion always contains cognition, I also assume that some cognitions are impersonal and, therefore, relatively cold or nonemotional; many of the things we know do not have a special emotional significance on any given occasion. Therefore, the obverse that cognition also includes emotion does not always follow. When we say that emotion affects cognition, we are saying, in effect, that thoughts are also part of the emotions they cause. As Sartre (1948) said so clearly and eloquently:

All psychologists have noted that emotion is set in motion by a perception, a representation-signal, etc. But it seems that for them the emotion then withdraws from the object in order to be absorbed into itself. Not much reflection is needed to understand that, on the contrary, the emotion returns to the object at every moment and it's fed there. For example, flight in a state of fear is described as if the object were not, before anything else, a flight from a certain object, as if the object fled did not remain present in the flight itself, as its theme, its reason for being, that from which one flees . And how can one talk about anger, in which one strikes, injures, and threatens, without mentioning the person who represents the objective unity of these insults, threats, and blows? In short, the affected subject and the affective object are bound in an indissoluble synthesis. Emotion is a certain way of apprehending the world. (pp. 51—52)

To say that emotion influences cognition, which is also its cause, seems at first to be a meaningless tautology, particularly if we do not specify very clearly what the emotion includes or does not include and the thoughts or feelings it causes. Social scientists get uncomfortable about the treatment of appraisal as both a cause and a part of the (emotional) effect. One reason for the discomfort is the traditional reliance on the Aristotelian logic that A cannot also be B. We are in the habit of defining categories as inviolately separate. This same reasoning, by the way, encourages the notion that emotion and cognition must involve separate systems of the brain, which I will say more about later.

The solution is to recognize that emotion is a superordinate concept that includes cognition, which is its cause in a part—whole sense. Cognitive activity, A, about the significance of the person's beneficial or harmful relationships with the environment, is combined in an emotion with physiological reactions and action tendencies, B, to form a complex emotional configuration, AB. To say that appraisal of blame, A, causes the angry reaction, AB, which includes the A, seems confusing until we realize that a component of the whole can result in the next whole and, in turn, the next A of the new AB can shape subsequent As and ABs.

When we say that an emotion such as anger is an AB because it includes the cognitive appraisal, A, as well as the behavioral and physiological response patterns, B, as integral features, and that this AB also influences the next A, hence the subsequent emotional state, AB, which flows from it, we are simply saying that what has happened earlier affects everything afterward, including the appraisal process that follows and the emotional state it produces, in a complex causal time series. This, incidentally, is what the systems principle of reciprocal determinism implies (see Bandura, 1978 ; Lazarus, DeLongis, Folkman, & Gruen, 1985 ; Phillips & Orton, 1983 ).

The preceding analysis is formally no different from the germ theory of disease. If a disease-causing microbe is present in a vulnerable organism, there is a high probability that the disease will occur. The disease, which is here analogous to emotion, also includes the microbe that caused it. If the microbe is destroyed by the body's defenses and disappears, the person is no longer sick. In effect, one of the causes, the microbe, remains present during the disease; in fact, it must be present for the person to continue to be sick. If it disappears or becomes dormant like a spore, so does the disease. Analogously, as in the quote from Sartre, if the cognitive cause of an emotion is no longer present or salient, the emotion is made moot. If we keep the temporal relations in mind and recognize that we are dealing with a part—whole relation in a continuous flow as in a motion picture rather than a still photo, there is no unacceptable tautology. A cause can be a necessary and continuing feature of the effect without specious reasoning.

Because cognition is a very broad concept, we must also consider the kinds of cognition that are relevant to emotion. Research and theory about the emotion process have generally failed to distinguish between knowledge and appraisal. Knowledge consists of what a person believes about the way the world works in general and in a specific context. Without a personal stake in a transaction with the environment, knowledge is relatively cold or nonemotional (see Lazarus & Smith, 1988 ). I am, of course, aware that knowledge involves a sense of oneself and world and the relation between self and world; therefore, others might suggest more emotive tone in knowledge than I have allowed for. Nevertheless, I would say that knowledge can be cold in a relative sense compared with appraisals that reflect what is personally relevant.

Appraisal , on the other hand, is an evaluation of the significance of knowledge about what is happening for our personal well-being. Only the recognition that we have something to gain or lose, that is, that the outcome of a transaction is relevant to goals and well-being, generates an emotion. Because the motivational principles states that without a stake in the outcome of a transaction no emotion will occur, knowledge should be viewed as a necessary but no sufficient condition of emotion, whereas appraisal is both necessary and sufficient.

The distinction between knowledge and appraisal in the emotion process illustrates a serious limitation of attribution theories of emotion. These theories, the foremost being that of Weiner (1985 , 1986) , have been able to explain a considerable amount of emotion variance. However, causal attributions are built on knowledge, with appraisal implicit, and so they leave a major gap in understanding and prediction resulting from cultural and individual differences in the implications of attributions (knowledge) about oneself and the world. Thus, in Japan, lack of effort appears to be more of an indictment for personal failure than lack of ability, whereas in the United States we feel badly about lack of ability, but lack of effort is often used as an ego-saving excuse for failure. Weiner (1985) acknowledged this problem without, however, offering a correction based on a more individualized understanding of the bases of appraisal in emotion. He wrote,

A word of caution...is needed....Given causal ascription, the linked emotion does not necessarily follow....Hence, the position being espoused is that the [attributional] dimension—affect relations are not invariant, but are quite prevalent in our culture, and perhaps in many others as well. (p. 564)

The failure to see that attribution (knowledge) is often not an appraisal makes it difficult to predict how any individual will react to an encounter except in a probabilistic sense. I say probabilistic because although many people within a culture will share values and commitments in common, these values and commitments also vary greatly among individuals. Without an appraisal of the personal significance of what we know, knowledge about causation will be insufficient to bring about a particular emotion dependably. To the extent that individual differences in what is personally important, and in beliefs about how things work to which we are committed, shape appraisals and the resulting emotion, we will miss the point that an emotion requires an evaluation of the personal significance of what is happening–in other words, that this appraisal is causal and necessary.

It is not at all unreasonable to argue that meaning, hence cognitive activity, is always (the high-risk word) involved in emotion, because most, if not all, living creatures are capable of simple evaluations of elemental categorical distinctions between harm, threat, the benign, and benefit. Whether or not one wants to include Tinbergen's (1951) fowl, which discriminate danger from nondanger on the basis of genetically built-in releaser mechanisms, or to limit appraisal to learned cognitive evaluations–which seems to me to be the most sensible choice–there is no logical or empirical reason why cognitive activity should not be regarded as a necessary condition of emotion, and it is theoretically more parsimonious to take a strong position on this.

What are we to think about emotions in children and animals, for whom the possibilities of complex cognitive processes are limited? Does this imply that emotion can occur without cognitive activity? Our thinking about cognition and emotional experience in infants and young children has been rapidly changing during the past five years. What has happened is that a growing volume of research is showing that children have much more understanding of the social and psychological rules of emotion than has hitherto been supposed, and that this understanding starts early, even before there has been language development and certainly before they can verbalize feeling rules.

In speaking developmentally about emotion, it is important to distinguish between emotional experience and expression ( Lewis & Saarni, 1985 ), because the latter may give earlier evidence of occurrence than the former. For example, because any peripheral response serves many masters that are nonemotional as well as emotional ones, early smiles may have little to do with emotion until more coordination has occurred among the various components of an emotional state. Unless we take Izard's (1978 , 1984 , 1990) position that there is a tight connection between emotional expression and experience, it is difficult to know with confidence and without converging operations what emotions infants experience (see Campos, Campos, & Barrett, 1989 ; also Campos, Barrett, Lamb, Goldsmith, & Stenberg, 1983 ).

Evidently, infants are not capable of experiencing a full range of emotions at birth, but this capacity emerges from about two to four months of age, with some emotions appearing earlier than others (e.g., Emde, 1984a , 1984b ; Fischer & Pipp, 1984 ; Izard, 1978 , 1984 ; Sroufe, 1984 ). The earliest emotion seems to be diffuse upset, with happiness, joy, anger, then fear emerging later. Shame and guilt probably arise much later because of their dependence on a grasp of complex social relationships, for example, the ability to grasp the significance of social approval and disapproval, but there is not much developmental observation about these emotions on which to rely.

A substantial case can be made that children are capable of making appraisals of the significance of encounters for their well-being very early and, in fact, rapidly learn important lessons in their early months, even in the absence of the ability to give verbal reports about what they know. Bertenthal, Campos, and Barrett (1984) , for example, have demonstrated that an infant's appraisal of the danger of height depends on experience with locomotion (crawling), which occurs developmentally very early. Only when the child has begun to crawl does it react with fear of height in the experimental testing situation of the visual cliff in which the child is directed to cross a transparent table top, which creates an illusion of height-induced danger. I take this to mean that the child must first appraise in an embodied sense that there is danger of falling and being hurt, which is grasped fully only after the perceptual—motor experience with the physical environment produced by crawling.

The fascinating observations of Stenberg and Campos (1990) on infants' reactions to arm restraint raise questions and suggest answers about appraisal in infancy. Three-month-olds seem to react with undifferentiated distress. Four-month-olds display marked anger and look at the restraining hand of the experimenter, suggesting that there is recognition of an external frustrating agency. The seven-month-old looks at the face of the experimenter. There may even be some rudimentary sense of self- or ego-identity if we take Bahrick and Watson's (1985) evidence that five-month-olds can recognize the contingency between their leg movements and what is shown of this moment on a TV moniter (see also Papousek & Papousek, 1974 ). In other words, very young babies may have far more cognitive capability than we tend to give them credit for, and this knowledge and its implications for personal well-being (i.e., appraisal) is essential for an emotional life. It is important to develop the chapter and verse of how knowledge and appraisal works in the emotions.

When we get to slightly older children, Dunn (1988) has observed the natural social transactions of children with each other and with their mothers in the second year of life. She suggested that the child's understanding is fired by encounters that engage strong adaptational self-interest. Eighteen-month-old children are described as recognizing and attempting empathically to alleviate the distress of their siblings, or to exacerbate it if they are resentful rivals. Teasing, in which deliberate attempts are made to upset or annoy someone else, was observed as early as 14 to 16 months. If a child can consistently succeed in "getting another's goat," that child certainly must understand intuitively that the other person, say, a mother or another child, is personally vulnerable and likely to get upset, and knows how to touch that vulnerability.

Others who have been helping to chart the developmental course of cognitive activity in emotion include Harris (1989) , Stein, Trabasso, and their colleagues ( Stein & Levine, 1987 ; Trabasso, Stein, & Johnson, 1981 ), Michalson and Lewis (1985) on the development of the self-concept, as well as a growing list of researchers–too many to cite here–who are studying the developmental course of knowledge, appraisal, and the emotions in children.

On the basis of what we know now, it is not unreasonable to argue that a grasp of the personal significance of what is happening in an adaptational encounter is necessary for an emotion to occur. Nothing in the data contradicts the claim that appraisal is necessary condition of the experience of emotion, even when the cognitive activity is inferred from observations made before the child has the verbal capacity to explain its appraisals.

A comparable case may also be made that nonhuman animals have for greater cognitive capacities than has been assumed in the past to evaluate the significance (or meaning) of what is happening for their will-being and to engage in strategic planning. Griffin's (1984) observations of feigned broken-wing displays in birds suggest well-or-chestrated conditional complexities and maneuvers, some of which represent elaborate deceptions that are flexible and adaptively responsive to the situation. Although I do not have the space to elaborate on this, here too, there is nothing to contradict the view that cognitive evaluation is necessary for an emotion.

I see no reason to challenge the common observation that physical conditions such as fatigue, illness, and drugs are capable of influencing emotions and mood states, but I do challenge that this demonstrates effects on emotion that bypass cognitive mediation. In one prominent theory of the effects of cocaine, microbiologists have proposed that the drug produces euphoria by flooding the synapses (the spaces between nerve cells) with chemicals such as dopamine and norepinephrine, which facilitate neural transmission.

However, the question remains whether the effect on mood is direct or mediated by the way these chemicals influence the appraisal process. For example, cocaine is said to facilitate neural functioning as a result of an increased rate of neural transmission and, therefore, to result in enhanced energy, self-confidence, personal power, and sense of security. No wonder the person feels euphoric and on top of the world for a little while, given the temporarily enhanced and subjectively perceived functional power. This is, of course, a totally subjective and phoney high because nothing basic has really changed in the person—environment relationship, and after a short time–when the neurotransmitters are back to normal or at subnormal levels–the person "crashes" and becomes dysphoric.

The point I am making about cocaine–and it could also be made about any other drug including caffeine, alcohol, or those used in treating depression–is that drugs do affect acute emotions and moods, but the process involved in this effect is unclear. Humans and other sentient creatures are constructed so that, except when unconscious under surgical anesthesia (see Symington, Currie, Curran, & Davidson, 1955 , and Gray, Ramsey, Villarreal, & Krakaner, 1956 , for evidence that, although it does not by itself produce adrenal cortical changes, unconsciousness as a result of anesthesia eliminates the adrenal effects of psychological stress), they are continually evaluating what is happening from the standpoint of its significance for their well-being.

It is perfectly reasonable, therefore, to propose as I do that drugs and other influences on the brain of a sentient being never operate independently of cognitive activities such as appraisal. Without a demonstration that drug effects on mood occur in the absence of mediating cognitions, evidence of these effects does not in any way undermine the proposition that cognitive mediation is a necessary condition of emotion. Although many of my colleagues would prefer me to temporize, there is clarity and parsimony in taking the position that cognitive mediation is necessary for emotion to occur. Otherwise, we are forced to consider two principles of emotion generation, one operating through cognitive mediation, the other producing direct effects neurochemically without appraisal. For methodological reasons, the issue is unlikely to be resolved easily, and I think we will continue to find theoreticians on both sides of the fence.

The alert reader will have spotted the underlying mind—body problem in the preceding discussion, as well as the problem of reduction from molar to molecular levels of analysis. In reduction, thoughts and actions are reduced to physiology, which is the next lower, more elemental level. There are many arguments against reduction of social science concepts to biological and physical science ones, and it would not be worthwhile to make a long digression about this. The main problem, in my view, is that concepts at the different levels of analysis are not parallel; one doesn't map the other, so that meaningful functional links might be drawn between them. For example, the concept of the reticular activating system made a meaningful connection with earlier thinking about drive, attention, and mobilization at the behavioral level, but in current thinking, activation seems inadequate to explain complex processes of adaptation, especially when cognitive mediation is considered.

With respect to the mapping of one level of analysis by another, too little is known about the physiology of the brain to match well with the psychological concepts of appraisal and coping. However, there is now a trend toward greater interest in cortical and limbic system research on the cognitive processes involved in emotion (cf. LeDoux, 1986 , 1989 ; Panksepp, 1982 ; see also Fox & Davidson, 1984 ), but it is too early to tell how far this will go in narrowing the gap between the physiological and behavioral levels of analysis. The further one goes in the molecular direction away from molar concepts of mind and behavior, the more remote seems to be the theoretical connection between them.

I don't want to be misunderstood about this, however. In spite of the problems presented by reductionism, I believe it is quite appropriate and valuable to build bridges across scientific levels of analysis–for example, the psychological and the physiological. Nevertheless, we must not try to explain the psychological on the basis of the physiological, and vice versa, or to adopt the pernicious view that the higher level is "nothing but" what is going on at the lower level, as Parisi (1987) has put it, or can be adequately explained thereby. Although Parisi is an opponent of reduction, he suggested that finding a functional link is often very useful, as when a chemical or surgical treatment is the best way to deal with a psychological dysfunction. When, however, one literally seeks to explain what is happening at a molar level by reference to the molecular, it only demeans the study of the molar level by making an implicit if not explicit claim that the theory of mind and behavior must conform to what is believed at any given time about the brain.

Some of those who have accepted reduction also accept the idea that emotion and cognition are really separate systems of the brain and mind, governed by separate anatomical structures of the brain (cf. Zajonc, 1980 , 1984 ). This position is especially appealing to those who think of emotion as a hard-wired, innate process. For example, LeDoux (1989) , a neuropsychologist, favors a view of the anatomical and functional independence of emotion and cognition. When he wrote that "the systems can also function independently" (p. 278), I had less trouble with his analysis than when he wrote that "affective computations can be performed without the assistance of cognitive computations" (p. 279). I think he is wrong in this unless he also qualifies that affective computations are evaluations and, therefore, are cognitive operations even when they are primitive and hasty. In the same article he also wrote that "the core of the emotional system is thus a mechanism for computing the affective significance of stimuli" (p. 271).

Expressions such as computing , and affective significance , imply meaning, evaluation, or cognitive processing of some sort even when conducted at the neural level of the amygdala, and I think it is a careless usage to speak of an affective system as though it involved no cognitive activity. Thus, part of the problem is to be much clearer about what we mean by cognition and to overcome the notion that there is only one way in which evaluation or meaning can occur.

One of the difficulties with the separatist solution is that emotion and cognition are each so complex and their mechanisms are spread so widely over the central and peripheral nervous system that, in my opinion, it is difficult to argue convincingly for separate systems as though there were a special brain organ for each. It is also true that emotions and cognitive activity can be functionally disconnected as when we are engaged in ego-defensive maneuvers such as distancing or intellectualization, or in some conditions of brain damage (cf. Sachs, 1987 ). Anatomical separatism could be carrying neural localization and specificity too far, a sort of deus ex machina of neuropsychology.

A second problem is that the functioning of humans and nonhuman animals is generally organized and coherent rather than fragmented. Mind is usually rapidly coordinated and directed, with each adaptational function operating interdependently rather than going off in diverse directions at once. There is a long and reasonable tradition of thinking of health as integration and pathology as conflict and disintegration. In my view, the separatist solution keeps us from asking the right questions, which have to do with the kinds of cognitive activities that are causally involved in the emotion process, how they operate in emotion and adaptation, and how they are coordinated in the mind.

It is noteworthy that the functional separation of cognition and emotion has a long cultural tradition in the Western world, going back to the ancient Greeks and continuing through the middle ages in the Catholic Church and in the present era. In the Apollonian Greek ideal, which the medieval church also adopted, rationality was enthroned as godlike. Passion was regarded as animallike, and people were enjoined to control their animal natures by reason. In the counterculture of the 1960s and in the romanticism of Rousseau and others, the relationship was briefly reversed. The outlook was that in the industrial world humans were overly constrained by rational values and societal rules at the expense of their humanity; therefore, individuals had to struggle to get in touch with their emotions, which were being suppressed. Experiential and humanistic therapeutic schools today also center on this position.

Perhaps we have become entrapped by this ideology when we think of mind and brain today. Psychophysiological research on emotions up to the present has centered on lower and midbrain structures such as the reticular activating system and the hypothalamus (these ideas have been thoroughly explored by Averill, 1974 ). Emotion is treated as a primitive gut reaction, so to speak. Our cultural traditions and philosophical biases have reinforced the concept of emotion and cognition as separate systems, with emotion as primitive and cognition as advanced in both a phylogenetic and ontogenetic sense, despite the very real possibility that this is not the best model for thinking of the emotion process.

Nevertheless, although I would not consider cognition, motivation, and emotion as separate systems, I would say that emotion itself functions as a special, organized, functional system when it is called into play by appropriate circumstances. Otherwise, we would be forced to take Duffy's (1941) position that emotion is no different than all of life itself, which is a struggle to adapt, and should have no special status of its own. To the contrary, I argue that much adaptation is nonemotional, although it overlaps with and often looks similar to emotion in response terms. There is a world of difference between what is emotional and nonemotional, and among the diverse kinds of emotion. The emotion system follows rules of its own, which is what we are trying to understand.

In Heidegger's analysis of knowing, we grasp the situation directly in terms of its meaning for one's being, which is what I believe very commonly happens in appraisal and which is what Baron and Boudreau (1987) seem to mean in their use of Gibson's (1966 , 1979) concept of affordances. Although people can think reflectively or conceptually, the process of contextual knowing for Heidegger is effortless and nonreflective, and stems from merely "being in the situation." Those who are interested in the self need to realize that for Heidegger a sense of being may exist even before there is an articulated concept of self, which implies an abstract, analytic attitude.

Baron's (1988) suggestion that modern approaches to cognition have emphasized complex reasoning and inference at the expense of immediate recognition by human beings and subhuman animals of affordances in the environment, as described by Gibson (1966 , 1979) seems also to recognize that much of the time we know instantaneously about what is good for us or bad for us without complex and time-consuming inference processes. We know it automatically on the basis of the optic stimulus array, which also makes his perceptual analysis, almost like radical behaviorism, quite different from a cognitive—mediational outlook (cf. Neisser, 1990 ).

In any case, either as a part of species inheritance or learned, these perceptual processes result in the rapid appraisals of harm or benefit that emotion theory must take into account. As used by emotion theorists, the concept of affordance suggests organismic and elemental appraisals rather than those that are abstract and symbolic in character. It is also probably related to, but perhaps not the same as, embodied intelligence ( Merleau-Pontyy, 1962 , 1968 ) and tacit knowledge ( Polanyi (1966) .

Shepard (1984) has used the term resonance as a metaphor in connection with Gibson's affordances (see also Trevarthen, 1979 ; Trevarthen & Hubley, 1978 , on intersubjectivity). About this, Shepard wrote,

Instead of saying that an organism picks up the invariant affordances that are wholly present in the sensory arrays, I propose that as a result of biological evolution and individual learning, the organism is, at any given moment, tuned to resonate to the incoming patterns that correspond to the invariants that are significance for it....Up to this point I have not departed significantly from what Gibson himself might have said. Moreover, with the notion of selective tuning I can encompass the notion of affordance and thus explain how different organisms, with their different needs, pick up different invariances in the world. (p. 433)

If we are to fully comprehend how emotions are generated, we may need to give much more attention to relatively inarticulate processes like resonances–for want of a better term–between wishes or fantasies and what is actually encountered, as well as to complex matchings and functional equivalences. I mean by this an amorphous or ineffable sense of connection between what is in us or what we are like and the outer world. I believe the process of emotion generation is often automatic rather than deliberate and volitionally controlled. It is not possible to say with any confidence what proportion of appraisals and emotions are based on either mode of cognitive activity, and perhaps most appraisals involve a mixture of both. I suspect that automatic ones probably dominate much of our emotional lives.

Cognitive psychologists are beginning to think about at least two modes of cognitive activity for processing information and knowledge, one that operates automatically and without volitional control , the other deliberate and under volitional control . Most writers who make this distinction propose or imply that the former occurs at a preconscious or unconscious level without necessarily implying Freudian defenses, the latter being conscious (cf. Kihlstrom, 1987 ). In making this distinction we must also not make the mistake of equating automatic with primitive, which is sometimes done, as what is automatic can also be quite abstract and symbolic. The same distinction applies to appraisal.

What does it mean to speak of automatic modes of appraising in contrast with deliberate ones? The terms primary process and secondary process were used by Freud (1957) to distinguish primitive ideas and images characteristic of elemental, instinctual forces and reality-oriented, ego processes that developed later through maturation and experience with the world, some of which might be conscious. The rules of operation of the latter are adult and rational, whereas those of the former are childish and irrational; they make use of magical thought and condensation, which are also more characteristic of dreams and of psychosis.

Werner (1926/1948 , 1956) , too, distinguished between primitive and advanced cognitive processes for achieving the same adaptational outcome, although unlike Freud he did not emphasize unconscious mental activity. For example, a child may be able to "tell time"–when the mother says to get up at 6:30 a.m., the child reads the clock with a sophisticated grasp of the additivity of numbers–when that child says it is 6:30 a.m., it understands what a half hour means, or what five minutes means, and that if you add five minutes to the time, it becomes 6:35 a.m. Another child also can "read" the clock and rises on cue, perhaps even being able to say it is 6:30 in the morning. However, it has little understanding of what this really means; this child has simply memorized significant times and recognizes them on the clock. Both children achieve a performance outcome that looks the same if one doesn't inquire too closely, but in one case the process is developmentally simple, in the other it is advanced.

We are also accustomed to thinking of Piaget (1952) as a pioneer in a point of view that emphasizes epigenetic stages of cognitive development, also without reference to unconsciousness. In consequence of the broad processes of accommodation and assimilation whereby children adjust to the requirements of the physical world and also transform that world in accordance with their own needs, early concrete forms of thinking are ultimately followed by abstract forms that permit us to gain cognitive distance from the here and now. This developmental process permits realistic thought to replace impulsive action, giving the adult a sense of past, present, and future, more control over person—environment relationships, and the possibility of future planning, all of which make possible there remarkable patterns of adaptation of which the human animal is uniquely capable.

The idea of different levels of cognitive processing can also be found in several modern treatments of the cognition—emotion relationship. Buck's (1985) distinction between analytic and synthetic cognition is an example.

In analytic cognition, there is a buildup of meaning from originally meaningless bits in a stimulus display through linear scanning and digital analysis. In syncretic cognition , there is analogue detection of ecologically significant information, which is similar to Gibson's (1966 , 1979) view of how perception works. Here adaptational meaning is instantaneously achieved without reflection or cognitive stages.

A comparable view can also be found in Leventhal's (1984) distinction between schematic and conceptual processing. (Leventhal also adds a third, sensori-motor process.) In schematic processing , which is the more elemental of the two, when knowledge has been consistently appraised in the past in a particular way by a person, probably as a result of positive or negative adaptational outcomes, connections between knowledge and appraisal are formed that are functionally inseparable and seemingly fused and instantaneous. The evaluation occurs automatically and without complex cognitive activity.

In conceptual processing , knowledge structures and appraisals shape emotions through abstract, conscious, and deliberate forms of reasoning. This could well follow predefined stages, although the idea of such stages, as in Scherer's (1984a , 1984b) concept of stimulus evaluation checks, is problematic. In any case, cognitive activity can occur in simple and very rapid fashion, or in more complex and multiple process fashion, which might also be slower.

The idea that there is more than one mode or level of cognitive processing has gained added credibility as a result of attempts to trace neurophysiological pathways of evaluative cognitive activity in the brain, illustrated by the work of LeDoux ( 1986 , 1989 ; see also Lazarus, 1986 , and other commentaries on LeDoux, 1986 ). LeDoux (1989) wrote,

Regardless of whether one favours a cognitive, feedback, or central theory of emotion, the core of the emotional system is thus a mechanism for computing the affective significance of stimuli. As this mechanism is the precursor to conscious emotional experience, it operates, by definition, outside of conscious awareness. (p. 271)

Thus, the amygdala receives sensory inputs from the thalamus both directly and by way of the cortex. The thalamoamygdala projections appear to be involved in the processing of the affective significance of relatively simple sensory cues, whereas the cortico-amygdala projections are necessary when complex stimuli are processed. (p. 274).

In effect, the study of neurophysiological pathways suggests that there is a primitive or subcortical neural pathway for emotional processing from the thalamus to the amygdala of the limbic system, which can function independently of neocrotical involvement. This permits rapid, crude, and even hasty judgments about danger in the environment, a defensive reaction that can later be aborted if it proves false in more detailed cognitive analysis, which is a kind of neurophysiological analogue of appraisal and reappraisal.

If, indeed, one takes seriously that there is more than one kind of cognitive activity in emotion, the automatic and the deliberate, one would no longer need to equate cognitive processing with the relatively slow, progressive, stepwise generation of meaning from meaningless stimulus bits, or with the deliberate, volitional, and conscious reasoning that is often found in the garden variety of emotion process, but would recognize that emotional meanings can be generated in more than one way. I believe this possibility strengthens my position that appraisal is a necessary and a sufficient condition of emotion.

The difference between automatic and deliberate processing may be illustrated by an experiment conducted by Folkins (1970) in my laboratory. At the time it was done, I could not have thought of this dissertation as relevant to how meaning is generated, because there was not much interest in this question then. In Folkins research, different subjects awaited an electric shock for various periods of time, 30 seconds, a minute, several minutes, and more, and their subjective distress and psychophysiological reactions were recorded. At the outset, each group was flashed a sign saying "shock in 30 seconds," "shock in 1 minute," and so on. Stress reactions were greatest in the briefest time period for waiting, but less in three- or five-minute waits.

Postexperimental interviews and reports from a control sample that was interrupted at various points during the waiting process to evaluate memory distortions, showed that alarm was generated in the short waiting periods, which was mitigated by reappraisals when there was enough time for them. For example, subjects in the groups with three- and five-minute waiting periods began to reappraise what was going on in the experiment, saying to themselves that an academic researcher truly would not dare to injure them and that shock from a battery-operated inductorium was not a valid cause for alarm. In effect, the meaning (appraisal) that could be generated when time was short was elemental dread, but when there was sufficient time for deliberation (reappraisal), subjects were able to generate all sorts of realistic and reassuring thoughts that reduced the experienced stress and distress. I would now be tempted to call the latter cognitive coping.

In the early 1950s there was also much interest in experimental demonstrations of what was often called in those days autonomic discrimination without awareness, or subception , as McCleary and I called it in one of these demonstrations ( Lazarus & McCleary, 1951 ). The word subsception , in contrast with conscious perception, was used to indicate discrimination of danger without the person being aware that it had occurred. Our study followed several others, most notably one by McGinnies (1949) , who had demonstrated that galvanic skin responses (GSRs) were greater for emotional or threat words such as whore, bitch , and raped , compared with neutral words, both presented at speeds too fast for conscious recognition using a tachistoscope. This finding seemed to have demonstrated that the threatening message of the threat words had somehow been recognized below the level of conscious awareness. The problem with McGinnies' research was that the subjects might well have been motivated to withhold their report of the socially taboo words or even to doubt that they were presented, thereby accounting for the difference in GSR in spite of erroneous reports.

So McCleary and I ( Lazarus & McCleary, 1951 ) painstakingly conditioned a set of nonsense syllables to a painful electric shock, and in a later session compared GSRs with shock syllables and syllables never associated with shock. We found that even when subjects were incorrect in their report about which syllable had been flashed on the screen–because the speed was below their recognition thresholds–they gave a larger GSR to the shock than to the nonshock syllables. This demonstrated that there could be discrimination of threat and nonthreat even when there was no awareness of the presented stimulus, as indicated by an incorrect report. Moreover, there was no reason to withhold reports of what had been seen, as the stimuli were nonsense syllables rather than socially taboo or threatening words. Therefore, the subception experiment seemed to leave little doubt that the discrimination had occurred at an unconscious level as evidenced by the differential GSR. Put differently, subjects were making an unconscious appraisal even when a correct deliberate evaluation was not possible, as evidenced by the fact that it was below the threshold of conscious recognition and in error.

What followed from this research is interesting. There were virtually no challenges to the methodology and findings of the subception research. ¹ However, there were numerous challenges to the interpretation that an unconscious process was somehow involved. The most important and influential of these were by Bricker and Chapanis (1953) , Eriksen (1956) , and Howes (1954) , to which I dutifully responded ( Lazarus, 1956 ).

As a callow beginner, I learned from this experience that research findings rarely, if ever, settle theoretical issues in psychology, especially those that depend on epistemological and ideological differences. Although I think the subception research was the most effective experimental demonstration of the two modes of cognitive processing that I have discussed, and perhaps of the unconscious nature of some evaluations, psychologists in those days were simply not ready to assimilate comfortably the idea of unconscious processing, especially if it contained the dynamic implication of Freudian thought that sometimes ideas are kept out of awareness as an ego-defense.

Actually, when McGinnies summarized some of the research in this area, called the "new look in perception," he phrased the concept of unconscious in dynamic terms as follows: "It seems well established, then, that the perceptual 'filtering' of visual stimuli serves, in many instances, to protect the observer as long as possible from an awareness of objects which have unpleasant emotional significance for him" (cited in Lazarus & McCleary, 1951 , p. 114). Reservations about this kind of analysis still remain among psychologists today, although the more benign version of unconsciousness as what is not paid attention to–that is, as preconscious–seems now to be widely accepted.

Now, roughly 40 years later, there is a ground swell of interest in preconscious modes of evaluation, in contrast with conscious modes, particularly in social cognition. This research and theory seems to have begun with concern about how people form evaluative impressions of others and how these impressions affect social interaction. There is growing evidence that evaluations are often made without awareness, automatically. Evidence and the theory involved has been reviewed in Uleman and Bargh ( 1989 ; see also Bargh, 1990 ) and discussed by others (see also Fazio, Sanbonmatsu, Powell, & Kardes, 1986 ; and Fiske, 1982 ). I comment later in the article only on features that might be salient for this discussion.

Preconscious evaluation of social events seems to be categorical; that is, it is made for coarse distinctions such as good or bad rather than for finely graded analogical distinctions. This is consistent with the classical idea that conscious, deliberate, and volitional evaluations are more finely graded than unconscious, automatic, and involuntary ones, which are apt to be hasty and vague because they do not have the advantage of language. The latter have been regarded as primitive, developmentally earlier, and subcortical (cf. LeDoux, 1989 ).

That there are two modes of meaning generation also helps us understand how it is possible for values and goals to influence what is attended to and the appraisals of harm, threat, the benign, and benefit automatically and without reflection in an adaptational context. Social psychologists Jones and Thibaut (1958) have expressed this as follows:

If we can successfully identify the goals for which an actor is striving in the interaction situation, we can begin to say something about the cues to which he will attend, and the meaning he is most likely to assign them. (p. 152).

The underlying theme that emerges from all this, which in social cognition is not usually explicitly directed at the emotion process but should be, is the question of how plans and goals (cf. Miller et al., 1960 ; Schank & Abelson, 1977 ) influence evaluations (or what I refer to as appraisals) in ordinary adaptive social interchanges. I want the reader to see here that what psychologists concerned with social cognition are dealing with is fundamentally the same as what cognitive—motivational—relational theorists of emotion are–namely, how personal meaning is extracted from person—environment relationships. Allow me to press this parallel further by quoting Bargh (1990) , who brings the relevant variables together as follows:

Therefore, the mechanism proposed here by which the social environment may possibly control judgments, decisions, and behavior is through the formation of direct and automatic mental links between representations of motives and goals in memory (and consequently the goals and plans associated with them) and the representations of the social situations in which those motives have been frequently pursued in the past. The result of this automatic associative link is for the motive/goal/plan structure to become activated whenever the relevant triggering situational features are present in the environment. The activated goals and plans then would presumably guide the social cognition and interaction of the individual, without the person's intention or awareness of the motive's guiding role. (p. 100)

In sum, the earlier rejection of the idea that cognition is a necessary condition of emotion arose, I believe, from a number of misunderstandings, some of which were fostered by ambiguities in the recent cognition—emotion debates (e.g., Lazarus, 1982 , 1984 ; Zajonc, 1980 , 1984 ; the reader should also see other contributions to this argument, including those of Baars, 1981 ; Ellis, 1985 ; Emde, 1984b ; Kleinginna & Kleinginna, 1985 ; Leventhal & Scherer, 1987 ; Scheff, 1985 ; Slife, 1981 ; and Tetlock & Levi, 1982 ). In the light of what I have said about knowledge and appraisal, the existence of more than one mode of meaning generation, appraisal in children and infrahuman mammals, as well as my cognitive—mediational analysis of physical states that affect acute emotions and moods, I hope that most or all of these misunderstandings will by now have been cleared up.

Although I have long been at pains to point out that appraisal is not coextensive with consciousness, deliberateness, and rationality ( Lazarus & Folkman, 1984 , 1987 ), the way it has been discussed seems to have encouraged an image of a developmentally advanced and even conscious set of cognitive processes. Although I am well aware that the ideas I shall be discussing later are difficult to test empirically and will not be accepted by some, I would like to reconcile appraisal theory with depth psychology and the unconscious.

As Erdelyi (1985) and many others have pointed out, there are at least two meanings to unconsciousness. One meaning is the more or less bland idea that we are simply not aware of many of the thoughts and thought processes that influence how we act and feel. More often than not, the term preconscious tends to be used for this meaning. The second is the more controversial Freudian concept of ego defense, in which cognitive processes are kept out of awareness to cope with threat and to avoid emotional distress. In common usage, the term unconscious tends to be used for this meaning.

Herein lies a theoretical dilemma. Freud's (1926/1936) concept of signal anxiety is an interpretation in which ego defense is treated as a response to emotion rather than a cause. But what tells the mind to activate a defense? Psychoanalytic theorists have assumed that the concept of defense requires postulating the existence of unconscious anxiety, which acts as the signal of danger and triggers a defense mechanism. The notion of an unconscious emotion, in contrast with an unconscious (or preconscious) appraisal or coping process, is logically and empirically very awkward. The awkwardness comes from obscurity about how an emotion, which is an experience with mental content as well as a process, can be unconscious. There would be no problem if we assumed that the defensive process and its ideational contents were unconscious, but that the emotional experience was conscious. As Oatley (1988) put it, "It is ideas that are stored, not affect. Goals may be unconscious. Emotions are not usually repressed in this way (although there are phenomena like being angry but not knowing it)" (p. 15).

Gillett (1987a , 1987b , 1990) has attempted to refute Freud's signal theory (as I did earlier in Lazarus, 1966 , pp. 67—68). Gillett (1990) suggested a distinction between the "triggering of defense" and the process of "motivating the defensive effort." He makes use of my distinction between primary and secondary appraisal, which makes it possible to dispense with unconscious anxiety in the theory of intrapsychic conflict. All it takes to initiate a defense is an appraisal of threat.

Consider what it means to speak of threat. When we anticipate a harmful transaction, we engage in secondary appraisal, which has to do with options for coping. These options are also relevant to anticipatory coping with the upcoming harm. However, on the basis of what we have learned, we also anticipate the probable success or failure of our coping effort, which is a feature of secondary appraisal. Although threat is sufficient to activate the coping process, both the anticipation of danger (primary appraisal) and the anticipation of our probable success or failure in coping (secondary appraisal) are factors in anxiety. In those cases in which we anticipate with great confidence that we will be successful in avoiding the danger, there is no threat or anxiety and no signal that triggers the (defensive) reappraisal. However, in those cases in which we anticipate the actuality of harm, there is threat appraisal but no need for anxiety, as the appraisal itself, which may be preconscious or unconscious, is sufficient to generat coping or defense. Many years ago I used the term short-circuiting of threat (e.g., Lazarus & Alfert, 1964 ; Opton, Rankin, Nomikos, & Lazarus, 1965 ) to suggest, by analogy to an electrical short circuit that automatically shunts the current from its main route to another, that something similar happens when an appraisal of threat automatically triggers a defense.

Bargh (1990) drew on this idea too, quoting a passage from Wilensky (1983) that adds fuel to the idea that the entire meaning system inherent in an appraisal might be triggered automatically and without deliberate or voluntary processing:

If a standard plan is associated with a goal that occurs in a particular situation, it would be more efficient to associate this plan directly with that situation and select the plan at the same time the goal is detected. This would permit the planner to "short-circuit" part of the planning algorithm and suggest a plan immediately upon noticing a significant situation. (Wilensky, pp. 24—25)

This analysis is vital to emotion theory, and the connection needs to be made explicit and pointed up: To feel an emotion such as anger, anxiety, guilt, shame, sadness, joy, or pride, the process of knowing–whichever form it might take–must get the person to the specific relational meaning associated with the emotion. We may arrive at the appraisal pattern through an automatic process of knowing in the Heideggerian sense, even if the process is preconscious or unconscious, or via deliberate, self-controlled, abstract cognitive analysis. Regardless of the manner in which one gets there, however, to experience the particular emotion, the process of knowing must lead to the adaptational meaning psychobiologically connected with that emotion.

Most cognitive psychologists today seem to accept the idea that there is unconscious processing of information and ideas and that some degree of effort may be required for retrieval. This is Erdelyi's (1985) first meaning of unconsciousness. However, many of these same psychologists have doubts about or even reject the concept of ego defense as a basis of unconsciousness, which is the second meaning. Even clinical psychologists who accept the idea of ego defense differ among themselves about how accessible the unconscious contents of the mind are.

A third meaning of unconscious mind, following Gillett (1987a , 1987b) , is that one can only be aware of the contents of mental activity but not the mental processes by which these contents are achieved. Neisser (1967) seems to have made the same point; we cannot access most of the cognitive processes involved in registration, storage, and transformation of our experience, but we can be aware of some of their products. Gillett noted, historically, that Lashley, a neurophysiologist, too made the claim that all mental processes are unconscious and only the contents, which are the product of these processes, are capable of becoming conscious. Sandler and Joffe (1969) first introduced this distinction between experiential mental contents and nonexperiential mechanisms, structures, and energies into psychoanalysis; later, they and Wallerstein (1983) argued that the nonexperiential realm is not part of the unconscious.

Gillett (1987a , 1987b) proposed that defense mechanisms and decisions to activate them are unconscious in the sense of being nonexperiential. Freud's descriptive unconscious is limited to mental contents, either repressed (the dynamic unconscious) or preconscious. Thus, nonexperiential, repressed , and preconscious contents constitute three different meanings for unconscious. According to Gillett, the nonexperiential or process features have sometimes been confused with the dynamic unconscious and sometimes with the preconscious. When I speak of "unconscious appraisals," as in the decision to censor or defend, I am referring to a nonexperiential process.

Gillett (1987a , 1987b) also argued that nonexperiential mental entities and processes are incapable of becoming conscious under any conditions because of the nature of the mind—brain. Preconscious mental contents can usually become conscious, and repressed mental contents can become conscious if the repression is lifted by powerful experiences or psychotherapy and insight. Even in the case of potentially voluntary or conscious appraisals, the appraisal content could be either conscious or preconscious, that is, close to consciousness and probably accessible with effort.

Writers differ about how accessible to consciousness preconscious contents are, but it is common to differentiate the unconscious contents and processes into deeper, inaccessible layers and more surface, preconscious layers (see also Epstein, 1984 ). Brewin (1989) noted, for example, that these mental contents may be preconscious merely because we choose not to attend to certain things or because they are a part of the "enormous number of sensations, images, and so forth, which are automatically filtered out of the material potentially available to our limited consciousness" (p. 380). These contents will become readily accessible if we are alerted to their presence (see also Broadbent, FitzGerald, & Broadbent, 1986 ).

I assume that an encounter with the environment may be appraised as harmful or beneficial, hence emotion-generating, without the person being aware of the operative motivational and cognitive agendas, the environmental influences that have contributed to the appraisal or to the connections between them. From a depth psychology point of view, which centers on ego-defensive processes, we must consider individualized cognitive—motivational—relational configurations that operate idio-syncratically as hidden, irrational, or distorted meanings shaping our appraisals of interpersonal and intrapersonal relationships and, therefore, generating emotions that seem not to fit the consensual premises of observers about what is taking place.

Oatley (1988) has reviewed two psychoanalytic books on emotions and memory in psychoanalysis, one by Erdelyi (1985) , which I cited earlier, and another by Wegman (1985) . In this review, he made a nice analysis of the importance of goals or wishes in Freudian language, making the central point that Freud was laying the ground for a psychology of intention, and that it is intentions that serve as important sources of threat in connection with a disapproving environment or personal value system (superego). The way Oatley put this is worth quoting here in the context of threat and defense, and what I have said about motivation:

Memory does indeed have a central place in psychoanalysis, but the key to Freud's theorizing about it is his treatment of it as related to human goals, wishes as he called them: how we sometimes act as if we had an intention but deny it. Freud's methods were methods for investigating goals and plans, by listening to patients' stories. A story makes sense only when the goals and plans of the actors are understood. Yes, Freud was interested in restoring memories, but the interpretations that psychoanalysts offer to fill gaps in a story do not fill any old gaps. They fill specifically those gaps left by missing intentions. They suggest goals that might have been forgotten or denied, but which might make sense of otherwise incomprehensible sequences of action. (p. 11)

Expressed in terms of appraisal, persons reacting this way have appraised the relationship unconsciously, or automatically if one prefers, and differently from the way they view it consciously and deliberately or the way others view it. Therefore, the emotions experienced or displayed seem to make little or no sense, because the intentional premises are hidden. Even when conditions make for an appraisal that we are secure, we react with anxiety as if threatened; even when the social conditions seem to others to be benign or supportive, instead of a friendly response we observe anger; even when there is no evident reason for self-blame, guilt is nevertheless experienced; even when there seems to be reason for rejoicing, instead we react with sadness or depression; and so on for any and all kinds of emotion, both positive and negative. It is as if we are reacting to a different world than that indicated by the actual conditions as judged by others or even consciously by ourselves. Although there are many relatively simple reasons why this could be so–for example, cultural and individual differences or confusions about signals–a frequent and theoretically consistent explanation centers on the operation of ego defenses.

Defenses, I assume, change appraised meanings, but the meaning postulated for each emotion may still apply if it is to be drawn on in our analysis of emotional transactions, whether or not the person is aware of the other meaning. This possibility leads me to the striking proposition that two contradictory appraisals might occur at the same time, one unconscious and another conscious, and, in a sense, both could be valid. One way in which this might take place is expressed in the idea of contrary mental states for the same event ( Horowitz, 1976 , 1989 ). That is, in one state after a loss or trauma, a person engages in denial and experiences emotional numbing, whereas in another state, the traumatic experience surfaces and enters intrusively into the person's thoughts and actions.

Whatever conceptual language one uses for the distortion of the emotional process by images and ideas that lie below the surface, people carry around with them private and recurrent personal meanings that lead them to react inappropriately to an encounter with a sense of betrayal, victimization rejection, abandonment, inadequacy, or whatever. They appraise their relationships in accordance with private and commonly unrecognized meanings that engender the relational themes for anger, anxiety, guilt, and so on, which they alone conceive as applying to their transactions.

But what happens to the emotion-generation process when it is kept unconscious as a result of a process of defense? Does it remain silent? Does it go away? Does it influence what is going on consciously? Of course, we don't know, but can suppose one or another answer. Freud tried to explain neurotic symptoms as the return of the repressed, which adopts the position that some or all defenses are unsuccessful in containing the impulse or wish that is offensive. If there were no symptoms to note in the form of distressing emotions and dysfunctions, we could not know about the defense. A number of personality psychologists are currently seeking ways to distinguish among those who present themselves positively and are psychologically healthy and those who present themselves positively but who are personally unsound as a result of defenses or self-presentational strategies (see Weinberger, 1989). It is problematic whether the criteria used in this research in making this discrimination are good analogues of active defenses.

We know that clinicians use a number of clues suggesting contradictions between unconscious and conscious appraisals, for example, the presence of symptoms of intrapsychic struggle, seemingly irrational reactions, contradictions in what is reported by a client from time to time, or between what is reported and behavioral and physiological reactions. These clues generate only the suspicion that something obscure is going on in the appraisal process; further exploration is needed to reveal what it is. Most of the programmatic research done in the study of the cognition—motivation—emotion relationship has been directed at readily observed processes operating on the surface of awareness. The depth psychologist, however, who emphasizes unconscious determinants will doubt that the real meanings underlying emotions can be deciphered without in-depth, intensive clinical examination. The danger of an exclusively surface analysis of an emotion, especially if it is based solely on what a person reports, is that we will mistake what is seen on the surface for what is happening at a deeper, presumably unconscious or preconscious level.

The idea of conflicts in what is going on in appraisal at different levels of the mind makes good sense to me. We can believe simultaneously, for example, that flying in an airplane is safe; at the same time, we can act as if we believed that it is very dangerous. The thoughts of persons like me on takeoffs and landings are preempted by a different kind of processing that treats these events as dangerous, while knowing in a different sense that they are not. About this sort of thing, Brewin (1989) wrote, Such a system would be able to account both for conditioning in animals and for the irrational nature of some human fears and phobias. One would only have to assume that with irrational fears the person's conscious recollection of events did not correspond to the representation of them that was automatically created by the system mediating nonconscious causal perception. This might occur for a number of reasons. For example, the critical event or events might have occurred very early in childhood during the period of infant amnesia. Other events might be so distressing that they were defensively excluded from consciousness....Discrepancies might also arise because the nonconscious representation is theoretically based on the full array of sensory input, plus related material in memory, whereas because of processing limitations, the conscious representation would be based on a much smaller amount of information. [Here Brewin points out that under emotional conditions, or in shock, perception and memory might be impaired.] Furthermore, social pressures may operate to influence the conscious interpretations that people form of their experience. (p. 381)

In recent years, there has been a renewed interest both in early cognitive—motivational—emotional development and, more broadly, in life-span development, which extends from infancy to adulthood and old age. Labouvie-Vieff, Hakim-Larson, DeVoe, and Schoerberlein (1989) have recently pointed to some very important issues in life-span development. They draw on the idea–explored earlier–that there are two general modes for deriving adaptational meaning, one relatively primitive and concrete–similar to what I have spoken of as automatic and involuntary–and the other more abstract and symbolic–similar to what I have spoken of as deliberate and voluntary.

Labouvie-Vief et al. (1989) have suggested that these modes are usually considered by developmental theorists as organized hierarchically, the automatic mode being the earliest but later overridden by rational, distancing, emotion-regulating, and symbolic cognitive activities. Hierarchical analyses are found in Freud (1957) in his discussion of primary and secondary process, in Piaget (1976 , 1980) in his discussion of accommodation and assimilation, and in Werner ( 1957 ; see also Werner & Kaplan, 1963 ) in the distinction between physiognomic and formal-technical modes of thought. I don't think these conceptions of the two modes of meaning generation are equivalent–for example, in Freud they have a psychopathology versus mental health slant–but they certainly have much in common.

Although the two modes of meaning generation are often contrasted as primitive versus advanced–with the latter ultimately and smoothly overriding the former in the course of psychological development– Labouvie-Vief et al. (1989) have suggested that their integration is a dialectical result of a continuing struggle. Rather than being hierarchically organized, one may think of them as arranged in a parallel fashion, always operating together either in conflict or harmoniously integrated.

Labouvie-Vief et al. (1989) have also suggested that Piaget overemphasized accommodative forms of coping in development by suggesting that ultimately the child comes to subordinate inner fantasy and personal agendas to the requirements of external reality, in effect implying that with maturity one becomes more focused on the environment and its demands, constraints, and resources. Piaget considered inner speech as inferior or childish and, rather than establishing truth, it is private and not communicable, whereas with developmental advances, the person gives up private symbols and adopts collective ones ( Watkins, 1986 ). Freud (1957) , too, treated primary process as primitive and pathogenic, and it must ultimately be given up in favor of secondary process, realistic, and abstract thought, although some Freudians (e.g., Kris, 1952 ) wrote about regression in the service of the ego and creativity. In this view we may reach down into our primitive, childish selves without compromising our adult integrity, as long as we ultimately come back to secondary-process thinking and use the more primitive processes adaptively. Labouvie-Vief et al. (1989) summarized this traditional, hierarchical developmental view as follows:

Thus, most theories of development portray the progression from childhood to adulthood as the gradual dissociation of two forms of meaning, as the individual adapts to a collective language of symbols. Through this process of dissociation and hierarchization, meaning systems that originate in the organismic, the sensorimotor, the figurative, the dynamic, and the personal, are gradually displaced by those that are abstract, conceptual, stable, conventional, and impersonal. Indeed, it is often assumed that the former are a characteristic of immature thought. (p. 283)

Everything that falls outside the domain of analytical, propositional, and formal thought is merely classed as emotive, irrational, and animalian....All other things our minds do are dismissed as irrelevant to intellectual progress; they are residues, emotional disturbances, or throwbacks to animal estate [and indicate] regression to a pre-logical state. (p. 292)

The point is, of course, that although the two modes of thought may be in conflict, one of the tasks of development is to somehow integrate them to the extent possible, thereby permitting the person to draw adaptively on each in the ordinary course of living. The authors speak of this as "reconnection," or as Turner (1973) called it in Piagetian terms, "re-centering," that is, reconnecting logical thought with an organismic core of meaning. From a life-span point of view, in the earliest regulation of emotion in the infant, inner and outer are undifferentiated, but soon children come to regulate spontaneous expression of emotion in accord with social situations and with social conventions or norms. Children first define emotions by the conditions that elicit them, but gradually recognize the inner agency of mind. In adolescence they become more reflective and clearly recognize an inner and outer life and the rules that apply to them.

We have the option of viewing these modes of meaning generation as hierarchical, with one overriding the other as development proceeds, or as parallel with the potential for conflict or integration as complementary forms of knowing on the basis of a dialectical struggle, with living taking place at different stages of life. The latter is sometimes called parallel processing (e.g., Leventhal, 1984 ; Townsend, 1990 ). It makes sense to me to consider maturity or mental health as the harmonious integration of both ways of knowing, as involving harmony among the three constructs of mind, cognition, motivation, and emotion, and as being in good touch with one's inner self as well as the outer reality ( Lazarus, 1989 ).

Perhaps relevant here, too, are ideas about higher stages of consciousness, which have not gotten much attention from traditional cognitive or developmental psychology but which are cast in developmental, life-span terms. The notion of complete or unified knowledge, which is said to be gained only at the highest level of development, when knower, known, and process of knowing are fully integrated, seeks to go beyond the notion of separate levels of cognitive activity, the childish and the adult.

I conclude that to understand the emotions, whether developmentally over the life course or contemporaneously, we must give close consideration to the ways personal meaning is generated, both horizontally and vertically, and the contents of this meaning, which is what knowledge, appraisal, and emotion are all about. And although it is a daunting task, I believe we must explore the conscious appraisal components and patterns resulting in each of the emotions, as well as find effective ways of exploring what lies below the surface, how it relates to what is in awareness, and how it influences the entire emotion process. We need a reasonably sophisticated theory about all this and about strategies of research appropriate to it.

1

There is one exception to this statement, a study by Chun and Sarbin in 1968 , which led to a brief interchange ( Lazarus, 1968 ).