Organisational Dynamics and Behaviour

VOL. 69, No. 5 SEPTEMBER 1962

PSYCHOLOGICAL REVIEW COGNITIVE, SOCIAL, AND PHYSIOLOGICAL DETERMINANTS OF EMOTIONAL STATE1

STANLEY SCHACHTER A

Columbia University

The problem of which cues, internal or external, permit a person to label and identify his own emotional state has been with us since the days that James (1890) first tendered his doc- trine that "the bodily changes follow directly the perception of the exciting fact, and that our feeling of the same changes as they occur is the emotion" (p. 449). Since we are aware of a variety of feeling and emotion states, it should follow from James' proposition that the various emotions will be ac- companied by a variety of differentiable bodily states. Following James' pro- nouncement, a formidable number of studies were undertaken in search of the physiological differentiators of the emotions. The results, in these early days, were almost uniformly negative. All of the emotional states experi-

1 This experiment is part of a program of research on cognitive and physiological de- terminants of emotional state which is being conducted at the Department of Social Psy- chology at Columbia University under PHS Research Grant M-2S84 from the National Institute of Mental Health, United States Public Health Service. This experiment was conducted at the Laboratory for Research in Social Relations at the University of Minne- sota.

The authors wish to thank Jean Carlin and Ruth Hase, the physicians in the study, and Bibb Latan6 and Leonard Weller who were the paid participants.

FD JEROME E. SINGER

Pennsylvania State University

mentally manipulated were character- ized by a general pattern of excitation of the sympathetic nervous system but there appeared to be no clear-cut physi- ological discriminators of the various emotions. This pattern of results was so consistent from experiment to ex- periment that Cannon (1929) offered, as one of the crucial criticisms of the James-Lange theory, the fact that "the same visceral changes occur in very dif- ferent emotional states and in non-emo- tional states" (p. 351).

More recent work, however, has given some indication that there may be differentiators. Ax (1953) and Schachter (1957) studied fear and anger. On a large number of indices both of these states were characterized by a similarly high level of autonomic activation but on several indices they did differ in the degree of activation. Wolf and Wolff (1947) studied a sub- ject with a gastric fistula and were able to distinguish two patterns in the phys- iological responses of the stomach wall. It should be noted, though, that for many months they studied their subject during and following a great variety of moods and emotions and were able to distinguish only two patterns.

Whether or not there are physiologi- cal distinctions among the various emo- tional states must be considered an open

379

380 STANLEY SCHACHTER AND JEROME E. SINGER

question. Recent work might be taken to indicate that such differences are at best rather subtle and that the variety of emotion, mood, and feeling states are by no means matched by an equal vari- ety of visceral patterns.

This rather ambiguous situation has led Ruckmick (1936), Hunt, Cole, and Reis (1958), Schachter (1959) and others to suggest that cognitive factors may be major determinants of emo- tional states. Granted a general pat- tern of sympathetic excitation as char- acteristic of emotional states, granted that there may be some differences in pattern from state to state, it is sug- gested that one labels, interprets, and identifies this stirred-up state in terms of the characteristics of the precipi- tating situation and one's apperceptive mass. This suggests, then, that an emotional state may be considered a function of a state of physiological arousal2 and of a cognition appropriate to this state of arousal. The cognition, in a sense, exerts a steering function. Cognitions arising from the immediate situation as interpreted by past experi- ence provide the framework within which one understands and labels his feelings. It is the cognition which de- termines whether the state of physio- logical arousal will be labeled as "anger," "joy," "fear," or whatever.

In order to examine the implications of this formulation let us consider the fashion in which these two elements, a state of physiological arousal and cog- nitive factors, would interact in a vari- ety of situations. In most emotion in- ducing situations, of course, the two

2 Though our experiments are concerned exclusively with the physiological changes produced by the injection of adrenalin, which appear to be primarily the result of sym- pathetic excitation, the term physiological arousal is used in preference to the more specific "excitation of the sympathetic nerv- ous system" because there are indications, to be discussed later, that this formulation is applicable to a variety of bodily states.

factors are completely interrelated. Imagine a man walking alone down a dark alley, a figure with a gun sud- denly appears. The perception-cogni- tion "figure with a gun" in some fash- ion initiates a state of physiological arousal; this state of arousal is inter- preted in terms of knowledge about dark alleys and guns and the state of arousal is labeled "fear." Similarly a student who unexpectedly learns that he has made Phi Beta Kappa may ex- perience a state of arousal which he will label "joy."

Let us now consider circumstances in which these two elements, the physio- logical and the cognitive, are, to some extent, independent. First, is the state of physiological arousal alone sufficient to induce an emotion? Best evidence indicates that it is not. Maranons

(1924), in a fascinating study, (which was replicated by Cantril & Hunt, 1932, and Landis & Hunt, 1932) in- jected 210 of his patients with the sym- pathomimetic agent adrenalin and then simply asked them to introspect. Sev- enty-one percent of his subjects simply reported their physical symptoms with no emotional overtones; 29% of the subjects responded in an apparently emotional fashion. Of these the great majority described their feelings in a fashion that Maranon labeled "cold" or "as if" emotions, that is, they made statements such as "I feel as if I were afraid" or "as if I were awaiting a great happiness." This is a sort of emotional "deja vu" experience; these subjects are neither happy nor afraid, they feel "as if" they were. Finally a very few cases apparently reported a genuine emotional experience. How- ever, in order to produce this reaction in most of these few cases, Maranon (1924) points out:

3 Translated copies of Marafion's (1924) paper may be obtained by writing to the senior author.

DETERMINANTS OF EMOTIONAL STATE 381

One must suggest a memory with strong affective force but not so strong as to pro- duce an emotion in the normal state. For ex- ample, in several cases we spoke to our pa- tients before the injection of their sick children or dead parents and they responded calmly to this topic. The same topic pre- sented later, during the adrenal commotion, was sufficient to trigger emotion. This ad- renal commotion places the subject in a situ- ation of 'affective imminence" (pp. 307-308).

Apparently, then, to produce a genu- inely emotional reaction to adrenalin, Maranon was forced to provide such subjects with an appropriate cognition.

Though Maranon (1924) is not ex- plicit on his procedure, it is clear that his subjects knew that they were re- ceiving an injection and in all likeli- hood knew that they were receiving adrenalin and probably had some order of familiarity with its effects. In short, though they underwent the pattern of sympathetic discharge common to strong emotional states, at the same time they had a completely appropriate cognition or explanation as to why they felt this way. This, we would suggest, is the reason so few of Mara- fion's subjects reported any emotional experience.

Consider now a person in a state of physiological arousal for which no im- mediately explanatory or appropriate cognitions are available. Such a state could result were one covertly to inject a subject with adrenalin or, unknown to him, feed the subject a sympathomi- metic drug such as ephedrine. Under such conditions a subject would be aware of palpitations, tremor, face flushing, and most of the battery of symptoms associated with a discharge of the sympathetic nervous system. In contrast to Maranon's (1924) subjects he would, at the same time, be utterly unaware of why he felt this way. What would be the consequence of such a state ?

Schachter (1959) has suggested that

precisely such a state would lead to the arousal of "evaluative needs" (Festin- ger, 1954), that is, pressures would act on an individual in such a state to un- derstand and label his bodily feelings. His bodily state grossly resembles the condition in which it has been at times of emotional excitement. How would he label his present feelings ? It is sug- gested, of course, that he will label his feelings in terms of his knowledge of the immediate situation.4 Should he at the time be with a beautiful woman he might decide that he was wildly in love or sexually excited. Should he be at a gay party, he might, by comparing himself to others, decide that he was extremely happy and euphoric. Should he be arguing with his wife, he might explode in fury and hatred. Or, should the situation be completely inappropri- ate he could decide that he was excited about something that had recently hap- pened to him or, simply, that he was sick. In any case, it is our basic as- sumption that emotional states are a function of the interaction of such cog- nitive factors with a state of physiologi- cal arousal.

This line of thought, then, leads to the following propositions:

1. Given a state of physiological arousal for which an individual has no immediate explanation, he will "label" this state and describe his feelings in terms of the cognitions available to him. To the extent that cognitive factors are potent determiners of emotional states, it could be anticipated that precisely the same state of physiological arousal could be labeled "joy" or "fury" or "jealousy" or any of a great diversity

4 This suggestion is not new for several psychologists have suggested that situational factors should be considered the chief differ- entiators of the emotions. Hunt, Cole, and Reis (1958) probably make this point most explicitly in their study distinguishing among fear, anger, and sorrow in terms of situ- ational characteristics.

382 STANLEY SCHACHTER AND J E R O M E E. SINGER

of emotional labels depending on the cognitive aspects of the situation.

2. Given a state of physiological arousal for which an individual has a completely appropriate explanation (e.g., "I feel this way because I have just received an injection of adren- alin") no evaluative needs will arise and the individual is unlikely to label his feelings in terms of the alternative cognitions available.

Finally, consider a condition in which emotion inducing cognitions are present but there is no state of physio- logical arousal. For example, an indi- vidual might be completely aware that he is in great danger but for some reason (drug or surgical) remain in a state of physiological quiescence. Does he experience the emotion "fear"? Our formulation of emotion as a joint function of a state of physiological arousal and an appropriate cognition, would, of course, suggest that he does not, which leads to our final proposi- tion.

3. Given the same cognitive circum- stances, the individual will react emo- tionally or describe his feelings as emo- tions only to the extent that he experiences a state of physiological arousal.5

PROCEDURE

The experimental test of these propositions requires (a) the experimental manipulation of a state of physiological arousal, (6) the manipulation of the extent to which the sub- ject has an appropriate or proper explanation of his bodily state, and (c) the creation of situations from which explanatory cognitions may be derived.

In order to satisfy the first two experi- mental requirements, the experiment was cast

5 In his critique of the James-Lange theory of emotion, Cannon (1929) also makes the point that sympathectomized animals and pa- tients do seem to manifest emotional be- havior. This criticism is, of course, as ap- plicable to the above proposition as it was to the James-Lange formulation. We shall discuss the issues involved in later papers.

in the framework of a study of the effects of vitamin supplements on vision. As soon as a subject arrived, he was taken to a private room and told by the experimenter:

In this experiment we would like to make various tests of your vision. We are particularly interested in how certain vita- min compounds and vitamin supplements affect the visual skills. In particular, we want to find out how the vitamin compound called 'Suproxin' affects your vision.

What we would like to do, then, if we can get your permission, is to give you a small injection of Suproxin. The injec- tion itself is mild and harmless; however, since some people do object to being in- jected we don't want to talk you into any- thing. Would you mind receiving a Suproxin injection?

If the subject agrees to the injection (and all but 1 of 185 subjects did) the experi- menter continues with instructions we shall describe shortly, then leaves the room. In a few minutes a physician enters the room, briefly repeats the experimenter's instruc- tions, takes the subject's pulse and then in- jects him with Suproxin.

Depending upon condition, the subject re- ceives one of two forms of Suproxin—• epinephrine or a placebo.

Epinephrine or adrenalin is a sympathomi- metic drug whose effects, with minor excep- tions, are almost a perfect mimicry of a discharge of the sympathetic nervous system. Shortly after injection systolic blood pressure increases markedly, heart rate increases somewhat, cutaneous blood flow decreases, while muscle and cerebral blood flow in- crease, blood sugar and lactic acid concentra- tion increase, and respiration rate increases slightly. As far as the subject is concerned the major subjective symptoms are palpita- tion, tremor, and sometimes a feeling of flushing and accelerated breathing. With a subcutaneous injection (in the dosage ad- ministered to our subjects), such effects usu- ally begin within 3-5 minutes of injection and last anywhere from 10 minutes to an hour. For most subjects these effects are dissipated within 15-20 minutes after injection.

Subjects receiving epinephrine received a subcutaneous injection of % cubic centimeter of a 1 : 1000 solution of Winthrop Labora- tory's Suprarenin, a saline solution of epinephrine bitartrate.

Subjects in the placebo condition receivpd a subcutaneous injection of % cubic centimeter of saline solution. This is, of course, com-

DETERMINANTS OF EMOTIONAL STATE 383

pletely neutral material with no side effects at all.

Manipulating an Appropriate Explanation

By "appropriate" we refer to the extent to which the subject has an authoritative, un- equivocal explanation of his bodily condi- tion. Thus, a subject who had been informed by the physician that as a direct consequence of the injection he would feel palpitations, tremor, etc. would be considered to have a completely appropriate explanation. A sub- ject who had been informed only that the injection would have no side effects would have no appropriate explanation of his state. This dimension of appropriateness was ma- nipulated in three experimental conditions which shall be called: Epinephrine Informed (Epi I n f ) , Epinephrine Ignorant (Epi Ign), and Epinephrine Misinformed (Epi Mis).

Immediately after the subject had agreed to the injection and before the physician en- tered the room, the experimenter's spiel in each of these conditions went as follows:

Epinephrine Informed. I should also tell you that some of our subjects have ex- perienced side effects from the Suproxin. These side effects are transitory, that is, they will only last for about 15 or 20 min- utes. What will probably happen is that your hand will start to shake, your heart will start to pound, and your face may get warm and flushed. Again these are side effects lasting about IS or 20 minutes.

While the physician was giving the injec- tion, she told the subject that the injection was mild and harmless and repeated this de- scription of the symptoms that the subject could expect as a consequence of the shot. In this condition, then, subjects have a com- pletely appropriate explanation of their bodily state. They know precisely what they will feel and why.

Epinephrine Ignorant. In this condition, when the subject agreed to the injection, the experimenter said nothing more relevant to side effects and simply left the room. While the physician was giving the injection, she told the subject that the injection was mild and harmless and would have no side effects. In this condition, then, the subject has no ex- perimentally provided explanation for his bodily state.

Epinephrine Misinformed. I should also tell you that some of our subjects have ex- perienced side effects from the Suproxin.

These side effects are transitory, that is, they will only last for about 15 or 20 minutes. What will probably happen is that your feet will feel numb, you will have an itching sensation over parts of your body, and you may get a slight headache. Again these are side effects lasting IS or 20 minutes.

And again, the physician repeated these symptoms while injecting the subject.

None of these symptoms, of course, are consequences of an injection of epinephrine and, in effect, these instructions provide the subject with a completely inappropriate ex- planation of his bodily feelings. This condi- tion was introduced as a control condition of sorts. It seemed possible that the descrip- tion of side effects in the Epi Inf condition might turn the subject introspective, self-ex- amining, possibly slightly troubled. Differ- ences on the dependent variable between the Epi Inf and Epi Ign conditions might, then, be due to such factors rather than to dif- ferences in appropriateness. The false symp- toms in the Epi Mis condition should simi- larly turn the subject introspective, etc., but the instructions in this condition do not pro- vide an appropriate explanation of the sub- ject's state.

Subjects in all of the above conditions were injected with epinephrine. Finally, there was a placebo condition in which subjects, who w^re injected with saline solution, were given precisely the same treatment as subjects in the Epi Ign condition.

Producing an Emotion Inducing tion

Our initial hypothesis has suggested that given a state of physiological arousal for which the individual has no adequate ex- planation, cognitive factors can lead the in- dividual to describe his feelings with any of a diversity of emotional labels. In order to test this hypothesis, it was decided to manipu- late emotional states which can be considered quite different—euphoria and anger.

There are, of course, many ways to induce such states. In our own program of re- search, we have concentrated on social de- terminants of emotional states and have been able to demonstrate in other studies that people do evaluate their own feelings by com- paring themselves with others around them (Schachter 1959; Wrightsman 1960). In this experiment we have attempted again to manipulate emotional state by social means. In one set of conditions, the subject is placed together with a stooge who has been trained

384 STANLEY SCHACHTER AND JEROME E. SINGER

to act euphorically. In a second set of con- ditions the subject is with a stooge trained to act in an angry fashion.

Euphoria Immediately9 after the subject had been

injected, the physician left the room and the experimenter returned with a stooge whom he introduced as another subject, then said:

Both of you have had the Suproxin shot and you'll both be taking the same tests of vision. What I ask you to do now is just wait for 20 minutes. The reason for this is simply that we have to allow 20 minutes for the Suproxin to get from the injection site into the bloodstream. At the end of 20 minutes when we are certain that most of the Suproxin has been absorbed into the bloodstream, we'll begin the tests of vision.

The room in which this was said had been deliberately put into a state of mild disarray. As he was leaving, the experimenter apolo- getically added:

The only other thing I should do is to apologize for the condition of the room. I just didn't have time to clean it up. So, if you need any scratch paper or rubber bands or pencils, help yourself. I'll be back in 20 minutes to begin the vision tests.

As soon as the experimenter had left, the stooge introduced himself again, made a series of standard icebreaker comments, and then launched his routine. For observation purposes, the stooge's act was broken into a series of standard units, demarcated by a change in activity or a standard comment. In sequence, the units of the stooge's routine were the following:

1. Stooge reaches for a piece of paper and starts doodling saying, "They said we could use this for scratch, didn't they?" He doodles a fish for some 30 seconds, then says:

6 It was, of course, imperative that the se- quence with the stooge begin before the sub- ject felt his first symptoms for otherwise the subject would be virtually forced to interpret his feelings in terms of events preceding the stooge's entrance. Pretests had indicated that, for most subjects, epinphrine-caused symptoms began within 3-5 minutes after in- jection. A deliberate attempt was made then to bring in the stooge within 1 minute after the subject's injection.

2. "This scrap paper isn't even much good for doodling" and crumples paper and attempts to throw it into wastebasket in far corner of the room. He misses but this leads him into a "basketball game." He crumples up other sheets of paper, shoots a few baskets, says "Two points" occasion- ally. He gets up and does a jump shot saying, "The old jump shot is really on today."

3. If the subject has not joined in, the stooge throws a paper basketball to the subject saying, "Here, you try it."

4. Stooge continues his game saying, "The trouble with paper basketballs is that you don't really have any control."

5. Stooge continues basketball, then gives it up saying, "This is one of my good days. I feel like a kid again. I think I'll make a plane." He makes a paper airplane saying, "I guess I'll make one of the longer ones."

6. Stooge flies plane. Gets up and re- trieves plane. Flies again, etc.

7. Stooge throws plane at subject. 8. Stooge, flying plane, says, "Even when

I was a kid, I was never much good at this."

9. Stooge tears off part of plane saying, "Maybe this plane can't fly but at least it's good for something." He wads up paper and making a slingshot of a rubber band begins to shoot the paper.

10. Shooting, the stooge says, "They [paper ammunition] really go better if you make them long. They don't work right if you wad them up."

11. While shooting, stooge notices a sloppy pile of manila folders on a table. He builds a tower of these folders, then goes to the opposite end of the room to shoot at the tower.

12. He misses several times, then hits and cheers as the tower falls. He goes over to pick up the folders.

13. While picking up, he notices, behind a portable blackboard, a pair of hula hoops which have been covered with black tape with a few wires sticking out of the tape. He reaches for these, taking one for him- self and putting the other aside but within reaching distance of the subject. The stooge tries the hula hoop, saying, "This isn't as easy as it looks."

14. Stooge twirls hoop wildly on arm, saying, "Hey, look at this—this is great."

15. Stooge replaces the hula hoop and sits down with his feet on the table. Shortly thereafter the experimenter re- turns to the room.

DETERMINANTS OF EMOTIONAL STATE 385

This routine was completely standard, though its pace, of course, varied depending upon the subject's reaction, the extent to which he entered into this bedlam and the extent to which he initiated activities of his own. The only variations from this standard routine were those forced by the subject. Should the subject originate some nonsense of his own and request the stooge to join in, he would do so. And, he would, of course, respond to any comments initiated by the subject.

Subjects in each of the three "appropriate- ness" conditions and in the placebo condi- tion were submitted to this setup. The stooge, of course, never knew in which con- dition any particular subject fell.

Anger Immediately after the injection, the ex-

perimenter brought a stooge into the subject's room, introduced the two and after explain- ing, the necessity for a 20 minute delay for "the Suproxin to get from the injection site into the bloodstream" he continued, "We would like you to use these 20 minutes to answer these questionnaires." Then handing out the questionnaires, he concludes with, "I'll be back in 20 minutes to pick up the questionnaires and begin the tests of vision."

Before looking at the questionnaire, the stooge says to the subject,

I really wanted to come for an experi- ment today, but I think it's unfair for them to give you shots. At least, they should have told us about the shots when they called u s ; you hate to refuse, once you're here already.

The questionnaires, five pages long, start off innocently requesting face sheet informa- tion and then grow increasingly personal and insulting. The stooge, sitting directly op- posite the subject, paces his own answers so that at all times subject and stooge are work- ing on the same question. At regular points in the questionnaire, the stooge makes a series of standardized comments about the questions. His comments start off innocently enough, grow increasingly querulous, and finally he ends up in a rage. In sequence, he makes the following comments.

1. Before answering any items, he leafs quickly through the questionnaire saying, "Boy, this is a long one."

2. Question 7 on the questionnaire re- quests, "List the foods that you would eat in a typical day." The stooge comments,

"Oh for Pete's sake, what did I have for breakfast this morning?"

3. Question 9 asks, "Do you ever hear bells ? How often ? " The stooge remarks, "Look at Question 9. How ridiculous can you get? I hear bells every time' I change classes."

4. Question 13 requests, "List the child- hood diseases you have had and the age at which you had them" to which the stooge remarks, "I get annoyed at this childhood disease question. I can't remem- ber what childhood diseases I had, and especially at what age. Can you?"

5. Question 17 asks "What is your fath- er's average annual income?" and the stooge says, "This really irritates me. It's none of their business what my father makes. I'm leaving that blank."

6. Question 25 presents a long series of items such as "Does not bathe or wash reg- ularly," "Seems to need psychiatric care," etc. and requests the respondent to write down for which member of his immediate family each item seems most applicable. The question specifically prohibits the an- swer "None" and each item must be an- swered. The stooge says, "I'll be damned if I'll fill out Number 25. 'Does not bathe or wash regularly'—that's a real insult." He then angrily crosses out the entire item.

7. Question 28 reads : "How many times each week do you

have sexual intercourse?" 0-1 2-3 4-6 7 and over The

stooge bites out, "The hell with it! I don't have to tell them all this."

8. The stooge sits sullenly for a few mo- ments then he rips up his questionnaire, crumples the pieces and hurls them to the floor, saying, "I'm not wasting any more time. I'm getting my books and leaving" and he stamps out of the room.

9. The questionnaire continues for eight more questions ending with: "With how many men (other than your father) has your mother had extramarital relation- ships?"

4 and under : 5-9 : 10 and over

Subjects in the Epi Ign, Epi Inf and Placebo conditions were run through this "anger" inducing sequence. The stooge, again, did not know to which condition the subject had been assigned.

In summary, this is a seven condition ex- periment which, for two different emotional states, allows us (a) to evaluate the effects of "appropriateness" on emotional inducibility

386 STANLEY SCHACHTER AND JEROME E. SINGER

and (i>) to begin to evaluate the effects of sympathetic activation on emotional induci- bility. In schematic form the conditions are the following:

EUPHORIA

Epi Inf Epi Ign Epi Mis Placebo

ANGER

Epi Inf Epi Ign Placebo

The Epi Mis condition was not run in the Anger sequence. This was originally con- ceived as a control condition and it was felt that its inclusion in the Euphoria conditions alone would suffice as a, means of evaluating the possible artifactual effect of the Epi Inf instructions.

Measurement Two types of measures of emotional state

were obtained. Standardized observation through a one-way mirror was the technique used to assess the subject's behavior. To what extent did he act euphoric or angry? Such behavior can be considered in a way as a "semiprivate" index of mood for as far as the subject was concerned, his emotional be- havior could be known only to the other per- son in the room—presumably another stu- dent. The second type of measure was self- report in which, on a variety of scales, the subject indicated his mood of the moment. Such measures can be considered "public" indices of mood for they would, of course, be available to the experimenter and his associates.

Observation Euphoria. For each of the first 14 units

of the stooge's standardized routine an ob- server kept a running chronicle of what the subject did and said. For each unit the ob- server coded the subject's behavior in one or more of the following categories:

Category 1: Joins in activity. If the sub- ject entered into the stooge's activities, e.g., if he made or flew airplanes, threw paper basketballs, hula hooped, etc., his behavior was coded in this category.

Category 2 : Initiates new activity. A sub- ject was so coded if he gave indications of creative euphoria, that is, if, on his own, he initiated behavior outside of the stooge's rou- tine. Instances of such behavior would be the subject who threw open the window and, laughing, hurled paper basketballs at pass- ersby; or, the subject who jumped on a table and spun one hula hoop on his leg and the other on his neck.

Categories 3 and 4: Ignores or watches

stooge. Subjects who paid flatly no atten- tion to the stooge or who, with or without comment, simply watched the stooge without joining in his activity were coded in these categories.

For any particular unit of behavior, the subject's behavior was coded in one or more of these categories. To test reliability of coding two observers independently coded two experimental sessions. The observers agreed completely on the coding of 88% of the units.

Anger. For each of the units of stooge behavior, an observer recorded the subject's responses and coded them according to the following category scheme:

Category 1: Agrees. In response to the stooge the subject makes a comment indi- cating that he agrees with the stooge's stand- ardized comment or that he, too, is irked by a particular item on the questionnaire. For example, a subject who responded to the stooge's comment on the "father's income" question by saying, "I don't like that kind of personal question either" would be so coded (scored +2).

Category 2: Disagrees. In response to the stooge's comment, the subject makes a com- ment which indicates that he disagrees with the stooge's meaning or mood; e.g., in re- sponse to the stooge's comment on the "fath- er's income" question, such a subject might say, "Take it easy, they probably have a good reason for wanting the information" (scored -2).

Category 3: Neutral. A noncommittal or irrelevant response to the stooge's remark (scored 0).

Category 4: Initiates agreement or dis- agreement. With no instigation by the stooge, a subject, so coded, would have volunteered a remark indicating that he felt the same way or, alternatively, quite differ- ently than the stooge. Examples would be "Boy I hate this kind of thing^' or "I'm en- joying this" (scored +2 or — 2 ) .

Category S: Watches. The subject makes no verbal response to the stooge's comment but simply looks directly at him (scored 0).

Category 6: Ignores. The subject makes no verbal response to the stooge's comment nor does he look at him; the subject, paying no attention at all to the stooge, simply works at his own questionnaire (scored — 1 ) .

A subject was scored in one or more of these categories for each unit of stooge be- havior. To test reliability, two observers in- dependently coded three experimental ses- sions. In order to get a behavioral index of anger, observation protocol was scored ac- cording to the values presented in parentheses

DETERMINANTS OF EMOTIONAL STATE 387

after each of the above definitions of cate- gories. In a unit-by-unit comparison, the two observers agreed completely on the scoring of 71% of the units jointly observed. The scores of the two observers differed by a value of 1 or less for 88% of the units coded and in not a single case did the two observers differ in the direction of their scoring of a unit.

Self Report of Mood and Physical Condition

When the subject's session with the stooge was completed, the experimenter returned to the room, took pulses and said:

Before we proceed with the vision tests, there is one other kind of information which we must have. We have found, as you can probably imagine, that there are many things beside Suproxin that affect how well you see in our tests. How hungry you are, how tired you are, and even the mood you're in at the time— whether you feel happy or irritated at the time of testing will affect how well you see. To understand the data we collect on you, then, we must be able to figure out which effects are due to causes such as these and which are caused by Suproxin.

The only way we can get such informa- tion about your physical and emotional state is to have you tell us. I'll hand out these questionnaires and ask you to an- swer them as accurately as possible. Ob- viously, our data on the vision tests will only be as accurate as your description of your mental and physical state.

In keeping with this spiel, the questionnaire that the experimenter passed out contained a number of mock questions about hunger, fatigue, etc., as well as questions of more immediate relevance to the experiment. To measure mood or emotional state the follow- ing two were the crucial questions:

1. How irritated, angry or annoyed would you say you feel at present ?

! I ! I don't I feel I feel I feel I feel

feel at all a little quite very extremely irritated irritated irritated irritated irritated or angry and angry and angry and angry and angry

(0) (1) (2) (3) (4)

2. How good or happy would you say you feel at present?

I don't feel at all

happy or good

(0)

I feel a little happy

and good (1)

I feel quite happy

and good (2)

I feel very

happy and good

(3)

I feel extremely

happy and good

(4)

To measure the physical effects of epi- nephrine and determine whether or not the injection had been successful in producing the necessary bodily state, the following questions were asked:

1. Have you experienced any palpitation (consciousness of your own heart beat) ?

I

Not at all (0)

A slight amount

(0

A moderate amount

(2)

An intense amount

(3)

2. Did you feel any tremor (involuntary shaking of the hands, arms or legs) ?

I l l I

Not at A slight all amount (0) (1)

A moderate amount

(2)

An intense amount

(3)

To measure possible effects of the instruc- tions in the Epi Mis condition, the following questions were asked:

1. Did you feel any numbness in your feet? 2. Did you feel any itching sensation ? 3. Did you experience any feeling of head-

ache? To all three of these questions was at-

tached a four-point scale running from "Not at all" to "An intense amount."

In addition to these scales, the subjects were asked to answer two open-end questions on other physical or emotional sensations they may have experienced during the experi- mental session. A final measure of bodily state was pulse rate which was taken by the physician or the experimenter at two times—• immediately before the injection and immedi- ately after the session with the stooge.

When the subjects had completed these questionnaires, the experimenter announced that the experiment was over, explained the deception and its necessity in detail, an- swered any questions, and swore the subjects to secrecy. Finally, the subjects answered a brief questionnaire about their experiences, if any, with adrenalin and their previous knowl- edge or suspicion of the experimental setup. There was no indication that any of the sub- jects had known about the experiment before- hand but 11 subjects were so extremely sus- picious of some crucial feature of the experiment that their data were automatically discarded.

Subjects The subjects were all male, college stu-

dents taking classes in introductory psychol- ogy at the University of Minnesota. Some 90% of the students in these classes volunteer for a subject pool for which they receive two extra points on their final exam for every

388 STANLEY SCHACHTER AND JEROME E. SINGER

hour that they serve as experimental sub- jects. For this study the records of all po- tential subjects were cleared with the Student Health Service in order to insure that no harmful effects would result from the injections.

Evaluation of the Experimental Design

The ideal test of our propositions would require circumstances which our experiment is far from realizing. First, the proposition that: "A state of physiological arousal for which an individual has no immediate ex- planation will lead him to label this state in terms of the cognitions available to him" obviously requires conditions under which the subject does not and cannot have a proper explanation of his bodily state. Though we toyed with such fantasies as ventilating the experimental room with vaporized adrenalin, reality forced us to rely on the disguised in- jection of Suproxin—a technique which was far from ideal for no matter what the ex- perimenter told them, some subjects would inevitably attribute their feelings to the in- jection. To the extent that subjects did so, differences between the several appropriate- ness conditions should be attenuated.

Second, the proposition that: "Given the same cognitive circumstances the individual will react emotionally only to the extent that he experiences a state of physiological arousal" requires for its ideal test the ma- nipulation of states of physiological arousal and of physiological quiescence. Though there is no question that epinephrine effec- tively produces a state of arousal, there is also no question that a placebo does not pre- vent physiological arousal. To the extent

that the experimental situation effectively produces sympathetic stimulation in placebo subjects, the proposition is difficult to test, for such a factor would attenuate differences be- tween epinephrine and placebo subjects.

Both of these factors, then, can be expected to interfere with the test of our several propositions. In presenting the results of this study, we shall first present condition by condition results and then evaluate the effect of these two factors on experimental differ- ences.

RESULTS

Effects of the Injections on Bodily State

Let us examine first the success of the injections at producing the bodily state required to examine the proposi- tions at test. Does the injection of epinephrine produce symptoms of sym- pathetic discharge as compared with the placebo injection? Relevant data are presented in Table 1 where it can be immediately seen that on all items subjects who were in epinephrine con- ditions show considerably more evi- dence of sympathetic activation than do subjects in placebo conditions. In all epinephrine conditions pulse rate in- creases significantly when compared with the decrease characteristic of the placebo conditions. On the scales it is clear that epinephrine subjects experi-

TABLE 1

THE EFFECTS OF THE INJECTIONS ON BODILY STATE

PnnHif ion

Euphoria Epi Inf Epi Ign Epi Mis Placebo

Anger Epi Inf Epi Ign Placebo

N

27 26 26 26

23 23 23

Pulse

Pre

85.7 84.6 82.9 80.4

85.9 85.0 84.5

Post

88.6 85.6 86.0 77.1

92.4 96.8 79.6

Self-rating of

Palpitation

1.20 1.83 1.27 0.29

1.26 1.44 0.59

Tremor

1.43 1.76 2.00 0.21

1.41 1.78 0.24

Numbness

0 0.15 0.06 0.09

0.17 0 0.14

Itching

0.16 0 0.08 0

0 0.06 0.06

Headache

0.32 0.55 0.23 0.27

0.11 0.21 0.06

DETERMINANTS OF EMOTIONAL STATE 389

ence considerably more palpitation and tremor than do placebo subjects. In all possible comparisons on these symp- toms, the mean scores of subjects in any of the epinephrine conditions are greater than the corresponding scores in the placebo conditions at better than the .001 level of significance. Exami- nation of the absolute values of these scores makes it quite clear that subjects in epinephrine conditions were, indeed, in a state of physiological arousal, while most subjects in placebo conditions were in a relative state of physiological quiescence.

The epinephrine injection, of course, did not work with equal effectiveness for all subjects; indeed for a few sub- jects it did not work at all. Such sub- jects reported almost no palpitation or tremor, showed no increase in pulse and described no other relevant phys- ical symptoms. Since for such subjects the necessary experimental conditions were not established, they were auto- matically excluded from the data and all further tabular presentations will not include such subjects. Table 1, how- ever, does include the data of these sub- jects. There were four such subjects in euphoria conditions and one of them in anger conditions.

In order to evaluate further data on Epi Mis subjects it is necessary to note the results of the "numbness," "itch- ing," and "headache" scales also pre- sented in Table 1. Clearly the subjects in the Epi Mis condition do not differ on these scales from subjects in any of the other experimental conditions.

Effects of the Manipulations on Emotional State

Euphoria: Self-report. The effects of the several manipulations on emo- tional state in the euphoria conditions are presented in Table 2. The scores recorded in this table are derived, for each subject, by subtracting the value

of the point he checks on the irritation scale from the value of the point he checks on the happiness scale. Thus, if a subject were to check the point "I feel a little irritated and angry" on the irritation scale and the point "I feel very happy and good" on the happiness scale, his score would be +2. The higher the positive value, the happier and better the subject reports himself as feeling. Though we employ an in- dex for expositional simplicity, it should be noted that the two com- ponents of the index each yield results completely consistent with those ob- tained by use of this index.

Let us examine first the effects of the appropriateness instructions. Compari- son of the scores for the Epi Mis and Epi Inf conditions makes it immediately clear that the experimental differences are not due to artifacts resulting from the informed instructions. In both con- ditions the subject was warned to ex- pect a variety of symptoms as a conse- quence of the injection. In the Epi Mis condition, where the symptoms were inappropriate to the subject's bodily state the self-report score is al- most twice that in the Epi Inf condition where the symptoms were completely appropriate to the subject's bodily state. It is reasonable, then, to attribute dif- ferences between informed subjects and those in other conditions to dif- ferences in manipulated appropriate- ness rather than to artifacts such as introspectiveness or self-examination.

It is clear that, consistent with expec- tations, subjects were more susceptible to the stooge's mood and consequently more euphoric when they had no ex- planation of their own bodily states than when they did. The means of both the Epi Ign and Epi Mis condi- tions are considerably greater than the mean of the Epi Inf condition.

It is of interest to note that Epi Mis subjects are somewhat more euphoric

390 STANLEY SCHACHTER AND JEROME E. SINGER

TABLE 2

SELF-REPORT OF EMOTIONAL STATE IN THE EUPHORIA CONDITIONS

Condi- tion

Epi Inf Epi Ign Epi Mis Placebo

N

25 25 25 26

Self- Report scales

0.98 1.78 1.90 1.61

Comparison

Epi Inf vs. Epi Mis Epi Inf vs. Epi Ign Placebo vs. Epi Mis,

Ign, or Inf

#"

<.01 .02

ns

All p values reported throughout paper are two- tailed.

than are Epi Ign subjects. This pat- tern repeats itself in other data shortly to be presented. We would attribute this difference to differences in the ap- propriateness dimension. Though, as in the Epi Ign condition, a subject is not provided with an explanation of his bodily state, it is, of course, possible that he will provide one for himself which is not derived from his interac- tion with the stooge. Most reasonably he could decide for himself that he feels this way because of the injection. To the extent that he does so he should be less susceptible to the stooge. It seems probable that he would be less likely to hit on such an explanation in the Epi Mis condition than in the Epi Ign condition for in the Epi Mis condi- tion both the experimenter and the doc- tor have told him that the effects of the injection would be quite different from what he actually feels. The effect of such instructions is probably to make it more difficult for the subject himself to hit on the alternative explanation de- scribed above. There is some evidence to support this analysis. In open-end questions in which subjects described their own mood and state, 28% of the subjects in the Epi Ign condition made some connection between the injection and their bodily state compared with the 16% of subjects in the Epi Mis condition who did so. It could be con- sidered, then, that these three condi- tions fall along a dimension of appropri-

ateness, with the Epi Inf condition at one extreme and the Epi Mis condition at the other.

Comparing the placebo to the epi- nephrine conditions, we note a pattern which will repeat itself throughout the data. Placebo subjects are less eu- phoric than either Epi Mis or Epi Ign subjects but somewhat more euphoric than Epi Inf subjects. These differ- ences are not, however, statistically significant. We shall consider the epinephrine-placebo comparisons in de- tail in a later section of this paper fol- lowing the presentation of additional relevant data. For the moment, it is clear that, by self-report manipulating appropriateness has had a very strong effect on euphoria.

Behavior. Let us next examine the extent to which the subject's behavior was affected by the experimental ma- nipulations. To the extent that his mood has been affected, one should ex- pect that the subject will join in the stooge's whirl of manic activity and initiate similar activities of his own. The relevant data are presented in Table 3. The column labeled "Activity

TABLE 3 BEHAVIORAL INDICATIONS OF EMOTIONAL

STATE IN THE EUPHORIA CONDITIONS

Condition

Epi Inf Epi Ign Epi Mis Placebo

N

25 25 25 26

Activity index

12.72 18.28 22.56 16.00

Mean num- ber of acts initiated

.20

.56

.84

.54

P value

Comparison

Epi Inf vs. Epi Mis Epi Inf vs. Epi Ign Plac vs. Epi Mis, Ign,

or Inf

Activity index

.05 ns ns

Initiates

.03

.08 ns

a Tested by Xz comparison of the proportion of sub- jects in each condition initiating new acts.

D E T E R M I N A N T S OF EMOTIONAL STATE 391

index" presents summary figures on the extent to which the subject joined in the stooge's activity. This is a weighted index which reflects both the nature of the activities in which the subject engaged and the amount of time he was active. The index was devised by assigning the following weights to the subject's activities: 5—hula hoop- ing; 4—shooting with slingshot; 3— paper airplanes; 2—paper basketballs; 1—doodling; 0—does nothing. Pre- test scaling on 15 college students ordered these activities with respect to the degree of euphoria they repre- sented. Arbitrary weights were as- signed so that the wilder the activity, the heavier the weight. These weights are multiplied by an estimate of the amount of time the subject spent in each activity and the summed products make up the activity index for each sub- ject. This index may be considered a measure of behavioral euphoria. It should be noted that the same between- condition relationships hold for the two components of this index as for the index itself.

The column labeled "Mean number of acts initiated" presents the data on the extent to which the subject deviates from the stooge's routine and initiates euphoric activities of his own.

On both behavioral indices, we find precisely the same pattern of relation- ships as those obtained with self-re- ports. Epi Mis subjects behave some- what more euphorically than do Epi Ign subjects who in turn behave more euphorically than do Epi Inf subjects. On all measures, then, there is con- sistent evidence that a subject will take over the stooge's euphoric mood to the extent that he has no other explana- tion of his bodily state.

Again it should be noted that on these behavioral indices, Epi Ign and Epi Mis subjects are somewhat more

euphoric than placebo subjects but not significantly so.

Anger: Self-report. Before present- ing data for the anger conditions, one point must be made about the anger ma- nipulation. In the situation devised, anger, if manifested, is most likely to be directed at the experimenter and his annoyingly personal questionnaire. As we subsequently discovered, this was rather unfortunate, for the subjects, who had volunteered for the experiment for extra points on their final exam, simply refused to endanger these points by publicly blowing up, admitting their irritation to the experimenter's face or spoiling the questionnaire. Though as the reader will see, the subjects were quite willing to manifest anger when they were alone with the stooge, they hesitated to do so on material (self- ratings of mood and questionnaire) that the experimenter might see and only after the purposes of the experi- ment had been revealed were many of these subjects willing to admit to the experimenter that they had been irked or irritated.

This experimentally unfortunate situ- ation pretty much forces us to rely on the behavioral indices derived from ob- servation of the subject's presumably private interaction with the stooge. We do, however, present data on the self-report scales in Table 4. These figures are derived in the same way as the figures presented in Table 2 for the euphoria conditions, that is, the value checked on the irritation scale is sub- tracted from the value checked on the happiness scale. Though, for the rea- sons stated above, the absolute magni- tude of these figures (all positive) is relatively meaningless, we can, of course, compare condition means within the set of anger conditions. With the happiness-irritation index employed, we should, of course, anticipate pre- cisely the reverse results from those ob-

392 STANLEY SCHACHTER AND JEROME E. SINGER

TABLE 4 SELF-REPORT OF EMOTIONAL STATE IN

THE ANGER CONDITIONS

Condition

Epi Inf

Epi Ign

Placebo

N

22

23

23

Self- Report scales

1.91

1.39

1.63

Comparison

Epi Inf vs. Epi Ign

Placebo vs. Epi Ign or Inf

t

.08

ns

tained in the euphoria conditions; that is, the Epi Inf subjects in the anger conditions should again be less suscep- tible to the stooge's mood and should, therefore, describe themselves as in a somewhat happier frame of mind than subjects in the Epi Ign condition. This is the case; the Epi Inf subjects average 1.91 on the self-report scales while the Epi Ign subjects average 1.39.

Evaluating the effects of the injec- tions, we note again that, as antici- pated, Epi Ign subjects are somewhat less happy than Placebo subjects but, once more, this is not a significant difference.

Behavior. The subject's responses to the stooge, during the period when both were filling out their question- naires, were systematically coded to provide a behavioral index of anger. The coding scheme and the numerical values attached to each of the categories have been described in the methodology section. To arrive at an "Anger in- dex" the numerical value assigned to a subject's responses to the stooge is summed together for the several units of stooge behavior. In the coding scheme used, a positive value to this index indicates that the subject agrees with the stooge's comment and is grow- ing angry. A negative value indicates that the subject either disagrees with the stooge or ignores him.

The relevant data are presented in

Table 5. For this analysis, the stooge's routine has been divided into two phases—the first two units of his be- havior (the "long" questionnaire and "What did I have for breakfast?") are considered essentially neutral revealing nothing of the stooge's mood; all of the following units are considered "angry" units for they begin with an irritated re- mark about the "bells" question and end with the stooge's fury as he rips up his questionnaire and stomps out of the room. For the neutral units, agreement or disagreement with the stooge's remarks is, of course, mean- ingless as an index of mood and we should anticipate no difference between conditions. As can be seen in Table 5, this is the case.

For the angry units, we must, of course, anticipate that subjects in the Epi Ign condition will be angrier than subjects in the Epi Inf condition. This is indeed the case. The Anger index for the Epi Ign condition is positive and large, indicating that these subjects have become angry, while in the Epi Inf condition the Anger index is slightly negative in value indicating that these subjects have failed to catch the stooge's mood at all. It seems clear that providing the subject with an ap-

TABLE 5

BEHAVIORAL INDICATIONS OF EMOTIONAL STATE IN THE ANGER CONDITIONS

Condition

Epi Inf Epi Ign Placebo

N

22 23 22»

Neutral units

+0.07 +0.30 -0.09

Comparison for anger units

Anger units

-0.18 +2.28 +0.79

P

Epi Inf vs. Epi Ign Epi Ign vs. Placebo Placebo vs. Epi Inf

<.05 ns

« For one subject in this condition the sound system went dead and the observer could not, of course, code his reactions.

DETERMINANTS OF EMOTIONAL STATE 393

propriate explanation of his bodily state greatly reduces his tendency to interpret his state in terms of the cog- nitions provided by the stooge's angry behavior.

Finally, on this behavioral index, it can be seen that subjects in the Epi Ign condition are significantly angrier than subjects in the Placebo condition. Be- haviorally, at least, the injection of epi- nephrine appears to have led subjects to an angrier state than comparable subjects who received placebo shots.

Conformation of Data to Theoretical Expectations

Now that the basic data of this study have been presented, let us examine closely the extent to which they con- form to theoretical expectations. If our hypotheses are correct and if this ex- perimental design provided a perfect test for these hypotheses, it should be anticipated that in the euphoria condi- tions the degree of experimentally pro- duced euphoria should vary in the following fashion:

Epi Mis ̂ Epi Ign > Epi Inf = Placebo

And in the anger conditions, anger should conform to the following pat- tern :

Epi Ign > Epi Inf = Placebo

In both sets of conditions, it is the case that emotional level in the Epi Mis and Epi Ign conditions is con- siderably greater than that achieved in the corresponding Epi Inf conditions. The results for the Placebo condition, however, are ambiguous for consist- ently the Placebo subjects fall between the Epi Ign and the Epi Inf subjects. This is a particularly troubling pattern for it makes it impossible to evaluate unequivocally the effects of the state of physiological arousal and indeed raises serious questions about our entire theo- retical structure. Though the emo-

tional level is consistently greater in the Epi Mis and Epi Ign conditions than in the Placebo condition, this differ- ence is significant at acceptable prob- ability levels only in the anger con- ditions.

In order to explore the problem fur- ther, let us examine the experimental factors identified earlier, which might have acted to restrain the emotional level in the Epi Ign and Epi Mis con- ditions. As was pointed out earlier, the ideal test of our first two hypotheses requires an experimental setup in which the subject has flatly no way of evalu- ating his state of physiological arousal other than by means of the experimen- tally provided cognitions. Had it been possible to physiologically produce a state of sympathetic activation by means other than injection, one could have approached this experimental ideal more closely than in the present setup. As it stands, however, there is always a reasonable alternative cognition avail- able to the aroused subject—he feels the way he does because of the injec- tion. To the extent that the subject seizes on such an explanation of his bodily state, we should expect that he will be uninfluenced by the stooge. Evidence presented in Table 6 for the anger condition and in Table 7 for the euphoria conditions indicates that this is, indeed, the case.

As mentioned earlier, some of the Epi Ign and Epi Mis subjects in their answers to the open-end questions clearly attributed their physical state to the injection, e.g., "the shot gave me the shivers." In Tables 6 and 7 such subjects are labeled "Self-in- formed." In Table 6 it can be seen that the self-informed subjects are con- siderably less angry than are the re- maining subjects; indeed, they are not angry at all. With these self-informed subjects eliminated the difference be- tween the Epi Ign and the Placebo

394 STANLEY SCHACHTER AND JEROME E. SINGER

TABLE 6

THE EFFECTS OF ATTRIBUTING BODILY STATE TO THE INJECTION ON ANGER IN THE

ANGER EPI IGN CONDITION

Condition

Self-informed subjects Others Self-informed vs.

Others

N

3 20

Anger index

-1.67 + 2.88

P

ns ns .05

conditions is significant at the .01 level of significance.

Precisely the same pattern is evident in Table 7 for the euphoria conditions. In both the Epi Mis and the Epi Ign conditions, the self-informed subjects have considerably lower activity indices than do the remaining subjects. Elim- inating self-informed subjects, compari- son of both of these conditions with the Placebo condition yields a difference significant at the .03 level of signifi- cance. It should be noted, too, that the self-informed subjects have much the same score on the activity index as do the experimental Epi Inf subjects (Table 3).

It would appear, then, that the ex- perimental procedure of injecting the subjects, by providing an alternative cognition, has, to some extent, obscured the effects of epinephrine. When ac- count is taken of this artifact, the evi- dence is good that the state of physi- ological arousal is a necessary com- ponent of an emotional experience for when self-informed subjects are re- moved, epinephrine subjects give con- sistent indications of greater emotion- ality than do placebo subjects.

Let us examine next the fact that consistently the emotional level, both reported and behavioral, in Placebo conditions is greater than that in the Epi Inf conditions. Theoretically, of course, it should be expected that the two conditions will be equally low, for

by assuming that emotional state is a joint function of a state of physiologi- cal arousal and of the appropriateness of a cognition we are, in effect, as- suming a multiplicative function, so that if either component is at zero, emo- tional level is at zero. As noted earlier this expectation should hold if we can be sure that there is no sympathetic activation in the Placebo conditions. This assumption, of course, is com- pletely unrealistic for the injection of placebo does not prevent sympathetic activation. The experimental situa- tions were fairly dramatic and cer- tainly some of the placebo subjects gave indications of physiological arousal. If our general line of reasoning is correct, it should be anticipated that the emo- tional level of subjects who give indi- cations of sympathetic activity will be greater than that of subjects who do not. The relevant evidence is pre- sented in Tables 8 and 9.

As an index of sympathetic activa- tion we shall use the most direct and unequivocal measure available—change

TABLE 7

THE EFFECTS OF ATTRIBUTING BODILY STATE TO THE INJECTION ON EUPHORIA IN

THE EUPHORIA EPI IGN AND EPI Mis CONDITIONS

Epi Ign

Self-informed subjects Others Self-informed vs.

Others

8 17

Activity Index

11.63 21.14

ns ns .05

Epi Mis

Self-informed subjects Others Self-informed vs.

Others

N

5 20

Activity Index

12.40 25.10

P

ns ns .10

DETERMINANTS OF EMOTIONAL STATE 395

TABLE 8

SYMPATHETIC ACTIVATION AND EUPHORIA IN THE EUPHORIA PLACEBO CONDITION

Subjects whose :

Pulse decreased Pulse increased

or remained same Pulse decrease vs. pulse

increase or same

N

14 12

Activity index

10.67 23.17

t

ns ns

.02

in pulse rate. It can be seen in Table 1 that the predominant pattern in the Placebo condition is a decrease in pulse rate. We shall assume, therefore, that those subjects whose pulse increases or remains the same give indications of sympathetic activity while those sub- jects whose pulse decreases do not. In Table 8, for the euphoria condition, it is immediately clear that subjects who give indications of sympathetic activity are considerably more euphoric than are subjects who show no sympathetic activity. This relationship is, of course, confounded by the fact that euphoric subjects are considerably more active than noneuphoric subjects—a factor which independent of mood could ele- vate pulse rate. However, no such fac- tor operates in the anger condition where angry subjects are neither more active nor talkative than calm subjects. It can be seen in Table 9 that Placebo subjects who show signs of sympathetic

TABLE 9

SYMPATHETIC ACTIVATION AND ANGER IN ANGER PLACEBO CONDITION

Subjects whose :

Pulse decreased Pulse increased

or remained same Pusle decrease vs. pulse

increase or same

/Va

13 8

Anger index

+0.15 + 1.69

P

ns ns

.01

*N reduced by two cases owing to failure of sound system in one case and experimenter's failure to take pulse in another.

activation give indications of consider- ably more anger than do subjects who show no such signs. Conforming to expectations, sympathetic activation ac- companies an increase in emotional level.

It should be noted, too, that the emo- tional levels of subjects showing no signs of sympathetic activity are quite comparable to the emotional level of subjects in the parallel Epi Inf condi- tions (see Tables 3 and 5). The similarity of these sets of scores and their uniformly low level of indicated emotionality would certainly make it appear that both factors are essential to an emotional state. When either the level of sympathetic arousal is low or a completely appropriate cognition is available, the level of emotionality is low.

DISCUSSION

Let us summarize the major findings of this experiment and examine the ex- tent to which they support the proposi- tions offered in the introduction of this paper. It has been suggested, first, that given a state of physiological arousal for which an individual has no explana- tion, he will label this state in terms of the cognitions available to him. This implies, of course, that by manipulating the cognitions of an individual in such a state we can manipulate his feelings in diverse directions. Experimental re- sults support this proposition for fol- lowing the injection of epinephrine, those subjects who had no explanation for the bodily state thus produced, gave behavioral and self-report indications that they had been readily manipulable into the disparate feeling states of euphoria and anger.

From this first proposition, it must follow that given a state of physiologi- cal arousal for which the individual has a completely satisfactory explanation, he will not label this state in terms of

396 STANLEY SCHACHTER AND JEROME E. SINGER

the alternative cognitions available. Experimental evidence strongly sup- ports this expectation. In those con- ditions in which subjects were injected with epinephrine and told precisely what they would feel and why, they proved relatively immune to any effects of the manipulated cognitions. In the anger condition, such subjects did not report or show anger; in the euphoria condition, such subjects reported them- selves as far less happy than subjects with an identical bodily state but no adequate knowledge of why they felt the way they did.

Finally, it has been suggested that given constant cognitive circumstances, an individual will react emotionally only to the extent that he experiences a state of physiological arousal. With- out taking account of experimental arti- facts, the evidence in support of this proposition is consistent but tentative. When the effects of "self-informing" tendencies in epinephrine subjects and of "self-arousing" tendencies in pla- cebo subjects are partialed out, the evidence strongly supports the proposi- tion.

The pattern of data, then, falls neatly in line with theoretical expectations. However, the fact that we were forced, to some extent, to rely on internal analyses in order to partial out the effects of experimental artifacts inevita- bly makes our conclusions somewhat tentative. In order to further test these propositions on the interaction of cog- nitive and physiological determinants of emotional state, a series of additional experiments, published elsewhere, was designed to rule out or overcome the operation of these artifacts. In the first of these, Schachter and Wheeler (1962) extended the range of manipu- lated sympathetic activation by em- ploying three experimental groups— epinephrine, placebo, and a group in- jected with the sympatholytic agent,

chlorpromazine. Laughter at a slap- stick movie was the dependent variable and the evidence is good that amuse- ment is a direct function of manipu- lated sympathetic activation.

In order to make the epinephrine- placebo comparison under conditions which would rule out the operation of any self-informing tendency, two ex- periments were conducted on rats. In one of these Singer (1961) demon- strated that under fear inducing condi- tions, manipulated by the simultaneous presentation of a loud bell, a buzzer, and a bright flashing light, rats in- jected with epinephrine were consider- ably more frightened than rats injected with a placebo. Epinephrine-injected rats defecated, urinated, and trembled more than did placebo-injected rats. In nonfear control conditions, there were no differences between epinephrine and placebo groups, neither group giving any indication of fear. In another study, Latane and Schachter (1962) demonstrated that rats injected with epinephrine were notably more capable of avoidance learning than were rats injected with a placebo. Using a modi- fied Miller-Mowrer shuttlebox, these investigators found that during an ex- perimental period involving 200 massed trials, 15 rats injected with epinephrine avoided shock an average of 101.2 trials while 15 placebo-injected rats averaged only 37.3 avoidances.

Taken together, this body of studies does give strong support to the proposi- tions which generated these experi- mental tests. Given a state of sympa- thetic activation, for which no imme- diately appropriate explanation is available, human subjects can be readily manipulated into states of euphoria, anger, and amusement. Varying the intensity of sympathetic activation serves to vary the intensity of a vari- ety of emotional states in both rats and human subjects.

DETERMINANTS OF EMOTIONAL STATE 397

Let us examine the implications of these findings and of this line of thought for problems in the general area of the physiology of the emotions. We have noted in the introduction that the numerous studies on physiological differentiators of emotional states have, viewed en masse, yielded quite incon- clusive results. Most, though not all, of these studies have indicated no dif- ferences among the various emotional states. Since as human beings, rather than as scientists, we have no difficulty identifying, labeling, and distinguishing among our feelings, the results of these studies have long seemed rather puz- zling and paradoxical. Perhaps because of this, there has been a persistent tendency to discount such results as due to ignorance or methodological in- adequacy and to pay far more atten- tion to the very few studies which demonstrate some sort of physiological differences among emotional states than to the very many studies which indi- cate no differences at all. It is con- ceivable, however, that these results should be taken at face value and that emotional states may, indeed, be gener- ally characterized by a high level of sympathetic activation with few if any physiological distinguishes among the many emotional states. If this is cor- rect, the findings of the present study may help to resolve the problem. Ob- viously this study does not rule out the possibility of physiological differences among the emotional states. It is the case, however, that given precisely the same state of epinephrine-induced sym- pathetic activation, we have, by means of cognitive manipulations, been able to produce in our subjects the very disparate states of euphoria and anger. It may indeed be the case that cognitive factors are major determiners of the emotional labels we apply to a common state of sympathetic arousal.

Let us ask next whether our results

are specific to the state of sympathetic activation or if they are generalizable to other states of physiological arousal. It is clear that from our experiments proper, it is impossible to answer the question for our studies have been con- cerned largely with the effects of an epinephrine created state of sympa- thetic arousal. We would suggest, however, that our conclusions are gen- eralizable to almost any pronounced in- ternal state for which no appropriate explanation is available. This sugges- tion receives some support from the ex- periences of Nowlis and Nowlis (1956) in their program of research on the effects of drugs on mood. In their work the Nowlises typically administer a drug to groups of four subjects who are physically in one another's presence and free to interact. The Nowlises de- scribe some of their results with these groups as follows:

At first we used the same drug for all 4 men. In those sessions seconal, when com- pared with placebo, increased the checking of such words as expansive, forceful, coura- geous, daring, elated, and impulsive. In our first statistical analysis we were confronted with the stubborn fact that when the same drug is given to all 4 men in a group, the N that has to be entered into the analysis is 1, not 4. This increases the cost of an already expensive experiment by a considerable fac- tor, but it cannot be denied that the effects of these drugs may be and often are quite con- tagious. Our first attempted solution was to run tests on groups in which each man had a different drug during the same session, such as 1 on seconal, 1 on benzedrine, 1 on dramamine, and 1 on placebo. What does seconal do? Cooped up with, say, the ego- tistical benzedrine partner, the withdrawn, indifferent dramimine partner, and the slightly bored lactose man, the seconal sub- ject reports that he is distractible, dizzy, drifting, glum, defiant, languid, sluggish, dis- couraged, dull, gloomy, lazy, and slow! This is not the report of mood that we got when all 4 men were on seconal. It thus appears that the moods of the partners do definitely influence the effect of seconal (p. 350).

It is not completely clear from this description whether this "contagion"

398 STANLEY SCHACHTER AND JEROME E. SINGER

of mood is more marked in drug than in placebo groups, but should this be the case, these results would certainly support the suggestion that our findings are generalizable to internal states other than that produced by an injec- tion of epinephrine.

Finally, let us consider the implica- tions of our formulation and data for alternative conceptualizations of emo- tion. Perhaps the most popular cur- rent conception of emotion is in terms of "activation theory" in the sense em- ployed by Lindsley (1951) and Wood- worth and Schlosberg (1958). As we understand this theory, it suggests that emotional states should be considered as at one end of a continuum of activa- tion which is denned in terms of degree of autonomic arousal and of electro- encephalographic measures of activa- tion. The results of the experiment de- scribed in this paper do, of course, suggest that such a formulation is not completely adequate. It is possible to have very high degrees of activation without a subject either appearing to be or describing himself as "emotional." Cognitive factors appear to be indis- pensable elements in any formulation of emotion.

SUMMARY

It is suggested that emotional states may be considered a function of a state of physiological arousal and of a cogni- tion appropriate to this state of arousal. From this follows these propositions:

1. Given a state of physiological arousal for which an individual has no immediate explanation, he will label this state and describe his feelings in terms of the cognitions available to him. To the extent that cognitive factors are potent determiners of emotional states, it should be anticipated that precisely the same state of physiological arousal could be labeled "joy" or "fury" or "jealousy" or any of a great diversity

of emotional labels depending on the cognitive aspects of the situation.

2. Given a state of physiological arousal for which an individual has a completely appropriate explanation, no evaluative needs will arise and the in- dividual is unlikely to label his feelings in terms of the alternative cognitions available.

3. Given the same cognitive circum- stances, the individual will react emo- tionally or describe his feelings as emo- tions only to the extent that he experi- ences a state of physiological arousal.

An experiment is described which, together with the results of other studies, supports these propositions.

REFERENCES Ax, A. F. Physiological differentiation of

emotional states. Psychosom. Med., 1953, 15, 433-442.

CANNON, W. B. Bodily changes in pain, hunger, fear and rage. (2nd ed.) New York: Appleton, 1929.

CANTRIL, H., & HUNT, W. A. Emotional effects produced by the injection of adren- alin. Amer. J. Psychol, 1932, 44, 300-307.

FESTINGER, L. A theory of social compari- son processes. Hum. Relat., 19S4, 7, 114- 140.

HUNT, J. McV., COLE, M. W., & REIS, E. E. Situational cues distinguishing anger, fear, and sorrow. Amer. J. Psychol., 1958, 71, 136-151.

JAMES, W. The principles of psychology. New York: Holt, 1890.

LANDIS, C, & HUNT, W. A. Adrenalin and emotion. Psychol. Rev., 1932, 39, 467-485.

LATANE, B., & SCHACHTER, S. Adrenalin and avoidance learning. /. comp. physiol. Psychol., 1962, 65, 369-372.

LINDSLEY, D. B. Emotion. In S. S. Stevens (Ed.), Handbook of experimental psychol- ogy. New York: Wiley, 1951. Pp. 473- 516.

MARANON, G. Contribution a 1'etude de 1'ac- tion emotive de 1'adrenaline. Rev. Fran- cmse Endocrinol, 1924, 2, 301-325.

NOWLIS, V., & NOWLIS, H. H. The descrip- tion and analysis of mood. Ann. N. Y. Acad. Sci., 1956, 65, 345-355.

RUCKMICK, C. A. The psychology of feeling and emotion. New York: McGraw-Hill, 1936.

DETERMINANTS OF EMOTIONAL STATE 399

SCHACHTER, J. Pain, fear, and anger in hy- pertensives and normotensives: A psycho- physiologic study. Psychosom. Med., 1957, 19, 17-29.

SCHACHTER, S. The psychology of affiliation, Stanford, Calif.: Stanford Univer. Press, 1959.

SCHACHTER, S., & WHEELER, L. Epi- nephrine, chlorpromazine, and amusement. /. abnorm. soc. Psychol, 1962, 65, 121- 128.

SINGER, J. E. The effects of epinephrine, chlorpromazine and dibenzyline upon the fright responses of rats under stress and

non-stress conditions. Unpublished doc- toral dissertation, University of Minnesota, 1961.

WOLF, S., & WOLFF, H. G. Human gastric function. New York: Oxford Univer. Press, 1947.

WOODWORTH, R. S., & SCHLOSBERG, H. Ex- perimental psychology. New York: Holt, 19S8.

WRIGHTSMAN, L. S. Effects of waiting with others on changes in level of felt anxiety. J. abnorm. soc. Psychol., 1960, 61, 216-222.

(Received February 17, 1961)