attributive data rather than conscious entities. The suggestion is made, therefore, that a sensation can exist only as a percept, and that its analysis is made on the same level as that of any other percept. May we not, then, give up the attempt to list and enumerate sensations, and, likewise, their corresponding images? Instead, a program of investigation is favored which seeks to study the nature of the attributive aspects of consciousness and their primary and secondary integrations. Affection, too, may fall into this scheme, not as a conscious entity, but as an attribute of a higher order, attaching to definite integrations of those qualitative and quantitative attributes which constitute both the perceptual and the ideational types of experience.

Concerning the Sensation Quality (A Behavioristic Account). EDWARD C. TOLMAN, University of California.

A careful analysis seems to indicate that in addition to the immediate presence of the quality or quale itself the sensing of a sensation-quality also involves, or is at any rate accompanied by, a capacity or readiness on the part of the individual experiencing the quality to indicate the relations of similarity and difference of that quality to other qualities.

We may designate this capacity or readiness to indicate the relations of the given quality to other qualities-a readiness of the "term-character" of the quality to function in behavior. That is, by the term-character of the quality we will mean the latter in so far as it receives a purely formal definition in terms of its relations of similarity and difference with respect to other qualities.

Now our main thesis is that it is only the term-character of the quality which "gets across" either in introspection or in any other form of behavior. The subject's quale as such is never known to the observer. The method of procedure and the final results which get into our science are in essentials no different whether we ask a fellow human to introspect and describe his sensation-qualities or whether we try out a lower animal or a "man from Mars" in a Yerkes Discrimination Box. In either case the only thing that we finally learn-that we finally get into our science-is what standardized stimulus situations the given organism will treat as similar and what ones he will treat as different, and to what degrees they will be similar and to what degrees they will be different. The organism's qualia corresponding to these standardized stimulus situations we will never be certain of knowing-of having any immediate acquaintance of. If what looks green to you looks the complementary red to me, I doubt if there would be any way of our ever discovering this fact.

In conclusion, then, I wish to urge the advantages of the behavioristic formulation. For whether our particular job concerns us primarily with introspection or primarily with the more gross forms of behavior, all that ever gets into our science is not the qualia as such, nor even ideas, awareness, and the knowing self, but merely the behavioristic implications of such entities.

The Functional Aspects of Structural Process. CHRISTIAN A. RUCKMICK, Wellesley College.

Almost a decade ago, when the author reviewed the variety of concepts that underlay the use of the term "function" in English textbooks of psychology, he promised later to treat the broader aspect of the problem of mental functions in systematic psychology. At the present time when fundamental concepts and methods are being discussed and delimited it seems not inopportune to offer a possible interpretation of the functional phases of mental processes. There are three ways in which a mental process may function: (1) It may be an essential constituent of another more complex process. When a perception is determined to consist of a group of sensations and images, the structural analysis ends with the enumeration of their qualitative and quantitative characterizations. The assignment of their degree of importance relative to each other is a functional aspect. Also in larger complexes we speak of the "rôle" of a process.

(2) It may serve as a substitute for another process. This refers to the phenomena of vicarious functioning as well as the replacement of processes in the abstract ideas of thought.

(3) It may imply a reference to the mental life in part or as a whole. In this group fall not only the processes that are regarded as serving a purpose for the mental life beyond their mere existence, such as attending, perceiving, remembering, imagining, etc., but also the reference that one group of processes bears to another as "context" or meaning.

It is important to review these concepts because, having forgotten or disregarded the earlier discussions raised by the structuralists, modern psychology has a tendency to discard that point of view entirely in favor of the supposedly newer psychology of dynamic force and biological purpose. In reality these issues are old. The literature shows, indeed, that the strongest supporters of structural psychology have accepted the doctrine of mental functions as plausible and profitable, but have chosen experimentally to follow their own path in order to keep their view unobscured. In a systematic treatise or textbook, however, the description of the mental life is incomplete without the adoption of the functional point of view.

A Physiological-Genetic Theory of Feeling and Emotion. FLOYD H. ALLPORT, Harvard University.

Although our knowledge of the physiology of emotional states has advanced considerably in recent years, there has been little progress in combining physiological and psychological viewpoints. This paper aims at a synthesis of the contributions of James, Lange, Cannon, Watson, and others, and a re-interpretation of physiological data at hand. Specifically, the writer hopes to supplement the James-Lange theory in respect to the physiological differentia of fear, anger, and other states, and also to throw light upon the nature of feeling and its relation to emotions. An emotion is analyzable into (1) an experience either of pleasantness or unpleasantness, and (2) some quality which serves to differentiate it from other emotions having the same feeling tone. Anger, for example, is introspectively different from fear or other unpleasant states. In the autonomic nervous system we have a mechanism logically suited to the antagonistic nature (pleasantness vs. unpleasantness) of the first-mentioned content. The cranio-sacral division, together with the cerebro-spinal, of which it is really a part, innervates movements whose return sensations constitute the feeling of pleasantness. The sympathetic division, in opposition to the craniosacral, functions in unpleasant experiences.

Evidence for this view is derived from a consideration of the rôle of the sympathetic division in unpleasant emotions (fear, anger, pain, etc.), and that of the other division in the pleasantly toned food and sex activities. Further evidence is at hand from the relative latent periods of these two classes of experience, and from careful introspective reports concerning their localization and temporal aspects.

The differentiation of emotions within one affective group is accounted for by the difference of facial, skeletal, and other postures characteristic of the different responses, and added by cerebrospinal innervation to the sympathetic core of feeling. Evidence for the elementary dichotomy on the basis of affection, with subsequent development of the differentiated emotions, is present in the infant. The earliest emotive state (protopathic) is probably pure "feeling." Indications of emotion first appear with the use of the prepotent somatic reflexes of defense and avoidance.

The conditions controlling the arousal of an unpleasant emotional experience are strength, suddenness, and insistence of the stimulus, the blocking of the somatic response, and the state of visceral preparation.

The theory proposed is supplementary rather than antagonistic to central theories. The facts of behavior however necessitate a peripheral differentiation as a necessary element in the total emotional event.

A Concept of Compensation. E. S. ROBINSON, University of Chicago.

Compensatory behavior can be defined in a very general way as a type of activity which grows out of conflicting and mutually modifying impulses. These impulses are tendencies toward action and, as such, vary from mere undetected dispositions toward action to dispositions which have got out into anything short of complete, unmodified expression. Such impulses enter into conflicts which result in the effectual inhibition of action in all but one direction, in emotion, in thinking, in some form of dissociation, and in compensation, or in some combination of two or more of these. Compensatory behavior is thus one form of conflict resolution.

In any particular compensatory act the nature of the conflict which that act resolves may or may not be evident on the face of the act itself. Since a multiplicity of receptors guarantees a constant conflict it might be claimed that all acts are to some degree compensatory. This is true just as it is true that all acts are to some degree rational, perceptual, et cetera. It will be useful to emphasize the compensatory aspect of only those acts which are most evidently the result of conflicting and mutually modifying tendencies.

Correlations of Four Intelligence Tests with the Grades of Students and with Each Other. A. M. JORDAN, University of Arkansas. Purpose of the Investigation:

(a) To find out the test or the elements of the test which furnishes the best prognosis of the standing of pupils of the high school in their subjects of instruction.

(b) To investigate the effect of time on the correlations.

(c) To discover by means of the regression equation the weights to be attached to the score in each test in order to obtain the optimum correlation.

Method:

A. Four intelligence tests-Army Alpha, Terman, Otis, Millerwere given to seventy high-school students.

B. Army Alpha was given to 315 university students.

C. Correlations with probable errors were computed between: (1) Each of the four tests and the subjects of instruction taken together. The value of each subject being determined by grade points. (2) Each of the tests and English, science, mathematics, and history individually. (3) Each of the elements of each test (31 elements in all) and (a) total subjects, (b) science, (c) English, (d) history, (e) mathematics, (b), (d), and (e) were not completed.

D. Correlations were made between Army Alpha and the subjects of instruction in the University for a period of two years (only one completed).

E. The computation of regression equations with at least four variables in the case of the highest correlations of the elements of each test and with the total tests (not completed).

Conclusions:

A. For general prognosis Terman has the highest correlation, .492; for history, Terman, .356; for science, Terman, .508; for English, Miller, .428; for mathematics, Army, .371.

B. Among the elements of each test there are considerable variations in the coefficients of correlations with the various subjects of instructions and with each one. Those elements having the highest correlations with the average of grades are: Terman-1, .555; Army-5, .514; Terman-7, .492; Otis-5, 479; Miller-1, .466.

C. With English the following elements correlate highest: Miller-1,.594; Terman-3, .572. Therefore elements that require less than ten minutes to give and a little while to score give as good a prognosis of the pupils' standing in English as the whole test.

D. Otis-5 gives a correlation of .67 with mathematics. E. There is a general reduction of the size of the coefficient of correlation during the three terms of the first year in the case of Army Alpha and the grades of the University students.