(This is the "text" page. If you already have the "guide" page open, please DO NOT click here.)

I. Confusion Due to Data from Two Different Frames of Reference

No visitor to a meeting of any psychological society in the last few years can fail to be impressed by the wide diversity of opinion there expressed. After fifty years of apparent progress American psychology has reached a point where two psychologists wishing to carry on a professional discussion must spend most of the time defining the terms they are using. As Hull has said, "One of the most striking things about the theory of learning and of psychological theory in general is the wide disagreement among individual psychologists" (8). The confusion is so great that already leading psychologists have predicted the impending dismemberment of psychology.

Controversy over clear-cut and well defined points of view is a necessary prelude to progress in the development of any science. It does not appear, however, that the 'wide disagreement' cited by Hull above holds the promise of an impending step forward in the history of psychology. It is more likely that the confusion is a symptom of defeat and of the present inability of the science to live up to its early promise, a failure that is implicit in Pratt's statement that " For a long time to come psychology should remain in the laboratory and the library" (13).

Some writers have sought to gloss over this chaotic situation by insisting that the disagreements concern theory only and that there is little difference over questions of fact. This evasion can be of little consolation to the vast number of parents, teachers, physicians, and other workers in the applied fields who must depend upon theory for guidance in their attempts to predict and control human behavior.

Prediction of new phenomena is necessarily based upon theory, not directly upon the results of previous observations. Knowledge of the results of a large number of discrete experiments, without theoretical interpretation, offers no way of predicting results when the experimental situation is different in the least degree. It may be anticipated that the differing conditions may lead to different behavior, but the kind and degree of difference are unpredictable without some theory of causation. Such circumstances are universal in any field of applied psychology where the teacher, the clinician, or the psychiatrist is always endeavoring to predict and control behavior in situations which are in some degree new and unique. Applied psychology is in vital need of principles of prediction that are simple, accurate, and unambiguous, that can function as principles for prediction rather than as myths for explanation, and that can be applied with precision in situations never before investigated.

It is precisely at these points that the current body of psychological theory is defective. It is complex, cumbersome and contradictory, and each of the conflicting systems is itself a welter of contradictory principles of causation and description. Catching phenomena coming and going, these systems serve nicely as explanatory concepts after the event but, because of their antagonistic character, are relatively useless for prediction. Ego-id, primacy-recency, similarity-contrast, differentiation-integration, facilitation-inhibition, learning-forgetting, self preservation-death wish are only a few examples. In the effort to explain all known facts every current system has been forced into the use of conflicting principles such as these. Each system has a large number of principles because it must account for a large number of facts; and in each system some of the principles are conflicting because the facts which they attempt to explain are conflicting. It is the thesis of this paper that this apparent conflict is an artifact. The facts are conflicting because they are derive from two mutually exclusive frames of reference.

It has become a common-place in physics that the obtained figures for the mass, velocity, and path of any object are

dependent upon the point from which they are observed. it is less widely recognized that the same principles apply to the collection of psychological data. For whatever purpose behavior is to be studied it must be observed from one of two distinct points of view. It may be studied objectively, as by an outside observer; or it may be studied phenomenologically, from the point of view of the behaving organism itself (17). The facts derived from these two points of view are nonidentical and are often completely contradictory. As seen by an outside observer, for instance, learning is a process of progressive change in the learner's response to a static situation. During this process the situation or task remains unchanged while the behavior becomes more relevant and efficient. Since the situation remains unchanged, improvement is ascribed to hypothetical changes within the learner. Thus the objective approach inevitably includes among its derived facts random molar behavior and improvement with repetition, usually by association or integration. Educational procedures based upon objective facts customarily stress frequency, drill, reward and punishment.

From the phenomenological point of view, that is to say, from the point of view of the learner, the facts are quite different. The learner remains unchanged. It is his experience of the situation or task which changes. From his point of view his behavior is always insightful, that is to say, it is always relevant to the situation as he interprets it at the moment. Improvement is concurrent with changes in the observed nature of the task, usually described as differentiation, individuation, or increase in detail. These data are in direct contradiction to the data derived from the objective approach; the observed facts of one frame of reference completely contradict the observed facts of the other. Educational procedures based upon phenomenological facts also differ from those based upon objective facts since they stress understanding of the individual child, pacing, and clear presentation of material.

Twenty years ago the point of view in American psychology was predominantly objective. An attempt was being made

by one school, the behaviorists, to construct a methodologically pure theory of learning from that point of view. Since then, largely due to the influence of the Gestalt movement, the trend has been toward an increasing acceptance of facts derived from the phenomenological point of view. Unfortunately the trend has, in most cases, manifested itself in the attempted assimilation of phenomenological facts into a basically objective theory.

An analogous situation might have arisen during the shift of astronomers from the Ptolemaic to the Copernican frame of reference. If we take our place of residence as the fixed unmoving point of observation, it follows as a necessary and undeniable fact that the sun revolves around the earth. It is a fact that can be verified by independent observation on any clear day. On the other hand, if the sun is taken as the fixed point of reference it is an equally observable fact that the earth revolves around the sun. Both facts are not only verifiable by observation but can be used in the prediction of phenomena, for example eclipses. They are not, however, co-existent facts since they are derived from two mutually exclusive points of view. It is not likely that during the long controversy over the 'truth' of the two systems any eclectic ever had the hardihood to predict an eclipse from a synthesis of Ptolemaic and Copernican facts, with the earth going around the sun while the sun went around the earth and both stood still.

Precisely such a synthesis, however, is continually attempted in psychological discussions of learning. A glance at any current text will show how error, reflexes, stimuli, frequency, and other data derived from the objective point of view are complemented with such phenomenological concepts as trial, insight, and belonging. The statement found in a widely used outline for students, that we learn usually by trial and error but sometimes by insight, is only an extreme example of this generous and uncritical point of view. Nor should it be supposed that the indiscriminant amalgamation of objective and phenomenological data is confined to the works of avowed eclectics. 'Effect' and 'consequences' have long done duty in both camps, referring on the same page to changes in the

structure of the organism and to changes in the way the learner experiences the talc. Even Gestalt psychologists have failed to recognize the non-simultaneity of the derived data. Koffka's invocation of the 'geographic' environment (9), necessitated by his desire to assimilate data derived from an objective approach, is a case in point. By its introduction into an essentially phenomenological system it is possible for him to take cognizance of items which exist in the phenomenal field of the experimenter but not in that of the behaving organism.

Given a theory of learning based upon one of the two points of view, an assimilation of facts derived from the other can be achieved only by introducing additional and necessarily conflicting laws and concepts. This was done by Thorndike (effect-belonging, 21) and, more recently, by Maier (learning-reasoning, 11) in adding phenomenological data to objective systems. Koffka and Freud, in adding objective data to phenomenological systems, attempted to avoid the uncertainty involved in a multiplicity of independent processes but found it necessary to postulate new loci of action (geographic environment, unconscious) not open to direct observation by either the observer or the participant. Uncertainty in prediction arises, on the one hand, from the multiplicity of independent laws, and on the other, from the circumstance that the postulated law is held to be operating in a field inaccessible to observation. In either case the increased possibility of a plausible explanation for any specific act is more than counterbalanced by the decrease of certainty in prediction. Mixed systems are therefore better suited for explanation in retrospect than as principles for prediction in advance. Data derived from the two frames of reference are not co-existent and a genuinely predictive psychology will be impossible of attainment until we stop treating them as if they were. Such a psychology must be either completely objective or completely phenomenological. It is the purpose of this paper to examine the value of data derived from each type of approach for their comparative value in prediction.

II. Inadequacy of the Objective Approach for Prediction of Human Behavior

Of the two possible points of view it is probable that most psychologists would prefer to use the objective approach because of its record of success in the physical sciences. It does not appear, however, that any observational approach is adequate at the present time to furnish the required principles for prediction. Whatever their possibilities for the future, attempted objective systems have up to now shared the defects of mixed systems by leaning heavily upon hypothetical loci of action or by requiring an unwieldy number of independent causal principles.

As viewed by outside observers, the behavior of living organisms varies even when the environment remains unchanged. This variability has been explained in the past by postulating mind as an unseen determiner of behavior. It is more frequently explained at present by assigning the same determining function to hypothetical changes in the organism, usually in the nervous system. If instruments can be invented which will make organic states more accessible to observation, the latter concept may prove to be a very fruitful one; but up to the present time both concepts have functioned almost exclusively as explanatory concepts, being relatively useless for prediction. Accurate prediction is possible only when the causal entities are open to inspection.

A tempting alternative to the use of unexplored causal fields is that adopted by the early behaviorists who attempted to refrain completely from causal inferences and to restrict themselves to data which could be objectively observed. Because of the notorious variability of animal behavior under objectively identical conditions such attempts to discover purely descriptive laws have not, however, been very successful. A common and necessary assumption of the objective way of search has been that the apparent irresponsibility of living organisms to physical causation is due to the gross character of the units studied. Further analysis, it is hoped, will show the parts of the organism functioning in ways

predictable by an adequate physics. By withdrawing from the study of organismic behavior into the study of part behavior, of reflexes, or of S-R bonds, it is possible to maintain the concept of lawful causation of events and at the same time maintain the objective approach. This procedure, however, involves an indefinite multiplication of causal processes, with attendant confusion in prediction. Since it is possible to investigate the relation between an animal's behavior and any feature of a situation which can be experienced by the experimenter, present day objectivists are embarrassed by a plethora of causal factors. Buel (4) has reviewed eighty-three factors which have been found to affect the pathway chosen by a white rat approaching a point of bifurcation in a maze. He points out that the list is not exhaustive and the eighty-third factor is 'chance.' The hopelessness of using such a large number of independent principles as bases for accurate prediction is obvious.

The situation, then, is this: From the objective point of view, behavior which is not pertinent to the situation as viewed by the experimenter is random, indeterminate, fortuitous. To accept this indeterminism, however, as final would involve the abandonment of all hope for accurate prediction. Any science which hopes to predict must postulate lawfulness. Lawfulness in an objective system, however, can be maintained only by postulating additional causal agents unseen by the experimenter, as mind, past experience, instincts, or organic change. As long as these entities remain inaccessible to the experimenter they can be endowed with any necessary characteristics and are ideally suited to function as explanatory concepts. But, conversely, as long as these agents remain inaccessible the systems of which they are parts will have vital gaps in their causal fields with consequent inaccuracy in prediction. Since the unseen agents are usually invoked to explain individual, variations in behavior, objective systems are apt to restrict themselves in practice to the prediction of normative behavior, concerning themselves chiefly with the establishment of norms and coefficients of correlation. This knowledge of what 'most people,' the average individual,' or 'the typical three-year

old' is most likely to do in a given situation 'other things being equal' is of little value, however, to the applied worker, the clinical psychologist, or the classroom teacher who must predict and control the specific behavior of particular individuals. If the analysis of this paper is correct, the accurate prediction of such specific individual behavior, from an objective point of view, will have to wait until one of the explanatory agents, most probably the physical organism, is laid open to observation by methods and instruments not yet devised.

Pending the perfection of these devices, it appears desirable that an attempt be made to explore the possibilities of the alternative point of view, that of the behaving organism. The remainder of this paper is devoted to a discussion of a phenomenological system that has been used with some success in predicting previously unobserved behavior (18, 19, 20). The discussion is restricted roughly to the field of learning, which is the most crucial to the problem of prediction.

The reader will bear in mind that the ' facts ' of such a system will necessarily conflict with those derived from the objective point of view and that the validity of any frame of reference must be judged, not by the degree to which its facts correspond to the facts derived from other approaches, but by its usefulness in prediction.

III. The Characteristics of a Phenomenological System

1. The basic postulates.-Assuming that the task of psychology is the prediction and control of behavior, a phenomenological system must rest upon three basic assumptions ¹ and three principles.

A. All behavior is lawful. This is a necessary assumption of any system, since chance behavior would be unpredictable.

B. Behavior is completely determined by and pertinent to the phenomenological field of the behaving organism. By phenomenological field, hereafter abbreviated to p.f., is meant the universe, including himself, as experienced by the behaver at the moment.

C. There is some relationship between the phenomenological fields of different individuals. This is a necessary assumption, since control is impossible if one individual is unable to affect another's field. The locus of the relationship, usually presumed to be an underlying reality, is not open to observation.

D. Greater precision of behavior (learning) is concomitant with greater differentiation of the phenomenological field. Another characteristic of p.fs. is that they are fluid and shifting; their phenomena are continually reshaped and given new meanings by the character of the total configuration. Memories, for example, are strongly affected in this way (1). Maier (11) found that the crucial act of solutions was forgotten as soon as the solution was made; and Wees and Line (23) found that school children, in the act of reading a story, distorted its details in ways that made it more meaningful and pertinent to their own experiences. Since behavior is part of the field, taking part in the field's interaction, principle E is in some ways a restatement of the sec and postulate B.

E. The characteristics of the parts of the phenomenological field are determined by the character of the field itself. More specifically, the direction and degree of differentiation are determined by the phenomenological needs of the behaver.² The reader may find, for example, that in reading this paper he has been particularly aware of the points which substantiate his own views. The fundamental need in a phenomenological system appears to be the preservation of the organization and integrity of the p.f. and especially of that part of the field which is the phenomenal self, whence our tendency to remain unaware of, or to reject with emotion, data inconsistent with our own beliefs.³

F. Differentiation takes time. It follows from this principle that the way to accelerate learning is to arrange the situation so that the required differentiations are either more obvious or are unnecessary. For instance, in a black Warden multiple-U maze of the LRRRLLRLLR pattern which had been learned by a group of white rats in a median of 29 trials, the application of white paint to the critical 2, 5, 7, 8, and 10 sections, where changes in procedure were necessary, enabled an experimental group to learn the maze in a median of 12 trials (19). When the differentiation of individual sections from one another was made completely unnecessary by painting the blind alleys white and the correct pathway black, or vice-versa, the median number of trials required for learning was lowered to 7 (18).

2. The problem of prediction.-By postulate B the determining locus of action is the behaver's p.f. This is not open to direct observation by any outside observer. The process of prediction therefore involves two steps : (1) the securing of an understanding of the subject's field by inference or reconstruction, (2) the projection of the future field.

The first operation is of the common "Now why did he do that?" or "Under what circumstances would I have done that?" character. Much of the topological work of Lewin is of this type and essentially the same procedure was used by Shepard (15) when from the behavior of his rats he inferred the existence of floor cues which he himself was unable to experience. The teacher who hears his pupil report that 3 X 0 is 3 and infers that his reasoning is "Zero is nothing so it does nothing to the three" has taken this step. The operation acquires its validity in this system by the postulate (B that behavior is completely determined by the p.f., whence it follows that variations in behavior are always indicative of concurrent variations in the field. The complete operation of prediction imposes two important conditions. To reconstruct an individual's field from his behavior it is necessary to have some idea what fields are like, and to

project the future field it is necessary to understand how fields change.

3. The nature of the field.-The p.f. is simply the world of naive, immediate experience in which each individual lives, the everyday situation of self and surroundings which the unsophisticated person takes to be real. Studies on the nature of this field indicate that all parts of the field are not equally distinct. The field consists of figure and ground, or focus and margin; there are not two definite, static levels, but one level may shade into the other so that the figure may be large and relatively indistinct or small and highly differentiated. Experience in any sense field can be figure. Pain, fatigue, or the disturbed organic states involved in emotion may emerge so sharply as the focus of the field that all the rest of the field fading into the homogeneity of ground, that the individual will lose touch with his surroundings and become unconscious. Since by postulate B behavior is completely determined by the p.f., a highly detailed and differentiated field will include definite and precise behavior, while, as anyone who has tried to find a snap switch in a strange room in the dark will agree, behavior in a vague and undifferentiated field is vague and confused. This leads to principle D.

4. How fields change.-Principle D of our system identified differentiation with learning; principle E made the determinants of differentiation somewhat explicit. Differentiation may be defined as knowing a difference, the basic act of knowledge. It is the manifestation of the continuous process by which the integrity and organization of the field are maintained. "When an individual, rat or human, is confronted with a task . . . the general procedure is determined by his initial perception of the nature of the problem; it is a gross response to a relatively undifferentiated situation. Should the first procedure, the response to the gross situation, prove inadequate the task is differentiated perceptually into segments each of which is solved by simple procedures" (19).

Although he is aware that his own field may be affected (E by his desire to maintain the predictive advantages of

having only one process in the system, the writer feels that differentiation may be safely assumed to be the only process of change in the p.f. The emergence of a new entity or character into figure implies the lapse of other characters into ground. Both are necessary for the existence of a difference and are not two independent processes but complementary aspects of the same process, which might be called 'change.' Since, however, it is the newly emerged figure, the focus of the behaver's field, which is the most directly potent in determining behavior, it seems more practical to emphasize the more effective aspect of the process and call it 'differentiation' or 'individuation' rather than the nonvaluative 'change.'

The basic assumptions and principles are now complete. There remains to be discussed principle F, which is subsidiary and based on experimental evidence (3, 16).

5. Use of the system in prediction.-Several examples of the use of this system for the prediction of previously unreported animal behavior have been published (2, 18, 19, 20), the last of which may serve as an example. The purpose of the study was to test the comparative usefulness of phenomenological and objective principles in the prediction of maze behavior. Since DeCamp's experiment (5), reported in 1920, it had been almost axiomatic that rats tend to adopt the shorter of two alternative paths to food. Now suppose that the structure of a rat's field is such that a path which is objectively the longer of two alternative paths to food appears to the rat to be the shorter. If, as we have postulated (B, the animal's behavior is completely determined by his p.f. he will choose this phenomenologically shorter path in spite of the fact that it appears to the experimenter to be the longer. According to objective principles, on the other hand, the effective determinant would be the physical lengths of the two paths 'other things being equal,' irrespective of how the situation was experienced by the rat.

Such a situation was arranged in a ten section Warden U LLLLLLLLLL maze, in which an alternative food-box was attached to the right hand alley of the third section. Thus

it was possible for the rats to take either a short three section LLR path or a long ten section LLLLLLLLLL path to food. If the behavior were governed by the objective situation it would be predicted that they would choose the shorter LLR path. However, previous experiments from a phenomenological point of view had led to the inference that "maze learning is a process of increasing differentiation in the total situation, during which the maze is first perceived as a general path to the food-box and is then differentiated into sub-mazes. . . ." (19). It was therefore predicted that in this maze the animals would at the end of the first trial experience the path to either food-box as a general leftward path. As a consequence they would take the longer LLLLLLLLLL path on later trials. The experiment was made by running a group of rats in this maze once a day for 100 days. During the first 75 days the animals verified the prediction by taking the longer path on 74 per cent of the runs. During the last 10 days the longer path was taken on 89 per cent of the runs. This behavior conformed to the inferred p.fs. of the rats, thus demonstrating the usefulness of the postulates used in its prediction.

In the same way, it was predicted that animals in a similar maze where the alternative food-box was on the right hand alley of the second section, which is directly in front of the maze entrance, would experience this LR path as something like 'ahead' or 'around the corner.' This would not lead them to take the long leftward path. This prediction, also, was verified. The rats in this maze took the shorter path on 64 per cent of the runs during the first 75 days, and on 66 per cent of the runs during the last ten days.

6. The conditioned response.-The only real and valid test of any system in science is its effectiveness in predicting previously undiscovered phenomena, as in the case just cited. Ability to explain phenomena already known is, of course, no criterion of usefulness in prediction. It is true, nevertheless, that inability to explain known phenomena is presumptive evidence of inadequacy in prediction as well. For this reason it is interesting to test the adequacy of the proposed

system to explain some of the phenomena of the conditioned response, upon which objective systems commonly base their description of learning. A phenomenological explanation would be something like the following, the capital letters referring to the principles involved:

(1) Conditioning.

a. Objective description: Given an animal with a need (e.g., for food), and a means of affecting that need (e.g., food). Present a signal (e.g., a tone) of a kind that in sufficient volume and under favorable circumstances is capable of eliciting a response from the animal, then, in fairly close temporal and spatial contiguity, present the means of affecting the need (the 'unconditioned stimulus'). After one or more presentations the signal will elicit the same response as the unconditioned stimulus or a response similar to it.

b. Explanation: Given time (F and a need to be satisfied (E, the signal and the unconditioned stimulus will be differentiated as a unit from a relatively homogeneous field. Since the degree of differentiation required is determined by the precision of response required (D signals for diffuse bodily and postural activity may be effective at a very low level of awareness. The closer the signal and unconditioned stimulus in the experimenter's time and space, the more apt they are to be differentiated as a unit from the rest of the learner's field (C. It would not be expected (B that the response to the signal-stimulus unit would be exactly the same as to the stimulus alone. In a class demonstration with a human subject, using a strong buzzer tone one second before a strong shock to the right fore-finger from a curved finger rest, the unconditioned response was an extension of the finger. The conditioned response, however, was flexion together with a lifting of the finger, the whole action taking it off the grill. Phenomenologically the subject was bracing himself for the expected shock.

(2) Irradiation.

a. Objective description: If the signal is altered within a varying range the established response will still be made.

b. Explanation: If the signal is incompletely differentiated it may be confused with other signals. If it is experienced as a vague feeling of discomfort or expectancy a large number of signals in different sense fields might elicit the response. If it is more clearly individuated there will be less opportunity for confusion with other signals.

On the basis of this inference that 'irradiation' and 'generalization' are incomplete differentiation, it would be predicted that continued presentations of the signal-stimulus unit will, by giving more opportunity for precise differentiation (F of the signal, lessen the number of signals with which it might be confused and diminish the probability of 'irradiation' responses. This prediction corresponds to the results reported by Razran (14, see pages 7 and 8), although it does not agree with his interpretation.

(3) Differentiation.

a. Objective description: Both signal A and signal A' elicit the response. If signal A is given with the unconditioned stimulus and signal A' without it, the latter signal will cease to elicit the response.

b. Explanation: Presentation of signal A' without food makes it necessary (E and possible (F for the subject to differentiate it from signal A.

(4) Unconditioning.

a. Objective description: If the signal is given without the unconditioned stimulus or some time after it, it will, after a varying number of presentations, cease to elicit the unit response.

b. Explanation: Separate presentation provides the subject with an opportunity for differentiating them from one another (C (F. The signal from a signal-food unit should have no effect even the first time it is given after food if the need for food has been completely satisfied (E. The signal from signal-punishment units, however, would be effective for several presentations when given after the punishment,

since the need to escape punishment is not satiable. Repeated presentations, however, will give an opportunity for differ

entiation and the signal will eventually emerge as a cue that punishment has ceased and will then evoke a different response.

(5) Pseudo-conditioning.

a. Objective description: "In some cases a response is elicited by a formerly inadequate stimulus (signal) which has been preceded by an unconditioned stimulus" (7).

b. Explanation: In cases where the unconditioned stimulus is one, like punishment, that leads to a continuing state of need and tension, the animal will differentiate out and respond to features of the field previously ignored. The signal is effective because it is unfamiliar, that is to say, it is incompletely differentiated from the shock situation. "The animals gave the impression of responding as if a shock were expected and as if they knew no appropriate response to make" (7, p. 372). In cases where the unconditioned stimulus satisfies the subject's need pseudo-conditioning will not occur.

7. Other concepts of learning.-One of the greatest advantages of the phenomenological method is that it is able to bring the experimenter within the scope of its system and to apply its laws to the observer himself. When examined in this manner, most of the conventional concepts of learning prove to be products of the relationship between the p.fs. of the observer and the learner. For example: If the observer's field is more highly differentiated than the learner's, the latter's less precise behavior (D, since it does not conform to the situation as experienced by the observer, is said to be error. Phenomenologically errors are recognized only in retrospect, that is to say, when an individual compares his past behavior with his present more highly differentiated field.

If the learner's field closely approximates the field of the observer, so that the learner does what the observer would do, the behavior is said to be correct or insightful.

If the observer's field is less differentiated than the learner's there are two possibilities: (1) The learner's more precise and efficient behavior may lead the observer to discover features of the situation of which he had previously been unaware; in which case a third party might infer that

the observer had learned by imitation. (2) The learner's mysteriously precise behavior in what to the observer is a relatively undifferentiated field may lead to the assumption that the behavior is determined by instinct. The farther removed an animal is from the human in sensory and behavioral possibilities, and the more difficult the reconstruction of its field, the greater the chances that instinct will be invoked as an explanation for its behavior.

8. Association or differentiation?-If an observer in a highly differentiated field, which he naively takes to be real, watches the behavior of a learner he is certain to interpret the process as one in which the highly individuated items in his own field are combined, organized, associated, or integrated by the learner. Because he believes his own field to be real, he postulates that the learner's field is made up of the same items, but in an inferior state of organization. Objective theories of learning, therefore, begin with an unorganized field and attempt to show how it becomes organized.

Cross section studies of the learner's p.f. at different stages of learning, however, show that the field is always organized. The change lies in increasing detail rather than organization. Entities are thus 'associated' when they are incompletely differentiated, having emerged together out of a common ground to satisfy a need. They are experienced as contiguous when they are incompletely differentiated in space or time and as like or opposite when they emerge in some mutual relationship. The so-called laws of association: contiguity, similarity and contrast thus become descriptive of. incomplete differentiation in the phenomenological field.

9. Frequency.-Frequency is the basic but unreliable causal factor of most objective systems. From the phenomenological point of view it may afford an opportunity for learning (F but cannot cause or guarantee it. What is learned depends upon the phenomenological needs of the learner (E. This accounts for the success of Dunlap's method of breaking habits by practicing them (6, 22).

10. Relationship to the physical sciences.-One of the most interesting implications of the phenomenological approach is

the status it assigns to the physical sciences. Physicists secure regularity and lawfulness by restricting themselves to a common and rigorously limited phenomenological field. They share this common p.f. (sometimes called the physical universe) by the process of taking all measurements from a standard position in front of their instruments and confining their observations to the pointer-readings thus derived. This process gives to physical scientists the unique advantage of a common field; but it is a field where all the characteristics of phenomena are eliminated except those reducible to length; where light, color, temperature, taste, and odor exist, if at all, only as readings on a scale; and where song, oratory, poetry, and propaganda are alternating areas of condensation and rarefaction. As long as the physicists can remain in this common field their behavior is identical. They make the same calculations and predictions because they are living in identical fields.

This use of mediate observation in the physical sciences has led to spectacular agreement in some fields. It is, however, essentially a search for agreement by elimination of phenomena and is therefore limited in its possibilities. Since the phenomena that must be excluded are essentially those necessary for the prediction and control of human behavior, it is becoming apparent that the methods of the physical sciences cannot be taken over bodily by the psychologists.

11. Advantages of the phenomenological approach.-Philosophically a phenomenological system has a number of advantages, several of which have been enumerated or implied. From the point of view of practical use for the prediction and control of behavior, however, four are outstanding:

(1) A phenomenological system is anthropomorphic. Its data are stated in terms of immediate experience and require no translation to make them meaningful.

(2) It is concerned with the prediction and control of individual behavior, a field closed to objective systems because of their necessary assumption of variability in individual behavior. For this reason psychiatrists, applied psycholo

gists, and teachers when dealing with individuals commonly adopt a phenomenological view. This accounts for the great use by these groups of psychoanalysis, Gestalt, and private non-academic systems which have large phenomenological components. The use of a general field can result in the prediction of general, normative behavior only.

(3) As compared with the objective approach, the phenomenological approach is more inclusive. Individual behavior cannot be predicted from normative behavior. On the other hand, accuracy in predicting individual behavior makes possible the prediction of normative behavior as well.

(4) The particular system outlined in this paper has the predictive advantage of postulating only one process, which is descriptive, rather than causal or explanatory.

12. Relation to other systems.-It is impossible to say whether this system is the only possible phenomenological system. The first three principles are certainly basic and the next three appear to follow. Gestalt psychologists and their pupils have made the major recent contributions in this field and have failed to achieve a purely phenomenological system only because of their failure to appreciate its necessity. As soon as non-phenomenological data are eliminated from Gestalt psychology, it becomes the system that has been described.

Psychoanalysis, the other semi-phenomenological system, differs basically in its interpretation of the nature of the field. Freud recognized the figure-ground character of the field and the importance of the ground in behavior but, lacking experimental evidence, distorted the figure-ground relations into an antagonistic dichotomy in which the ground (unconscious) had all the characteristics of figure (conscious) including a self (id). Since these characteristics are not experienced by the individual the system ceased to be purely phenomenological. The methods of analysis are directed toward securing an understanding of the patient's field and have resulted in the accumulation of important data, especially in the area of motivation. Since the conscious-unconscious dichotomy is the cornerstone of the system it is impossible to see, however,

how psychoanalysis can evolve into the purely phenomenological system that is needed.

IV. Summary

The current confusion in psychology is largely due to the uncritical combination of data from two different frames of reference. Facts derived from a phenomenological point of view are non-identical with and often completely contradictory to facts derived from an objective point of view. Systems attempting to combine facts derived from both frames of reference are forced into a multiplicity of conflicting laws and concepts or into the postulation of loci of action inaccessible to observation. In either case accurate prediction of behavior is impossible. Objective systems, although philosophically more satisfactory than mixed systems, have been forced in practice to confine themselves to the prediction of normative behavior. For this reason psychiatrists, applied psychologists; and teachers, when dealing with individuals commonly adopt a phenomenological view. This accounts for the great use by these groups of psychoanalysis, Gestalt, and private non-academic systems having large phenomenological components. The best immediate prospects for a psychology able accurately to predict individual behavior lie in the development of Gestalt psychology along purely phenomenological lines.

[MS. received December 2, 1940]