No matter what our philosophy of behavior may be, we are not likely to deny that the world about us is important. We may disagree as to the nature or extent of the control which it holds over us, but some control is obvious. Behavior must be appropriate to the occasion. Failure to keep in touch with reality leads to the kinds of difficulties often observed in psychotic behavior. Even when a man is engaged in rejecting the world, in systematically reducing certain forms of its control over him, he is physically interacting with it.
Many theories of human behavior, nevertheless, neglect or ignore the action of the environment. The contact between the organism and the surrounding world is wholly disregarded or at best casually described. This is almost always true in clinical psychology, for example. The clinician often speaks of people, places, and things as “facts” entering into the interpretation of his patient’s behavior, without further specifying their action. This practice may be adequate for certain purposes of communication, but it must be expected to fail at some point. Some of the problems of clinical psychology show that that point is often reached. A case history may inform us, for example, that on a given day the patient saw an acquaintance who was approaching him cross the street, and this event may be regarded as significant in interpreting the patient’s behavior. But the report “X saw Y cross the street” does not prepare us for many possibly relevant questions. For example, what are the important properties of visual patterns which lead X to say, “That is Y”? Was X’s report of this event determined by a clear visual stimulus possessing these properties—in other words, was it really Y or did X merely “think it was Y”? In the latter case how plausible was the mistake? How much of the effect upon X was due to the appearance of Y as a person and how much to Y’s behavior in crossing the street? Upon what past occasions had similar stimuli affected X, and what conditioning had taken place with respect to people who cross streets, whether or not they were Y? To what extent was X’s reaction due to a condition which we may describe by saying that X was “afraid that Y was avoiding him”? Did earlier conditioning with respect to people who cross streets really involve Z, who resembles Y, and if so, may we say that Y was serving as a “symbol” for Z?
Questions of this sort are frequently treated in the later discussion of a case history, but often they would not arise if the earlier analysis of the contact between organism and environment had been adequate. An improved analysis would mean, not necessarily more information in any particular instance, but rather an understanding of the ways in which stimuli generally work. The casual account ignores many important points.
In studying the extremely important independent variables which lie in the immediate environment, we may begin with a physical description. What is the structure of the world which we see, hear, touch, smell, and taste? We should not prejudge these events from their effects upon the organism. They are to be described in the usual terms of the physics of light and sound, the chemistry of odorous or tasteful substances, and so on. We are interested, of course, only in conditions or events which have an effect upon behavior. The electromagnetic radiation of radio and television has no effect upon the unequipped organism, except perhaps at very high energy levels. We do not say that the radiation is “not a stimulus because it does not stimulate.” We simply ignore it just as we ignore the color of the apparatus we use in the study of mechanics as soon as we find it to be irrelevant.
The kinds of events which stimulate the organism are effective only within certain limits. We hear sound, but only of certain pitches and intensities. We see light, but only of certain intensities and wave lengths. The limits of stimulation, and also the smallest differences in stimuli which make detectable differences in behavior, have been extensively investigated. The normal individual differs from the blind or color-blind in his reaction to visible radiation, from the deaf or partially deaf in his reaction to tones, from the anosmic in his reaction to odors, and so on. Smaller differences between normal individuals may be equally important. Research of this sort often emphasizes the action of the organ where the interchange with the environment takes place—the eye, the ear, the taste buds in the tongue, and so on—but the whole organism may be involved. What appear to be simple sensory reactions often depend upon variables in the fields of conditioning, motivation, and emotion.
Several important problems concerning stimulation are relatively independent of the particular physical properties of stimuli and of their range of effectiveness. In attacking these problems it does not matter whether the receiving organ is the eye or ear, for example, and we may work with values of stimuli which do not raise the problem of limits. In discussing the stimulus functions of elicitation, discrimination, and reinforcement, it was not always necessary to specify the nature of the stimulation, and we shall see in Chapter IX that this is also true of another function of stimuli in the field of emotion. There are even more general processes which may be studied not only without considering the particular form of energy exchange at the periphery of the organism, but also without specifying whether the stimuli are eliciting, discriminative, reinforcing, or emotional. In the following discussion the discriminative stimulus will be emphasized, but each process could presumably be demonstrated in the other functions as well.
When we have once brought behavior under the control of a given stimulus, we frequently find that certain other stimuli are also effective. If a pigeon has been conditioned to peck a red spot on the wall of the experimental chamber, the response will also be evoked, though not with the same frequency, by an orange or even a yellow spot. The property of redness is important, but not exclusively so. Spots of different shapes or sizes, or spots against different colors of background, also may be effective. To evaluate the full extent of the change brought about through reinforcement, we need to survey the effects of a large number of stimuli. The spread of the effect to other stimuli is called generalization or induction. The process suggests that a discrete stimulus is as arbitrary a notion as a discrete operant. The “identical elements” of a response have their parallels in the values or properties of a stimulus which are separately effective. If we reinforce a response to a round, red spot one square inch in area, a yellow spot of the same size and shape will be effective because of the common properties of size and shape; a square, red spot of the same area will be effective because of its color and size; and a round, red spot half a square inch in area will be effective because of the common properties of color and shape.
The effectiveness of a single property of a stimulus when combined with novel properties is shown when we are uneasy in the presence of a new acquaintance because he resembles someone whom we dislike. The very subtle property responsible for the resemblance is sufficient to arouse an emotional reaction. The Freudian argument that early emotional conditioning affects later personal adjustment presupposes such a process, in which the subtle property by virtue of which an acquaintance resembles one’s father or mother, for example, is said to be independently effective. The Freudian “symbol” presupposes the same process: a piece of abstract sculpture which generates an emotional response because it resembles the human body demonstrates the effectiveness of the property responsible for the resemblance. As Freud pointed out, the resemblance may be effective whether or not it is recognized by the individual.
In literature the same process is exemplified by the device of metaphor. The emphasis in the usual rhetorical analysis is inverted, for the active control is assigned to the organism rather than to the stimulus. The speaker is said to transfer a description from one state of affairs to another which resembles it. We should say here that the metaphorical response is evoked by a stimulus which shares some of the properties of the stimulus to which the response is normally appropriate. Thus when Romeo compares Juliet to the sun, we need not suppose that he is engaging in an act of creative imagination; we need only suppose that Juliet’s effect upon him shares some of the properties of the effect of the sun and that the verbal response “sun” is therefore strengthened. (The elaboration of the metaphor must be distinguished from an explanation of its ingredients. The first step is to account for the appearance of the metaphorical term. This can usually be done by pointing to a property of a current stimulus which is possessed also by the customary stimulus for the verbal response.)
We check the importance of any dimension of a stimulus by examining the effect of different values. After building up a strong tendency to respond to a red spot, we may examine the rate of response during extinction to orange-red, orange, yellow-orange, and yellow. An experiment of this sort yields a generalization or induction gradient. The responding during extinction is most rapid whenever the spot is red. It is slightly slower to orange-red and much slower to yellow. An experimental animal such as the pigeon may not respond at all if the color is as different as, say, green, even though the two spots have common properties of shape, position, illumination, and being visual rather than, say, auditory stimuli. For the pigeon, therefore, color is obviously an important property. A color-blind organism, on the other hand, would not show this gradient; the rate would not change with color, if differences in brightness, texture, and so on were eliminated. Other properties of stimuli yield similar gradients when systematically explored. This procedure enables us to answer such a question as whether a given change in color is as important for the organism as a given change in size, or even whether color is as important a property of visual stimuli as pitch is of auditory stimuli. Not all the dimensions of stimuli, however, are continuous in this way.
Induction (or generalization) is not an activity of the organism; it is simply a term which describes the fact that the control acquired by a stimulus is shared by other stimuli with common properties or, to put it another way, that control is shared by all the properties of the stimulus taken separately. A particular combination of properties comprises what we speak of as a stimulus, but the expression does not represent the control exercised by the environment very accurately.
The discrimination described in Chapter VII is also not a form of action on the part of the organism. When we establish a discrimination between red and orange spots of light, we simply sharpen a natural gradient. By continuing to reinforce red spots while extinguishing orange spots, the control of the property of redness is consistently strengthened, while that of the property of orange is consistently weakened. In such an experiment, other properties of the stimuli—for example, size, shape, and location—are both reinforced and extinguished. Those who work with pigments, dyes, or other colored materials are affected by contingencies in which slight differences in color make a great deal of difference in the consequences of behavior. We say that they become “highly discriminating” with respect to color. But their behavior shows only processes of conditioning and extinction.
Behavior may be brought under the control of a single property or a special combination of properties of a stimulus while being freed from the control of all other properties. The characteristic result is known as abstraction. The relation to discrimination may be shown by an example. By reinforcing responses to a circular red spot while extinguishing responses to circular spots of all other colors, we may give the red spot exclusive control over the behavior. This is discrimination. Since spots of other colors apparently have no effect, it would appear that the other dimensions which they possess—for example, size, shape, and location—are unimportant. But this is not quite true, since it is less likely that the response will be made to a red object of another size and shape. We have, in other words, brought the response under the control of circular red spots but not of the “property of redness” alone. To achieve the latter, we must reinforce responses to many objects, all of which are red, but which differ widely in their other properties. Eventually, the organism responds only to the property of redness. The case is exemplified by the verbal response “red.” It should be remembered, however, that a perfectly abstract response is probably never achieved. Stimuli which possess the required property but which are quite extraordinary in other respects may not evoke the response. Stimuli without the required property which resemble especially common instances which possess it may also exert some control.
Abstraction, too, is not a form of action on the part of the organism. It is simply a narrowing of the control exercised by the properties of stimuli. The controlling property cannot be demonstrated upon a single occasion. In other words, a single instance of an abstract response will not tell us very much about its “referent.” The controlling relation can be discovered only through a survey of a large number of instances.
We are likely to overlook the history required for an abstract response, and we make many mistakes in interpreting behavior when we do so. When a child is taught to call a red ball red, we are surprised to find him calling a green ball red. In our own behavior, the response has long since come under the control of a particular color, but in the behavior of the child the properties of size, shape, and manipulability remain important until a program of differential reinforcement rules them out.
An organism will not acquire an abstract response until a reinforcing agency sets up the required contingency. There are no “natural” contingencies which reinforce a response in the presence of a single property without respect to other properties. The necessary contingency apparently requires the mediation of other organisms. Abstraction, therefore, appears to have become possible only with the development of verbal behavior. It does not follow that, if this was the case, abstract responses could never have arisen; for it is not impossible to conceive of events in a group of individuals which could have given rise to the rudiments of a verbal environment from which abstract verbal behavior could then have sprung. The matter, however, is highly speculative.
We are in a better position to see how abstractions grow and change. Verbal behavior, perpetuated by the verbal community, has succeeded in isolating more and more subtle properties of nature. Sometimes we can watch this happen. Sometimes we can make plausible speculations as to how it might have happened. Etymology often supplies valuable clues. The word “chance,” for example, comes from a word which referred to the fall of a die or coin. A conspicuous feature of such an event is the indeterminacy of the result, which is similar to the indeterminacy of other events in which nothing falls—for example, of the suit of a card drawn from a deck. The metaphorical transfer of the term for falling, on the basis of indeterminacy, is the first step in isolating this important property. The referent of the term is further refined—perhaps through centuries of changing practices in a verbal community—until in the hands of the modern mathematician the term comes under the control of a very special property of nature, the modern referent of the word “chance.”
Cross-modal induction. We sometimes find that a response is under the control of two stimuli which have no physical properties in common. If it has been conditioned to each of the stimuli separately, no explanation is required; but apparently this is not always the case. “Induction” appears to occur although common properties are lacking. Sometimes an intermediate connection can be discovered. Pins and pains are both called “sharp.” That sharp pins cause sharp pains may be relevant. It is only a short step from “the pin is sharp” to “the pain caused by the pin is sharp.” Once this verbal practice has been established in a community, the response is usually learned separately in the two cases and is then no longer an issue.
Common mediating behavior supplies another possible explanation. When Samuel Butler once saw the Wetterhorn, he caught himself humming an aria from Handel. “The big shoulder of the Wetterhorn seemed to fall just like the run on [the word] ‘shoulder.’” Here an auditory response appears to have been made to a visual stimulus which in some way resembled it. Presumably Butler had not heard the one while looking at the other; and we may suppose, for the sake of the example, that he had also not seen the musical phrase in visual form. We may account for the result if we assume that the two stimuli were capable of generating similar behavior. If Butler had learned to execute certain spatial responses to the “ups” and “downs” of pitch—say, in playing an instrument—and if, as the amateur artist he was, he had learned to respond to visual patterns with the copying responses described in Chapter VII, then the two stimuli could have evoked a common form of behavior, self-stimulation from which might have served as the basis for the response. The melodic line of the aria could have evoked a response which generated stimulation often followed by the response “Wetterhorn.” Conversely, the profile of the Wetterhorn could have evoked a response which in turn generated stimulation often followed by imitative humming or the verbal response “Handel.” In this particular instance the verbal response “shoulder” provides a clear cut example of mediating behavior. The shoulder of the mountain strengthens the verbal response “shoulder” which has been part of the auditory pattern of the aria. Speculation of this sort proves nothing, but it does suggest a possible solution of the problem of induction from one sensory field to another. An adequate solution would require an experimental analysis of the various auxiliary processes through which stimulus control can be extended.
Responding to a relation. If an organism has been conditioned to choose a five-inch disk rather than a three-inch disk when the two are presented together, it may choose a seven-inch disk if this is paired with the five-inch. This fact has frequently been offered as a criticism of the principle of the stimulus. If the five-inch disk is the controlling stimulus, why is it not effective in the new combination? Actually it is possible to condition an organism either to choose the larger of two objects or to choose a particular size no matter what the size of an accompanying object. Similar conditioning begins very early in the history of the individual, and the behavior which predominates when a test is made will depend upon such a history. The relational case is important in most environments. As the organism moves about in space, reinforcements are generally contingent upon relative, rather than absolute, size.
Stimulus induction on the basis of a “relation” presents no difficulty in a natural science if the relation can be described in physical terms. Where this appears not to be the case, we have to turn to other possibilities—for example, the mediating behavior just discussed. Even such relatively simple organisms as the pigeon may respond appropriately to new stimuli on the basis of relative size, relative intensity, relative position, and so on. They can also be conditioned to ignore any of these properties and to transfer a response on the basis of some other property. The relevant properties are all capable of physical specification.
The “interpreted” stimulus. Another problem in stimulus control has attracted more attention than it deserves because of metaphysical speculations on what is “really there” in the outside world. What happens when an organism responds “as if” a stimulus had other properties? Such behavior seems to indicate that the “perceptual” world—the world as the organism experiences it—is different from the real world. But the difference is actually between responses—between the responses of two organisms or between the responses of one organism under different modes of stimulation from a single state of affairs. Thus I may “think” I have found my coat on the coat rack of a restaurant, though I discover upon examining the contents of the pockets that I am wrong. I may “think” that an object in the sky is a plane only to see a moment later that it is a soaring bird. I may “think” that an object is square only to find when I shift my position that it is not. I may “think” that a spot of light has moved from one point to another, although an examination of the wiring circuit which has produced the spot convinces me that it merely disappeared from one position and reappeared at the other. There is no reason to regard the first of each of these pairs of reactions as “perceptual” and the second as a form of contact with the real world. They are different responses made at different times to a common source of stimulation.
Usually, objects are capable of generating many different kinds of stimuli which are related to each other in certain ways. Responses to some forms of stimulation are more likely to be “right” than responses to others, in the sense that they are more likely to lead to effective behavior. Naturally these modes are favored, but any suggestion that they bring us closer to the “real” world is out of place here. As we saw in Chapter VII, the visual and tactual properties of objects in space lead us to develop an effective repertoire in which we approach and reach for objects successfully. To take a specific case, the visual stimuli generated by a square object are usually accompanied by other visual stimuli when the object is seen from another angle or placed alongside measuring scales, as well as by certain tactual stimuli when the object is manipulated. Now, we can construct an object which, seen from a given point of view, supplies the stimulation characteristic of a square object, although it supplies very different stimuli when handled, measured, or viewed from other angles. Once we have responded to such an object in apparently inconsistent ways, we may be less confident in saying “square” to any one set of visual stimuli, but we have no reason to argue that our original visual response was not to the object “as it really is.” We operate in one world—the world of physics. Organisms are part of that world, and they react to it in many ways. Responses may be consistent with each other or inconsistent, but there is usually little difficulty in accounting for either case.
To take another example, suppose we observe a faint haze in the distance at the edge of a forest. This stimulus is appropriate to either of two large classes to which we emit the verbal responses “fog” and “smoke,” respectively. The appropriate nonverbal responses are very different: in one case we simply pass on; in the other we dash to give the alarm. We may do neither until we have “decided which it really is.” We “interpret” the stimulus before taking specific overtaction. But “interpretation” is like the “attention” discussed in Chapter VII; we need not find a particular form of behavior to be identified with it. We “interpret” a stimulus as smoke insofar as we tend to respond with behavior appropriate to smoke. We “interpret” it as fog insofar as the probability of a different repertoire is increased. It is only when specific behavior has occurred that we can say that a stimulus has been “interpreted” in a given way, but we may still speak meaningfully of both probabilities. A given stimulus may have two different effects simultaneously when they are compatible, and two different effects in rapid alternation when they are not. A complex condition of indecision may prevail until the matter is resolved either by clarifying the stimulus or in some other way. (What happens when we make a decision will be discussed in Chapter XVI.)
The functional control exerted by a stimulus enables us to distinguish between sensing and certain other activities suggested by such terms as “seeing,” “perceiving,” or “knowing.” “Sensing” may be taken to refer to the mere reception of stimuli. “Seeing” is the “interpretive” behavior which a stimulus controls. The term “seeing” characterizes a special relation between behavior and stimuli. It is different from “sensing” just as responding is different from being stimulated. Our “perception” of the world—our “knowledge” of it—is our behavior with respect to the world. It is not to be confused with the world itself or with other behavior with respect to the world or with the behavior of others with respect to the world.