Another Look at New Look 1

Jerome Bruner School of Law, New York University

New Look 1 was not initially about the unconscious. It was the new mentalism on its way to becoming the Cognitive Revolution. Its subsequent concern with "unconscious defense mechanisms," although useful, was not its main theoretical thrust. Its principal questions have always been how and where selective processes operate in perception. Obviously, many such processes are unconscious, for consciously guided attention and search become automatized easily in use. And they are fairly flexible as well. So how smart is "the unconscious"?Not very, but a big help anyway.

New Look 1, as Greenwald (1992, this issue) rightly notes, did not begin by taking a position on the distinction between the conscious and the unconscious. Nor was it the intention of its early proponents to make a case for psychoanalytic dynamics operating unconsciously through perception. Eventually it became embroiled in (or even pushed into) such issues by the very nature of the psychological arena of its day. Two competing radical voices dominated that arena: antimentalist behaviorism, on the one side, insisting that mental processes did nothing, and magical-realist psychoanalysis, on the other, insisting that unconscious mental processes did everything. Alas, the din drowned out the cognitive message that the first New Look was trying to get across. It was simply that perception was not, in the positivist sense, a mere neutral registration of what was "out there," but was, rather, an activity affected by other concurrent processes of thought, memory, and so on. Nor was this an entirely new agenda. Just as Bartlett (1932) had fought to rid us of the copy version of the memory trace, we early warriors of the first New Look were going to rid psychology of the "pure percept." Indeed, it was a constructivist view of perception, offered in the same constructivist spirit in which Bartlett had approached memory. Its constructivism gave the first New Look a strong instrumentalist bias: Perception was seen as an instrument of adaptation subserving fundamental adaptive functions. For me, the two primary functions were the minimization of surprise and the carrying out of perceptual search. The first required guidance by working mental models of the world's invariances and redundancies. The second required the operation of selective programs that could be switched in and out according to circumstances. Perception should be both highly sensitive to the stimulus geography of the world of input and capable of looking for needles in haystacks, if required to do so. The first was 780

easy enough, if one granted that the construction of the perceptual world was guided both by the perceiver's expectancies (representation was still a rather scandalous concept in the 1950s) and by the operation of stable autochthonous processes in perception that yielded the figure-ground structures, contrast, the constancies, and so on. But selectivity guided by search requirements was a harder nut to crack. What were those requirements? No one doubted the existence of set effects, but could one say something more than that they were simply there? I recall that we got into this issue in a most indirect way. It had little, if anything, to do with the unconscious or with Freud. It was concerned, rather, with perceptual accentuationwhat made some features of the world more phenomenally salient than others. The size of more valuable coins was overestimated more than that of less valuable ones, and was done more by poor kids than by rich ones (Bruner & Goodman, 1947), given the right conditions. The effect worked best when size and value are correlated (as they are in most mintages), when one imposed a small delay between looking at the coin and matching its size, and so on. Indeed, if you arranged experimental conditions psychophysically just right, you could even make the effect go away. But leave all that aside for the moment. So far, the unconscious had not appeared on the scene, and the closest we had got to psychodynamics was that constant errors (we should have called it phenomenology) might have something to do with wish fulfillment. The establishment response was, of course, that the findings were not really about perception but about "something else"apperception, memory, or whatever. But soon the tachistoscope entered the scene, and all of us shifted over to investigating recognition thresholds, investigating the selectivity of perception for different kinds of stimulus input. Immediately it became clear that a recognition threshold was no all-or-none matter. But even so, we were nowhere near considering the psychodynamics of the unconscious. Those early studies (e.g., Postman, Bruner, & McGinnies, 1948) dealt with such issues as the threshold for recognizing equal frequency words that represented different values in the subject's psyche, as measured on the Allport-Vernon Study of Values (Vernon & Allport, 1931). It quickly became apparent that the time required for recognizing a word depended

Correspondence concerning this article should be addressed to Jerome Bruner, School of Law, New York University, 249 Sullivan Street, New York, NY 10012.

June 1992 American Psychologist

Copyright 1992 by the American Psychological Association, Inc. 0OO3-O66X/92/S2.O0 Vol. 47, No. 6. 780-783

on where it stood in terms of one's measured hierarchy of values (a finding that today seems utterly banal, but which in the late 1940s and 1950s stirred the dovecotes more than a little). But something new came out of that study that presaged stirring times ahead. Psychologists found a weak but reliable and systematic tendency for subjectsprior to recognizing certain target wordsto report seeing words that were their synonyms or, more accurately, were within the same value domain as the not-yet-recognized visual targets. Indeed, it was also the case (weakly but reliably) that lowest value target words produced denigrative hypotheses. Subception was on its way in, but there is an issue that needs to be considered before I turn to that. These results, and several like them, suggested that there was something at work that depended on the prerecognition processing of semantic and even value features of the input. This caught us all rather aback, but it was not long before there were all manner of experiments afoot to discover what other surprising features of input the judas eye of the perceptual process could register in exercising perceptual selectivity. Postman and I (Bruner & Postman, 1947)neither of us particularly doctrinaire about the unconscious in its psychoanalytic sensehit on the idea of relating associative reaction times with perceptual recognition thresholds. We chose for visual targets those words (again controlled for frequency) that elicited the longest, the midmost, and the shortest associative reaction times, and sometime afterward presented them to the subjects whom we had tested, this time in the usual tachistoscopic recognition experiment. We fully expected associative reaction time to correlate with recognition thresholds. Instead, the words producing the longest associative reactions either produced the highest or the lowest recognition thresholds. We duly attributed delay of recognition to a process of "perceptual defense," with the reduced recognition times attributed to "perceptual vigilance." Not long after, Leo Postman and I (Postman & Bruner, 1948) discovered that if one stressed one's subjects by hassling them while they were engaged in a tachistoscopic recognition task, their recognition thresholds were also elevated. So we began to conceive of defense as a response to stress induced by subception of a disturbing word. Very soon, however, the corner was turned into psychodynamics. One study (Lazarus, Eriksen, & Fonda, 1951) reported, for example, that outpatients at a clinic who were judged by their therapists to use denial and repression as ego defenses showed more of a tendency to exhibit perceptual defense, whereas those who used intellectualizing and counterphobic defenses were more prone to perceptual vigilance. This was followed by a study showing that nonsense syllables associated with stress elicited galvanic skin responses before correct recognition occurred (Lazarus & McCleary, 1951). The question thus naturally arose concerning how deeply perceptual input was processed prior to its conscious recognition. June 1992 American Psychologist

Psychoanalytically committed investigators were, of course, inclined to take the view that everything was "seen" through the judas eye of the unconscious mind prior to getting into consciousness, but the more conservative critics (i.e., the main body of American experimental psychologists) would have none of it. They tried either to explain it all away by frequency effects (without ever quite explaining how such effects worked) or to dismiss it as an instance of response withholding by embarrassed subjects faced with dirty words in the presence of a clean psychologist. But the fat was in the fire. The more psychoanalytically inclined New Lookers came to dominate the debate, concentrating on the search for a typology of perceptually focused forms of ego defense to account for how patients managed their perceptual experiences psychodynamically. In its own right, that work (particularly the research of George Klein, e.g., 1964, and his students) did much to promote a new "ego psychology" into the old psychoanalysis. What resulted, I realize in retrospect, was a split between those who were principally interested in psychodynamics and those who were committed to the study of cognitionfor 1956, the mythical birthday of the cognitive revolution, was approaching. I went on to study information and redundancy as they affect perceptual processing time (Miller, Postman, & Bruner, 1954); Postman soon moved on to study inferential and memory processes; and very soon a British phase of the New Look began to emerge, dominated by an amalgam of ideas derived from information theory, from Kenneth Craik (1943), and from Bartlett (1957). The principal players were Donald Broadbent (e.g., 1971) and his students, Neville Moray (1969), Colin Cherry (1957), and several others, mostly at Cambridge and Oxford Universities. They were soon deep into studies of dichotic listening and the distribution of attention. The new banner was information processing, and it might have been called the second New Look but for the fact that it rapidly became mainline psychology. William Hirst (in Le Doux & Hirst, 1986) has provided an acute account of this new direction, and I shall not try to condense his subtle assessment here. I want to mention only one line in this work, for it bears on Greenwald's (1992) question about how smart the unconscious is. Broadbent (1957) proposed, in trying to explicate the results of studies on dichotic listening, that the task was to find the selectivity bottleneck in perceptual processing. His first proposal was that it was something like a filter and that it operated far downstream close to input, and it operated principally on relatively unprocessed sensory input. For example, it is easier to tune out a voice coming into the ignored ear if it is of a different pitch than if it is of the same pitch. Then Anne Treisman (1969) discovered that the chance of crossing over to the unattended ear increased at points at which the semantic content in the two ears were related. This implied some cognitive processing by the neglected ear and suggested that the bottleneck in the processing was higher up than was originally proposed by Broadbent. 781

This led Treisman to propose that what sat athwart the bottleneck was more like an attenuator than a fixed filter. Eventually, new work (Marcel, 1983) led to the view that the filter or attentuator operated surprisingly far upstream and did a surprising lot of processing before an input was ever permitted to enter consciousness. And after a decade of quite varied and often puzzling results in studies of selective attention, it began to be the general consensus that words such as bottleneck and filter were really unfortunate and that, in fact, the level of processing varied quite widely for different types of tasks. Any number of studies indicated, for example, that semantic processing was taking place at the preconscious level and that a presented word not in awareness could interfere semantically with the recognition threshold of another. Indeed, there were even studies (e.g., Corteen & Dunn, 1972) indicating (as in the old subception studies) that words associated with a noxious stimulus would produce a galvanic skin response, even though they entered through the unattended ear in a dichotic listening experiment. Indeed, if the associated words were all "city" words, another city word not before associated with the shock would also produce a galvanic skin response by virtue of its generalized semantic status. Then, as if to confound matters still further, the blindsight studies appeared on the experimental scene, indicating that in certain injuries to the visual cortex, a patient could guess the identity of a stimulus impinging on the injured segment (given forced choice) without being able to "see" it (see Weiskrantz, 1986, for overview). At the same time, priming studies revealed that, beyond question, a visually subthreshold primer could affect memory recovery of semantically related target words and could even alter the hedonic tone of a picture that they had primed (Niedenthal, in press). It wasn't everything, but it was something. So, how smart is the unconscious? (Note that we no longer have to ask the question about whether unconscious processes enter into perceptual selectivity; it now seems downright barmy to assume that everything that affects any mental process, perceptual or otherwise, has to be accessible to consciousness.) But the broader, deeper question of the functions of unconsciousness still remain. So, how smart is it? And what kind of smart is it? I agree with Hirst (LeDoux & Hirst, 1986) that the answer lingers in the neighborhood of "not very." Hirst takes the view, and I share it, that preconscious perceptual processing is carried on as deeply as necessary, and as soon as something is no longer needed for upstream processing decisions, it is dropped. This conforms to a principle that seems to me perfectly reasonable biologically. It is this: Keep the complex levels of processing unencumbered with detail carried over from prior decisions. Schematize, accentuate, note only exceptions from canonicity, and honor the magic number 7 2. The purpose of the fragile buffer store in information-processing systems is to hold input features just long enough to decide what may be relevant to the next level up, be it another higher order buffer or long-term memory. Another way 782

of saying this is that, although one does not have to operate with the speed and automaticity of Avia, there are time constraints that argue against prolonging decision processes by loading upstream transducers with questionable junk. I would be very surprised indeed if some forthcoming article on perceptual recognition carried tidings to the effect that there is evidence that a whole subceived sentence, replete with imbeddings and a cleft construction, could interfere with my recognition of another sentence that parses the same way or that ends up bearing the same propositional message. We would have to be much more thoroughly wrapped in thought as a species to bear that depth of processingor have even more connections in our cortex than the 5 X 109 we already have. So, what kind of smart is it? All I can do is to hazard a guess about another principle. It is that smartness is never fixed: Intelligence is always self-generative. As Kahnemann (1973) has pointed out, the amount, kind, and place of preconscious activity that operates in attentional selectivity varies with experience and expertise. Even that evanescent thing, effort, turns out to matter. I am struck by the speed with which a gifted composer can write out the notation for an entire orchestral piece or by the uncanny sense with which a gifted New York magazine editor can circulate cannily among the aspiring and promising at a cocktail party, while bypassing the disgruntled and rejected. Neither of them, the composer or the editor, can begin to tell you how they do it. Indeed, I rather suspect that the conscious subroutines we develop and overlearn for doing our own expertise "thing" very soon go underground and operate with a very large preconscious component. So, why not a neurotic, hanging on to an overlearned, overdetermined adjustment to his or her own needs and the world's threat to them? Why not a psychoanalyst primed to see countertransference in every flicker of a patient's independence? I think generally, then, that the kind (and depth) of preconscious processing we use in coping with our complex cultural environmentwhile it is never likely to be massiveis likely to reflect the degree of specialization that goes into our adaptation to the world. The more, the more. I rather suspect that one of the reasons that experimental results in this domain have been so shaky and so hard to replicate is that we do our research precisely on the unspecialized, the uncommitted, the very ones who, in the totally appropriate French expression, lack une deformation profesionelle. So when you ask how smart the unconscious is, be prepared to answer it with two provisos in mind, (a) It depends on how hard you have worked in mastering and organizing a domain of knowledge-cum-activity. The harder you've worked at it, the less you will depend on fully conscious processing, (b) If you are generally "smart" you will know better than to overload your unconscious with impossible jobs of information processing. You will rely instead on some sort of signal that informs you that you have breached the limits of unconsciousness and that it is time to wake up. June 1992 American Psychologist

REFERENCES Bartlett, F. C. (1932). Remembering. Cambridge, England: Cambridge University Press. Bartlett, F. C. (1957). Thinking: An experimental and social study. Cambridge, England: Cambridge University Press. Broadbent, D. E. (1957). A mechanical model for human attention and immediate memory. Psychological Review, 64, 205-215. Broadbent, D. E. (1971). Decision and stress. San Diego, CA: Academic Press. Bruner, J. S., & Goodman, C. C. (1947). Value and need as organizing factors in perception. Journal of Abnormal Social Psychology, 42, 3 3 44. Bruner, J. S., & Postman, L. (1947). Emotional selectivity in perception and reaction. Journal of Personality, 16, 69-77. Cherry, C. (1957). On human communication: A review, a survey, and a criticism. New \brk: Wiley. Corteen, R., & Dunn, D. (1972). Shock associated words in an nonattended message: A test for momentary awareness. Journal of Experimental Psychology, 94, 308-314. Craik, K. J. W. (1943). The nature of explanation. Cambridge, England: Cambridge University Press. Greenwald, A. G. (1992). New Look 3: Unconscious cognition reclaimed. American Psychologist, 47, 766-779. Kahnemann, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-Hall. Klein, G. S. (1964). Need and regulation. In M. R. Jones (Ed.), Nebraska Symposium on Motivation. Lincoln: University of Nebraska Press.

Lazarus, R. S., Eriksen, C. W., & Fonda, C. P. (1951). Personality dynamics and auditory perceptual recognition. Journal of Personality, 19, 471-482. Lazarus, R. S., & McCleary, R. A. (1951). Autonomic discrimination without awareness: A study of subception. Psychological Review, 58, 113-122. LeDoux, J. E., & Hirst, W. (1986). Mind and brain: Dialogues in cognitive neuroscience. Cambridge, England: Cambridge University Press. Marcel, A. (1983). Conscious and unconscious perception: Experiments on visual masking and word recognition. Cognitive Psychology, 15, 197-237. Miller, G. A., Bruner, J. S., & Postman, L. (1954). Familiarity of letter sequences and tachistoscopic identification. Journal of General Psychology, 50, 129-139. Moray, N. (1969). Attention and listening. Baltimore: Penguin. Niedenthal, P. M. (in press). Affect and social perception: On the psychological validity of rose-colored glasses. In R. Bornstein & T. Pittman (Eds.), Perception without awareness. New York: Guilford. Postman, L., & Bruner, J. S. (1948). Perception under stress. Psychological Review, 55, 314-323. Postman, L., Bruner, J. S., & McGinnies, E. (1948). Personal values as selective factors in perception. Journal ofAbnormal Social Psychology, 43, 142-154. Treisman, A. M. (1969). Strategies and models of selective attention. Psychological Review, 76, 282-299. Vernon, P. E., & Allport, G. W. (1931). A test for personal values. Journal of Abnormal Social Psychology, 26, 231-248. Weiskrantz, L. (1986). BlindsighvA case study and implications. Oxford, England: Clarendon Press.

June 1992 American Psychologist

783