Perception 1996, volume 25


Irvin Rock, 1922--1995

Irvin Rock died on 18 July 1995 after struggling for six months with pancreatic cancer. He was a man of wide theoretical influence, a brilliant experimentalist, yet a modest and warm human being. Born in 1922, Rock grew up in New York City, receiving his BA in psychology in 1947 and his MA in 1948, both from the City College of New York. As a student of Hans Wallach, he completed his PhD with highest honors in 1952 from the New School for Social Research. He taught at the New School and at Yeshiva University, moving in 1967 to Rutgers University where he worked first at the Newark campus and later at New Brunswick. He officially retired from Rutgers in 1987, but moved to California where he continued his very active research and teaching career as an Adjunct Professor of Psychology at Berkeley.

During his long career, Rock worked on an impressively broad range of topics. Perhaps the most widely known work of Rock is the series of experiments he conducted with Lloyd Kaufman on the moon illusion. That work decisively overturned the generally accepted explanation of that time, and demonstrated that the moon illusion is a logical outcome of the rules by which the visual system processes size and distance in general. Rock's ingenuity can be seen in that work. Several of those experiments were conducted on the rooftop of a New York City building, with an artificial moon that could be projected to appear anywhere in the sky.

The topic that recurs most consistently throughout Rock's entire career is the dependence of the perceived form of a figure on its orientation. Rotate a square by 45° and it appears to be a diamond, not a square. This work begins with his doctoral dissertation, and includes an authoritative book and many papers, several published shortly before his death. It was in this work that he first applied the kind of question that he would turn into a powerful research tool: Is it egocentric orientation or environmental orientation that determines the perceived form? He went on to apply this question to a wide range of phenomena, often with surprising results. Max Wertheimer had argued that the stroboscopic motion one sees in the space between two alternately flashed dots is based on a kind of short circuit between the two locations of stimulation in the neural tissue. By asking observers to move their eyes in synchrony with the flashing dots, Rock, with Ebenholtz, showed conclusively that the perceived motion depends on whether the flashes appear to come from two locations in the environment, not whether they stimulate two locations on the observer's retina.

Rock applied this approach to the Gestalt grouping principles in several experiments. The Gestaltists had observed that black dots on a piece of paper appear to form groups based on their proximity to one another. By the simple and ingenious technique of rotating the paper to an angle oblique to the observer's line of sight, Rock, working with Brosgole, showed that it is the perceived distance between dots that underlies the groupings, not the distance between the dots in the retinal projection of the display. This finding challenged the Gestalt view that grouping factors operate very early in visual processing. Rock's mentor, Hans Wallach, had shown that the perceived blackness or whiteness of surfaces depends on the intensity of light reflected by that surface relative to the intensity from its adjacent neighbor. In my own dissertation work, under Rock, I showed that it is the perceived adjacency between two surfaces in the environment that is crucial, not the adjacency of their retinal images. I myself did not appreciate the impact of Rock's thinking on this work until much later.

Work on anorthoscopic perception done with Halper and others showed the extent of the discrepancy between the retinal pattern and what is perceived; even for something as fundamental as form perception, an extended retinal image is not necessary. The test of whether a perceptual factor operates at the level of the retinal image or at the level of the perceived environment has become a basic and powerful tool of psychophysics. It remains a concrete part of Rock's legacy.

It is not possible to survey all the topics on which Rock worked, so wide is the range. He established the dominance of vision over touch. It had often been argued that the ambiguity in retinal images is resolved, at least in early visual experience, by using the sense of touch. Submitting the issue to empirical test, Rock demonstrated that, when visual and tactile information contradict one another, it is vision that dominates, not touch. And vision dominates touch to such a degree that it literally changes the way things feel to the touch, a phenomenon that Rock referred to as visual capture. Beyond what has been mentioned, Rock made substantial contributions to our understanding of reversible figures, induced motion, anorthoscopic perception, and the role of attention in perception. One should not forget that, in addition to his main work in perception, several experiments he published on one-trial learning in the late 1950s had an enormous impact on research in the field of learning. The intense debate and enmity aroused by that work mark it as an important landmark in the post-Gestalt development of cognitive theories of learning. The controversy he raised there has not been resolved so much as set aside.

Many of Rock's theoretical ideas have foreshadowed important developments in computational vision, including his early proposal that form perception is based on an internal description, his ideas about the role of frames of reference in shape perception, and his principle of avoidance of coincidence in perceptual outcomes. During his career, Rock became perhaps the leading advocate of the position that visual perception is based, as Helmholtz had suggested, on unconscious inferences. To the consternation of the reductionists, Rock repeatedly showed the thought-like character of perception and the difficulty of reducing it to the processing of relatively local information by peripheral mechanisms. Just as one example, in a series of experiments, Rock showed that the perceived motion normally caused by the alternate flashing of two lights does not occur if, because of other information in the context, the flashing can be perceptually attributed to occlusion and disocclusion by other surfaces.

Rock became a central figure in the vigorous debate over direct versus indirect perception and an outspoken critic of the glib reductionism that has characterized far too much work in the tradition of sensory physiology. As Arien Mack has noted, Rock's cognitive-inferential theory of perception is the most articulate and complete version of an indirect theory of perception that we have. Even those who disagree with Rock's theoretical ideas seem to agree that he was a brilliant experimentalist. Rock had a deep feeling for the scientific method. He was often impatient with wordy debates and anxious to get on with the empirical test. I cannot resist including a revealing anecdote from my graduate student days with Rock. A fellow graduate student arrived one morning with the announcement that he had a solution to the problem of lightness constancy: pupil size. When the illumination level becomes higher your pupil contracts, letting less light into the eye. I reacted by launching into a set of logical arguments: pupil size varies by a factor of only eight whereas illumination varies by a factor of perhaps a billion; there can be more than one level of illumination within a single visual scene, but only one pupil size; etc. Rock's response was simultaneously more gracious than mine, and more devastating. He observed: "So, if I understand you, it would follow that lightness constancy would be absent when looking at the world through a pinhole (artificial pupil)". As this story suggests, Rock seemed at times to literally think in terms of experiments. It has often been observed that Rock's experiments speak for themselves. Once you understand what was done and what was found, you often understand the importance of the work, even without reading the text. Rock approached issues with an almost childlike simplicity. He had an extraordinary ability to take a fresh look at an issue, often seeing something important in what other people considered merely obvious. I endorse what Lloyd Kaufman has written in the preface to his book Sight and Mind: "By his example I learned to recognize boldness, creativity, and independence in scientific thought".

Anyone who knew Rock knows that part of his greatness lies in the character of the man himself. He was a man of great kindness and generosity. Rock's manner was always dignified, yet he was warm and caring, with an intellectual modesty that is rare. Rock was a gifted teacher and a superb mentor. He instilled the highest scientific values in his students. Many of Rock's students have distinguished themselves in the field of perception and several other well-known figures consider themselves honorary students of Rock, such has been their experience with him.

Rock served in the infantry during World War II. During fighting in Europe near the end of the war, Rock vowed to himself that, if he survived, he would put his life to constructive use. In the next half-century he published 70 journal articles, 16 chapters in books, 8 articles in Scientific American, 9 entries in encyclopedias, and 9 books. Three of these books, including one still in press, were edited by Rock. Five of the books were authored solely by Rock: The Nature of Perceptual Adaptation; Orientation and Form; An Introduction to Perception; The Logic of Perception; and Perception. A final book, written with Arien Mack on their joint studies of inattention, will appear soon.

Rock was an elected fellow of the Society of Experimental Psychologists and a two-time recipient of the Research Scientist Award from the Public Health Service. And he was honored in 1986 at an APA symposium entitled: "Computation, Inference, and Perception: Theoretical Essays in Honor of Irvin Rock". Nevertheless, the scope of formal recognition he received was probably limited by several of his personal characteristics. First, Rock was an iconoclast. He never hesitated to speak out against what he considered to be scientific foolishness, even if that meant confronting powerful vested interests such as the neuroscience movement. During the one-trial learning controversy, vigorous and effective efforts were made by influential learning theorists to suppress Rock's work. Second, Rock was not a careerist. He refused to cultivate his reputation in the manner that has become so commonplace. He was genuinely interested in the scientific questions themselves and he was uncomfortable with too much praise. He flatly vetoed an attempt by Steven Palmer and several others to organize a festschrift in his honor, in part because he felt such a book would not serve a useful enough scientific purpose.

Rock is survived by his wife Sylvia, and five children: Peter, Alice, Lisa, and David Rock, and Rayna Shilling-McCallum. I speak for many others when I say that my life has been impoverished by the passing of Irvin Rock.

Alan Gilchrist, Rutgers University, New Brunswick, NJ, USA

Return to contents of current supplement

Return to main contents

© 1996 Pion Ltd