Richard Gregory

Introduction

Richard Langton Gregory (July 24, 1923 – May 17, 2010) was the most articulate modern defender of the idea that seeing is not a matter of receiving an image but of making a guess. The claim has a long pedigree — it descends from Hermann von Helmholtz's notion of unconscious inference — but Gregory gave it its modern form and its modern audience, arguing that perception is built the way science is built: by generating hypotheses against fragmentary evidence and testing them fast enough that we never notice it happening.

He is also one of the few psychologists of the twentieth century whose influence ran through four channels at once. He wrote Eye and Brain (1966), which a University of Bristol obituary describes in terms normally reserved for the Feynman lectures — the book that pulled a generation of students into perception research. He invented things, including techniques for viewing microscopic objects in three dimensions and for taking sharp telescope images through an unstable atmosphere. He founded Perception, the journal that set the terms of the field for decades. And he founded Britain's first hands-on science centre. Gregory left school at seventeen with no qualifications and no university place, and ended with the Royal Society, the Royal Society of Edinburgh, and a CBE.

Early Life and Education

Gregory was born in London, the son of Christopher Clive Langton Gregory, an astronomer and the first director of the University of London Observatory. The family habit of building optical instruments stayed with him; he later attributed his interest in optics partly to having grown up around his father's work. He was educated at King Alfred School in Hampstead, where he was drawn to electronics in the school radio club.

War interrupted everything. He left school at seventeen without any A-levels, because of the outbreak of the Second World War. In 1941 he joined the Royal Air Force, but he was not accepted for pilot training because of a childhood ear operation, and was enrolled in the Signals branch instead, working on communications and radar, largely in Canada.

The assignment that followed is worth dwelling on, because it is the whole of his later career in miniature. The Air Ministry put him on the bombed-out site of the John Lewis store in Oxford Street and told him to explain how radar worked to anyone walking past. Somewhere around four million people stopped over six months. That job — standing in rubble, translating a technical system into something intelligible to ordinary passers-by — is what he then did for forty years.

A demobilisation scholarship to Downing College, Cambridge, followed in 1947. He read moral sciences; his first two years were largely taken up with philosophy, and he was tutored by Bertrand Russell. Only in his final year did he turn to experimental psychology, where he was taught by Frederic Bartlett. Gregory described himself as one of Bartlett's last pupils, and said he had been inspired by him. Bartlett asked him to stay on.

Academic and Professional Career

From 1950 to 1953 he worked as a research worker at the MRC Applied Psychology Unit in Cambridge. During this period he was seconded for a year to the Royal Navy establishment at Whale Island, Portsmouth, to devise experiments about escape from disabled submarines; he built what he called his first gadget, an automatic recorder of the behaviour of trapped submariners preparing to escape into worsening air.

He then lectured in the Cambridge Department of Experimental Psychology through the 1950s and 1960s, producing Eye and Brain in 1966 and giving the Royal Institution Christmas Lectures on "The Intelligent Eye" in 1967.

In 1967, with Donald Michie and Christopher Longuet-Higgins, he co-founded the Department of Machine Intelligence and Perception at the University of Edinburgh — an institution that anticipated academic artificial intelligence in Britain. He was head of its Bionics Research Laboratory, Professor of Bionics, and department chair from 1968 to 1970. In 1970 he was persuaded to come to the University of Bristol by the neurophysiologist Grey Walter, to establish the Brain and Perception Laboratory in the Medical School, and he remained Professor of Neuropsychology there until his formal retirement on his sixty-fifth birthday in 1988. A "Gregoryfest" was held for the occasion. Retirement was "purely notional": he kept his desk in the Department of Experimental Psychology and was working within days of his death.

Institutionally he was relentless. He founded the journal Perception in 1972 and edited it for decades, deliberately placing it with a small publisher and keeping its remit broad enough to include art and philosophy alongside psychophysics; the open-access sibling i-Perception appeared the year he died. He was a founding member of the Experimental Psychology Society and its president in 1981–82, and he founded the Exploratory hands-on science centre in Bristol — dated variously as 1978 or 1981 depending on the source — the first of its kind in Britain, later to evolve into Explore@Bristol. He was appointed Osher Visiting Fellow of San Francisco's Exploratorium in 1989 and received the Michael Faraday Prize in 1992.

One small episode connects him to another psychologist on this site. Gregory collaborated with William Edmund Hick as a subject in Hick's classic "On the rate of gain of information" experiment, and later recalled that he was the only participant besides Hick himself to complete the course — Hick having abandoned it when the apparatus fell apart.

Major Contributions to Psychology

Perception as hypothesis

The core claim is an argument from underdetermination. The pattern of light on the two retinas is consistent with infinitely many arrangements of the world, is frequently occluded, and is corrupted by noise, so the input cannot specify what is out there. Gregory's move was to treat the brain's resolution of this ambiguity as analogous to scientific inference: the visual system brings stored knowledge to bear, tests it against incoming evidence, and delivers an interpretation. The interpretation is not delivered as a report — it is projected outward and experienced as reality itself, which is why disagreement about what we are seeing is so rare and so difficult to produce deliberately.

This framing made visual illusions central rather than incidental. If perception is inference, then illusions are not curiosities; they are the laboratory in which normally invisible assumptions become visible. Gregory estimated that about ninety percent of the information available at the eye is lost by the time signals reach the brain, which is why so much of what we see must be supplied from memory — though the figure is his own rough estimate, and one that is easier to repeat than to define.

Illusions and their explanations

He worked across a wide range of cases, including the Kanizsa figure, whose illusory contours he read as evidence of Bayesian-style processing, the Müller-Lyer, which he explained through size scaling invoked by features that read as inside or outside corners of rooms, and the café wall illusion, which he re-encountered on a café wall in St Michael's Hill, Bristol, and took as his own.

His later writing pressed further into the relationship between perception and knowledge. He distinguished two senses of intelligence — the stored answers available to a system and the active processing needed to use them — and argued that as knowledge grows, top-down contribution dominates; he noted that there are more fibres running from the cortex down to the lateral geniculate relay station than running up from the eyes.

The public dimension

He wrote and edited books that reached well beyond the field, including The Intelligent Eye (1970), the Gombrich collaboration Illusion in Nature and Art (1973), Mind in Science (1981), and The Oxford Companion to the Mind (1987), which he edited and to which he contributed heavily. He also recorded Desert Island Discs in 1993 and picked Beethoven's Piano Sonata No. 30, listing his hobbies as punning and pondering.

Major Publications

  • Gregory, R. L. (1966). Eye and Brain: The Psychology of Seeing. Weidenfeld & Nicolson. Fifth edition, Oxford University Press, 1997.
  • Gregory, R. L. (1970). The Intelligent Eye. Weidenfeld & Nicolson.
  • Gregory, R. L., & Gombrich, E. H. (1973). Illusion in Nature and Art. Charles Scribner's Sons.
  • Gregory, R. L. (1981). Mind in Science: A History of Explanations of Psychology and Physics. Weidenfeld & Nicolson.
  • Gregory, R. L. (1997). Knowledge in perception and illusion. Philosophical Transactions of the Royal Society B, 352(1358), 1121–1127.

Impact and Legacy

The great counter-position. Gregory's work is largely a reply to J. J. Gibson, whose ecological optics held that the ambient array of light specifies the environment directly enough that no inference is needed and no stored knowledge consulted. Gregory's answer, sharpened in a 1993 paper, was physiological: given the number of stages between retina and perception, "direct" accounts cannot be sustained, and we should abandon them in favour of indirect ones. The debate is the spine of late-twentieth-century perception research. Most contemporary workers think neither side had it wholly — that perception involves both the pickup of rich structure in the environment and constructive processes that resolve genuine ambiguity — but the terms both parties set remain in use.

Whose hypothesis is it? The strongest objection to the account is that "the brain tests hypotheses" relocates the puzzle rather than solving it. Scientific hypothesis testing has a tester, a hypothesis space, and a criterion for rejection. Perception has none of these in any obvious form, and there is a persistent worry about a small scientist sitting inside the head doing the work. Critics in the modularity tradition argued the opposite of Gregory's analogy: perceptual processes look encapsulated and informationally sealed off from the rest of cognition, precisely unlike the open-ended enterprise of science.

Absorbed into something more quantitative. A standard modern reading treats Gregory as a proto-Bayesian, and it is easy to see why — he was writing about prior expectations combined with imperfect evidence long before the formalism was standard. But that reading is also the criticism. The current versions are models with specified priors and likelihoods that generate quantitative predictions; Gregory's versions are verbal, and can often be fitted to whichever result emerged. The framework accommodates outcomes rather than anticipating them.

Specific explanations did not settle. The explanation of the Müller-Lyer through misapplied size scaling has never displaced alternatives, including accounts built on low-level spatial filtering, and after decades there is still no agreed explanation for many geometric illusions. Much the same is true of the café wall illusion. The awkward consequence is that the method Gregory bet on — read the mechanism off the illusion — presupposes that individual illusions have single, localizable causes, and that remains contested. Even his account of why illusions occur has been reframed: if perception is optimal inference given the statistics of natural scenes, an illusion is not really an error at all, and calling it one imports a judgement the theory does not license.

Set against all this is a fair assessment of where his influence actually landed. Gregory's most durable contributions may be institutional and pedagogical rather than theoretical — Perception, the science centre, a book that recruited a generation, and a career path that the Royal Society memoir describes with some understatement as unconventional. Theorists of perception no longer cite him for the mechanism. They still work in the field he organised.

Learn More

References

  • Gregory, R. L. (1968). Perceptual illusions and brain models. Proceedings of the Royal Society of London B, 171(1024), 279-296.

Last updated: 2026-10-20

Menu
Loading