From Charles Lamb's "Dissertation on Roast Pig" to Max Beerbohm's "The Pervasion of Rouge" the English tradition of the essay flourished. Alas, it is for the most part moribund. Nowadays it is only school children and Bernard Levin - to make a fine distinction - who practice it regularly.

At its best the essay should have charm, wit, irony, and a lightness of touch. Richard Gregory has now produced his own second collection of essays. At his best he succeeds extremely well. Such pieces as those on traffic lights or on the luminaries who have dignified the city of Bristol are a delight to read. Although perhaps not notable for their irony, they live up to the great traditions of their predecessors. They have charm, a gentle humour and make much of little by picking up unconsidered trifles. Gregory comments, for example, on Marconi's luck in sending radio waves across the Atlantic, an achievement that according to contemporary scientific theory should have been impossible, for radio waves do not bend: his signals were in fact reflected back to earth by an as yet undiscovered ionised layer in the atmosphere. Again, Gregory himself noticed in a physics paper set for school children that the examiner assumed that different weights fall at different speeds: as Gregory observes, it is little wonder that with such obtuseness in science teachers and examiners, today's children have difficulty in acquiring even a smattering of science.

Despite their deftness of touch, some of Gregory's essays are marred by heavy punning, particularly the one on "Cracks", which would be enough to drive most readers crackers - a small thing but mine own addition to his. He displays a lack of insight when he asks why of all forms of humour "puns . . . evoke the loudest groans''. Surely he should know.

Perhaps unfortunately, most of the essays in Even Odder Perceptions are devoted to more serious subjects such as fractals, inertia and consciousness. These topics are tackled with the same magnificent and wholly admirable zest as the more trivial ones, but Gregory's clarity and accuracy are no match for his enthusiasm. Nobody not already familiar with the principles of inertial motion could follow his account of why gyroscopes remain in the same position: his analogy with a bicycle wheel is a case of obscurum per obscurius.

ÁñÁ«ÊÓƵ

In his repeated onslaughts on the problem of mind and brain Gregory falls into the usual trap that ensnares intellectuals. he seems to think that if a computer could solve cognitive problems as well as people can, it could be conscious, but he ignores the fact that, as J. Weizenbaum pointed out nearly 20 years ago, no computer or computer program could bear much resemblance to the human mind unless it had emotions. Even if a simulation of emotions and basic drives were built in, the program would still differ in a major way from people. We feel hungry because we genuinely need food, we make love because we enjoy sex (or want someone else to enjoy it) and we become angry, afraid and so on all for good biological reasons. Even if a computer could simulate these emotions and drives, it would not be the genuine thing - a computer could never have real needs and feelings of this kind, though it might perhaps genuinely be made to reach for the nearest electric plug.

Some of Gregory's observations are sensible. In discussing whether the brain is digital or analogue, he argues that the fact that we can simulate some of the behaviour of the brain on a digital computer does not mean that it is itself a digital machine. The behaviour of any analogue system, such as the weather, can be simulated digitally. On the other hand, he is surely rash to declare that the nervous system is analogue, for the on-off firing of a nerve cell is clearly digital. The only reasonable view is that the brain contains elements of both kinds: neurotransmitters and postsynaptic potentials are analogue.

ÁñÁ«ÊÓƵ

Although Gregory sometimes puts received ideas well, there are too many errors and misapprehensions in the more serious essays. John Broadus Watson was not born in 1918 but in 1878, and Hubel and Weisel discovered cells sensitive to the orientation of lines and edges not in 1962 but in 1959 (the date of their first publication on this topic). Such errors are misleading but trivial. But one hopes that when Gregory writes that "atomic statements must be verified'' (instead of "verifiable") it was his pen that slipped rather than his mind. Among other misdemeanours he discusses "sensation'', an extremely vague word, without making any attempt to define it, and he appears to equate aniseikonia with astigmatism. Such confusions are no doubt caused by carelessness rather than ignorance, but they are likely to bewilder the reader.

He proclaims the importance of visual illusions as tools for investigating the workings of the mind, but he does not tell us what has been learned from them. In fact there is no existing explanation that is satisfactory for the most famous illusion of all - the Muller-Lyer - and it could be argued that we have learned nothing from the traditional geometric illusions, or at least nothing that we did not know already, for there are some illusions that are based on effects that were discovered only when much simplified patterns were used. Although much of Gregory's scientific work has been devoted to illusions, his explanation of some of them are baffling. For example, his account of the Ames Window illusion, a trapezoid that physically rotates always in the same direction but that is seen to rotate first one way and then another. The key to the illusion is that the image of the trapezoid foreshortens along the horizontal axis as it passes through the frontal parallel plane where because of false perspective it is seen tilted away from the observer. At this apparent tilt it could foreshorten only if the direction of motion were reversed and this is exactly what is seen.

Even Odder Perceptions gives rise to disturbing thoughts. Most writers of popular books on philosophy and science now are obscure and prone to error. There are exceptions such as Richard Dawkins or Richard Feynman, but Daniel Dennett or Gerald Edelman are more typical. Even writers with something to say, like Stephen Hawking and Roger Penrose, can be unnecessarily obscure, yet their books are widely bought if not - one suspects - widely read. Why has this passion for scientific gobbledygook arisen? Perhaps, in the absence of the mysticism of religious belief, people can only have faith in books they cannot understand, for even the novel has gone the same way. If something is clearly set out, readers may feel they might have thought of it themselves, but if something is incomprehensible they admire the minds of its authors for having thoughts too deep to grasp. The worry about Gregory's book is that it will be read for the serious parts, because they are - at least in part - unintelligible, and not for his delightful resuscitation of the art of expounding trivia. But we should forgive him his sins, for whatever he writes he displays the same charmingly innocent enthusiasm.

Stuart Sutherland is emeritus professor of experimental psychology, University of Sussex.

ÁñÁ«ÊÓƵ