A popular genre, a chat about the brain: on Hugh Groucutt's interesting idea that Homo sapiens took shape slowly and gradually, without revolutionary upheavals

In essence, the cognitive revolution of 40,000 to 50,000 years ago is one of the textbook stories. Homo sapiens supposedly grew clever all of a sudden, drew a bull on a cave wall and set off to conquer the planet. The trouble is that anatomically modern humans appeared some 300,000 years ago, which means that behavioural modernity turned up a quarter of a million years late. Two hundred and fifty thousand years of dim-wittedness, and then, off they went! The argument for a cognitive leap rests on European material and on the painful status quo of finds that simply are not there.

At any rate, Groucutt argues quite persuasively that there may have been no flick of a switch in the head at all. The development of intellect was cumulative, with experiments, dead ends, regional inventions and long plateaus of stagnation.

Here, as an Americanist, I recognise my favourite awkward hypotheses. The whole Clovis theory is the same cognitive revolution, only in the New World. For decades the firm foundation of the theory of the peopling of the Americas was the Clovis points, about 13,000 years old. The ice-free corridor in North America was named as the main route. The key assumption was the swift settlement of both Americas by a single pioneering people. The whole picture was held together by the fact that Clovis in the United States has been studied inside out, for stone points survive beautifully in the ground and look handsome in museum display cases. Then came discoveries that did not fit the pattern: Monte Verde in Chile, a good fifteen hundred years older, followed by Paisley and Meadowcroft, and the unified Clovis-first hypothesis began to founder.

The amusing thing is that quite recently, in March of this year, Monte Verde itself went under the knife. It began to be said, rather convincingly, that the organic material had been washed by water into the lower layers, and that the site itself is therefore thousands of years younger. Not everyone agrees, and the debate is not yet closed. But one of the cornerstones of a large theoretical edifice has already been called into question. That is why Groucutt's argument is perfectly clear to me. Paradigms of settlement are revised again and again as new data appear, as new ways of working with old data emerge, and as dating is refined. At present, what is truly stable is neither the antiquity, which can leap about by thousands of years, nor the interpretation of the finds, but only the map of where they lie.

With the African Stone Age, Groucutt is trying to do what pre-Clovis archaeology did with Clovis. And it really does turn out that, from the caves of the Dordogne to the bogs of southern Chile, science has time and again tried to pass off the geography of its excavations as the real chronology of human development.