IT IS ONE of the spookiest skills that humans possess - the ability to predict which day of the week February 14, 2073, will fall on, or to calculate whether there were five Wednesdays in July 1991.
Most of us would take a month of Sundays to answer these questions. But a handful of autistic savants can rattle off the answers in seconds. How do they do it?
Some psychologists believe that this astonishing, if largely useless, skill is due to practice. But as Richard Cowan and Neil O'Connor of the Institute of Education in London say in a new study, there are several grounds for doubting this explanation. The ability often emerges suddenly between the ages of five and eight; the savants are apparently untaught and they are usually unable to explain how they do it.
Dr Cowan has studied an adult and two children with the skill. The adult is a professor of maths at Princeton University - known only as JC - who devised his own method of "calendrical calculation" when he was a teenager.
The professor was tested on 20th-century dates and found to be faster than all but one of six savants. In a second test involving the years 2079-2099 he outshone even the fastest savant. In a third test that included dates in the unimaginably distant future - up to the year 820,000 - he matched the best of his rivals.
He struggled to name years with particular features, such as having March 1 on a Wednesday, even though the more able savants had reeled off numerous years in chronological order.
Nevertheless, Dr Cowan elieves that the professor's skills match or exceed those of the savants.
But he remains unconvinced that "calendrical calculation" is simply a matter of practice. He has monitored the performance of two boys aged five and six. Their calculating skills, though impressive, are below the adult savants' level: their range is less than 10 years and they take more than 10 seconds to answer questions.
"Both intentionally learn calendars but their skill does not dep-end on memory alone," Dr Cowan says. "They have detected calendrical regularities that they can use to answer questions about more remote years. While some features of their skill fit the 'practice' view, they are also exceptional in their arithmetical skills."
Disappointingly, Dr Cowan concludes that there is only one way to resolve the talent v practice debate: "That is to get a group of children, provide them with minimal instruction and calendars and then monitor their development." He accepts, however, that no one is likely to volunteer their child for such an experiment. Not, at least, until the 32nd day of the month.
Calendrical calculation: talent or skill? by Richard Cowan and Neil O'Connor, Institute of Education, University of London
* Education researchers who wish to disseminate their findings in The TES should send summaries of no more than 750 words to David Budge, Research
Editor, The TES, Admiral House, 66-68 East
Smithfield, London E1W 1BX. Tel 020 7782 3276. E-mail: David.Budge@tes.co.uk