'It's a pity,' he says. 'There are so many other more practical subjects now. So why would the government turn to the abacus, which shows no immediate benefit?'

Ng's intuition is right. There are longer-term cognitive benefits, especially among children, as recent neurological and brain-imaging research has shown.

In a 2001 Japanese study, the brain activity in abacus champions while they were using the device was found to be concentrated more in the right side - the part associated with visual and creative thinking - rather than the left side, used for logic, language and computation.

Dr Kimiko Kawano - a researcher at Japan's Nippon Medical School, who led the study - wrote that, while stimulation of the right side of the brain might not be enough to improve learning capacity outright, 'image thinking' could prove beneficial in a society filled with logic- and problem-based learning.

Other studies of abacus users using fMRI (functional magnetic resonance imaging) - a kind of neuro-imaging that tracks blood flow, and therefore activity, in the brain - lend support to Kawano's claims.

William Wang Shi-yuan, a linguist and cognitive scientist at Chinese University, believes the monosyllabic and highly regular - or systematic - number system for the low integers in the Chinese language gives an advantage to young children learning numerical skills. These advantages may be reinforced by learning to use the abacus.

Aaron Seitz, an assistant psychology professor at the University of California in Riverside, agrees that the abacus is ideal for teaching maths to children. 'If you have information coming in from multiple senses, that activates more of your brain and accelerates learning,' he says.

'[With the abacus], you have this combination of the tactile information, you have actually auditory information - because when you move the beads, there are sounds associated with that - and then you also have the visual-spatial. The key is that the entire experience is much richer in terms of the sensory information you're getting than if you're just doing a standard maths problem with just paper and pencil - or worse, the calculator. Maths is not strictly verbal, tactile, or a language; it is kind of an abstract representation - and so if that abstract representation is built upon information from multiple sensory modes, it's going to be more developed and more effective than coming from a single sense.'

In a Chai Wan classroom, a stream of clicking fills the air. The teacher is firing the maths problem - '68,375 plus 97,523 minus 28,560' - at the class. Upon the last digit, the students shout out the answer simultaneously. They are all correct.

Such calculations are commonplace at abacus training centres such as this one, which drill children in handling enormous streams of numbers via deft finger movements. The abacus association runs a centre that trains about 1,000 students.

Ng says intensive contact between fingertips and the beads during training stimulates the development of connections among neurons, which may be key to learning.

In the association's classroom, amid the clatter of abacuses, one student sits in silence with no abacus on his desk. Yet Lee Chi-chun replies as fast as the others. For Lee, 15, who has studied the abacus for nine years, the device is not in his hands, but in his head. 'When I hear the numbers I move the beads up and down on the abacus in my mind,' he says.

This sort of 'mental abacus' is present in expert abacus users, who translate numbers immediately to pictures in their head. Studies have shown that, usually, we have to translate numbers to language in our heads before being able to comprehend them; people like Lee are able to skip this step, allowing him to be even quicker in mental calculations.

In another Japanese study, in 2002, Sophia University researchers in Tokyo compared abacus experts with non-experts in digit recitation, and found advanced abacus users had better short-term memories.

One of the first studies into the pedagogic value of the abacus was published in 1986.

The US study of primary-school children in Taiwan found that students with abacus training earned better grades in both maths and reading. The researchers compared the reading and maths grades of 618 students from the end of the first grade, when no one had any abacus training, to the end of the fifth grade, when the children with abacus training did better.

'Abacus skill has no direct effect on perceived cognitive competence, but abacus does have an indirect effect via its impact on reading and mathematics performance,' wrote researcher James Stigler at the time in the American Journal of Education.

The abacus is believed to have been invented by the Chinese in about 1200; it spread to Japan and Korea in about 1600. The abacus is still used around the globe today - but for a different purpose. Parents are sending children to abacus training centres and competitions to marvel at their progeny's speed at doing mathematical operations.

Ng says that abacus training is gaining popularity in North America, where it was being touted as a way to improve overall achievement. In Japan, the study of the abacus - known as soroban - has been compulsory in elementary school education since 1945. Previously taught only in the third grade, from this year it is being taught in the third and fourth grades; normally it takes two to three years to master an abacus.

On then mainland, the abacus has been missing from the compulsory curriculum since 2001, and in Hong Kong the subject has never been mandatory.

Although some nations have recognised the benefits of the abacus, it has given way to more practical academic subjects. Perhaps cutting-edge neuroscience may help revive the ancient computational device.