Basic neuroscience and educational theory have, until now, ploughed largely separate academic furrows. But that hasn't stopped overenthusiastic individuals from designing ‘brain-based’ learning aids — often making healthy profits in the process.

Many of these tools have been built on gross misrepresentations of the science. Take the industry spawned by the idea that there is a ‘critical period’ for learning in early childhood, when the brain has the highest density of synapses. This ignores evidence that ‘pruning’ of synaptic connections is a necessary part of brain development.

Now the US National Science Foundation (NSF) is getting serious about the science of learning and its application in the classroom. Cognitive neuroscientists, psychologists, computer scientists and educationalists are being melded into huge collaborative teams (see Big plans for little brains).

Hopes are deservedly high. But questions need to be asked about whether the time is ripe for some of the links between basic science and educational practice that are now being proposed.

The computer-based ‘cognitive tutors’ being developed at Carnegie Mellon University in Pittsburgh are among the most solidly grounded aspects of the NSF initiative. The first tutor, for algebra, has already proved a boon for overstretched teachers. What's more, the Pittsburgh team's strategy for involving teachers in its ongoing research is both innovative and practical.

How ‘looking under the hood’ at the development of numeracy and reading will inform educational practice remains, for now, largely a matter of speculation.

However, things get a little less convincing when it comes to basic neuroscience. Researchers are planning to use magnetic resonance imaging to ‘look under the hood’ at the development of skills such as numeracy and reading. It's fascinating stuff, but how the results will inform educational practice remains, for now, largely a matter of speculation. Making meaningful connections between brain activity and behaviour is difficult, even under controlled lab settings.

Brain imaging is seductive, and has an unfortunate tendency to spawn breathless, overreaching media coverage. Care will be needed to ensure that these projects don't encourage ill-informed ‘experts’ to design yet more pseudoscientific educational tools.

That's not to say that scientific advances can't already help to inform educational policy. For instance, there is now a solid body of evidence that sleep patterns change significantly with age — and that, as a result, it makes little sense to wake teenagers up early to go to school, when their attention will be low as a natural consequence of their daily rhythms. Education authorities and schools are starting to hear this message, and some are adjusting their schedules accordingly.

There's also a strong case for putting the educational tools derived from research in neuroscience to more rigorous empirical tests. For instance, researchers who have evidence that dyslexics have problems with auditory processing have developed a program called Fast ForWord to help them learn to read. But the scientists' company is now marketing the software as a learning aid for children with no specific reading deficits, before they have gathered evidence that it helps anyone other than dyslexics. For now, providing this sort of evidence is where the emphasis should remain.