Thursday, March 18, 2010

What brain scans can’t teach us

(Cross-posted, with minor edits, at Liberal Conspiracy – they get rather more traffic than me, so feel free to post comments there instead.)

This month’s Prospect magazine has a section on neuroscience, and in particular its political implications. As well as being a political junkie, I find psychology and the brain really interesting, so this should have been manna from heaven for me. And there’s some decent stuff there, although most of the contributors don’t get the chance to make any extended arguments. There are also plenty of banal, incoherent and irrelevant points made.

But one thing came up in their roundtable discussion that always gets my goat: the idea that neuroscience is going to be a good way of telling what effects on people different policies will have. Barbara Sahakian, a clinical neuropsychologist at Cambridge, says:

For years we changed our education system again and again, but these changes weren’t based on evidence about how we learned. Instead, wouldn’t it be useful if we thought about how the brain really works, and how children learn best, and in turn formulated educational policy based on that?

And the RSA’s Matthew Taylor adds, in a similar but more nakedly political vein:

I am confident that, as we find out more about our brains, it will strengthen the progressive case, in the sense that children learn best when they are actively involved, not being passive.

No, no, no.

Think about it: how could you use neuroscience to tell which teaching methods promote the best learning?

Well, what you’d do is get a load of kids, try out different teaching methods on them and then test how much they’ve learned, and give them brain scans to see what neurological changes have taken place. This should let you identify the neural correlates of learning. So then, in the future, you can get another load of kids, try out another bunch of teaching methods on them, and use brain scans to see what the results are.

The only weakness here is that the brain scans are completely redundant. You’ve already got ways of testing how much children have learned – and however fallible these might be, bear in mind that the neurological method you develop is of necessity based on these already existing methods for its validity.

So, unless brain scans become quicker, easier and cheaper to administer than pencil-and-paper exams, there’s nothing being added here. Yet very many intelligent people are still drawn to neuroscientific evidence as something so much more impressively real then mere psychological or behavioural phenomena (i.e. what we say and do and think).

As Zoe Drayson, a doctoral researcher at Bristol, explains:

Getting back to the issue of why we think neuroscience is so compelling, there have been studies done that show if you give people explanations of behaviour on a purely psychological level, and then you add a bogus additional neurological explanation which is logically irrelevant, people still think that explanation is better. That’s not to say that there are no good neuroscientific explanations, but it does mean we must be careful.

As I say, I find neuroscience compelling. But I think its main practical benefits are going to be in medicine, not policy design.

And the reason there’s so little progress in the popular educational policy debate is that teaching is something we’re very ideological about while also thinking that our own experience and common sense have already given us the right answers. It’s not because the academic evidence that already exists isn’t of the right kind.

1 comment:

CS Clark said...

'For years we changed our education system again and again, but these changes weren’t based on evidence about how we learned.'

This reminds me somewhat of people calling for debate, or even for serious debate, when there's been a debate going on for decades - it relies on denying the existence of things that aren't going the way you want them to. Of course people have used evidence to back up their pedagogical theories. Does she think educators were just picking methods at random?

I'd also add that evaluating the outcomes of one method of learning isn't even as simple as testing at the end of the course - you can't, for example, test long-term retention, or transferable skills. That's surely been a great driver of trying too many methods, and I can't see how it'll be helped by scannning brains, and then waiting twenty years.