People are attracted to attempts to generally boost our cognitive abilities for obvious reasons. Traditionally, the way to improve at mathematics or piano playing or writing essays has been to pair teaching of the relevant knowledge and skills with plenty of practice. Such a plan is painstaking and at the end of it, those with a purely instrumental view of education are likely to point out that there is no guarantee that this particular ability is one that potential employers will be looking for in the future.
It is this reasoning that leads to attempts to try to teach more generally applicable abilities such as ‘critical thinking’ or ‘learning how to learn’. Unfortunately, when we examine the specifics of such notionally generic abilities, we find that they are embedded in subject knowledge. You can think critically about something you are knowledgeable about but, bar a few moderately useful heuristics, little of this will help you to think critically about are area you are less familiar with. This is why we become frustrated when scientists or engineers or actors or singers or entrepreneurs tell teachers how to improve education. And this is, of course, why these amateurs tend to wear their ignorance like a pink ten-gallon hat.
One approach to generally improving cognitive abilities is inspired by the field of intelligence research. Intelligence researchers note that all cognitive abilities tend to be closely correlated with each other and so they suggest the existence of some underlying factor, G, or ‘general intelligence’. This can be further divided into Gf or ‘fluid intelligence’ and Gc or ‘crystallised intelligence’. Fluid intelligence is something like raw processing power – how quickly an individual can pick-up and use new ideas. Crystallised intelligence, in contrast, is the ability to use prior knowledge to bring to bear on problems, such as the knowledge of relevant strategies or solutions that have worked in the past.
Fluid intelligence, in turn, seems to be correlated with working memory capacity and so researchers have investigated the possibility of training the working memory – a bit like exercising a muscle – in order to produce a general improvement in fluid intelligence. At best, this looks like intelligence researchers conducting controlled studies with fully informed participants and at worst this looks like a sham therapy marketed by quacks to parents of children who are struggling to learn to read or do mathematics.
A new paper sheds more light on this approach. It does not contradict the now familiar finding that such training fails to improve fluid intelligence. Typically, subjects become really good at the task they complete as part of the training, but this does not lead to much if any discernible improvement in the ability to carry out alternative tasks. It does not transfer.
However, in the new study, researchers have investigated why this may be the case.
One suggestion is that as we practise the training task, we shift the approach that we use to complete it. Familiarity with the type of task means that some elements become more predictable and we can develop strategies to address these elements. Given that humans are predisposed to minimise cognitive effort, a shift from approaches that rely heavily on working memory to alternative strategies may be inevitable.
Unfortunately, these strategies are specific to the training task and that is why they do not result in transfer.
The researchers predicted that if such a shift occurred then, over time, the ability to complete the training task would become less dependent on fluid intelligence and so the correlation between participants’ fluid intelligence scores and their task performance would decrease.
Oddly, this did not happen. If anything, the correlation between fluid intelligence and task performance slightly increased.
How do we square this? How come improvements in these training tasks do not transfer to other tasks and yet they do seem to be closely related to fluid intelligence?
The researchers discuss a number of post hoc hypotheses. Perhaps fluid intelligence and working memory capacity are not the same thing and are instead both affected by some other factor such as the ability to create new connections between neurons? If so, improving working memory would not lead to an improvement in fluid intelligence. It would be like lifting weights with your right arm and expecting your left arm to grow stronger.
Another possibility is that individuals with higher fluid intelligence are better at noticing and applying task specific strategies – they are better at discovery learning. If this is the case then this represents yet another Matthew Effect, where those who are already cognitively wealthy get richer while those who are not get poorer. This seems to be a common motif in attempts to develop generic skills.
As a teacher, I cannot help wondering what would have happened if participants had all been taught these strategies. Would the link to fluid intelligence still have been so strong?
Because I think we are ideologically blind to at least fifty percent of the equation here – what about crystallised intelligence? No, we may not be able to wave a magic wand and improve people’s fluid intelligence, but we have developed highly efficient systems for growing their crystallised intelligence, with schools being one such system.
Instead of barking up the wrong tree of trying to improve fluid intelligence, perhaps we should focus on improving our approach to teaching a broad base of knowledge about the world.