On near and far skill transfer

Note: I wrote this as a response to a comment on one of my articles about how brain games don’t have enough evidence to really back up their claims that they improve general cognition, and it ended up getting weighty enough that I thought it merited its own (slightly edited) blog post.

Most of the confirmed effects of video games on cognition that I know of are near-transfer, mostly on a physically measurable scale. Strengthening hand-eye-coordination is a little more concrete and closer to being near transfer than far transfer, since the skill you’re training by reacting to an enemy on screen (quick response to visual stimuli) is very similar to skills you would need in other quick-response settings.

The problem with cognitive effects of games and other mental activities is that not only are they harder to measure by default, but they’re almost all examples of far transfer, or skill transfer between dissimilar domains. The existence of near transfer tasks, like playing video games and getting better at other video games, seems pretty intuitive, but extending that intuition to far transfer gets pretty hazy. It’s totally possible that playing chess does strengthen a certain part of our brain and leads to something neurologically measurable, but whether that then translates to improved performance at something like math or stock trading is difficult to figure out at best, and most studies so far haven’t found much evidence for it.

For example, here’s a recent meta-analysis on chess and music: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5724589/

So it’s really not just brain training games where the skills are non-transferable, it’s pretty much any cognitive activity. Playing one instrument will probably help you play other instruments better, but maybe it won’t be as helpful for learning Chinese. The issue with brain-training games isn’t that they’re useless, it’s more that they rely on the idea that far transfer works, and the only skill they really build is peoples’ ability to get good at their games. If the skill you want to train is puzzle-solving, you’re good to go; it’ll give you puzzle training + a few other cognitive perk-ups.

General brain activity is pretty scientifically proven to be a good thing (lots of studies have found correlations between high levels of brain activity leading to lower levels of dementia, for example), so brain training games are probably a fine way to keep your brain ticking. If you’re doing brain training hoping it’ll boost your general intelligence and help you pass an exam or something in the future, though, you’re better off just studying that thing. As far as we know, one type of cognitive activity is as good as another, so you might as well just directly target the skill you want.

Actual exercise might actually be a better option for improving overall brain function and memory, though! There’s a big literature coming out these days on how especially cardiovascular activity is strongly linked to better cognition, which points again to a lot of these more general factors being mostly physical, and perhaps marginally affected by cognitive input.

Georgian teachers and Indonesian experiments

I just came across a fascinating piece on a Georgian (the country, not the state) economics blog, Tbilinomics. It won’t really come as a surprise to anyone that teachers in Georgia are chronically underpaid, as they tend to be in many countries, but what’s really surprising is maybe just how much they are underpaid. They make under 60% of the average national of 1000 GEL (372 USD) at around 580 GEL (220 USD)–less than agricultural workers, retail workers, hotel workers… you get the picture.

This salary clearly doesn’t attract the best and brightest into the profession, and it shows in the test scores: Georgia is among the lowest-scoring countries on the international PISA ranking. Scores have been going up steadily, perhaps due to some earlier pay increases for teachers (yes, it used to be worse), and a new proposal would actually see teachers earning substantially more, up to 2 or 3 times their current salary, but would this actually have an effect?

Regardless of how incentivized they are to do better, even if they quit their side job and focus on teaching, pay increases won’t automatically create more qualified teachers, and the teacher’s union in Georgia is apparently pretty powerful. It’s unlikely that pay increases would pay off in the short term, since the money would be going to existing teachers without an accompanying incentive to actually improve instruction quality. Frankly, even if they’re motivated by the pay increase, old habits die hard–education systems all over the world struggle to reform with any sort of speed simply because it’s such a highly personal issue.

And, as Eric Livny points out in the article, this has been tried before: Indonesia ran a large multi-year experiment in teacher pay raises, working with the world bank to collect and analyze the data. When all was said and done, the study found no effect. Even with a custom-designed randomized controlled trial, they couldn’t find evidence that the pay increases improved instruction quality–though the teachers were substantially happier. That sort of goes to support my intuition: people don’t necessarily change because their pay does.

But I’d argue that this finding doesn’t conclusively mean that higher teacher salaries won’t improve teacher quality and student grades in the long run. As long as teacher salaries are coupled with more access to educational resources and student support services, there could certainly be a substantial effect down the line.

Max Planck once said that “A new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it.” What he meant, of course, was that old ways are held by the old guard, and until they’re gone, the ways will just stick around. There’s not much we can do about it, and there’s not even much they can do about it–sticking to a paradigm isn’t just a psychological comfort mechanism; old ways are built on heuristics that we’ve been developing our entire lives, and wiping the slate clean isn’t so easy if you haven’t made a habit of doing it previously. 

The logic here, though, is that if we establish higher pay standards now, teaching will become a more attractive career for new entrants. Current teachers will probably stick to what they’re doing and we won’t see much improvement, but future teachers might decide to choose education because it’s a legitimately decent career option. Low salaries just don’t attract the best and brightest candidates (though PhDs seem to be an exception to this rule, particularly in the adjuncting phase), whereas a salary that’s at least average will probably attract a decent mix of above-average people who are passionate about education and people who weighed their financial options and personal aptitude and chose education over, say, nursing or fishing. I’m curious to see if the Indonesian World Bank experiment pays off by attracting more qualified candidates to job openings.

Unfortunately, this means we have to wait. Possibly a while. The old teachers will have to clear out of their jobs (perhaps this could be hastened by a retirement bonus) and the new teachers will have to go to school and get into the workforce.

That’s a lot of outlay for very little near-term benefit, though. The across-the-board pay increase might be the most politically effective way to get the teachers union and others on board, but it’s probably not the most efficient. Another scheme, such as offering different pay levels for different levels of qualification, might be more effective, followed by a gradual phase-in of higher pay across the board as average teacher quality rises.

If the increase is pushed through, however, I’d hope that it would be leveraged for its full research potential as was the case in Indonesia.