Increasing IQ with Synesthesia
I found this paper: Adults Can Be Trained to Acquire Synesthetic Experiences.
The goal of the study was to see if they could induce synesthesia artificially by forcing people to associate letters with colors. But the interesting part is that after 9 weeks of training, the participants gained 12 IQ points. I have read that increasing IQ is really difficult, and effect sizes this large are unheard of. So I found this really surprising, especially since it doesn't seem to have gotten a lot of attention.
EDIT: This is a Cattell Culture Fair IQ which uses 24 points as a standard deviation instead of 15. So it's more like 7.5 IQ points.
They made each participant do 30 minutes of training every day of 9 weeks, which involved a few different tasks to try to form associations between colors and letters. They also assigned colored reading material to read at home.
They took IQ tests before and after and gained 12 IQ points after the training. A control group also took the tests before and after but did not receive training, and did not improve. The sample sizes are small, but the effect sizes might be large enough to justify it. They give a p value of 0.008.
In the paper there are some quotes from subjects, and they describe thinking about words visually. E.g. ‘‘I see the colors like on a monitor in my head and its very automatic’’ or ‘‘The color immediately pops into my head… When I look at a sign the whole word appears colored according to the training colors… it is just as automatic for single letters’’.
I speculate that this might be the cause of the effect, something about using more of the visual system when thinking. That's just weak speculation though.
I tried to do some more research to see if there was any correlation between synesthesia and IQ. I did not expect there to be, but perhaps it does correlate. This paper suggests it might:
In addition, a neuropsychological test battery was employed, in which all subjects performed superior on tests of general intelligence (mean IQ = 120 ± 17) [out of 9 subjects]
The data from this study shows 10 synesthetes had the same average IQ scores as the controls (but greater standard deviation if that means anything.)
Same story with this study of 10 female synesthetes:
The two subject groups were matched for... IQ, as assessed by the Multiple Choice Vocabulary Test version B... synesthetes = 117 ± 10.2, controls = 116 ± 13.2.
But on second look, it looks like the last two studies intentionally selected the control group to have the same IQs to avoid confounders. If that's the case then it does support the hypothesis as the reported IQ is greater than average.
Here is another study with more of the same:
Control subjects (n = 14) and synesthesia subjects (n = 14) were matched for age..., gender..., and general intelligence (IQ values for synesthetes: 119 ± 13 and controls: 112 ± 17) as assessed by the MWT-B – “Mehrfach–Wortschatz Test” (Lehrl et al., 1995).
So now I want to try the experiment on myself. I'm considering how to do this. I want to make some kind of tool or browser extension that could color text to match the desired associations. I want to know if it would be better to try letter level associations or word level ones.
I think that word level coloring would be more semantically meaningful and therefore likely to help. But the paper used letter coloring. Most of the subjects in those papers reportedly had grapheme–color synesthesia. They weren't very specific on the details, or I didn't look too closely.
Second whether to just use random colors, or try to assign them meaningfully. Like grouping nouns together, or using something like word2vec to find semantically similar words and optimize them to be close in color space if possible. If I do that it's more complicated and there are a lot of technical decisions to make.
And then how to actually color text in a readable way. Perhaps limiting the color space to what can be read on a white background, or somehow outlining the letters.
EDIT: I found a chrome extension that has some of these features. Only does letter level associations. And the source is available!
Discussion on lesswrong.
