Health knowledge made personal
Join this community!
› Share page:
Search posts:

A Self-Experimental Near-Miss

Posted Dec 29 2008 10:00pm

I am developing tests to measure how well my brain is working. Brief tests I can use daily. My experiences with flaxseed oil make me suspect that sometimes our brains work better and/or worse than usual for many hours or days at a time and this goes unnoticed. If these instances of better or worse function could be detected, maybe we could figure out their causes — and thereby improve how well our brains work by getting less bad stuff and more good stuff. In the case of flaxseed oil, I noticed that one morning my balance was much better than usual. I noticed this only because I was doing something unusual: putting on my shoes standing on one foot. I verified that observation with a better test of balance and later found that flaxseed oil improved my performance on several mental tests, such as speeded arithmetic.
One test I am using is a typing test: On each trial I type a random sequence of six letters four times. For example, if the sequence is “rksocn” I would type rksocnrksocnrksocnrksocn”. At the moment the measure of performance is how fast I type the 24-letter sequence. Each session consists of ten trials.

Here are the results so far:

old analysis

Each point is a mean over the ten trials; the error bars show standard errors. I was glad to see there was little sign of learning after the first few sessions. Having to correct for learning would make comparison of different days more difficult.

Because I am collecting a lot of data, I could look at these data more carefully. It took me a little while to do an analysis where I corrected for the difficulty of each string: Some will be easier to type than others. My first attempt at correction involved adding a factor for each letter: does the string contain an “a” (factor 1)? Does the string contain a “b” (factor 2)? And so on. This correction made a big difference: The residual mean square was almost cut in half (= sensitivity was doubled). After correcting for this, I got new estimates and standard errors for each test session:

new analysis

Uh-oh! The new analysis revealed there had been something unusual about the second-to-last test session — my typing had been distinctly slower than usual. Something I ate? Unfortunately, by the time I did this analysis I could no longer remember what might have been different.

Post a comment
Write a comment:

Related Searches