For instance, regarding "learning subjects". You believe that some change in eating habits has improved your ability to learn. Does your belief have a rational basis?
Subjects (let us call them "topics", to distinguish from "test subjects": people being experimented upon) are not some amorphous substance that you "pour" into your mind; they are all different. You also cannot erase a topic from your mind so you can fairly re-test the pouring after making some change, like diet or whatever.
If you have difficulty with a topic X today, you can find the same topic X easier after some passage of time simply because you were exposed to X, and let it "settle in your brain" for a while.
There is no way to to have a valid, objective test using the same person and the same topic, where we can determine that the same topic X became easier to learn due to some external factor, like not eating bread. The test is confounded by the subject's prior experience with tackling X: the "after" situation is different from "before" because of prior experience.
There is no way we can test different topics X and Y, either. If you have difficulty with learning topic X, and then some external factor changes (like not eating bread), and you have an easier time with topic Y, that doesn't conclusively show anything. The test is confounded by both similarities between X and Y (applicability of the prior experience with tackling X to tackling topic Y) and by their differences (difficulty in Y is unrelated to difficulty in X).
If you want to study this properly, you need large, random samples of people. You need control groups as well as test subjects. One group doesn't consume bread (or anything with glutent) and another does. Both learn similar topic in a similar setting. Does the test group show different progress from the control group? Different in a significant way? Even studies done at this level still have confounding problems that give us reason to be skeptical of their claimed results: such as improper sampling methods, lack of proper control over information leakage in experiments, poor accountability for the round-the-clock behavior of human subjects, and abuse of statistical methods to draw unwarranted conclusions.
Then there is the poor amount of data you have gathered (if we can call it data). How many times have you quit eating bread and observed the difference on learning or other cognitive tasks? Sounds like you've only done this once. You've done this once, and, what is more, you're biased: prior to the experiment you already believed the "gluten-brain" hypothesis already from your Internet readings, and were looking to confirm it. In honest research, even if we personally believe in a hypothesis, we must work on focus on eliminating the serious objections which tend refute the hypothesis, and not simply look for a few shreds of observations which suggest that it might be true. If we cannot eliminate the objections, we have to face the fact that the hypothesis may not be true after all.