Obviously there's no way to conduct a high quality controlled experiment and eliminate bias. But you can play around with adding or removing certain foods from your diet. Look for correlations with how you subjectively feel, and objectively perform. You might find some effects that are large enough to be significant even with n=1. Or not.