The greatly increased variance, but only somewhat increased mean, is consistent with nicotine operating on me with an inverted U-curve for dosage/performance (or the Yerkes-Dodson law): on good days, 1mg nicotine is too much and degrades performance (perhaps I am overstimulated and find it hard to focus on something as boring as n-back) while on bad days, nicotine is just right and improves n-back performance.

In avoiding experimenting with more Russian Noopept pills and using instead the easily-purchased powder form of Noopept, there are two opposing considerations: Russian Noopept is reportedly the best, so we might expect anything I buy online to be weaker or impure or inferior somehow and the effect size smaller than in the pilot experiment; but by buying my own supply & using powder I can double or triple the dose to 20mg or 30mg (to compensate for the original under-dosing of 10mg) and so the effect size larger than in the pilot experiment.

Today was the first day that I tried this, and it definitely works as far as what the description for the product says. I am studying for a very important exam and I thought judging by the reviews left by previous users that this would be something worth trying, and I totally agree. Its a great substitute if you don't like the feeling of adderrall, which for me I didn't like because my heart would be racing and I couldn't sleep, and just overall was irritable. With this product you get the focus you need and youre mentally ready for what task needs to be done. I will continue to take it and will write another review on an update after how I feel after this. The only thing is I would really appreciate if this product was FDA approved and researched more.

Or in other words, since the standard deviation of my previous self-ratings is 0.75 (see the Weather and my productivity data), a mean rating increase of >0.39 on the self-rating. This is, unfortunately, implying an extreme shift in my self-assessments (for example, 3s are ~50% of the self-ratings and 4s ~25%; to cause an increase of 0.25 while leaving 2s alone in a sample of 23 days, one would have to push 3s down to ~25% and 4s up to ~47%). So in advance, we can see that the weak plausible effects for Noopept are not going to be detected here at our usual statistical levels with just the sample I have (a more plausible experiment might use 178 pairs over a year, detecting down to d>=0.18). But if the sign is right, it might make Noopept worthwhile to investigate further. And the hardest part of this was just making the pills, so it’s not a waste of effort.


Ngo has experimented with piracetam himself (“The first time I tried it, I thought, ‘Wow, this is pretty strong for a supplement.’ I had a little bit of reflux, heartburn, but in general it was a cognitive enhancer. . . . I found it helpful”) and the neurotransmitter DMEA (“You have an idea, it helps you finish the thought. It’s for when people have difficulty finishing that last connection in the brain”).

As you may or may not know, curcumin has become a darling of the nutrition world in the last several years, thanks to a flurry of research that indicates the turmeric derivative can do everything from support the brain to reduce painful body-wide inflammation to even support positive mood. You can learn more about the research behind curcumin here:

So with these 8 results in hand, what do I think? Roughly, I was right 5 of the days and wrong 3 of them. If not for the sleep effect on #4, which is - in a way - cheating (one hopes to detect modafinil due to good effects), the ratio would be 5:4 which is awfully close to a coin-flip. Indeed, a scoring rule ranks my performance at almost identical to a coin flip: -5.49 vs -5.5420. (The bright side is that I didn’t do worse than a coin flip: I was at least calibrated.)


All of the coefficients are positive, as one would hope, and one specific factor (MR7) squeaks in at d=0.34 (p=0.05). The graph is much less impressive than the graph for just MP, suggesting that the correlation may be spread out over a lot of factors, the current dataset isn’t doing a good job of capturing the effect compared to the MP self-rating, or it really was a placebo effect:

Methylphenidate – a benzylpiperidine that had cognitive effects (e.g., working memory, episodic memory, and inhibitory control, aspects of attention, and planning latency) in healthy people.[21][22][23] It also may improve task saliency and performance on tedious tasks.[25] At above optimal doses, methylphenidate had off–target effects that decreased learning.[26]
×