Statistical Worries in Cognitive Enhancement

Cognitive Enhancement is a growing, vibrant research topic. The excitement surrounding the field of Cognitive enhancement is enormous, but not without contention, statistically and ethically. In our first article, I will illustrate key statistical pitfalls that cognitive enhancement research fall into, which are also applicable to other disciplines of psychology and beyond. These are 4 misuses of statistics that you should bear in mind while reading or analysing papers on cognitive enhancement.

1. Regression to the Mean

Performance is never ‘ordinary’ one day to another according to statistics, participants in experiments might vary in performance one day compared to another, corrected around a mean.

A classic example is from Kahneman’s observations in the Israeli Air Force.

The observation that the army had was that when Israeli Air Officer extending praise to subordinates, this led to worse performance. However, scolding led to improvement in subsequent efforts; hence officers should more generous with negative feedback and yell at more subordinates.

But this could be a classic example of regression to the mean, it might not be the scolding that helped with the improvement, but rather change in performance might occur naturally, and nothing can be proven about feedback given by officer.

We need to have a control group to exclude what we see is a regression to the mean, and in the long term perhaps they will regress towards the mean and we can compare the two groups. Regression to the mean is simply unavoidable provided that the correlation is not perfect.