Follies of the "One Rat Per Scientist" Researchers

The recent junk science paper "Hippocampal theta sweeps indicate goal direction during navigation" is another joke of a neuroscience research paper, one in which the main study group size is the same size as the number of authors, which should make us laugh hard and say: "What was the rule here: only one rat per researcher?"

We have in this study a study group size of only five rats, which is at least 300% too small for any reliable result to be claimed. No control group was used (an error as bad as the use of a way-too-small study group). The number of paper authors is five. So it is very definitely a "one rat per scientist" situation. 

You can do a Google image search for "Effect size versus sample size" to get an idea about the relation between the two, and another thing called statistical power. Some of the visuals you will see may be hard to understand. The visual below describes the situation in an easy to understand way. 

The smaller the effect size, the larger the study group size needed to show something in a convincing manner (such as a statistical power of 80%).  Most effect sizes in neuroscience are small.  That means in the great majority of cases a study group of 25 or more is needed. No experimental neuroscience study should be taken seriously if it uses a smallest study group size smaller than 15 subjects. It is conceivable that a neuroscience experimental study using 15 or 20 subjects might provide modest evidence for something, but only in the very unlikely case of a high effect size. For the much more likely case of an effect size that is only medium or low, then at least about 35 subjects per study group are needed. 

As a general rule, we should be dismissing as junk science any experimental neuroscience study that fails to either use at least 15 or 20 subjects per study group or fails to do a sample size calculation to determine whether the number of subjects was adequate to achieve a good statistical power such as 80%. If the smallest study group size is between 15 and 30, we should only regard the paper as being possibly modest evidence for something if the authors did a sample size calculation to show that with the effect size of the type they are dealing with, the study group size they used was adequate to achieve a good statistical power. 

The authors of this one-rat-per-scientist paper confess, "No statistical methods were employed for determining the sample size for the experimental data." A sample size calculation would have shown how way-too-small the study group size was. 

The paper is being promoted as evidence of goal representations in the brain. It is no such thing, because of its severe defects discussed above.  What went on was that brain waves were read from five moving rats who were moving toward some goal.  It is well-known that muscle movement produces all kinds of severe contamination of brain-wave readings (which is why when a neurologist takes an EEG reading), you are told to stay very still). The authors have analyzed the data, and claimed to have found something they think is evidence of goals held in the brains of the rats. This is just "see what you want to see" pareidolia. like someone seeing the face of Jesus in his toast. Nowadays neuroscientists are very abundantly making groundless uses of the words "represent" and "representation." The situation has got so bad (with so very many groundless uses of those words by neuroscientists) that it is now a good general rule of thumb to never trust a neuroscientist researcher using the word "represent" or "representation," unless he produces a best-practices experimental study of a type that would be the "quality opposite" of the study I have discussed.