Thanks for inviting me to review the manuscript entitled "Short-term memory, attentional control and brain size in primates" submitted to Royal Society Open Science. The authors provide a re-analyses of data published by the ManyPrimates consortium under the rationale that the original study did not test for relationships between STM and overall brain size. This, according to the authors, would help elucidate whether STM reflects a domain specific vs general cognitive ability, as expressed by its relationship to overall brain size and not any specific functional area. The manuscript presents many interesting points and discusses the problems associated with comparative analyses of brain and cognition at length. Generally, it warrants publication in your journal but I believe there are a few important points that need to be addressed before its suitability can be determined - importantly, these are related to the statistical analyses - the choice of cut-off points and data filtering and selection, choice of phylogenetic tree and analytical method without multiple testing correction. Hereby I provide feedback on some specific points:

ln 47 - missing parenthesis after 'technical [2]'

ln 75 - overall not over all

ln 127 - .to chech. ?

ln 158 - what is meant by 'females are the ecological sex'? This needs clarification. Additionally, this is given as the only rationale behind using female brain size only. I think it will be useful to either make a very convincing point for this choice ('ecological sex' does not provide any convincing argument) or alternatively, and probably better, include analyses on male, or male and female brains, too. It would be intriguing if the results are different, and it will warrant an interesting explanation which will benefit the overall quality of the manuscript.

ln 170 - 'size of their cognitive brain' needs clarification. This is non-standard terminology.

ln 191 - once introduced (much earlier in the manuscript), the abbreviation of short term memory (STM) does not need repeated explanation.

ln 194 - The choice of phylogeny with a single consensus tree from 2009 is not appropriate. The primate phylogeny has been reshuffled multiple times since (i.e. Wisniewski 2022, among many others) and the use of a single consensus tree is generally being discouraged. Given that the authors are employing MCMCglmm, they could easily run their analyses on hundreds or even thousands of trees using the mulTree package. This is needed in order to control for the uncertainty in resolving certain clades (i.e. Catarrhini) and overall, provides better control for phylogenetic relationship at the current state of primate taxonomy.

ln 207 - MCMCglm should be MCMCglmm. Additionally, have the authors used some multiple testing correction? I believe this is warranted as they perform analyses on the same dependent variable multiple times using different predictors and sample sizes. This might change the result and some of the p-values reported might change drastically.
ln 213 - given that the resulting sample size of species with more than 10 individuals was 11(!) I think it will be useful if the authors show some measure of effect size. Regression analyses on such small sample sizes are notorious for producing skewed results and/or overfitting, so I believe this step warrants additional indicators of substantiation for the observed result.

Additionally, the choice of cut-off point at N=10 seems arbitrary and the 'increasing R2' does not warrant such a cut-off point per se. Looking at Fig 2 the authors could have easily chosen a different number. How would the results look like if the authors chose 9 or 8 as a cut-off point? The fact that the R2 increases up to N=11 does not warrant using an arbitrary number as a cut off point. I believe it will be reasonable if the authors report the results with N= 8, 9 etc. or alternatively make a very strong point justifying the chosen cut-off point. I think this will inevitably strengthen the manuscript.

ln 314 - I really like the construct validity point that is made in this paragraph!

How are the confidence intervals (?) on page 17 and 19 calculated. I am not sure what are they depicting.

Regards,
Orlin S. Todorov

Dear authors,

in its current form the manuscript has improved, and besides some minor language issues that can easily be addressed during typesetting, I believe it warrants publication in RSOS.

Kind regards,
Dr Orlin S. Todorov