Leveraging brain-body scaling relationships for comparative studies

In Horschler et al. (Anim Cognit 22(2):187-198, 2019), we found that two components of executive function (short-term memory and self-control) were strongly associated with estimated absolute brain weight across dog breeds, and argued that dogs present a powerful model for studying evolutionary links between cognition and neuroanatomy due to their extraordinary degree of intraspecific morphological variation. In a commentary on this work, Montgomery (Anim Cognit, 2019) raises concerns about the practice of estimating brain weights from brain-body scaling relationships. Montgomery explores the practical significance of this approach, ultimately concluding that such estimations should be avoided. In this response, we point out some limitations of the analyses presented by Montgomery and consider his conclusions in light of these issues. We then explore the extent to which body weight serves as a valid proxy for brain weight under varying conditions. Through simulations, we show that the consequences of using body weight as a proxy for brain weight depend on parameters including effect size, the correlation between brain and body weight, and the variance in brain and body weight within a sample. Under conditions approximating those in Horschler et al. (Anim Cognit 22(2):187-198, 2019), we find that body weight is a reliable proxy for brain weight, and that statistical results from models using either brain weight or body weight as predictor variables are highly convergent. Nonetheless, we wholeheartedly agree with Montgomery that empirical data on brain weight, structure, and cellular composition will be critical for creating new opportunities to investigate the relationships between neuroanatomy and cognition in dogs.