The potential for multi-level phylogenetic meta-analyses in studies of animal coordination

Abstract

A primary goal of studies of the evolution of cognition and behaviour is to assess how cognitive capacities and behaviours have evolved using data from various studies of different species. One promising method to achieve such a goal is to use phylogenetic and meta-analytic statistical models to collate and assess data at individual, group and species levels. This will allow researchers to ascertain the effects of shared inheritance vs. independent adaptation to similar selective pressures on the evolution of a specific cognitive capacity, whilst accounting for differences at both the individual and group levels. However, there are several issues which make such a goal difficult to achieve at present. To name three specific issues: studies may be too distinct in design to be truly comparable; they may use measures which do not measure the same thing; and they may not make their datasets and analyses transparent and available enough to make certain that these data can be included in a large scale statistical model such as the one mentioned above. The present thesis determines to what extent studies in the field face these issues, and how the field may need to adjust in order to meet these pressing challenges. I use studies of animal coordination as a case study for how the scientific process is affected by these issues. Firstly, chapter 1

provides a detailed history and review of the relevant literature, from theory relating to the evolution of cooperation, to the cognitive perspective on cooperation and coordination, to the state of the field at the time of writing. In this chapter, I also introduce the relevant methodological issues facing comparative cognition and some potential remedies for these problems emerging in the literature. Chapter 2 builds a dataset composed of studies that are comparable enough to each other for meta-analysis, and assesses the quality of that dataset. The study builds a small dataset comparable enough to ask some meta-analytic questions, although the dataset remains too small to ask any questions of evolutionary interest. In chapter 3, I use human children as a model species to collect new data that determines whether different observational measures of social relationship quality are comparable in a study of behaviour coordination. I also assess whether methods used in developmental psychology are comparable to those used in comparative psychology and animal behaviour. Ultimately, the data suggest that measurements assessed do - to an extent - measure the same things, and methods used by developmental psychologists turn out to be the most predictive of decision making in young children. However, much more observational data will be required to test these hypotheses further. Finally, chapter 4 presents an assessment of the prevalence of data and analysis code sharing practices in the field of animal coordination. The study finds that data sharing is low, and sharing of analysis code is even lower. However, there is some evidence that such practices are increasing. Ultimately, studies of animal coordination lack the comparability and data sharing practices to make large scale phylogenetic analyses possible. This makes evolutionary questions difficult to

assess with currently available data. However, initiatives such as the Many Primates, Many Birds, and Many Dogs projects are making such analyses more plausible. Studies from these projects are already turning up some interesting results. Comparative cognition as a whole faces the methodological issues presented in this thesis, and this thesis can guide researchers towards resolving these issues