While capacities such as episodic memory, behavioral innovation, and mindreading have occupied an exceptional role in the comparative evolutionary study of the mind, it has not always been clear what unifies them. Concepts like “sophisticated,” “advanced,” or “higher” cognition are often employed and either obfuscate what they are intended to capture or traffic in confounding scala naturae style implications. These concepts have also been the source of difficult verbal disputes in comparative cognition, particularly when they are contrasted with associative learning explanations, and have plausibly contributed to deadlocking numerous research programs. By contrast, “complex cognition” as a unifying concept holds promise; it can tell us something specific about the function, development, and evolution of cognition, without leaning on vague or misleading terminology.

Committing to a rigid definition of “complex cognition” prior to substantive empirical investigation and iterative conceptual development would plausibly stifle inquiry into the major transitions in the evolution of complex cognition (Allen Reference Allen2017), which has proven to be a difficult set of phenomena to capture (Barron, Halina, & Klein Reference Barron, Halina and Klein2023). However, given the central role that the term “complex” holds in “complex cognition,” it would also be a mistake to leave it up to an intuitive interpretation. After all, the choice is not between an unspecified conception on the one hand and a highly extra-empirical one that approaches analyticity in character on the other. Tentatively defined concepts have an elucidating potential as an exploratory tool for inquiry. Given the ambiguity of current usage, this strategy also promises to ensure that an ontological conception of “complex” is being invoked rather than an epistemic one, which would use “complex” as a placeholder for an epistemic situation characterized by opacity.

Counting-based heuristics, such as simplicity and complexity, are commonly invoked as guides in the biological and cognitive sciences. For example, building on Kuhn’s (Reference Kuhn1997) discussion of the different senses of simplicity in the evaluation of theoretical virtue, Dacey (Reference Dacey2016) argues, within the context of comparative cognition, that debates around simplicity do not target simplicity simpliciter. Instead, they often involve distinct features of a target system, such as processes, energetic demands, structure, or input data, among others. This type of domain-specific conceptual clarification of what is at stake in a claim to simplicity has been genuinely progress-generating.

And the same seemingly goes for complexity. “Complex” is a multi-dimensional adjective (Sassoon Reference Sassoon2013). This means that there are multiple senses in which a given target might be more or less complex when compared to another, and increased complexity can entail an increase in the number of tokens of something, the types of things, or both. In order to interpret a system as complex in a sense that bears on the evolution of cognition, we need to know what to quantify over, what to compare it to, and why it is relevant.

Coombs & Trestman (Reference Coombs and Trestman2024) implicitly center complexity as a feature of traits and claim that it is of central importance. They argue that as pivotal traits are acquired, cognitive complexity increases in a biological system. The difficulty lies in the fact that there are multiple senses of “complex” implicitly at play here, and increased complexity in one dimension could be a path to simplicity on another, or vice versa. Complexity plausibly increases and/or decreases along different features of a system in evolutionarily non-linear ways. What is important for the account is to specify why particular increases in complexity matter for the evolution of cognitively complex lineages. Admittedly, this is a difficult task, but it is one that the concept of “complex cognition” is in principle well positioned to direct.

In the context of evolutionary biology, McShea (Reference McShea2001), for example, has distinguished between horizontal complexity (cell types) and vertical complexity (structural hierarchies), and Bonner (Reference Bonner1988) has argued for a relationship between organism size and complexity, where complexity is implicitly conceived as an all-things-considered feature of a system. Moreover, given that cognition is embodied on Coombs and Trestman’s account, ecological dimensions might also feature prominently. Hinegardner,& Engleberg (Reference Hinegardner and Engelberg1983) characterize the notion in terms of the minimal description necessary to capture the system’s behavior, and Elliott-Graves (Reference Elliott-Graves2018) counts heterogenous parts when determining complexity.

While these are just a few of the features that might be quantified over, the local precisification of complexity within a larger multi-dimensional space is potentially progress-generating. This is not just because it will allow for verbal disputes to be avoided. By iteratively specifying a local notion of complexity within a multi-dimensional graded space, the causal structure of the phenomenon of interest can be better understood; more specifically, what capacities such as episodic memory, behavioral innovation, flexibility, and mindreading have in common, how they evolved and develop, and how complexity bears on this. Again, this does not require dogmatically committing to a particular concept in advance.

Still, this proposal does not come without its own difficulties. Accepting the multi-dimensional nature of the adjective “complex” might reveal that all-things-considered comparative evaluations necessarily employ disparate scale types that cannot be aggregated (Hedden & Nebel Reference Hedden and Nebel2024) or that they make use of concepts that are necessarily incommensurable. This would imply that while a system might be more complex than another along one dimension, it might also be less complex than that system along another dimension in a way that makes determining which of the two is all-things-considered more complex unintelligible. It is not obvious that an appeal to relevance will be enough to resolve this problem, and results from social choice theory seem to provide reason to be skeptical that it will be easily resolved (Khmelnitskaya Reference Khmelnitskaya, Van Deemen and Rusinowska2010). If this is the case, it could turn out that the role of complexity in the major transitions of cognitive evolution is necessarily underdetermined. However, even so, we will be in a better epistemic situation after a substantial empirical attempt to identify what makes complexity important in this context.