Burkart et al. conclude that many or all vertebrate minds consist of two sets of modular skills: one hardwired and the other ontogenetically constructed – primary and secondary skills respectively. They seem to define general intelligence operationally in terms of its facilitatory role in the process by which an organism obtains secondary skills. If this is correct, the next step should be to identify and understand the neurobiological mechanisms that underlie this process. Studying the evolution of general intelligence in terms of the evolutionary history of its constituent neural structures should provide valuable direction to the comparative research recommended by Burkart et al. Here, we offer a mechanistic explanation for these two skill types and why they could be indistinguishable from one another; this centers around the basal ganglia, a set of subcortical nuclei that are good candidate structures for the foundations of general intelligence. In particular, learning and automatizing secondary skills appears to be what cognitive neuroscientists describe as habit formation, a process reliant on basal ganglia circuitry (Smith & Graybiel Reference Smith and Graybiel2016).

Certain aspects of almost any ecological context cannot be directly addressed by natural selection simply because those aspects are too transitory to support intergenerational selection, which is required for specific adaptive solutions to evolve. Although inaccessible to evolved primary skills, some transitory ecological factors nevertheless remain stable for significant portions of an individual organism's life span – the specific geographical features surrounding a given animal's den, for example, or the physical layout of the controls in a personal automobile (and the associated actions required to operate those controls). Transitory stability is itself an ecological factor that is susceptible to adaptive exploitation via natural selection. Species that evolved the general capacity to identify, learn, and exploit instances of transitory ecological stability will have obtained organismal objectives more successfully and more efficiently than species lacking this capacity. Secondary skills as described by Burkart et al. are the exploitative products of such a capacity; that is, although evolved primary skills exploit the stability of specific perennial factors, the capacity to form secondary skills exploits transitory stability as a perennially general characteristic of temporary factors (Nordli Reference Nordli2012). From this perspective, as primary and secondary skills have each arisen to exploit ecological stability (whether long-term or transitory), it is unsurprising that their shared properties – speed, efficiency, automaticity – render them effectively indistinguishable (absent knowledge of their ontogeny), as Burkart et al. point out.

Supporting this perspective, research suggests that primary and secondary skills are each encoded within basal ganglia-based circuits (Graybiel Reference Graybiel1995), such as the fixed sequence of grooming behaviors that is ubiquitously exhibited in rats (Aldridge et al. Reference Aldridge, Berridge and Rosen2004) and learned paths rats take in a maze (Barnes et al. Reference Barnes, Kubota, Hu, Jin and Graybiel2005). These nuclei also appear to be integral to goal-directed action selection, stringing behaviors together in service of achieving contextualized reinforcement (Graybiel Reference Graybiel2008). As a sequence of reinforced behaviors (e.g., a maze path) is repeated within a specific context (e.g., a chocolate reward at the end), the entire string is encoded within the basal ganglia as a single behavioral “chunk” that then exhibits quick and efficient cue-based automaticity (Jin et al. Reference Jin, Tecuapetla and Costa2014). The basal ganglia contribute centrally to cognition as well, through working memory, attention, decision making, and other processes (Stocco et al. Reference Stocco, Lebiere and Anderson2010). In this context, it is intriguing that mice with a humanized version of their Foxp2 gene – the so-called “language gene” – develop neurons in the basal ganglia with increased plasticity and significantly lengthened dendrites (Enard et al. Reference Enard, Gehre, Hammerschmidt, Hölter, Blass, Somel, Brückner, Schreiweis, Winter, Sohr, Becker, Wiebe, Nickel, Giger, Müller, Groszer, Adler, Aguilar, Bolle, Calzada-Wack, Dalke, Ehrhardt, Favor, Fuchs, Gailus-Durner, Hans, Hölzlwimmer, Javaheri, Kalaydjiev, Kallnik, Kling, Kunder, Mossbrugger, Naton, Racz, Rathkolb, Rozman, Schrewe, Busch, Graw, Ivandic, Klingenspor, Klopstock, Ollert, Quintanilla-Martinez, Schulz, Wolf, Wurst, Zimmer, Fisher, Morgenstern, Arendt, de Angelis, Fischer, Schwarz and Pääbo2009), and also exhibit accelerated habit learning relative to normal mice (Schreiweis et al. Reference Schreiweis, Bornschein, Burguière, Kerimoglu, Schreiter, Dannemann, Goyal, Rea, French, Puliyadi, Groszer, Fisher, Mundry, Winter, Hevers, Pääbo, Enard and Graybiel2014).

If general intelligence is the set of processes that allow organisms to discover, learn, and automatize secondary skills, the basal ganglia may be largely responsible for much of what qualifies as intelligence: (1) modulating rewards to direct or “canalize” attention and motivate goals (e.g., inclining to attend toward and imitate conspecifics); (2) exploring a potential action space and achieving targeted objectives by selecting goal-directed behaviors (e.g., practicing/refining an approximation of what conspecifics do); and (3) automatizing contextualized behavioral sequences that have been repeatedly reinforced (e.g., reproducing efficient skill behavior that may now be imitated by others). This is consistent with the cultural intelligence perspective advocated by Burkart et al., but the coevolutionary enhancement of social learning and general intelligence does not itself explain the evolutionary origins of general intelligence. Instead, we should look to basal ganglia circuitry, which is functionally conserved across all vertebrate species, and which likely evolved over 560 million years ago (Reiner Reference Reiner, Steiner and Tseng2010; Stephenson-Jones et al. Reference Stephenson-Jones, Samuelsson, Ericsson, Robertson and Grillner2011).

The most basic function of basal ganglia circuitry is adaptive motor control, directing goal-oriented motor sequences (Grillner et al. Reference Grillner, Robertson and Stephenson-Jones2013). Energy demands and reproductive success are the main fitness pressures, and most organisms adapt to that pressure by moving about through space – foraging for food and mates – as efficiently and effectively as possible (see Stephens & Krebs Reference Stephens and Krebs1986); competition over limited resources in these domains likely resulted in an evolutionary arms race. The capacity to automatize stereotyped patterns of learned motor behaviors into secondary skills is a powerful weapon in that war, enabling the execution of learned motor patterns with the same speed, efficiency, and specificity of evolved motor patterns. We suspect that general intelligence initially coevolved with mechanisms of adaptive motor control to facilitate the search for and learning of new adaptive motor skills. This intelligent search capacity may subsequently have been generalized through exaptation to facilitate the search for new adaptive cognitive skills: spatial foraging and searching through memory space appear to be expressions of the same general exploratory capacity, the goal-directed nature of which is modulated by basal ganglia circuitry via the dopaminergic reward system (Hills et al. Reference Hills, Todd and Goldstone2008; Hills et al. Reference Hills, Todd, Lazer, Redish, Couzin, Bateson, Cools, Dukas, Giraldeau, Macy, Page, Shiffrin, Stephens and Wolfe2015). Comparative investigations of interspecies differences in basal ganglia-based circuitry may provide further clues regarding the evolution of general intelligence; future studies should pay attention to these structures and the mechanisms of habit formation to which they contribute.