In their admirably broad discussion, Burkart et al. review many important distinctions in the study of human cognition, including fluid versus crystallized intelligence and domain-general versus domain-specific mechanisms. Nonetheless, by focusing on g, the authors did not acknowledge that individual aspects of human intelligence – some of which presumably evolved separately – may have been particularly important for the evolution of human intelligence. In our view, the capacity to decide when to develop and use intellectual skills is not only a crucial aspect of human intelligence, but also it may in fact be unique to human intelligence. Human metacognition of this sort was not discussed by Burkart et al.

Consider the following problem (Frederick Reference Frederick2005):

A bat and ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?

Most educated adult humans are capable of generating a response to this question intuitively and automatically (namely: 10 cents). This likely occurs through a domain-general canalization process (see “The urgency problem” in the target article, Table 1). However, the automatic response is not the correct answer (if the ball cost 10 cents, the bat would have to cost $1.10 and together they would cost $1.20 – the correct answer is 5 cents). Although the majority of people simply give the incorrect intuitive response to this question (e.g., 64.9% of University of Waterloo undergraduates; Pennycook et al. Reference Pennycook, Cheyne, Koehler and Fugelsang2016a), some are capable of answering it correctly. This exercise of intelligence requires not just the capacity to solve the problem, but also the willingness to apply effortful cognitive processing to a problem despite the presence of what initially appears to be a suitable response (Stanovich & West Reference Stanovich and West1998; Reference Stanovich and West2000). There is now a great deal of evidence that human rationality (however imperfect, see Kahneman Reference Kahneman2011; Kahneman & Frederick Reference Kahneman, Frederick, Holyoak and Morrison2005) involves not simply computational cognitive operations (i.e., g), but also algorithmic-level operations that determine the course of reasoning and decision making (see Stanovich Reference Stanovich, Evans and Frankish2009a; Reference Stanovich2009b; Reference Stanovich2011).

Moreover, recent research indicates that the propensity to think analytically as a means to override automatic responses has consequences for our everyday lives (Pennycook et al. Reference Pennycook, Fugelsang and Koehler2015b). For example, more analytic individuals have less-traditional moral values (Pennycook et al. Reference Pennycook, Cheyne, Barr, Koehler and Fugelsang2014; Royzman et al. Reference Royzman, Landy and Goodwin2014) and are less likely to hold beliefs that are religious (Gervais & Norenzayan Reference Gervais and Norenzayan2012; Pennycook et al. Reference Pennycook, Cheyne, Seli, Koehler and Fugelsang2012; Reference Pennycook, Ross, Koehler and Fugelsang2016b; Shenhav et al. Reference Shenhav, Rand and Greene2012), paranormal (Pennycook et al. Reference Pennycook, Cheyne, Seli, Koehler and Fugelsang2012), and/or conspiratorial (Swami et al. Reference Swami, Voracek, Stieger, Tran and Furnham2014). Analytic thinking disposition has also been linked with increased acceptance of science (Gervais Reference Gervais2015; Shtulman & McCallum Reference Shtulman and McCallum2014) and lowered acceptance of complementary and alternative medicine (Browne et al. Reference Browne, Thomson, Rockloff and Pennycook2015) and pseudo-profound bullshit (Pennycook et al. Reference Pennycook, Cheyne, Barr, Koehler and Fugelsang2015a). Analytic thinking can also undermine cooperation and prosociality (Rand Reference Rand2016; Rand et al. Reference Rand, Brescoll, Everett, Capraro and Barcelo2016; Rand et al. Reference Rand, Peysakhovich, Kraft-Todd, Newman, Wurzbacher, Nowak and Greene2014; Rand et al. Reference Rand, Greene and Nowak2012), as well as punishment (Grimm & Mengel Reference Grimm and Mengel2011; Halali et al. Reference Halali, Bereby-Meyer and Meiran2014; Sutter et al. Reference Sutter, Kocher and Strauß2003).

Consideration of the evolutionary dynamics of metacognition is, therefore, of key importance for understanding the evolution of human intelligence (Bear & Rand Reference Bear and Rand2016b). Recent work using formal evolutionary game theory models has begun to shed light on this issue from a theoretical perspective, both in the domains of intertemporal choice (Tomlin et al. Reference Tomlin, Rand, Ludvig and Cohen2015; Toupo et al. Reference Toupo, Strogatz, Cohen and Rand2015) and cooperation (Bear & Rand Reference Bear and Rand2016a; Bear et al. Reference Bear, Kagan and Rand2016). These models illustrate how the willingness to override intuitive responses can be favored by natural selection in settings where flexibility and planning are particularly useful, and also how complex cyclical dynamics of automatic versus controlled cognition can emerge. This growing body of theoretical work calls for empirical examination of cognitive control in nonhuman animals (e.g., MacLean et al. Reference MacLean, Hare, Nunn, Addessi, Amici, Anderson, Aureli, Baker, Bania, Barnard, Boogert, Brannon, Bray, Bray, Brent, Burkart, Call, Cantlon, Cheke, Clayton, Delgado, DiVinventi, Fujita, Herrmann, Hiramatsu, Jacobs, Jordan, Laude, Leimgruber, Messer, Moura, Ostojic, Picard, Platt, Plotnik, Range, Reader, Reddy, Sandel, Santos, Schumann, Seed, Sewall, Shaw, Slocombe, Yanjie, Takimoto, Tan, Tao, van Schaik, Viranyi, Visalberghi, Wade, Watanabe, Widness, Young, Zentall and Zhao2014; Rosati & Santos Reference Rosati and Santos2016).

Burkart et al. discuss executive functions like inhibitory control, working memory, and cognitive flexibility (sect. 1.1) and highlight the importance of “reasoning ability and behavioral flexibility” for human and nonhuman intelligence (sect. 1.1, para. 1). Thus, the human capacity for overriding intuitive outputs (such as 10 cents in the bat-and-ball problem) is clearly acknowledged. Nonetheless, treating these aspects of human cognition as other types of cognitive processes suppresses a distinction we think should be emphasized. Can humans alone decide when (or if) to initiate cognitive processes, as well as when (or if) to reflect upon their outputs? The findings highlighted previously suggest that the capacity to decide to think is a core intellectual skill that distinguishes humans from each other. We assert that this skill is also crucial to distinguishing humans from nonhuman animals.

Although we agree that the pursuit of g (and G) in nonhuman animals is worthwhile, it is not simply that the current body of work is preliminary (as the authors state). Rather, understanding the evolution of human intelligence requires a broader view of human rationality. Thus, unfortunately, we are even further from definitive conclusions than is intimated by the target article. Even if there is good evidence for g in nonhuman animals and this ultimately informs us about the evolution of cognitive skills in humans, we will still be left with major questions about how the human capacity to decide when to think (i.e., the disposition to think analytically, over and above g) can evolve.