I agree with the authors' basic thesis, depicted in the target article's Figure 3, which argues for combinations of heritable and learned abilities that result in general intellectual achievements, whether in humans or nonhumans. However, in their discussion of nonhuman subjects, the emphasis on nonhuman primate, and to a lesser extent mammalian, species is discouraging. The few references to avian cognition do not do justice to the wide variety of abilities and studies – sometimes in a single individual or species – that provide evidence for generalized intelligence. Thus, the point of my commentary is to advocate strongly for avian g.

I do, of course, understand that formal g analyses (i.e., batteries of different tests on numerous individuals within and across species) are lacking for avian cognitive capacities, as compared to the several existent analyses on nonhuman primates and rodents. Nevertheless, I hoped that Burkart et al. would have discussed the large number of studies on a wide variety of topics performed on avian species, particularly on corvids and psittacids … and maybe would have attempted some kind of review, if not analysis, of their own. For example, early in the article the authors suggest that transfer of knowledge from one domain to another novel context provides evidence for g, yet little discussion exists of instances of such behavior in avian subjects (for example, transfer of the trained use of the label “none” from describing the absence of similarity and difference of specific attributes between objects [Pepperberg Reference Pepperberg1988] to the spontaneous use of the label for describing the absence of a size differential [Pepperberg & Brezinsky Reference Pepperberg and Brezinsky1991] to the additional spontaneous use for describing the absence of a specific numerical set in a collection [a zero-like concept; Pepperberg & Gordon Reference Pepperberg and Gordon2005]).

The authors note that g has “robust correlates to brain structure and function” (sect. 1.1, para. 3), yet do not mention significant research on avian brains. Recent studies, although not always performing the correlations themselves, have shown that avian brain anatomy would correlate with many g-related abilities; see, for example, in addition to those references cited, Iwaniuk et al. (Reference Iwaniuk, Lefebvre and Wylie2009) or Jarvis et al. (Reference Jarvis, Güntürkün, Bruce, Csillag, Karten, Kuenzel, Medina, Paxinos, Perkel, Shimizu, Striedter, Wild, Ball, Dugas-Ford, Durand, Hough, Husband, Kubikova, Lee, Mello, Powers, Siang, Smulders, Wada, White, Yamamoto, Yu, Reiner and Butler2005). Specifically, Olkowicz et al. (Reference Olkowicz, Kocourek, Lučan, Porteš, Fitch, Herculano-Houzel and Němec2016) have found that parrots and corvids have forebrain neuron counts equal to or greater than primates with much larger sized brains, and argue that this finding likely explains the advanced cognitive abilities found in these avian species.

In general, the authors only partially consider parallel/convergent evolution of g with respect to avian species, again particularly concerning corvids and parrots. For example, K-selected traits (e.g., long lives, few offspring that are relatively slow to mature, etc.), are discussed at various points in the target article as being associated with the evolution of g; these traits are found in most corvid and parrot species. The authors highlight the importance of sociality in driving intelligent behavior (sensu Jolly [Reference Jolly1966] and Humphrey [Reference Humphrey, Bateson and Hinde1976]): Notably, the majority of parrot and corvid species live in complex social groups, and evidence exists for several types of learning that are enhanced via conspecific or allospecific social influences (e.g., corvids: Miller et al. Reference Miller, Schiestl, Whiten, Schwab and Bugnyar2014; New Zealand kea parrots: Heyse Reference Heyse2012). Myriad papers on corvid social cognition have been published by Bugnyar and his colleagues (e.g., Bugnyar & Heinrich Reference Bugnyar and Heinrich2006). Acquisition of referential use of human speech by Grey parrots occurs through social learning (e.g., Pepperberg Reference Pepperberg1981; Reference Pepperberg1999); similarly, “bilingual” songbirds learn the form and likely use of heterospecific vocalizations via intense social interaction (e.g., Baptista Reference Baptista1981).

Furthermore, as with nonhuman primates, the dominance hierarchies that are prevalent in social groups of corvids (e.g., Chiarati et al. Reference Chiarati, Canestrari, Vera, Marcos and Baglione2010) require an understanding of advanced cognitive processes such as individual recognition (e.g., Izawa & Watanabe Reference Izawa and Watanabe2008) and transitive inference (e.g., Paz-y-Miño et al. Reference Paz-y-Miño, Bond and Kamil2004). Granted, the references I cite involve different corvid species; nevertheless, the cognitive requirements across species would likely be quite similar given their similar ecology/ethology. And, although dominance hierarchies in parrot flocks have not been studied in the wild, hierarchies have been observed in captivity (Szabo et al. Reference Szabo, Bugnyar and Auersperg2016; parrots in my lab also exhibit a hierarchy), and understanding linear ordering can also be related to cognitive capacities such as the spontaneous comprehension of ordinality (Pepperberg Reference Pepperberg2006).

Tests used as evidence of general intelligence, even by the authors' admission, are mostly basic, but other tests, even if performed on only a limited number of subjects, strongly demonstrate advanced avian capacities. For example, evidence for executive function (planning, delayed gratification) is evident in corvids (Hillemann et al. Reference Hillemann, Bugnyar, Kotrschal and Wascher2014; Raby et al. Reference Raby, Alexis, Dickinson and Clayton2007). One cannot argue that such behavior is modularly related to caching, as success on the same tasks can be seen in parrots that do not cache (Auersperg et al. Reference Auersperg, Laumer and Bugnyar2013; Koepke et al. Reference Koepke, Gray and Pepperberg2015). Grey parrots understand not only categories (e.g., what is or is not green), but also concepts such as “color,” “shape,” and “matter” (i.e., the existence of these hierarchical concepts, under which categories such as green and wood are sorted; Pepperberg Reference Pepperberg1983) and that two objects can be related based on just a subset of these concepts; that is, for second-order concepts of same-different (Pepperberg Reference Pepperberg1987). The authors mention reasoning by exclusion: for such abilities in parrots and corvids, see Pepperberg et al. (Reference Pepperberg, Koepke, Livingston, Girard and Hartsfield2013), Schloegl (Reference Schloegl2011), and Schloegl et al. (Reference Schloegl, Dierks, Gajdon, Huber, Kotrschal and Bugnyar2009). Likewise, for advanced avian understanding of number concepts, see Smirnova (Reference Smirnova2013) and Ujfalussy et al. (Reference Ujfalussy, Miklosi, Bugnyar and Kotrschal2014); these abilities are often at a level more advanced than those shown to date for nonhuman primates (e.g., Pepperberg Reference Pepperberg2006; Pepperberg & Carey Reference Pepperberg and Carey2012). Research papers on tool use by corvids that do not use tools in nature are too numerous to mention; for aspects of physical cognition in parrots, note van Horik and Emery (Reference van Horik and Emery2016).

The authors have, essentially, performed a meta-analysis on a number of meta-analyses, and I have no arguments about their basic thesis – my criticism is merely that readers interested in this thesis, particularly readers with little knowledge of nonhuman capacities, would unfortunately be left unaware of a large number of striking avian abilities that provide considerable evidence for g.