If Evans & Levinson (E&L) are right (and we believe they are), the dominant consensus among comparative psychologists is as specious as the Universal Grammar (UG) hypothesis in linguistics.
At present, most comparative psychologists believe that the difference between human and other animal minds is “one of degree and not of kind” (Darwin Reference Darwin1871; for recent examples of this pervasive consensus, see commentaries on Penn et al. [2008] in BBS, Vol. 31, No. 2). Among researchers willing to admit that the human mind might be qualitatively different, most argue that our species' cognitive uniqueness is limited to certain domain-specific faculties, such as language and/or social-communicative intelligence.
We believe this view of the human mind is profoundly mistaken. In our opinion, the discontinuity between extant human and nonhuman minds is much broader and deeper than most researchers admit. We are happy to report that E&L's target article strongly corroborates our unpopular hypothesis; conversely, the comparative evidence bolsters E&L's provocative argument.
The case for a domain-general discontinuity
Linguists can (and should) argue about which human languages employ constituency and which do not. Or about which employ syntactic recursion and which do not. But from a comparative psychologist's perspective, the spectacular fact of the matter is that any normal human child can learn any human language, and no human language is learnable by any other extant species. Why? Why are human languages so easy for us to learn and so unthinkable for everyone else?
The standard explanation is that only humans have a “language instinct” (Pinker Reference Pinker1994), which fits nicely with the presumption that the rest of the human mind is more or less like that of any other ape (e.g., Hauser et al. Reference Hauser, Chomsky and Fitch2002). But as E&L point out, the diversity of human languages suggests that our faculty for language relies largely on domain-general cognitive systems that originally evolved for other purposes and still perform these non-linguistic functions to this day.
If E&L are right, there should be significant differences between human and nonhuman minds outside of language. Conversely, E&L's case would be in bad shape if the nonverbal cognitive abilities of nonhuman apes turned out to be highly similar to those of humans.
We recently reviewed the available comparative evidence across a number of domains – from “same-different” reasoning and Theory of Mind (ToM) to tool-use and spatial cognition (Penn et al. Reference Penn, Holyoak and Povinelli2008). Across all these disparate domains, a consistent pattern emerges: Although there is a profound similarity between the ability of human and nonhuman animals to learn about perceptually grounded relations, only humans reason in terms of higher-order structural relations. Nonhuman animals, for example, are capable of generalizing abstract rules about “same” and “different” relations in terms of the perceptual variability between stimuli. But unlike human children (Holyoak et al. Reference Holyoak, Junn and Billman1984), they appear incapable of making analogical inferences based on structural, logical, or functional relationships. Nonhuman animals are able to predict the actions of others on the basis of the past and occurrent behavior of these others. But they are incapable of forming higher-order representations of others' mental states or of understanding others' subjective experience by analogy to their own (Penn & Povinelli Reference Penn and Povinelli2007b; in press; Povinelli & Vonk Reference Povinelli and Vonk2003). Nonhuman animals can infer the first-order causal relation between observable contingencies and reason in a means-end fashion. But they show no evidence of being able to cognize the analogy between perceptually disparate causal relationships or to reason in terms of unobservable causal mechanisms (Penn & Povinelli Reference Penn and Povinelli2007a; Povinelli Reference Povinelli2000). And although nonhuman animals can learn the first-order relationship between symbols and objects in the world, they are incapable of cognizing combinatorial hierarchical schemas in any domain – causal, social, spatial, or symbolic.
So here is the central explanadum that most comparative psychologists and linguists seem to be avoiding: Why is it that the discontinuity between human and nonhuman forms of communication appears at the same degree of relational complexity as does the discontinuity between human and nonhuman cognition in every other domain?
The hypothesis that human and nonhuman minds differ only in their social-communicative intelligence (cf. Tomasello Reference Tomasello2008) does not explain why the discontinuities between human and nonhuman minds in non-communicative tasks (e.g., causal reasoning) are just as profound as those between human and nonhuman forms of communication; nor does it explain why the human learning system is the only one that can cope with the relational complexity of human language. Furthermore, the evidence from high-functioning autistic populations demonstrates that normal relational intelligence can be preserved in the absence of normal social-communicative intelligence, whereas the converse is not the case (see Morsanyi & Holyoak [in press] and references therein). This suggests that our unique social-communicative skills rely on our unique relational intelligence – not the other way around.
The hypothesis that the communicative and cognitive functions of language played an important role in rewiring the human brain makes good sense to us (Bermudez Reference Bermudez2005; Bickerton Reference Bickerton2009). And it is clear that language still enables, extends, and shapes human cognition in many profound ways (Clark Reference Clark2006; Loewenstein & Gentner Reference Loewenstein and Gentner2005; Majid et al. Reference Majid, Bowerman, Kita, Haun and Levinson2004). But the evidence from comparative psychology points to the same conclusion as does that from comparative linguistics: It is not language alone that made or makes humans so smart (see our discussion in Penn et al. Reference Penn, Holyoak and Povinelli2008). Rather, the diversity of human languages seems to have been shaped by the capabilities and limitations of the human mind (Christiansen & Chater Reference Christiansen and Chater2008). And our species' unique relational abilities seem to be a necessary precursor for the advent of even the most rudimentary human language. As Darwin himself put it: “The mental powers of some early progenitor of man must have been more highly developed than in any existing ape, before even the most imperfect form of speech could have come into use” (Darwin Reference Darwin1871, p. 57).
The model of a physical symbol system (PSS; Newell Reference Newell1980) provides a useful heuristic framework for understanding what happened to the human mind. According to our hypothesis (Penn et al. Reference Penn, Holyoak and Povinelli2008), animals of many taxa evolved the ability to represent first-order symbolic relations; but only humans evolved the ability to approximate the higher-order capabilities of a PSS. As a result, only humans can make the kind of analogical, modality-independent, role-governed inferences necessary to master the spectacular complexity of human language (Holyoak & Hummel Reference Holyoak, Hummel, Dietrich and Markman2000). This capability is not specific to language, nor did it originally evolve for language. Rather, our domain-general ability to reason about higher-order relations co-evolved with and continues to subserve all our uniquely human abilities.
An object lesson for comparative research
We cannot help but notice an unfortunate parallel between the current state of comparative psychology and generative linguistics: in both cases, researchers have gone to elaborate lengths to minimize any possible differences from English-speaking humans.
To our mind, this is a gross travesty of Darwin's legacy. Wasn't the entire point of Darwin's (Reference Darwin1859) magnum opus to show that the natural processes of variation and selection combined with “Extinction of less-improved forms” could create enormous differences between extant organisms, some so great as to constitute different species altogether? Human minds and human languages may both be bio-cultural hybrids. But, in our view, the principles of evolution that apply to bodies apply, mutatis mutandis, to minds and languages.
