A core claim of the target article by Christiansen & Chater (C&C) – that languages change, and that theories of language evolution must take this into account – is undoubtedly correct. Language change, or “glossogeny” (Hurford Reference Hurford and Roca1990), plays an important role in language evolution, recognized and explored by many scholars (Deacon Reference Deacon1997; Fitch Reference Fitch2005; Reference Fitch2007; Keller Reference Keller1994; Kirby Reference Kirby1999; Kirby et al. Reference Kirby, Dowman and Griffiths2007; Lieberman et al. Reference Lieberman, Michel, Jackson, Tang and Nowak2007; Pagel et al. Reference Pagel, Atkinson and Meade2007). C&C review some good reasons why. But their core assertion that “Language has been shaped by the brain, rather than vice versa” (sect. 4, para. 9) is based on a false dichotomy. The explanatory role of glossogeny is complementary to, not in competition with, that of biological evolution. The keyword is “co-evolution” (Deacon Reference Deacon1997; Dunbar Reference Dunbar1993; Durham Reference Durham1991; Feldman & Laland Reference Feldman and Laland1996). C&C's argument pervasively mixes metaphors, uses slippery terms like “language” and “Universal Grammar (UG)” chimerically and without definition, and fails as a contribution to this literature. The “logical” problem is not with the logic of evolutionary reasoning, but with C&C's understanding of it.

C&C's rejection of the possibility of biological evolution of a human-specific language capacity is grounded on a misguided rejection of what they mistake for “circular” reasoning. One of Darwin's core insights was that the biological past influences the evolutionary future. Despite the power of natural selection to tailor organisms to their environment, adaptation occurs within a limited range determined by past evolutionary history. Thus, inevitably, today's “effects” are tomorrow's “causes.” Despite the essential correctness of this way of thinking, it has been dogged since its inception by accusations of circularity and tautology (Mayr Reference Mayr1982), which periodically reappear due to a failure to fully understand the nature of biological causality (e.g., Moorhead & Kaplan Reference Moorhead and Kaplan1967; Peters Reference Peters1976; Stebbins Reference Stebbins1977). C&C's “logical problem of language evolution” (target article, Abstract) provides another example of this failure, addressing problems easily solved with standard evolutionary reasoning. But C&C rule out such solutions due to their supposedly “obvious” circularity. Like many previous critics of evolutionary theory, C&C have fallen into the “linearity trap”: thinking that the only logically permissible form of causal reasoning is a one-way movement from causes to effects. Such reasoning is inadequate for understanding biological causality, in development or evolution, which is far more complex and indeterminate than this: multiple effects at one point in time become the causes of subsequent effects in a never-ending cycle (Mayr Reference Mayr1961; Reference Mayr1988). Cyclicity is a core characteristic of the phenomenon under study; and by ruling out appropriately cyclical forms of explanation under the heading of “circular reasoning,” C&C create a problem where there is none.

C&C argue that languages change so fast that any biases to learning a particular language, selected in one generation, will be unsuited to a later version of that language generations later. If this “logical problem of language evolution” is indeed as severe as C&C claim, it is not just linguists who should be worried: any biologist interested in adaptation to a rapidly changing environment (e.g., individual learning) or culturally transmitted system (e.g., bird or whale song) should be alarmed. Birdsong is a well-studied example of “animal culture” (Catchpole & Slater Reference Catchpole and Slater1995; Marler & Slabbekoorn Reference Marler and Slabbekoorn2004). Many birds learn their song, and due to accumulated errors, the songs of local populations diversify, form dialects, which gradually change over time (Baker & Cunningham Reference Baker and Cunningham1985). Modern biologists universally accept the notion that birds have an “instinct to learn” not just song in general, but their species-specific song (Gould & Marler Reference Gould and Marler1987; Marler Reference Marler, Carey and Gelman1991), an outcome that should be impossible if C&C were correct. Have all these birdsong biologists fallen into the “circularity trap”? Of course not. Understanding gene/;culture co-evolution requires more sophisticated models, and more subtle reasoning, than the simplistic options considered by C&C (cf. Boyd & Richerson Reference Boyd and Richerson1985; Marler Reference Marler, Carey and Gelman1991; Nottebohm Reference Nottebohm1972).

Consider what happens when some new culturally transmitted display arises for non-genetic reasons (e.g., a copying error) and proves popular among display recipients. For whatever reason, some members of the current population may find this cultural variant easier to master than do others; and if there is any genetic basis for this difference, this biological variant will increase in frequency in subsequent populations: the alleles responsible may experience consistent positive selection. Crucially, some types of constraints (e.g., a proclivity for specific details) will have no long-term advantage, because of rapid dialectal change. For other constraints (e.g., better sequence memory), the constraint may be abstract and general enough to apply across time, to many different chronodialects. Rapid cultural change thus selects for abstract constraints on learning, just as rapid change in the physical environment selects for “generalist” species that solve problems via plastic responses to the environment (e.g., biased learning) rather than fixed instincts (West-Eberhard Reference West-Eberhard1989). The “instinct to learn” continues to evolve biases and constraints, but for abstract features rather than details. Similarly, language change does not entail a cessation of selection. Rather, consistent positive selection occurs only for variants capable of abstracting over rapid glossogenetic variation. A proto-constraint (“UG1”) shapes proto-language, which influences further evolution of the constraint (“UG2”), which shapes later glossogeny, and so on. Far from it being “inevitable” that UG would adapt to “specific word order, phonotactic rules, inventory of phonemic distinctions, and so on,” as C&C claim (sect. 3.2.3, para. 2, emphasis theirs), language change makes this outcome extremely unlikely. This conclusion does not involve “circular reasoning,” just a clear recognition of the essentially cyclical nature of evolution itself.

In a recent chapter inquiring whether language evolution is the “hardest problem in science,” Christiansen and Kirby correctly assert that language is the “most complex behaviour we know of” (Christiansen & Kirby Reference Christiansen and Kirby2003). If so, the real traps for evolutionary approaches to this problem are oversimplification and a reliance on inappropriate models, metaphors, and preconceptions. By oversimplifying the nature of genes, evolution, the brain, and language, C&C do a disservice to the hypothesis they intend to champion. Increases in our understanding of language demand sophisticated models of the co-evolutionary interplay between ontogeny, glossogeny, and phylogeny, not a simplistic replacement of one by the other.