Terry Deacon, Relaxed Selection, and the Evolution of Language

By Paul Mason & Daniel Lende

Terrence Deacon, professor of anthropology at Berkeley, has a 2010 paper in the Proceedings of the National Academy of Sciences, A Role for Relaxed Selection in the Evolution of the Language Capacity. It is part of an outstanding special PNAS issue “In the Light of Evolution,” which includes papers from Bernard Wood, Nina Jablonski, Kristen Hawkes, Richerson & Boyd, Tooby & Cosmides, and Steven Pinker (all full access!).

Deacon has a long-term interest in building explanations of language that combine evolutionary process with neurobiological analysis, most prominently in his excellent book The Symbolic Species. This paper focuses on the evolutionary side:

A common feature [of these evolutionary explanations] is an interplay between processes of stabilizing selection and processes of relaxed selection at different levels of organism function. These may play important roles in the many levels of evolutionary process contributing to language. Surprisingly, the relaxation of selection at the organism level may have been a source of many complex synergistic features of the human language capacity, and may help explain why so much language information is “inherited” socially.

You can hear Deacon lecture on the topic in Whence Language?, an audio file coupled to a PowerPoint, where he outlines his overall approach.

In the PNAS paper, Deacon describes natural selection as characterized by replication, variation and differential preservation, and hypothesizes that the social history of language should exhibit evolvability due to redundancy, dedifferentiation (degeneracy), and functional interdependency.

Deacon argues that as a system like language becomes more highly complex, it simultaneously becomes more functionally degraded by structural variation. In complex systems, the tight coupling of structure and function can be loosened, as selection maintains the overall function of the system even as processes like mutation introduce variation into the different components of the working system.

Complex systems also have functional redundancy, which allows for a relaxation of selective processes that provides the space for replicators to incrementally deviate from their antecedent function. The relaxation of selection helps favor the preservation of structural variants, and thus functional dedifferentiation.

According to Deacon’s reasoning, the involvement of neural circuitry and the increased importance of social transmission in the determination of vocalization could be a product of this functional dedifferentiation. The distribution of function onto an array of components effectively offloads a significant degree of genetic control onto epigenetic processes. The subsequent openness to experiential modification provides a greater degree of freedom for the influence of social transmission and the intersubjective exploration and experimentation with vocabulary and grammar.

In other words, the structural aspects of language, as they grow more complex and inter-linked, relax the process of selection, with its natural tendency to hone particular functional adaptations. In turn, this opens up new evolutionary spaces for the evolution of complexity. In the case of language, this relaxed selection opens up language to greater epigenetic influence and social and experiential learning.

It is a complicated but compelling argument, so please do check out the full paper (pdf), A role for relaxed selection in the evolution of the language capacity.

Deacon’s lecture Whence Language?

Deacon also has a piece over at On the Human, Rethinking the natural selection of human language.

Related Posts Plugin for WordPress, Blogger...
This entry was posted in Evolution, Language. Bookmark the permalink.

19 Responses to Terry Deacon, Relaxed Selection, and the Evolution of Language

  1. Pingback: Deacon featured on PLoS Neuroanthropology « Neuroanthropology

  2. the structural aspects of language, as they grow more complex and inter-linked, relax the process of selection, with its natural tendency to hone particular functional adaptations.

    I’m probably misreading your post, but I’m sure this is the other way around: as selection is relaxed, this allowed language to grow more complex. It’s not that language got more complex and then relaxed the selection pressures. Looking at the Finches, for example, Deacon points to how selection was relaxed when they were domesticated, in that there was no pressure to maintain rigidity in their mating song, which then resulted in functional complexity.

    • daniel.lende says:

      James, I think selection could work both ways. Each scenario is possible – systemic complexity, allowing for structural variants due to relaxed selection on the components of the system; or changed selection pressures, such as domestication, which provides for functional complexity, since there was less need for functional specialization.

      I think Deacon wants to invoke both systemic complexity and changed selection pressures in a two-part argument. In his abstract he writes, “the relaxation of selection at the organism level may have been a source of many complex synergistic features of the human language capacity.” So a relaxation at the organism level opened up structural variation; the domestication of people then permitted more synergistic features to evolve in language. At least that’s how I read it.

      I emailed Deacon, and he didn’t object to what I wrote, calling it a “nice capsule version of the argument.” In thinking about it more now, Deacon invoked the Baldwin effect in The Symbolic Species, that cloud of variants that can better explore an adaptive space. Here he’s trying to come up with a more robust explanation for how something like the Baldwin effect could come about.

      In the case of language, it can’t be only the relaxation of selection and more functional complexity, otherwise we’d probably see more complex communication systems out there. It’s also the organism level versus the component/genetic level, and the ability to have structural redunancy and variation that can then undergird new forms to communicate based on social learning and neural encoding. Something like that, I think. What are your thoughts?

      • Cheers for the clarification. I’m currently listening to the Whence Language? lecture to refresh my memory on his arguments.

        On your last point: I totally agree that it can’t simply be the relaxation of selection. Relaxation simply allowed the processes of dedifferentiation and redistribution to shift from an innate, localised function onto a more distributed array of systems. That is, we shifted towards a greater degree of developmental flexibility in language, previously not available to us as a species, which was then honed by natural selection. This then leads onto Deacon’s earlier arguments about our brain being adapted to the processing of language, if not in an innate capacity for arbitrary linguistic features, with language itself also exhibiting an evolutionary dynamic: adapting to become increasingly transmissible and learnable for the speaker community through social transmission/cultural evolutionary processes.

        In summary, I think there are two things I find really cool about Deacon’s current approach: (1) it appreciates that Darwinian processes at multiple levels of organisation; and, (2) this evolution-like process in development is shown to be commonplace throughout nature, and not just something that happened to be a unique human feature.

        On another note: Have you had chance to look at the Pinker paper yet?

        • daniel.lende says:

          James, ditto for me – Darwinian processes at multiple levels, and not necessarily the same type of evolutionary process at each level. And I agree, I think Deacon’s argument has the potential for broader application.

          And no, I haven’t seen the Pinker paper. What’s the cite?

          • Paul Mason says:

            James might be referring to this paper:
            Pinker, S (2010) The cognitive niche: Coevolution of intelligence, sociality, and language,
            PNAS (Supplement 2) 107, 8993-8999.


          • Sorry, I should’ve been clearer. Paul is right: I meant the Pinker paper in the same edition as Deacon’s.

          • daniel.lende says:

            I just glanced at the Pinker paper. Pinker is very good at taking what others are working on and restating it, often as his own. So his argument doesn’t strike me as quite as novel as Deacon’s. I’ve been hearing the same sort of stuff from my Notre Dame colleague Agustin Fuentes for years! I do think ideas about niches are interesting, and that mixing together sociality and cognition is crucial.

            However, like much of evol psych, this work just states that these adaptations do these things – they are sort of sui generis, given to us by evolution because they provided an adaptive benefit and because they are, since we just described them as adaptive traits, obviously heritable. The second part is not so obvious…

            But it’s really the first part that bothers me more, because it misses the really interesting and important work on understanding HOW these things actually happen. Rather than saying they do, and provide an adaptive benefit, I prefer work like Deacon’s that gets more at the whole complex mess of how!

  3. Paul Mason says:

    G’day James and Daniel,

    I would only add that there was a relaxation of selection that preceded language and that language further released selective pressures. In his article, Deacon points out that it is the structural redundancy itself that opens up the opportunity for variations to be explored. As long as the variants are not exposed to selective pressures, then they are not deleted from the system. The protection from elimination is two-fold: (1) the environmental pressure may be released, (2) the intraorganismic selective pressure is released. The statement we made, that “the structural aspects of language…relax the process of selection”, refers to intraorganismic evolution-like processes.

    If we look at the evolution of pidgins to creoles, for example, the needs of communication are met primarily through nouns, verbs, and adjectives. Articles, auxiliary verbs, conjunctions, prepositions, and pronouns do not feature strongly in pidgins. To my knowledge, pidgins are not used to express complex thought but are used simply to address the basic needs of communication, trade and economic interaction etc.

    When a Pidgin is adopted as a language by a generation of interlocutors, the communication system becomes more complex and evolves into a Creole; the vocabulary expands, the syntax stabilises, and a fuller repertoire of sounds is practiced. In general, Creoles employ consistent word order, conjunctions, relative clauses, and auxiliary verbs to express verb moods, aspects and anterior tense. (But even then, Creoles lack conjugations of verbs for tense and person, declensions of nouns for case and number, most prepositions, and the passive voice of verbs). Creoles will eventually complexify as evidenced by the English language which evolved over many centuries since the Battle of Hastings from the interaction of Germanic-speaking Anglo-saxon peasants and the French-speaking Norman upper-class. Bahasa Indonesia is another example of a Creole that has evolved to be capable of carrying complex and nuanced thought (as written, for example, by authors like Pramoedya Ananta Toer). The moral of the story is that language add-ons are included once the basic needs of communication are addressed by a Pidgin, and then the complexity is expanded as we see in a Creole. The expansion of that vocabulary affords more complex interaction between individuals and can further release selective pressure by leading to invention, the storage of adaptations, and the rapid distribution of adaptations.

    (The fact that Creoles such as English often have two words for things, e.g. pork, swine and pig; arse and butt; gut and stomach; snot and mucus, etc., is evidence of structural redundancy in the language).

    Thank you for the question, you helped me think about things in a new way. As you can see, I found Jared Diamond’s 1991 paper “Reinvention of Human language” (Natural History 5/91: 22-28) a really useful summary on pidgins and creoles.

    • Cheers for the reference and reply. Some of the work I’m hoping to do this academic year will involve looking pidgins and creoles.

      Also, whilst I agree on all your points (and found them very interesting), I would also mention that certain features languages may simplify based on their underlying speaker environment/linguistic niche (see Lupyan & Dale, 2010). In particular, languages spoken in exoteric niches (large number of speakers, large area, many linguistic neighbours) and esoteric niches (small number of speakers, small area, few linguistic neighbours) will produce languages adapted to these socio-demographic conditions. In Lupyan & Dale’s study, they found that those languages spoken in an exoteric niche are prone to simpler inflectional morphology, with an increased reliance on lexical strategies to encode certain linguistic features (e.g. evidentiality). Conversely, languages spoken in esoteric niches are morphologically complex, and as such show greater levels of redundancy.

      Here’s the reference:

      Lupyan G, & Dale R (2010). Language structure is partly determined by social structure. PloS one, 5 (1) PMID: 20098492

  4. Paul Mason says:

    Intriguing academic year ahead of you James!!! I hope you keep us all tuned in about your findings. Lupyan & Dale’s research seems to relate in some way to Sperber’s law of the epidemiology of representations.

    In thinking about esoteric and exoteric niches, I’m thinking about the differences in the Alsatian language spoken in Alsace (a territory that has passed hands between the French and Germans multiple times) and the English language that developed from the interaction of a French-speaking elite and a Germanic-speaking anglo-saxon populace. Although both English and Alsatian developed in communities in contact with Romance and Germanic languages, English took a very different form to Alsatian. In England there was a hierarchical divide between language groups. In Alsace any hierarchical divide was repeatedly turned on its head, and trade-groups from different language backgrounds may have had more socially equal standing. Am I understanding the ideas you wrote about if I ask: would the longstanding disparity between groups in England have led to a simpler language than in Alsace?

    • Oops, I’ve only just noticed your reply… I’ll let you know how I get on with my research (it’ll appear on my blog at some point).

      I’m not sure how related Lupyan & Dale’s suggestion is to that of Dan Sperber’s. I’ll admit that I don’t know much about the epidemiology of representations.

      As for your question: would the longstanding disparity between groups in England have led to a simpler language than in Alsace? I’m afraid the only answer I can give is: perhaps. I think the broad idea is that significantly different speaker demographics will result in a language adapting to those conditions in different ways. I’m not familiar with Alsace so it’d be disingenuous of me to comment on its complexity. I will look into it and get back to you if I find anything of note.

  5. Pingback: Food for thought: Cooking in human evolution | Neuroanthropology

  6. camelo says:

    Ahem, I´m new at this … but I would like to throw in with Daniel on the frustrating lack of generative theories for adaptive traits. Too often we are just told that a trait exists because its beneficial and heritable. I equate this roughly to the difficulty we have seeing what it is we want to explain with our theories. This is the classic problem of framing the problem correctly and along those lines I get the feeling that Deacon is showing us the trees when we want to see the forest.

    Ultimately, what we want to know here is how biological novelty is generated. Specifically we would like to use this knowledge to explain how “humaness” was generated and we all suspect this has much to do with the evolution of language. In Deacon´s case, his arguement is that novelty comes from internal processes, specifically genetic and epigenetic “darwinian processes”. He places novelty here because selection is not a positive process, something has to generate that from which adaptive traits can be selected. Do any of you know why “that from which adaptive traits can be selected” should be constrained to internal processes? It seems to me that behavior, as it mediates external interaction, is the first and most important stop in finding true novelty. Novelty in this sense is generated by the differential preservation of behavioral routines as they interact with environmental features (in this case other protohumans) .

    What Deacon implies by naming only internal processes is that language evolution started within protohumans and this excludes the possibility that it evolved between humans. He is looking for humaness in the tree not in the forest. I think it´s too early to discount cooperative generation of language in favor of individual generation.

  7. camelo says:

    P.S. Obviously Deacon considers behavior and external interaction (e.g. his affection for the Baldwin Effect) but do you think he gives it it´s due?

  8. daniel.lende says:

    I’ve always appreciated the Baldwin Effect as an idea ever since I first came across it, but it’s hard to operationalize, find evidence, and so forth. What I was struck by with Deacon’s new work is that this is his attempt to do some of that heavy lifting for language evolution, to get at how something like the Baldwin effect might work on the ground, while also incorporating new ideas from evo-devo and the like.

    But you’re probably right, Camelo, that in doing so, he moves away from the core point of the Baldwin effect, that “sustained behavior of a species or group can shape the evolution of that species” (Wikipedia), or that there is something close to a “learning” process in evolution (or at least something that looks like it, even if it might just be fitness hillclimbing, per Sewall Wright). These sorts of ideas are gaining ground through ideas about niche construction, and more could probably be done there.

    That’s one interesting aspect of Changizi’s interview, that he sees culture adapting to the brain. That gets us closer to what a cultural niche might look like, that the niche construction happens in relation to the biology, and not simply as some supra-organism.

  9. camelo says:

    Excellent Changizi interview! I presume both he and Alva Noe are in contact … but then I guess I´ll have to buy the book and find out. I happen to agree with both of them that we won´t find the missing keys to language or consciousness by concentrating on the brain. The brain is not only embodied, it is encultured. Beyond that it has a long history of being selected by various environments. It seems that the fashion these days is to keep extending the context and I applaud that. Context is key.

    You are absolutely right about Deacon doing some of the heavy lifting for language evolution but I lament his focus on ontological and neural processes. Ultimately all internal processes either are maintained or were maintained by external selection somewhere along the descent line. The key to language evolution is how it helped protohumans (probably groups of them) interact with their environments. It was a behavioral adaptation first. Once language existed, Deacon´s internal processes probably did a great job of canalizing successful symbolic interactions.

  10. Paul Mason says:

    I think you’ll find that Deacon offers a convincing alternative to strictly adaptationist models of language evolution.

  11. Pingback: Is seeing believing? | Replicated Typo

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>