Thursday, April 26, 2012

Report from the Consilience conference, part I


www.iasaglobal.org
by Massimo Pigliucci

I am in St. Louis these days, where the University of Missouri has organized a three-day, 19-speaker conference on “consilience,” or the unity of knowledge, in the somewhat idiosyncratic interpretation of E.O. Wilson in his popular 1998 book. Indeed, the proceedings started with a keynote by Wilson himself, as sharp and as sprite as ever. The danger with this sort of conference is that they either become a predictable and somewhat uncritical celebration of a central figure or idea (in this case, Wilson), or that they evolve into a hodgepodge of loosely (if at all) related talks with only a vague central theme. Nevertheless, I accepted to be a speaker here because I thought the conference was a good idea and because my fellow speakers are likely to provide interesting food for thought on a broad array of issues. So, let’s get started.

Wilson’s keynote covered a lot of known material, from the evolution of eusociality (it’s rare, and yet has huge consequences for the species that crosses the threshold from partial to eu-sociality) to the basic steps of the evolution of humans (bipedalism, larger brains, etc.). His overarching theme, however, was that there are three fundamental questions that neither religion nor philosophy can answer, and that science has begun to tackle: Where do we come from? What are we? Where are we going? Somehow Wilson thinks this is equivalent to asking about the meaning of life, though I submit that’s a bit of an unjustified leap (surely what meaning we construct or attribute to our life can be informed by those broad questions, but the meaning of human life being local and personal, those questions hardly take center stage and are more likely simply interesting background).

The crux of Wilson’s talk was that human eusociality appeared as a result of group selection, and that individual and group selection are constantly in unstable tension, which explains why it is in the nature of humanity to always struggle between what he called (borrowing from David Sloan Wilson, no relation) “sinful” (i.e., selfish) and “virtuous” (i.e., cooperative) drives. Wilson also added a provocative note — on which he did no elaborate — about a paper he co-authored with two mathematicians (published in November 2010 in Nature) where he shows that the concept of inclusive fitness (and hence of kin selection) is mathematically incoherent. Which pretty much would demolish, if true, an established explanation for the evolution of altruism and leave the field entirely to a newly resurrected group selection.

I cannot comment on the kin selection issue (anyone?), and I am generally sympathetic to the idea of multi-level selection. But Wilson provided no evidence or particular reason to accept the idea that group selection played a crucial role in the evolution of human eusociality. Furthermore, it seems obvious to me that to label some behaviors as good or bad (virtuous or sinful) on the basis of which selection mechanism (allegedly) evolved them is a flagrant violation of the ought/is divide (which, I know, is not impermeable according to people like Quine, but I am a Humean on this...).

Besides, even if we take Wilson’s highly speculative scenario at face value, we are immediately confronted by the problem — of which Wilson himself seemed aware — that group selected “virtue” only applies to members of the in-group, thus generating a variety of nasty inter-group behaviors, including xenophobia and, of course, war. And here, I think, is where biology reaches a dead end, providing no answer to some of the broadest human problems, which are, somewhat ironically, handed back to the humanities (including philosophy, political science, literature, history, and possibly even religion!). A colossal failure of consilience? (Incidentally, Wilson left the conference immediately after his talk. I understand about his age, but I thought that was a bit rude, considering that the whole event is supposed to be a commentary on his work. Oh well, I guess he won’t see my critical analysis of his take on consilience.)

The second talk of the first day was by John Hawks, on behavioral implications of ancient genomes. Lots of interesting stuff here on the comparative genomics of humans, Neanderthals, other hominids and the broad array of contemporary primates. I do not have much to say about this talk, however, because — although packed with fascinating suggestions about Neanderthals in particular — it seemed to me to have little to do with the theme of consilience (see the second danger mentioned above for this kind of conference). Indeed, the speaker must have been aware of it, since his only reference to the humanities was a half-joking remark to the effect that fiction writers need to drop their stereotype of Neandys as dumb beasts, because of all the evidence of their smarts. Given that the total number of novels featuring Neanderthals is pretty minuscule, I doubt this will have much of an impact on English Lit classes...

Next we have Dan McAdams, with a talk entitled “From actor to agent to author: human evolution and the development of personality.” His starting point is that human nature is an evolved psychological design, with personality psychologists being interested in the variations on the basic design. Of course there are a lot of hidden assumptions here (is there such a thing as human nature? Is it the result of genetic evolution, culture, or both?).

To make sense of variation in human personality McAdams invokes three nested layers of understanding (from the inner to the outermost): the individual as a social actor (you know, as in “All the world’s a stage...” and so forth), the individual as motivated agent (i.e., engaged in goal-oriented social behaviors), and the individual as an autobiographical author (we weave “personal myths” about our lives). Social acting is connected to the Big Five personality traits, and begins very early on. Motivated agency appears in children 7-9 years old. Autobiographical authorship begins in someone’s 20’s and 30’s.

The talk struck me as very interesting, but again with precious little to do with consilience. While the author did mention the word evolution a few times, these were both speculative and largely irrelevant to the main points: McAdams gave a good talk about the psychology of personhood, but evolutionary biology provided only a distant background condition. And, I would add, this is precisely the way it should be.

Ellen Dissanayake talked about “markmaking” as a human behavior. The author focused on early non-representational marks on rocks, which apparently constitute more than 99% of known Pleistocene rock “art,” thus making the famous cave paintings of animals look like anomalies.

Dissanayake considers a number of possible proximate explanations for rock marks, including accidental byproduct of other activities, utilitarian and/or communicative functions (record keeping, didactic, territory marking, etc.), and doodling for pleasure. She is not happy with any of these on the ground that they do not apply universally. But one could reasonably ask why one expects a universal explanation for such a wide variety of human artifacts.

The author then suggests a connection between Paleolithic rock art and modern aboriginal tribes’ body painting, on the basis of the similarity of (some of) the patterns. The commonality would be “ritual use” (a somewhat fuzzy category of human behavior, it must be said). This strikes me as fanciful to say the least, and Dissanayake herself quipped that a “nice” feature of this suggestion is that it is virtually impossible to test, so that she can’t be proven wrong. Okay, then!

What about the ultimate causes of rock art? Possibilities include sexual selection, display of prestige within the group, provision of “cognitive order” (induced by the geometries of the marks), reduction of stress, group selection for social order and group unification. Needless to say, there is absolutely no way to discriminate empirically among any of these.

Toward the end of her talk, Dissanayake  suggested that humans may have had a behavioral disposition to what she calls “artify” that actually preceded symbolic art. Artification would be an evolved behavioral predisposition to make ordinary reality extraordinary or special. Fascinating, but why would we have this artification tendency? And how do we know this to be the case? As usual with evolutionary psychology: it’s easy to make stories up, it’s exceedingly hard to test them scientifically.

We then moved to Herbert Gintis, and the evolution of morality. Gintis began with a cartoon model of the so-called Standard Social Sciences Model and the blank slate approach to human culture, which he (rightly) dismissed. He then — a bit simplistically — mentioned that “the” model in philosophy is the Hobbesian model of war of all against all, moving to Dawkins’ idea of morality as the result of a culture that rebels against selfish genes, and finally arriving at the economists’ assumption that selfish (“rational self interest”) behavior is at the basis of all human transactions.

Instead of all of the above, the author prefers a gene-culture co-evolution approach, along the lines of the now classical studies by Feldman, Cavalli-Sforza and others. Human morality, then, is seen as the product of a dynamic process in the course of which humans transform culture, and culture makes new behaviors fitness-enhancing.

Gintis suggests that morality emerged as a contribution to social harmony and efficiency once hominids had destroyed the basis of the standard primate dominance hierarchy. The latter was undermined, allegedly, by the invention of weapons, by which weaker individuals could kill the alpha male from a distance, or even in his sleep. This is certainly quite speculative, but it does agree with a generally emerging view of basic morality being the result of evolution of highly integrated social behavior in small primate groups. The problem, as usual (and as readily acknowledged by the author) is that it is hard to test the proposition according to which, for instance, leadership by persuasion replaced leadership by brute force.

Gintis understands that persuasion was certainly not enough, and that early morality had to be based on evolved pro-social sentiments. That is, there had to be a strong emotional component to the process. He maintains that these elements explain both the evolution of language (persuasion) and of moral sentiments. Presumably, the combination of the two gave us morality as we understand it today.

I am generally sympathetic to gene-culture co-evolution models, which certainly beat the crap out of simplistic evolutionary psychological hypotheses or of equally vacuous “memetics.” However, it seems to me that this can explain the very beginnings of human traits like language and morality. Pretty soon the speed of cultural change has far outpaced that of genetic evolution. This doesn’t mean that our genetic makeup is not important (as a background condition to all we do), but it seems to imply that we need a second-order theory of cultural evolution per se, and so far I haven’t seen much of a serious candidate for this.

The interesting evidence put forth by Gintis concerns honesty and game theory. As is well known, the only straightforward biological explanations of “altruism” are kin selection (favoring individuals with whom you share a substantial chunk of your genes) and reciprocal altruism (tit-for-tat). Notoriously, neither of these behaviors actually brings about genuine altruism, because it always comes down to your own (genetic, long-term) self-interest. But Gintis’ evidence shows that when people play one-round (i.e., not iterated) games involving fairness and honesty — even when played anonymously — a significant portion of subjects behave honestly (though that percentage goes down if the cost of the honest behavior increases sufficiently). The idea is that, just like Aristotle would have said (though Gintis didn’t mention him), people behave morally for the simple reason that they think (and feel) it is the right thing to do.

Next to last for the day, we had Robert Frank on “The Darwin Economy: competition and the common good.” (And he was the only speaker without slides! He thinks that’s more engaging, I think it’s so much easier to lose concentration and track of what’s going on. But that’s another conversation.) According to Frank, economists eventually will recognize Darwin, not Adam Smith, as the founder of their discipline. Smith was much less of a true believer in the virtues of the free marketplace than modern libertarians and many economists consider him to have been. Frank agrees that markets fail frequently, but he thinks that is for reasons different from those maintained by Smith. He also, apparently, doesn’t buy much into the tenets of behavioral economics.

The reason the invisible hand doesn’t work, according to Frank, is because of Darwin’s central insight that there is a tension between individual and group interests. An example is the fact that, historically, NHL (hockey) players never wore helmets, even though they all thought helmets were a good idea. The problem is that without a helmet the player sees better, thus gaining an edge on the others. So if even one player took the helmet off, pretty much all the others would follow, in a hockey version of the tragedy of the commons. Helmets were finally instituted because of a league-wide rule that was voted on near unanimously by the players themselves. This, incidentally, makes a perfect anti-libertarian argument... (Which was made by John Stuart Mill in the 19th century.)

The talk went on for quite a while with example after example of absurdities caused by market-enabled runaway competition among people who end up damaging society as well as their own long-term flourishing.

Though Frank makes very good points, it seems to me that there may be a couple of fallacies at work here. First of all, once again we see the desire for a totalizing explanation, even though it is perfectly reasonable to think that Adam Smith, the behavioral economists, and Darwin all have gotten pieces of the puzzle right. Second, these aren’t even independent explanations, since the human behavior repertoire evolved (in part) by Darwinian mechanisms, and markets are the result of cultural evolution that is affected in turn by the range of human behaviors.

Dulcis in fundo (so to speak), it was my turn. I will publish a separate essay on my talk, so I’ll give just the gist here. It is a criticism of Wilson-style consilience (in the sense of “unity of knowledge”), which I think is a reductionist approach and has actually little to do with consilience in the original (and still most widely used in philosophy) meaning of the term, elaborated by William Whewell and referring to a type of induction known as inference to the best explanation. My basic theses are that: i) “Knowledge” is a heterogeneous category that does not admit of Wilson-type consilience; ii) Applying the type of knowledge emerging from the natural sciences to (some) other domains is a category mistake and ought to be avoided; and iii) Wilson-type consilience is actually a scientistic and anti-intellectual enterprise.

I gave a few examples of where Wilson goes wrong with his consilience (i.e., let’s reduce humanities to biology) program. For instance, Wilson is fond of what he calls “epigenetic rules” which he defines as “the regularities of sensory perception and mental development that animate and channel the acquisition of culture.” I don’t know that any biologist has ever measured any such entities, which are just as vague as another of Wilson’s favorite, memes. On the latter, I’ll let my colleague Jerry Coyne comment: “[Memetics is] completely tautological, unable to explain why a meme spreads except by asserting, post facto, that it had qualities enabling it to spread.” Well said, Jerry.

Wilson is also fond of the Enlightenment and of its 20th century philosophical offspring, logical positivism. He hopes that positivism will be back, especially once neurobiology tells us more about how human beings reason. The latter is an obvious non sequitur, on which I think I don’t need to comment further. As for logical positivism, Wilson — who is openly dismissive of philosophy — apparently has never read Putnam or Quine, or a host of other critics of positivism. Incidentally, the demise of logical positivism is a good example of how philosophy makes progress: people find faults in certain views or arguments, and when these can no longer be patched or repaired they are abandoned and the field moves on.

In the end, I found evolutionary biologist Allen Orr’s critique of the 1998 Wilson book on consilience to be right on target. Among other things Orr says: “The real reason Wilson favors his consilient scenario isn’t because he finds it more plausible but because he finds it more attractive. For as he admits near the start of his book, consilience isn’t science, it is a philosophy, a metaphysical view that he obviously finds both beautiful and deeply satisfying. The irony, of course, is that Wilson’s own science of evolution gives every reason for questioning this metaphysic, every reason, that is, for doubting whether our brains — jury-rigged and riddled with blindspots — are the stuff from which certain knowledge and seamless consilience can be obtained.” Yup.

26 comments:

  1. Interestingly Orr brings up Colin Mcginn in his review. Massimo, what exactly do you think about mysterianism in general?

    ReplyDelete
  2. Amazing.
    Welcome to St. Louis!

    ReplyDelete
  3. Life *is* alive. Life *ought* to remain alive. This is the bridge between the is-ought divide. This is a basic belief of Evolutionary Philosophy. www.evphil.com

    ReplyDelete
  4. Sorry, more comments as I read further.

    Those nasty inter-group behaviors are sorted out as cooperation trumps competition on larger and larger scales. The broadest problems are solved by the broadest definition of life that must work together to stave off death.

    Human nature - comes from gene culture coevolution. Yes. (Oh, I see you kind of admit that later.)

    I believe cultural evolution simply adapts to allow the species to survive on longer and longer terms. The culture helps us learn and reinforce new behaviors, cognitive appraisals, emotions, and morals. They grow in timescale according to the breakdown of the divisions of biology: biochemical, molecular, cellular, organismal, social, ecological, evolutionary. There is no culture over the timescales of the first three divisions, but the culture is widening to allow consideration of the final four levels. Our evolutionary heritage leaves us with competitive instincts and emotions that are useful for morals on the organismic level, but culture is trying to teach us to evolve our morals to the social, ecological, and evolutionary timescales.

    Surely our brains cannot come up with certain knowledge and seamless consilience, but the beauty of the iterative scientific method is that we can work towards that goal on essentially infinite timelines. That does not undermine the principle of consilience, it gives it an inspirational home to channel our energies. Would you rather the relativistic or nihilistic word view won out?

    Thank you for sharing all of this information from the conference. I wish I could be there!

    ReplyDelete
  5. Dan,

    I don't like the word "mysterianism," sounds a bit too ocups pocus to me. But I think McGinn (and Chomsky, and Orr) are correct that there is every reason to expect that certain things are going to remain beyond the pale of human cognition. This isn't an excuse for giving up, it's a simple call to a bit of epistemic humility.

    ReplyDelete
  6. Wilson should drop the group selection act already, he's just making a fool of himself in the eyes of practically every evolutionary biologist else.

    ReplyDelete
  7. Massimo,

    > Applying the type of knowledge emerging from the natural sciences to (some) other domains is a category mistake and ought to be avoided; and iii) Wilson-type consilience is actually a scientistic and anti-intellectual enterprise. <

    Why is it not possible for a theory emerging from the natural sciences (e.g. Darwinian evolution) to have some applicability to another field or discipline (e.g. cosmology or psychology)?

    ReplyDelete
  8. Alastair,

    I did not say it is not possible to have *some* applicability. I object to wholesale reduction, which is what Wilson is after.

    ReplyDelete
  9. I'm skeptical of most claims in evolutionary psychology, though I find some well reasonable enough. String theory, at the moment, is pure math with no connection to empirical reality.

    ReplyDelete
    Replies
    1. All evolutionary psychological views of anything reduce to anti-scientific mythology.

      Delete
  10. "it seems to imply that we need a second-order theory of cultural evolution per se, and so far I haven’t seen much of a serious candidate for this."
    Check out adaptive mutation where the wider sharing of cultural experiences eventually effects the evolutionary change of behavioral strategies.

    ReplyDelete
  11. I'd like to hear more about the use of slides in talks or "powerpoint cognition".

    ReplyDelete
  12. "Second-order theory of cultural evolution:" Boyd, Richerson and colleagues have been developing this for decades now. Their "Origin and Evolution of Cultures" has some key papers - also on their websites.

    The Wilson paper with Nowak, in a nutshell, conflates "inclusive fitness theory" with Hamilton's rule - which is a purposefully very simple *model* of inclusive fitness theory. I mean, it has only three terms in it. They argue that the simplifying assumptions that go into Hamilton's rule never actually hold in the real world. To a theorist, this line of argument is silly since all models include simplifying assumptions and Hamilton's rule is just a way to show how inclusive fitness works in a straight-forward way. The reason why this annoyed so many evolutionists is that "inclusive fitness theory" is much more broad than r > b/c.

    Inclusive fitness and multilevel selection are just two ways of modeling/thinking about the same processes. Arguing which one is "right" is besides the point.

    ReplyDelete
  13. "... Wilson — who is openly dismissive of positivism — has apparently never read Putnam or Quine, or a host of other critics of positivism." Has Wilson read Proust, Kafka, or Joyce? In order to be an ultra-skeptic it might be necessary to understand literature, philosophy, or anthropology in order to have access to a multitude of different perspectives.

    ReplyDelete
  14. Matt Zimmerman has the basics covered, two posts above mine. For cultural evolution see also the free paper "Towards a unified science of cultural evolution" - from 2006.

    ReplyDelete
  15. Academia has mostly dealt with gene-culture co-evolution - especially in the 1980s and 1990s - perhaps in an attempt to distinguish their project from memetics, or perhaps as a side effect of how their projects were funded. However, pure cultural evolution, in these models, may be treated by holding genes constant. Cultural evolution is then seen as a degenerate case of gene-culture co-evolution. More recent academic work has started to redresses this strange imbalanace.

    ReplyDelete
  16. Hi Massimo, thanks for the summary! You say: "I am generally sympathetic to gene-culture co-evolution models, which certainly beat the crap out of simplistic evolutionary psychological hypotheses..." Can you elaborate on which evolutionary psychological hypotheses you refer to? (In general, I think it's problematic to distinguish between co-evolution models and evolutionary psychological models, since both groups are trying to fundamentally understand the operation of psychological products of natural selection.

    ReplyDelete
  17. Anthony,

    not all evopsych explanations are equally problematic. The difference with gene-culture coevolution is that usually the latter deals with more specific, and more testable, hypotheses. For instance the famous case of the co-evolution between the habit of drinking milk and the evolution of milk-digesting enzymes.

    ReplyDelete
  18. thanks Massimo. Most evolutionary psychologists I know use that same example themselves, but it's still not clear to me what makes ev psych explanations less testable. Especially since gene-culture coevolution is not really an analytic alternative to ev psych, but a particular application of it. In other words, what we are necessarily talking about in both instances are the selection effects of certain ancestral environments, and certain "types" of ancestral environments (e.g. culture), that shape biological adaptations. If ev psych explanations are difficult to test, then a fortiori, gene-culture coevolution hypotheses must suffer the same difficulty. Which is why I was hoping you could narrow it down from "simplistic evolutionary psychological hypotheses" to actual specific hypotheses that you view as problematic.

    ReplyDelete
  19. Anthony,

    the two are not at all the same. Take the classical gene-culture co-evolution study, the co-evolution of enzymes to metabolize milk and the habit of drinking milk. The former can be studied empirically with the tools of population and molecular genetics, the latter is well documented by archeological records. Now compare that with a classic example of evopsych: the hypothesis that rape evolved as a secondary sexual strategy. No genes are known to be involved, no archeologica record of the behavior is available...

    ReplyDelete
  20. Thanks Massimo, I think that's an interesting point and it helps me to understand where you're coming from. However, let's take the example of human aggression. First, there is plenty of archaeological and anthropological evidence of inter-specific aggression in humans. Second, there is sound evolutionary theory reasoning from first principles regarding the existence of adaptations for the regulation of aggression and how they should operate (e.g. Parker 1974). Third, there is a host of empirical studies that reveal the operation of such adaptations at the hormonal, psychological, and behavioral level (e.g. Jon Archer's excellent work on the challenge hypothesis, and also experimental work by Aaron Sell). Maybe you meant to use the rape example as an illustration of hypotheses you believe are weak, which is fine. Independently of whether the rape hypothesis is correct, this does not show that ev psych hypotheses in general are less testable than gene-culture coevolutionary models, or that they cannot be studied empirically.

    ReplyDelete
  21. Anthony,

    we may need to agree to disagree on this one, particularly because it's not really the main scope of the post. However:

    > let's take the example of human aggression <

    Let's not. "Aggression" is a bit too generic of a term, and it is common to a lot of species of vertebrates. While this may make it a good research topic for comparative behavioral studies, it isn't clear what specifically evopsych would add to it.

    > Independently of whether the rape hypothesis is correct, this does not show that ev psych hypotheses in general are less testable than gene-culture coevolutionary models <

    Of course, but you asked for an example. The broader point would have to be made systematically via a review of the technical literature, which is well beyond the scope of a blog post.

    Nonetheless, I must add that I think even gene-culture studies are pretty limited, and it isn't that easy to find too many successful examples outside of the milk case.

    ReplyDelete
  22. Thanks for the replies, Massimo, I appreciate it. And very interesting blog in general!

    ReplyDelete
  23. This comment has been removed by the author.

    ReplyDelete

Note: Only a member of this blog may post a comment.