review for Journal of Evolutionary Biology

Daniel C. Dennett


August 10, 2001

[3300 words]


Eytan Avital and Eva Jablonka, Animal Traditions: Behavioural Inheritance in Evolution, Cambridge University Press, 2000.


                For thousands of years, members of our species have been captivated by the cleverness of animals, the elegant ways in which birds of different species build their nests, the circumspection and efficiency with which predators stalk their chosen prey, and so forth. The genius of “instinct” comes in abundant variety, and breeds true. “It must be in the genes”–that’s what we tend to conclude. But when we do, we may be jumping to conclusions, because there are other possibilities: the clever behavior we observe could be the do-it-yourself invention or discovery of the individual behaver or it could be a clever trick copied from an elder member of its species, most likely one of its parents. This last possibility is an ancient doctrine, enshrined in folklore about animal parents sternly but lovingly training their young, and in countless anecdotes, but this appealing idea of animals benefitting from hand-me-down wisdom from earlier generations much the way we do has recently languished in the shadow of the genes, an oversight this book seeks to correct.  The folklore is not all fanciful; some of it can be supported by good science, which moreover will open up surprising vistas on the role of individual behavior in evolution. The book is fascinating on at least three levels: first, it provides a vivid and insightful survey of a  wealth of examples drawn from studies of literally hundreds of species (almost all mammals and birds); second, it forthrightly addresses the theoretical problems posed for evolutionary theory by cultural transmission and its interaction with genetic transmission; and third, by the very strenuousness of its efforts to overcome the prejudice against its main thesis, it inadvertently throws a powerful spotlight on the way received opinion in science can close investigators’ minds. As I began reading the book, I thought the authors were overdramatizing their position as underdogs, and caricaturing the bulwarks of opposition they saw themselves confronting (and in some instances–to which I will return below–they do overshoot their targets),  but a few sidelong inquiries of my own among biologists in recent weeks have convinced me that the mindset they seek to overturn is actually quite common. So this is an important book, potentially a major investigation-shaper in the years to come, for in addition to the widespread work they discuss, they point to a much larger array of still-to-be-done studies, eminently possible, that have never been done simply because nobody thought to do them.


                Sometimes, you can’t see the evidence lying all around until you somehow acquire theoretical permission to see it as evidence.  The evidence for continental drift was literally thick on the ground but couldn’t be taken seriously until the idea of plate tectonics showed geologists how to imagine whole continents sliding around on the surface of the planet, driven by the upwelling at the mid-ocean ridges. So the first task of the authors is to establish the theoretical possibilities of cultural transmission, which they do with a delightful thought experiment about an imaginary species of small mammals, tarbutniks (from the Hebrew word tarbut, meaning culture).  These are declared by fiat to be all clones of each other, with zero genetic diversity and hence zero genetic evolution by natural selection.  The population comes to be divided, as populations so often do, and in one group a pioneer digs a hole in the ground (it might just be an accident, or the result of a “bad” habit of this individual) and this novel act happens to inspire some of the onlooking conspecifics to do likewise. Why? Just because the tarbutniks are postulated to have a genetically maintained penchant for imitation, unlike a less fortunate strain of tarbutniks that were restricted to individual, risky, trial-and-error learning, and went extinct. Since it happens that hole-digging is beneficial in their local environment (it provides some protection, or gives access to a good underground food source), those who dig holes do better than those who don’t. The habit spreads, but not genetically: the young pick it up by “social learning” from their parents, or others in their community. (“Imitation” has become a controversial concept among researchers in animal cognition, who have offered grounds for distinguishing true imitation from other varieties of transmission of similarity: stimulus enhancement, emulation, and so forth. The authors wisely eschew these battles,  adopting a minimalist position: “All that is needed for social learning is that the presence of one relatively experienced individual increases the chances that a naive individual will learn a new behaviour.” (p90))  Hole-digging leads to tunnel-digging leads to giving birth in the underground tunnels, and in due course this lineage of tarbutniks have all adopted behaviors and a diet of the sort observed in moles, living most of the time underground, eating what is abundant there, and so forth. Those among them who had the misfortune to adopt alternative traditions (from their parents or neighbors) have tended to die childless.  The other population takes up berry-picking, living in the protection of the underbrush, and its diet and habits are likewise molded by natural selection of behavioral tradition, not genes. One group has a problem with the acidity of its diet, corrected by eating a bit of dirt, a trick pioneered by one and copied by the others, which then opens up other food sources heretofore toxic, and so on.  One group develops the tradition of a courtship offering of red berries, which would mean nothing to the members of the other group, who have fallen into their own courtship rituals. Eventually, thanks to the differences in habitats and hence habits and hence diet and hence physiology and hence development, the typical members of the two groups don’t just act differently; they have different size, color, physiology, diet–they have become as different as two closely related but distinct species can be–all without any genetic change at all. Possible in principle?  Absolutely. It all follows from uncontroversial facts about the knock-on effects of changes in behavior or environment, norms of reaction and developmental requirements for genes. Everybody pays lip service to these facts; the authors want us to notice what follows from them: there is a broad-band informational pathway that runs roughly parallel to the genetic pathway that can transmit adaptations just as well as–sometimes better than–genes can. 


                What is possible in principle may nevertheless not actually be an important factor, for one reason or another. The next task is to demonstrate that the contributions of transmitted tradition or culture to animal behavioral design are in fact substantial. Here the authors provide more informed speculations than conclusive demonstrations, but in addition to the wealth of circumstantial evidence they cite, they do point to key studies that point the way to confirmation. The way to test their claims is to interrupt one transmission channel or another and see what gets through. Cross fostering is the most obvious manipulation: to see if young exhibit the behaviors of their foster parents instead of their “biological” parents. (Notice how strong the association has become between genes and biology–as if there were nothing in biology except genes!)  The authors report the results of cross fostering studies already undertaken, as well as studies that block the paths of social learning in one way or another, and the results they cite certainly support their contention, but there is much more to be done.


                Before researchers can be enlisted to embark on these long-term investigations, they have to be persuaded that they are likely to hit paydirt. That is where the thought experiments and speculative scenarios come in. Time and again, the authors offer a persuasive redescription of the setting of some well-studied behavior–food preferences, predation techniques, nest-building, danger-avoidance, mating tactics–and consider what could be the case about how it is installed in each generation.  Along the way, they consider and disarm a host of objections, and present reasons for thinking that evolution ought to avail itself of cultural transmission whenever possible. In the first place, such social learning is clearly safer–less risky–than individual trial and error by novices, surely a large benefit that would be recognized by natural selection. Moreover, the cultural transmission of newly discovered Good Tricks (my term, not theirs) is orders of magnitude swifter than the incorporation into the genome of whatever it takes to specify the Good Trick genetically. Cultural transmission works as an enhancement of pure trial and error: when an old trick outlives its usefulness because of a change in environment, the lineage doesn’t have to wait many generations for the right new combinations or mutations to come along. The individual animals can revert to trial and error immediately, and as soon as one explorer finds a new trick in the right direction, others can copy it and abandon their riskier explorations. Of course they may also be led down the path to destruction by copying an innovation that is only apparently an improvement; social learning has its own risks.


                If this is all so obviously adaptive, why have investigators been so prejudiced against it as a major possibility? There are many reasons. Since “culture” is commonly taken to be one of the prime idiosyncracies distinguishing Homo sapiens from all other species, anything that smacks of the exploitation of culture by non-human species promises to blur a boundary many want to keep as hard-edged as possible.  Then there is the ever-present fear of  lapsing into one “Lamarckian” heresy or another by countenancing the transmission–by any path–of something acquired. (I wonder if some evolutionists are reluctant to leave a last will and testament, for fear of seeming to their colleagues to be countenancing the transmission of acquired wealth!) But the main source of covert resistance comes from the genocentric assumptions that have apparently swept to fixation in the minds of many biologists.


                Mother Nature (natural selection) is not as genocentric as those biologists. Natural selection doesn’t insist on moving all valuable information through the germ line. On the contrary, if the burden can be reliably taken over by continuities in the external world, that is fine with Mother Nature–it takes a load off the genome.  Consider the various continuities relied on by natural selection: uncontroversially, there are those supplied by the laws of physics (gravity, etc.) and by the long-term stabilities of environment that can be safely “expected” to persevere (salinity of the ocean, composition of the atmosphere, colors of things that can be used as triggers, etc.). But there are also the regularities that can be transmitted from generation to generation by social learning. These are “just” a special case of expectable environmental regularities, but they take on further importance since they are themselves subject to natural selection, directly and indirectly. How and why these regularities are there is itself an important evolutionary issue, and as the authors note, in the same way that the genetic informational pathways have themselves been subject to incessant refinement over billions of years, there has been a recursive or iterative process of enhancement of non-genetic pathways:  “The evolution of the transmission of mechanisms of transmission is of central importance for the evolution of learning and behaviour.” (p132)


                A further mistaken grounds for suspicion of the idea of transmission of behavioral tradition is the hunch that since there is no proprietary code (like A,C,G,T) in which such information is couched, transmission cannot be sufficiently high-fidelity to count as replication.  This popular idea trades on a confusion of levels: the handy digitality of the genetic code, which makes it straightforward to quantify the information (in bits), does not specify the information at the level at which natural selection acts. The semantic information in “coded” genes is just as hard to specify, just as open-ended and interpretation-prone, as the information embodied in external behaviors. An easy way to remind oneself of this awkward fact is that there is exactly as much digital information in a sequence of junk DNA as there is in a portion of a gene of the same length: when you confront the vexing question of what a gene is for–a protein, a brain structure, a behavior, a policy–you must leave the crisp digital world of A,C,G,T behind and venture out into what might be called organism hermeneutics. Poets don’t make the mistake of thinking that a poem transmitted in ASCII code is any less subject to quandaries of interpretation than a poem transmitted orally. Biologists need to appreciate the same point.


                Codes do make a big difference. We human beings have symbolic codes–natural languages composed of finite vocabularies of words anchored to norms of both production and meaning–and this gives our practice of cultural transmission a hugely different profile of competence compared to other species.   “Since animals cannot represent information symbolically, . . . . the focus must be the social and ecological conditions that lead to the manifestation and re-generation of essentially similar patterns of behaviour.” (p95)   Or, to turn the point around, symbolic representations (of behaviors and other topics) have built-in self-stabilizing features (the norms) that can permit them to survive drastic changes in supporting ecological conditions essentially intact, something quite impossible in the animal world.  (Marco Polo is said to have brought the idea of pasta from China to Europe; he didn’t have to become a pasta chef to do this.)


                The comparison with human cultural transmission presents a delicate problem for the authors. Choosing their battles carefully, they go somewhat overboard in distancing themselves from the controversial topic of memes, which arouses blind animosity in so many. What they are mainly concerned to argue is that transmitted tradition is another path to genetic fitness of behaving organisms–it is information that creates adaptations in the extended phenotype. They are not concerned with the fitness as replicators of such designed behaviors themselves.  One cannot blame them for wanting to secure a little good will by hastening to endorse some of the standard (but inconclusive) objections to memes, but in fact their own account is consistent with the more careful formulations about memes, and they make many of the favorite points of memeticists in their own terms. They point to the possibility of the spread of traits with no fitness advantage (p131-6), and stress (as Dawkins did) the fact that transmission is via copying the phenotype. They point to ease of remembering as a factor in transmission of a trick that may move it away from what otherwise would be the “engineering” optimum (p135). But right here they might benefit from the memeticists’ perspective, for it is easier to see that memorability is itself just as much a question of “engineering” if one is thinking of the fitness of the remembered items themselves, instead of their possible contribution to the fitness of their hosts. And by concentrating on vertical transmission of culture (from parent to offspring), they ignore the predictable prospect of virulent,  parasitic habits being more readily spread obliquely, an insight that beckons once one adopts the meme’s-eye point of view. (See also the review by Matteo Mameli of this book in Biology and Philosophy. 16(5), 2001.)       


                Occasionally, the authors overstate their best case. For instance, wishing to provide a striking alternative to the received wisdom about parent-offspring conflict in, e.g,. tits (p173) they ask “who really controls the allocation of resources in the tit family?”  and go on to argue that there is an alternative story to the story of genetic conflict that can be told, in which the parents are teaching their offspring, not being blackmailed by them.  There may well be something to their perspective on this, but the question they don’t address is this: is there also an argument against the existence of genetic conflict?  If we already have reason to believe that there should be genetic conflict, their alternative might be a useful supplement, but not an alternative. They say “Translating agonistic behaviour among family members into ‘evolutionary conflict’ may mislead us.”  (p182)  This is true. The question is: does it mislead us in fact? They say that the conflict theory’s assumptions “are not substantiated by any data” (p185)–since their alternative fits the data equally well–but then they are equally in no position to assert, as they do, “Looked at in this way, the squabbles between parents and young are not an outcome of evolutionary conflict, but are inevitable results of the learning process and the somewhat painful transition to the youngsters’ independence.”  (p184) The bland truth may turn out to combine both ideas, with the genetic conflict harnessed into an opponent process system of teacher and learner.                                                                                          

                To this reviewer, some of the most exciting suggestions appear in their discussion of the Baldwin effect, or Waddington’s genetic assimilation, and the possibilities of a sort of teamwork between cultural and genetic transmission in the design of elaborate adaptations. Novel (to me, at least) is their suggestion:


Since plasticity of higher animals can mask both environmental and genetic variations, many genetic variations are protected from selective elimination and can accumulate. The net effect is a large reservoir of genetic variation underlying the organization of the nervous system. This variation is exposed and recruited when the environment changes. (p323)


Another is their “assimilate-stretch principle.” Genetic assimilation can foster cumulative adaptations of lengthy behavioral sequences, such as nest building, by a process rather like knitting: Trial and error learning, at the knitting end, does the R&D for each new step, which then is held in place by cultural transmission (“tradition”) until it can be added to the already growing skein of automatized steps. “As formerly learnt behaviours are transformed into innate behaviours, and as new learnt behaviours are added to the sequence, the overall number of learnt elements in the sequence may be preserved.” (p331)


The transition from learned to innate behavior made possible by genetic assimilation is illustrated by another wonderful thought experiment: the literacy gene (p362-3). There are children today who learn to read without instruction. They are relatively rare nowadays, but if they have a survival advantage (and they may well), we can well imagine coming back in a thousand years and finding that our descendants are innately endowed with something a Chomskian might call a RAD (Reading Acquisition Device).


                One final comment, inspired by the authors’ occasional lapses of overselling. A brute fact about evolution that is rhetorically inconvenient, when confronting skeptics, is that it is . . . shy about displaying its powers. It works when it works, but usually it doesn’t.  Every time a parent gives birth to offspring, this is potentially the initiation of a speciation event, but it almost never is. Similarly, every time a habit is picked up by one animal from another, a potential cultural tradition is born, but it almost never is. Only slightly less rare, presumably, are ephemeral group habits, commonalities in behavior that spread through a neighborhood or population, but that are too minor and evanescent to count as traditions, passing fancies that don’t even rise to the status of fads. One shouldn’t be put off by this; it would be theoretically tidy to be able to say that whenever such and such happens, a tradition results. But it is a fool’s errand to try to prescribe the sufficient conditions for such momentous innovations. The likeliest candidates almost never pan out, in fact. But sometimes they do. When it happens, it happens. Tradition-creation in evolution, like speciation, can be both an all but invisibly rare event and a highly significant force in the design processes that create the cleverness we observe.[1]



















[1]I am indebted to Matteo Mameli for discussions of this book and an earlier draft of this review.