Sunday, August 21, 2005

A tale of one ratchet   posted by Razib @ 8/21/2005 08:11:00 PM

As some of you know, I am involved in the Cognitive Science Blog Reading Group where we are reading Michael Tomasello's The Cultural Origins of Human Cognition. As I stated on the group I view my participation on the list and posts on the weblog as synergistic, intersecting but not necessarily coterminous activities. The core readership of this weblog is focused on the biological sciences, and so a post I make here will not have the same flavor as a comment I might make on the reading group, because the goal and shared lexicon differ. Unlike other group members I decided to read the book in two sessions and will give a short review of the whole book here. I do not know if I will post follow up posts because I will state beforehand that most of the book is way outside the core of my knowledge base, biological-evolutionary concerns are simply background assumptions that Tomasello does not engage with after an almost perfunctory nod in chapter 2 (even here he focuses more on ethology, the study of animal behavior, than on evolutionary biology or palaeoanthropology). If you want more details, you can always check in on Chris' weblog where he will be commenting and linking copiously for weeks to come (I would characterize him as the John Hawks of cognitive science). Also, I will always update the "Related links" below with weblog posts that deal with the book indefinitely.

First, speaking directly to the audience of this weblog, I will offer a tangential digression about why I am interested in cognitive science. Though I had read Steven Pinker's popular books years ago, my interest in the field was triggered primarily by my encounter with Scott Atran's In Gods We Trust: The Evolutionary Landscape of Religion. I emphasized the evolutionary portion because I did not pick up that book with an understanding that it would be suffused with terminology and paradigms draw from cognitive science, rather, I was looking for a Darwinian account of the growth and development of what I perceive to be a human universal, religious expression. Engaging with Atran led me to other books like Explaining Culture by Dan Sperber and Religion Explained by Pascal Boyer. The jargon overhead in these books (especially Atran's) can be quite high, so I was driven to read Mind Readings by Paul Thagard just so I could get the gist of the terms that assault you in a never ending stream. Cognitive science is in many ways (I feel) philosophy + experiments, and unfortunately that triggers a schema which sometimes casts a pall over my attitude, because fundamentally I am not particularly interested in philosophy as an ends (I read philosophy for its insight into historically important ideologies as well as a dictionary for comprehending what other people are saying). But, nevertheless, I think cognitive science has some direct relevance to many areas of the human sciences that I am interested in, and so that's why I am still venturing into this strange territory.1 This was brought home to me by a concluding review chapter of Atran's book where he addresses "mind blind" theories of human behavior, like sociobiology or group selection. Atran's contention was that looking at humans as cognitive black-boxes and focusing on how they express their behavior, rather than how they perceive and conceive of the world around as well, is a theoretically hamstrung paradigm. I agree. To use an analogy, Dmitriy Mendeleev's periodic table is very useful in its own way for categorization and conceptualizing the relations between the elements (ie; electronegativity, nobility, etc.), but one needed a genuine post-Daltonian atomic theory grounded on the quantum level to make great deductive strides (even octect rules get you only so far!). Similarly, common theses like "proverty leads to terrorism" can be easily empirically refuted (at least in its deterministic strong form), but to go on to address this issue in all its manifestations we need to examine the phenomenon on all of its various levels of organization, starting with the cognitive level (psychology and neuroscience), and working up to macrosocial analyses. Too often economists spit out correlations and regressions, foreign policy analysts often concoct ad hoc verbal models while sociologists and historians try to squeeze behavior into "grand theories" (ie; Marxism, Post-Colonialism, etc.) by looking at the higher order group trends, insteading of starting from the basic level of the individual.

So there is the manifesto, on to the book. As I stated above, this might be my only review because most of the chapters are simply outside of the core focus of my knowledge base. Though I've read spottily in cognitive science (and am an enthusiastic miner of paradigms from it), I don't know much about developmental psychology, which is almost a background assumption beyond chapter 2. So I will touch upon the primary thesis and then go on to some biological implications and assumptions.

Tomasello makes a pretty bald and simple claim: there is basically one species-unique feature of our cognitive phenotype which is at the root of much of what differentiates us from other beasts, and that is our ability to conceive of conspecifics (other humans) as intentional agents like ourselves. This is the overarching necessary and sufficient condition that results in our humanity, our ability to empathize with others and model their behaviors. Of course, the ability to perceive other humans as intentional agents (goal directed) is irrelevant outside of a social matrix, so much of Tomasello's arguments rests upon our intentional modelling capacity being unleashed in distributed networks of information, the "culture." Individual acquisition of information via imitation and socialization leads to the cummulative "ratchet effect" of cultural evolution which results in our spieces' many unique implied traits.

Since the author's hypothesis rests on developmental psychology rather than evolutionary biology, neurobiology or palaeoanthropology, there are many reviews of literature relating to children and blocks and other people. As someone who has a positive affinity for children the descriptions were sometimes distracting because an experiment that recounts how a child pretends that a block is a plane by shouting "Vroom!!!!" induced a mental image of a cute toddler which broke my train of thought. Chapter 2, which ostensibly deals with biological points in favor of his thesis elucidates the differencess between chimpanzees and humans, in particular chimps and infants. Tomasello dismisses assertions of chimpanzee culture, at least in the way humans have culture which is rooted in instruction, imitation and accumulation of novel behaviors. He even implies at some point that part of the problem is that we tend to see intentionality in chimps where there might not be any intentionality (this is common in many fields, one might argue that Intelligent Design is the more prominent current example of this tendency). Tomasello suggests that even young infants tend to engage their parents in a way that chimpanzees simply do not, and offers that the way infants learn behaviors implies an understanding of intent and overall structure of interpersonal relations that chimpanzees have a difficult time conceiving of. For example, human infants and toddlers will often exactly mimic adults in a task even if it is less efficient than another method which is plainly obvious, while chimpanzee "mimics" tend to exhibit a lot more variation and are almost ad hoc in their "imitation." Tomasello gives the example of a chimp mother rolling over a log to get at some ants. Her infant subsequently "copied" this behavior, but Tomasello suggests that many of these behaviors are not imitation as much as a parent adding more information to the infant's data base, which it subsequently acts upon. Now that the infant chimp knows that ants are under the log, it will attempt to get at them, pushing the log out of the way just happens to be the easiest way to do this. In controlled experiments Tomasello points out that chimps often imitate in a scatter shot fashion that suggests that they are not fixed as much on the behavior of the chimp who serves as the model as opposed to the object of interest that the model chimp brings attention to via their behavior. Chimpanzees, in short, live in a world filled with moving animals (which includes other chimps, more or less) that might cause changes and display correlated behaviors, but they are not 3-dimensional Others who are worthy of understanding fundamentally, and, from whom one could learn.

In contrast to chimps, human infants around 9 months (though there are glimmers of species unique behavior before, especially in parent-child interactions) begin to behave in a fashion where other humans beings, as well as the object of attention, become important. In other words, the goals of other human beings and their motivations and tendencies as individuals like the infant itself become noticeable. Whereas before 9 months infants behave "dyadically," that is, in a one-to-one interaction, as opposed to "triadically," where there are multiple relations at work (usually involving the infant, another person, and an object). This ability to interact with other human beings as complex creatures with motives and intentions similar to one's own is the beginning of the ontogenetic ratchet, as humans begin to develop toward a mature cognitive phenotype, developing verbal and cognitive sophistication gradually because of the saturation of social input enabled by the initial spark of intentional thinking. This ontegenetic ratchet is nested within the historical-cultural ratchet, where humans leverage each other's information and spread memes throughout the population via instruction and imitation. Note that in The Meme Machine Susan Blackmore goes to great efforts to dismiss the idea that non-human creatures "really engage in imitation." Some of Toamsello's ideas are reminiscent of her arguments, though he focuses on the issues of "why" more than "how," that is, humans imitate so well because they can conceive of others as intentional agents that one can learn from. Some of Judith Rich Harris' talking points in The Nurture Assumption can also be slotted into Tomasello's paradigm, though he focuses more on parent-child interaction while Harris is fixed on peer groups. To be human is to be social, and sociality is enabled by the intentional concept of others. When it comes to language, Tomasello pretty clearly rejects the idea that language is an innate module, and explicitly says that unlike many people he is skeptical that if everyone over the age of 1 became autistic, but still retained the ability to feed and care for children, language would "naturally" emerge out of the social matrix. Rather, Tomasello, argues that language is a cultural device generated by humans via their symbolic-representational aptitudes which are themselves an outgrowth of their facility with modelling intentional relations. There are similarities with Terrence Deacon's ideas in The Symbolic Species, where much of the argument about language focuses on the thesis that language is simply a subset of the symbolic capacity and that "universal grammar" is just an artifact of the particular biases of the brain's wiring which all "invented" languages slowly converged upon through cultural selection and pruning. If you replaced "intentional agents" with a more generalized "symbolic" capacity much of Tomasello's argument matches that of Deacon's, Tomasello asserts at one point that our models of physical objects are simply mappings of concepts which were initially relevant only to social interactions with conspecifics. While Deacon would argue for a general symbolic ability which can be applied in different contexts, Tomasello seems to argue for a general intentionalizing capacity which is abstracted toward a general symbolic ability.

This is not a conversational popular work, I am eliding a great deal and compressing Tomasello's argument unfairly. But being a blog post, so it goes. Though a certain "Clark" seems to be referenced on every other page, I have no idea who this individual is, or the character of their life, as would be common in a popular work. Tomasello likely excised as much as possible from his argument to compress the prose into 216 pages, and I am not doing justice to the nuance of his argument, in part because I am not totally comfortable with terms like "intersubjectivity" or "perspectival." While I am comfortable with "ontegeny" and other biologically intelligible terms, cognitive science jargon is still a third language for me. So I will move on now to the issues which I know a bit more about, the biological frame that Tomasello works within. He asserts that the capacity for humans to behave as intentional agents arose sometime between 6 million and 250,000 years ago, roughly the period between our last common ancestor with chimpanzees and the emergence of anatomically modern humans in Africa. This is a large window, and he doesn't seem to want to get caught up with the details of when intentionality arose. Additionally, he argues against modularity of various cognitive phenotypes because 6 million years is not long enough for these to evolve. Finally, he throws in the point that 99% of our sequence is cognate with that of chimpanzees to argue that the genotypic difference is minimal. Tomasello emphasizes that it is development, ontegeny, which is at the heart of our differences with our nearest relatives.

First, I will point you to Bora Zivkovic's post on the first chapter of Tomasello's book. It is an understatement to assert that Bora and I have disagreed in the past, but as biologically oriented individuals it was no surprise to me that he basically asks the same questions that immediately came to my mind. Just as I find the Evolutionary PsychologicalTM argument that salient cognitive features must be monomorphic because of the importance of contingency ("half a trait is worthless," a mutation inserted into the mechanistic epistatic network would result in a collapse of the cascade of pathways which result in the phenotype) an exaggeration contradicted by the reality of human variation on many traits, I find the argument that 6 million years isn't "enough time" for evolution to work upon variation to generate species specific modules unpersuasive. I don't think the evidence is conclusive in either direction (calling all cognitive neuroscientists!), and evolution of phenotype does not proceed at a constant rate, but is contingent upon variation responding to selection (OK, at one locus, but I suspect it isn't like selection on correlated traits is always at a steady pace). The amount of variation within a population and the magnitude of selection can both change a great deal over time, ergo, I don't think that Tomasello's assertion at the beginning of the book really holds. Palaeoanthroplogy tells us that there has been a persistent increase in cranial volume over the past 2 million years, ceasing about 200,000 years ago, and, this rate has not been constant (there were spurts here and there). Bruce Lahn's recent work points to persistent directional selection on the genes which control for the size of the brain. There is also research which explores the matter of gene expression in the brain (vis-a-vi chimpanzees, for example). I don't know whether it is the sequence or the expression that matters, but I certainly don't think the question has been answered, and I don't think that 6 million years is piddling (frankly, I don't think a few hundred thousand years is piddling, depending on what traits you are talking about).

Though I am willing to grant that humans are unique insofar as they view others as intentional agents, and that this difference is one of the most important factors that generate our humanity, I think there are other issues that Tomasello gives short shrift to. For example, even if language started out as an emergent property of the social-informational complexity engendered by the ability to peceive others as intentional agents, I don't see why the Baldwin Effect wouldn't result in those who are able to be more linguistically eloquent having a selective advantage, so there was a shift toward innateness over time. Certainly, the fact that there is a "critical period" after which our ability to pick up language seems to dissipate suggests to me that our brain has been "retrofitted" toward "competence" for this phenotype. Additionally, Tomasello juxtaposes language with mathematics, and implies that the former is far more similar to the latter than we perceive it to be. Though I suspect bioengineering and cybernetics will make conventional selection moot soon enough, in some ways I think that mathematical fluency is a good model for what language must have been like before selection for competency on this ability allowed it to be a "natural" ability. The Number Sense chronicles many cases of "mathematical aphasia," but these cases are usually much milder, more diffuse, and more spotty than the linguistic aphasia which are center stage in The Language Instinct (unlike linguistic aphasia, you might not even find out via casual conversation that somone simply lacks the ability to count, and in fact, casual conversation will in all probability yield a lot of false positives as far as mathematical aphasias go). That suggests to me that the coupling between a genetic substrate (either rooted in sequence or expression differences, doesn't matter) and the phenotype is closer in language than in mathematics (which is a retrofit in its first stages and coopts many different areas of the brain in different people). And, since Tomasello's book was published there are tentative signs that language can be "invented" in a very short time by those who are isolated from parental inputs.

Tomasello does acknowledge the possibility for a behavioral module here and there when the fitness impact is totally obvious (he cites David Buss' hypothesis about jealousy for example). But his focus on developmental psychology tends to result in his neglect of many human abilities which seem to result from long term selection via "culture." For example, in neurobiologist William Calvin's latest book he points out that chimpanzees simply do not have the sensory-motor capacity to create tools, even to the sophistication of the Oldowan technology that erectus used between 2.5 and 1.5 million years ago. Perhaps the ability to view others as intentional agents resulted in symbolic thinking, which triggered ideas on how to use tools, but I suspect that these subsequent downstream developments almost certainly had a genetic impact, and resulted in additions to our "cognitive toolkit." The relatively static progression of tool use, with periodic spurts, until about 50,000 years ago is also peculiar. Obviously the ratchet need not be constant...but in terms of technology, and it seems expression of symbolism, its rate of ascendence is highly erratic. Many have used this fact to argue that there was a genetic change that must have precipitated the technological-cultural revolution of the past 50,000 years. Some think that the selective sweep that occurred around 100,000 years ago (give or take tens of thousands of years!) on the FOXP2 "language gene" has something to do with it. Of course, selection doesn't always occur how you'd expect, so evolution might have some more surprises for us.

After all this, I guess I can say that I'm not convinced by Tomasello's argument. I think he has found an essential cog in the whole artiface of humanity, but it isn't the master-cog. I'm not really sure there is a master-cog. There is evidence of both sequence changes and alterations in gene expression in reference to the human brain. It seems clear that selection occurred on our sensory-motor capacities that resulted in our competence in reference to tool use...but obviously, it seems strange that this would develop before we used tools, so why did we starting using tools a few million years ago anyhow? FOXP2, which is a regulatory gene that has some relevance to language fluency, as well as general intelligence, swept through our species a hundred thousand years ago, or earlier. Symbolic culture seems to have really taken off less than 100,000 years ago. This is a really knarly bush of theoretical contingencies and possibilities, and Tomasello has evaded the biological ones by simply moving past them very quickly. If I had to bet, I would guess that the ability to view each other as intentional agents is relatively recent, perhaps within the last few hundred thousand years. I do think it is a crucial change in our conception of the world, but, I believe it also set into motion changes which resulted in other selective forces being unleashed, and language arose as a competency in its wake. Our ability to use tools and manipulate physical objects in a complex manner though preceded the intentional mind, and in fact might have somehow set the stage for selection for this trait. Perhaps Tomasello has it backward, and somehow we began to imagine other human beings as tools, to be shaped and bent to our own uses. If my timeline is right the major push toward expansion of the human cranium and selection on those loci which affect brain size occurred before the major switch in our conception of the world.

The Cultural Origins of Human Cognition is a very good book. It is a bit dense if you are a cognitive science virgin, but if you've made it only to first or second base (like me) it won't offend your sensibilities and tax your level of expertise too much. It reinforces some points I have made many times on this weblog, for instance, that children learn many ostensibly cause-and-effect conceptual structures through imitation, that their causative nature is only apparent to the initiated and culturally fluent. Etiquette is a clear example, you behave like so because to not behave like so is "rude," but as we all know, why it is rude is not always easy to pinpoint (or, that's what I claim!). Ultimately, rudeness is often based on social deviation. The conception of others as intentional agents worthy of some empathy and consideration is certainly very important. But I doubt it is the whole story by a longshot, and I would take bets on that. But it is a big enough story that it deserves a lot of attention.

Related links: Cultural Origins of Cognition: Introduction and Context, and intro at Chris' weblog. Clark on chapter 1. Jesse riffing off Clark. Bora on chapter 1. Blar on Santayana and Tomasello. Chris on chapter 1. Here is a paper that comes close to being a precis for the book, and a recent paper which tests some of his hypotheses. A piece in Scientific American reports on research published in Nature which directly contradicts Tomasello's assertion that chimpanzees do not have imitative culture. Of course, the conclusion is up for dispute, but the quote at the end explicitly rejects the thesis that it was attention to the goal rather than the task itself that the chimps were focused on. Chris has a post on autism that readers might find interesting. Chris on chapter 2. Clark with chatper 1 bonus.

Related on this blog: Dusk of Human Culture, The Mating Mind, Grooming and Gossip and Mother Nature.

See more on CogBlogGroup Technorati tag, CogBlogGroup tag and Culturaloriginsofhumancognition tag

Technorati tag:

1 - I won't deny that philosophy also has relevance on higher orders, but I feel that "philosophy of...." fields are often too detached from the discipline they are devoted to analyzing (ie; science, religion, history, etc.).