The Cultural Origins of Human Cognition by Michael Tomasello was the first book (and still the only one so far) we were reading in the newly minted CogBlogGroup, a group of bloggers reading stuff about cognitive science. You can download the whole book in PDF or the first chapter only in html. This was the first of two parts (I never finished the book nor rerview!), originally posted on August 20, 2005:
Chris of Mixing Memory has written an introductory post providing a broader context and background. Some of the participants have already posted their commentaries on the First Chapter, which I will eagerly read once I finish posting this one. Here is Jesse on Chapter 1 and David Clark on Chapter 1. Here is also David Clark's collection of all relevant links so you can follow the discussion there if you do not like the Yahoo group linked above.
The first chapter is quite short and introductory. Tomasello sketches briefly the main thesis of the book by presenting the problem and offering a solution. My take on the first chapter may sound too critical, but I find the thesis intriguing and I hope that Tomasello answers some of my current questions later in the book. Also, my declarative sentences should be read as questions, or as provocations for the other participants in the discussion.
A Brief Summary of Chapter 1:
First, let me briefly summarize (actually mostly through quotes from the chapter), what Tomasello covers so far:
Problem, according to Tomasello:
The basic puzzle is this. The 6 million years that separates human beings from other great apes is a very short time evolutionarily, with modern humans and chimpanzees sharing something on the order of 99 percent of their genetic material - the same degree of relatedness as that of other sister genera such as lions and tigers, horses and zebras, and rats and mice (King and Wilson, 1975). Our problem is thus one of time. The fact is, there simply has not been enough time for normal processes of biological evolution involving genetic variation and natural selection to have created, one by one, each of the cognitive skills necessary for modern humans to invent and maintain complex tool-use industries and technologies, complex forms of symbolic communication and representation, and complex social organizations and institutions. And the puzzle is only magnified if we take seriously current research in paleoanthropology suggesting that (a) for all but the last 2 million years the human lineage showed no signs of anything other than typical great ape cognitive skills, and (b) the first dramatic signs of species-unique cognitive skills emerged only in the last one-hundred thousand years.
- There is a very close genetic relationship between humans and closest relatives, suggesting they do not differ in a whole suit of traits, i.e., "modules";
- 6 million years, 2 million years, and especially 100,000 years is too short period for evolution of a whole suite of cognitive abilities.
Solution, according to Tomasello:
There is only one possible solution to this puzzle. That is, there is only one known biological mechanism that could bring about these kinds of changes in behavior and cognition in so short a time - whether that time be thought of as 6 million, 2 million, or one-quarter of a million years. This biological mechanism is social or cultural transmission, which works on time scales many orders of magnitude faster than those of organic evolution.
- Humans differ from primate ancestors in only one biological trait;
- Humans evolved complex culture via the mechanism of cultural evolution.
What is the type of cultural transmission? Collaborative learning:
The basic fact is thus that human beings are able to pool their cognitive resources in ways that other animal species are not. Accordingly, Tomasello, Kruger, and Ratner (1993) distinguished human cultural learning from more widespread forms of social learning, identifying three basic types: imitative learning, instructed learning, and collaborative learning. These three types of cultural learning are made possible by a single very special form of social cognition, namely, the ability of individual organisms to understand conspecifics as beings like themselves who have intentional and mental lives like their own.
For collaborative learning, one needs a "theory of mind", thus the single newly evolved trait is the ability to, for lack of a better word, feel "empathy":
This understanding enables individuals to imagine themselves "in the mental shoes" of some other person, so that they can learn not just from the other but through the other. This understanding of others as intentional beings like the self is crucial in human cultural learning because cultural artifacts and social practices - exemplified prototypically by the use of tools and linguistic symbols - invariably point beyond themselves to other outside entities: tools point to the problems they are designed to solve and linguistic symbols point to the communicative situations they are designed to represent.
The complete sequence of hypothesized evolutionary events is thus: human beings evolved a new form of social cognition, which enabled some new forms of caltural learning, which enabled some new processes of sociogenesis and cumulative cultural evolution. This scenario solves our time problem because it posits one and only one biological adaptation - which could have happened at any time in human evolution, including quite recently. The cultural processes that this one adaptation unleashed did not then create new cognitive skills out of nothing, but rather they took existing individually based cognitive skills - such as those possessed by most primates for dealing with space, objects, tools, quantities, categories, social relationships, communication, and social learning - and transformed them into new, culturally based cognitive skills with a social-collective dimension. These transformations took place not in evolutionary time but in historical time, where much can happen in several thousand years.
That is, my specific hypothesis is that human cognition has the species-unique qualities it does because:
Phylogenetically: modern human beings evolved the ability to "identify" with conspecifics, which led to an understanding of them as intentional and mental beings like the self.
Historically: this enabled new forms of cultural learning and sociogenesis, which led to cultural artifacts and behavioral traditions that accumulate modifications over historical time.
Ontogenetically: human children grow up in the midst of these socially and historically constituted artifacts and traditions, which enables them to (a) benefit from the accumulated knowledge and skills of theIr social groups; (b) acquire and use perspectivally based cognitive representations in the form of linguistic symbols (and analogies and metaphors constructed from these symbols); and (c) intemalize certain types of discourse interactions into skills of metacognition, representational redescription, and dialogic thinking.
OK, now to my response. First, let's look at the problem, as seen by Tomasello. Let's, for now, take both parts of the problem at their face values, i.e., as if they are correct.
If there is very little difference between chimps and humans, there is no "space" for more than one new trait, no matter how much or how little time evolution has to create new cognitive traits.
Likewise, 6 million years is too short a time for evolution of more than one trait, no matter how close or distant relatives humans are to chimps.
Having both of these problems simultaneously is a double-whammy - it makes it even more plausible that the difference is only one cognitive trait. But, are the two problems true?
Are chimps and humans really that close?
I am surprised that Tomasello used that old figure of 99% similarity between humans and chimps. The number actually varies between about 95% and 99.9% depending on who is citing. That number was derived from a couple of old studies on DNA/DNA hybridization, i.e., a direct comparison between sequences of genomes of the two species.
But what does that number mean? There is a whole book devoted to the subject: What It Means to Be 98% Chimpanzee (a very readable, fun and smart book - highly recommended for lay people). Actually it means very little. We are almost as close to zebrafish and fruitflies. Have you seen a chimp lately? Does its anatomy looks 99% similar to human? How about its behavior? About 99% similar to human? Would you say that we look about 80% like fruitflies?
Gene sequences are bound to be similar because widely diverse organisms keep using same genes and same suites of genes for same purposes. If there is a suite of genes that is good for making a protuberance and you need a horn, and a mutation happens to switch on that suite during the development of the forehead, you WILL get a small horn that can be made larger or sharper or whatever by further selection. Unicorns are not impossible in principle, just highly unlikely (actually there are horses with two little bumps on their foreheads - those are invisible to selection because stallions do not fight by head-butting). A beetle can get a horn in the same place by switching on that very same suite of genes. There are groups of genes that are good at making tubes (like intestine), or making segments, or making solid transparent bodies (e.g., lens of the eye), etc. What makes species different from each other are small mutations in regulatory genes that affect development.
In short, we are different from chimps not so much because of differences in DNA sequence, but because of differences in patterns of expression of same genes during embryonic development.
Tomasello describes, a couple of times in this chapter, evolution as a result of "mutation and natural selection". That is a simplistic two-step understanding of the process: Step 1) random mutations occur in genes; and Step 2) better adapted phenotypes, if heritable, leave more offspring in the subsequent generation.
But, how do you go from mutations in genes (in DNA hidden in cells and invisible to selection) to phenotypes that can be selected? Tomasello does not appear to be aware of the three-step model: 1) mutations in genes result, via interactions with the environment, in 2) changes in developmental trajectories leading to new phenotypes that are 3) visible to natural selection. Genes are toolkits for developing an organism. Tools can be moved from one place to another (heterotopy), or used earlier or later than before (heterochrony), or somewhat modified (heterotypy), or affects the way developent responds to the environment (reaction rate - e.g., an organism may grow larger if it is warmer, and a mutation in a regulatory gene can switch it over, so the organism grows smaller if it grows in a warmer environment), etc. Thus, most of the differences between two closely related species (e.g., humans and its ancestors) are a result of developmental reorganization which depends on small and subtle mutations in a very few regulatory genes.
Anatomical traits are usually most "fixed", e.g., you will develop five fingers on each hand pretty much no matter what kind of environment you are developing in. Physiological traits are a little bit more malleable. Behavioral traits, even fixed action patterns, are much more flexible. Cognitive traits are to be expected to be the most flexible traits of all - highly dependent on environmental input.
Cognitive traits are unlikely to be based on a brand new gene, or even a brand new brain structure. It is much more likely to be based on changes in connectivity between neurons. During development, neurins migrated along concentration gradient of various products of developmental genes. Very subtle changes in placement, timing or concentration of these product can have relatively large effects on the final architecture of the brain. Also during brain development, neurons form many connections (synapses) - much more than needed. As the nervous system start being used, those synapses that are often used get reinforced, while other connections detach and die off. This is called Neural Darwinism. This process begins in the embryo - your baby kicks in utero not for fun, but because it is using the muscles, thus informing the brain which synapses are useful and which are not. In other words, in order to get a new cognitive trait, like "empathy", you do not need to evolve a whole new gene, just to reorganize the way the brain is developed by an unchanged suite of genes.
In theory (but highly unlikely in practice - I cannot immagine a selective regimen that would bring this about) it is possible to have two species that are almost identical in DNA sequence, the difference beeing just a few nucleotides (DNA base-pairs). Those two species would have identical anatomy, identical physiology, identical most of the basic behaviors, yet one species would be morons, and the others geniuses. That is, if those few mutations are in the regulatory genes that guide brain development.
So yes, chimps and humans are close, but not THAT close, i.e., the number 99% is meaningless. The remaining 1% of genetic difference can potentially account for an enormous phenotypic difference that can account for vast differences between species. Subtle shifts in brain development are all that is needed.
This in no way affects Tomasello's solution #1 - having just a single new trait - but makes me extra careful while reading the rest of the book, as he does not seem to be up to speed on evolutionary theory, while writing a book on evolution. I hope that this is just the introductory chapter and that later on he will get more sophisticated, describing both brain development and evolutionary process in a better way than here. If you want to get up to speed on the most current thinking on evolution, I strongly suggest this, easy-to-read book.
Was the time too short for evolution of multiple modules?
Again, relying only on mutation and natural selection (and calculations by population geneticists based on the 2-step model) one may come to the conclusion that 100,000 years is too little time to evolve a bunch of new traits. Two million also looks short. Even six million looks short. But, there are a number of ways the process can speed up. One, or more, or all of the following processes may speed up evolution of a trait.
Baldwin Effect is a process by which flexible, learned behaviors, become "fixed", i.e, incorporated in the genotype. Evolution of language is the only case hypothesis put forward and no empirical examples of Baldwin effect have been described. Terry Deacon's hypothesis for evolution of language utilizes Baldwin's effect to some extent and his book contains probably the clearest description of how the process MAY work. Loss of flexibility that the process results in is unlikely, in my mind, to be of selective value when one thinks about cognitive processes and language. So far, it appears that Tomasello is working toward some kind of process similar to this. We'll wait and see (Click on various links in this paragraphs for some discussions on Baldwin's effect in evolution of consciousness and language).
Niche construction is a term for an evolutionary process that entails a feedback (or feedforward) interaction between natural selection and the environment. In Niche Construction, an organism modifies its environment and this modified environment is now selective environment for the next generation that, in turn, further modifies the environment. In evolution of language one can imagine a situation in which greater and greater "eloquence" of people provides the selective environment in which natural selection would favor individuals who learn languge sooner and better and are more eloquent than the other individuals. This feedforward loop will greatly speed up the evolutionary process.
So far in the book, it appears that Tomasello is looking only at one prong of the feedforward loop - the cultural evolution - and I hope that later he also brings in the other prong - the natural selection - and comes up with something similar to niche construction, though he may give it another name.
Sexual selection can be a very fast process - akin to strongly directed natural selection. Being more emphatetic, or more eloquent, may be a target of sexual selection. A sweet-talking 'son of a preacherman' may be more attractive (and have better pick-up lines) than someone with a vocabulary of five grunts. An eloquent woman is more fun to be around ("Give it to me now" instead of just "Aaaaah!" - just joking). An eloquent man will naturally become a leader in the hunt, issuing orders to the others ("Grok, you go left, Grub, you go right, I'll be in the middle, and let's chase this mammoth over that ledge over there and into the abyss"), thus rising in the tribal hierarchy. An eloquent woman will, likewise rise in the hierarchy of the tribe. A person who is better able to "read" other people will be more able to manipulate other people and likewise rise in the hierarchy......You see how it goes: the more emphatetic and more eloquent individuals of both sexes winning the game of sexual selection and passing more of their genes into the next generation than those less endoved with such cognitive skills.
Elliott Sober abd David Sloan Wilson are reintroducing group selection to the human behavioral sciences. Their book, Unto Others develops the (so far) best mathematical model for group selection, then applies it to the evolution of altruism. That is one of the most important books in late 20th century evolutionary theory. Wilson then followed up on it with applting group selectionist thinking to the evolution and adaptive function of religion, another seminal and provocative work. Possibly the best and clearest explanation of multilevel selection (not just group selection) is Chapter 3 in Adaptation and Environment by Robert Brandon, one of the most important works in recent philosophy of evolutionary biology.
The thinking that human consciousness originated by group selection has an old and noble origin - Charles Darwin's Descent of Man. So, here are two (out of many similar) excerpts from that book describing possible evolution of consciousness:
Every one will admit that man is a social being. We see this in his
dislike of solitude, and in his wish for society beyond that of his own
family. Solitary confinement is one of the severest punishments which can
be inflicted. Some authors suppose that man primevally lived in single
families; but at the present day, though single families, or only two or
three together, roam the solitudes of some savage lands, they always, as
far as I can discover, hold friendly relations with other families
inhabiting the same district. Such families occasionally meet in council,
and unite for their common defence. It is no argument against savage man
being a social animal, that the tribes inhabiting adjacent districts are
almost always at war with each other; for the social instincts never extend
to all the individuals of the same species. Judging from the analogy of
the majority of the Quadrumana, it is probable that the early ape-like
progenitors of man were likewise social; but this is not of much importance
for us. Although man, as he now exists, has few special instincts, having
lost any which his early progenitors may have possessed, this is no reason
why he should not have retained from an extremely remote period some degree
of instinctive love and sympathy for his fellows.
And natural selection arising from the
competition of tribe with tribe, in some such large area as one of these,
together with the inherited effects of habit, would, under favourable
conditions, have sufficed to raise man to his present high position in the
So, even if genetic similarity between humans and chimps was a problem, the time is not neccessarily the problem - evolution can proceed faster than population genetics models predict. Still, the fact that none of the two problems that Tomasello cites are real problems, does not mean that he did not strike at a correct solution. So, let me now turn to Tomasello's proposed solutions.
A single novel trait (as opposed to multiple modules)
Just because there is enough genotypic/phenotypic space for many modules, and enough time to evolve multiple modules, does not mean that multiple modules actually evolved. I actually think that it is likely that a single change in brain development and functioning was all that was neccessary for the evolution of human-level consciousness. We'll see if Tomasello manages to persuade me further.
The new trait is 'empathy'.
Maybe yes, maybe no. Is "understanding others as intentional and mental agents like the self" neccessary? Thought experiment: you are an early hominid. You watch an elephant use an axe to cut down a tree (I know elephants don'a make axes, but I do not want the agent to be human). Do you really need to be able to see "through the elephant's eyes" in order to recognize that an axe is useful for cutting down a tree? Can't you just look at it and try to make one for yourself and test it on a nearby tree? Do you even need to watch an elephant doing it? If you found an axe on the ground, how long do you think it would take you to discover that it can be used to cut down a tree? A day or two, a few weeks perhaps?
If 'empathy' is what is selected for, why did we not evolve into sharp mind-readers instead of inventing a second mode of communicating intent and mental states, namely the language? Isn't language actually better in communicating that? You can talk about someone who is not even present and wonder what s/he is thinking or feeling. "My woman is at home and she must be hungry - I better bring her back a piece of this mammoth" - but birds and most mammals bring food home to family they do not see while hunting. So, what's new?
The outcome is that each child who understands her conspecifics as intentional/mental beings like herself--that is, each child who possesses the social group - can now participate in the collectivity known as human cognition, and so say (following Isaac Newton) that she sees as far as she does because she "stands on the shoulders of giants." Importantly, we may contrast this species-typical situation with that of both:
- children with autism, who grow up in the midst of cumulative cultural products but are not able to take advantage of the collective wisdom embodied in them because, for biological reasons, they do not possess the requisite social-cognitive skills; and
- an imaginary wild child who grows up on a desert island with a normal brain, body, and sense organs, but with no access to tools, other material artifacts, language, graphic symbols, writing, Arabic numerals, pictures, people who could teach her things, people whose behavior she could observe and imitate, or people with whom she could collaborate.
For the child with autism there are cognitive shoulders to stand on, if only she could, whereas for the imaginary wild child there are no cognitive sholders to stand on. In either case the result is, or would be, the same: something other than species-typical cognitive skills.
But, a bird that is heavily parasitized and under-nourished will not be able to learn his species-specific song (or just rudiments) and thus will not be able to defend a territory or attract mates and thus will be eliminated by both natural and sexual selection. What's new in autistic humans?
A bird in perfect health that is raised in captivity also never learns its species-specific song and, if let loose, would be selected against. What's new in wild-child humans?
This first post is likely to be the longest as I put out all of the heavy artillery up front. After briefly sumarizing problems that Tomasello is trying to solve and the solutions to those problems that Tomasello is propsing, I have listed a number of potential criticisms of Tomasello's hypothesis. As I read the rest of the book, I will keep checking the list to see if Tomasello manages to eliminate some of the criticisms from the list. Whatever still remains at the end of the book will still remain a problem for Tomasello.
I have also placed everything here in the first post so I, as well as other participants in the reading group, can refer back to this when we discuss potential criticues in the future. If you are not familiar with some of the terms or concepts, please click on the links for more information.
Update 2: So did Razib.
Technorati tag: cogbloggroup