Cracking Open Overimitation
Your ability to retain and reproduce without understanding gives you time to reach that understanding.
Again from How Traditions Live and Die. As I mentioned here, this book is arguing against many of the ideas put forward by prevailing theories of cultural transmission (with which I mostly agree). So, I will likely find myself occasionally using this space to quote from that book in order to argue back and sharpen my thinking—as I do here.
Authors like Tomasello, Boyd, or Richerson usually point out that rudimentary animal traditions may be preserved without true imitation, but cannot be gradually improved. Such progress is the privilege of cumulative cultures: cultural transmission only sustains cumulative traditions when it is based on true imitation or teaching.
There are a couple of issues to disentangle in this before we go further. First, let's look at how the author talks about 'true imitation' (emphasis mine):
The standard experimental test to determine whether an animal readily imitates in this particular sense ['truly imitative'] is the "two actions method". . . . One of the most famous uses of this method (a not quite standard case, but a simple one) comes from one of Andrew Meltzoff's experiments. The subjects, toddlers, are faced with an adult model and a half-sphere of plastic that lights up when pressed upon. Instead of activating the toy with her hand (which seems the most direct and simple gesture), she does it with her head. For another group of toddlers (the controls), the regular hand gesture is used. Psychologists have deemed human toddlers willing and able to "truly" imitate, because toddlers in the test group use their head much more often than those of the control group. An apparently useless gesture was slavishly imitated.
The language used in this quote (setting aside slavishly imitated for the time being) is more than a little sneaky. What experiments like the above reveal is, simply, that humans naturally overimitate (not 'truly' imitate) from a very young age. The experiments are not designed or employed as filters or tests to weed out 'true imitators' from others (usually non-human primates).
Second, the fact that human toddlers overimitate—remember and reproduce causally opaque moves demonstrated by a model (moves that are manifestly unnecessary to achieving a goal)—does not mean, contrary to what the author seems to suggest in the first quote, that overimitation is the mechanism by which cumulative cultural transmission is sustained. Overimitation points to a set of unique human capacities that allow the cultural ratchet to do its work. It is not a comprehensive description of how that work is done in the real world with adults. The absurdities that emerge as the author continues are, to my mind, more a consequence of this misunderstanding rather than of underlying problems in prevailing theories of cultural transmission.
Why should true imitation [overimitation] have this effect? Because, the argument goes, true imitation retains "good tricks" that would vanish without it. This may sometimes be true; but it is also dangerously close to what we might call "the Hoarder's Fallacy." I once had as a neighbor a man who kept every single wrapping, box, tin can, old newspaper he ever had; the police eventually forced him to empty his basement (which had become a fire hazard). When asked about all this, he invariably replied that "those things might serve some day." Surely, they might—but it would not hurt to be a little more selective in keeping things. True imitation, as defined so far, is such a hoarder. Of all the social learning mechanisms that have been described, it is one of the stupidest. Suppose I am learning to crack nuts from a master of that trick. If I try to copy as many gestures as I can, I will reproduce many useless (but voluntary) gestures (like scratching my head in such and such a way). Of all the things that I reproduce, there will of course be useful gestures that I can recognize as such; there will also be, perhaps, a handful of opaque gestures, the use of which is not immediately perceptible. The key contribution of "true imitation" is supposed to lie in enabling the persistence of such opaque tricks; but what a wasteful form of conservation it is! The utter lack of selectivity implied by standard definitions of "true imitation" is not only inefficient: it is hard to see how such faithfulness could be sustained in a reasonably extended transmission chain.
Again, in the real world, people are not routinely passing along "opaque tricks" via the mechanism of overimitation used by toddlers. On the contrary, by the time you are learning to crack nuts—from an elder in your tribe, say—you are presumably no longer a toddler. Other more 'sophisticated' techniques, such as inference, selection, and the use of language, now can interact in complex ways with your capacity for overimitation. For example, your ability to retain and reproduce some of your model's gestures without understanding gives you and others time to reach that understanding or a partial understanding (via those other techniques) without slowing down cracked-nut production within the tribe.
