Of Minds and Language (54 page)

That is the idea in a nutshell. Now I want to take it a little bit further. The main idea in Noam's opening remarks (
Chapter 2
) and alluded to in Gabby Dover's paper (
Chapter 6
), is that rather than having thousands and thousands of variants, we have one animal, one blueprint. I know that Gabby is somewhat opposed to this metaphor, but the idea is that we have some kind of conserved process that is generating all the variation. Of course one of the nicest stories to come out in the last twenty or so years is the account related to the Hox gene system, where we can see a direct mapping, remarkably similar from the drosophila embryo all the way up to the mouse embryo, between genes that are basically building segmentation in the body forms. Patterns of segmentation are being driven by evolutionarily ancient genetic mechanisms that have been conserved over evolutionary time. Part of the reason why this is important is because it has changed the nature of the way we think about the notion of homology. If you were simply to focus on anatomical form, and note how different things look, you would be missing the underlying genetic similarity that is extremely conserved and is homologous in that sense.

A second is a recent paper by Bejan and Marden (2006) which has received a lot of attention, especially from functional morphologists, claiming that all patterns of animal locomotion can be explained by equations relating force, energy, frequency, and mass. And just to give you one example of a beautiful fit, they consider the relationship between body mass on the X-axis and force on the Y-axis. In such a plot, it becomes evident that all the animal species that move by
running (mammals, reptiles, and insects), all those that fly (birds, bats, and insects), and those that swim (fish, marine mammals, and crayfish) all align beautifully. Now much of the controversy surrounding these results is that experts in the field claim the analyses fail to account for complexities and variations observed. I am sure this is correct, but my guess is that the history of this debate will end up looking a lot like the debates in linguistics, where there are going to be some battles about the details, but what seems to be captured here are generalities.

The move that I want to make is that, given the kinds of depth of investigation that have gone on in biology over the last thirty to forty or more years, when variation has in some sense been put to the side for the purpose of looking at explanatory mechanisms, there is a common theme that seems to keep emerging. What I want to ask now is basically whether that kind of move can be adopted in thinking about the nature of the human mind. Thus, for example, it certainly appears to be the case that there is limitless cultural variation. Can we account for it by some simple, primitive mechanisms, and then use pruning as a mechanism for selecting among the possible, biologically given variants? To test this question, we need to run the universal minimalist program of research. We first look for a core set of rules or mechanisms with a generative power of expression, interfacing with specific forms of knowledge. We next explore whether these mechanisms are present in other animals and the degree to which their presence in humans is unique to a domain of knowledge or more domain-general. Then we run the comparative analysis from genes to behavior, attempting to understand what limits the phenotypic space.

This kind of approach raises a paradox to keep in mind. I don't think either Randy Gallistel or I want to be taken as saying that there is nothing unique about humans at all; only that the comparative evidence we have presented shows there are extraordinary abilities in animals and it is important to keep this in mind. Here, however, is the paradox I want to point to today. Gabby Dover mentioned the genomics of humans, and of course one of the most interesting things about the study of genomics today is the fact that if you look at the genetic relationships or similarities between chimpanzees and humans, they are far more similar to each other than are chimpanzees and humans as a cluster to gorillas. Now that is surprising again if you think about their anatomy. Chimps look much more like gorillas than they do like human beings, and yet at the level of genetic similarity chimpanzees cluster with humans and not gorillas. That said, if we leap now from the anatomical level and genetic levels to the psychological level, we are faced with a fundamental problem. If we take some of the towering intellectual achievements in our history (and even some of the less towering intellectual achievements), the gap between us and them is extraordinary; in fact I would say it is larger than the gap we see between gorillas and
chimpanzees on the one side, and the humble beetle on the other. So we have to somehow come to grips with the fact that the genetic level of similarity is not accounting for the psychological variation and differences we see.

So here is the outline for what I want to say in the rest of the present paper. I want to run through three examples. I am going to first come back to language in the way that I spoke about in my previous paper here, and I will flesh out a little bit more of the argument and present some new data that bear on the conceptual richness in non-human animals. Then I will turn to some parallel arguments about the nature of the moral faculty. Then I will turn to music as another domain in which we can basically begin to ask similar kinds of questions, and finally I will end with some summary points about nature's solution to the various kinds of problems about variation and unity.

So first, language. As I described earlier in this conference, I am going to think about language as
a mind-internal computational system designed for thought and often externalized in communication
, and as such,
language evolved for internal thought and planning and only later was co-opted for communication
.
²
What I want to do now is use this hypothesis as a wedge to pinpoint a disagreement in the literature which I think unfortunately misses the point. But I am going to use it as a way of showing some data that I think actually bear on the argument, capturing the difference between the way that Noam has talked (in this conference and elsewhere) about the internal computational system of language, and the way that Steven Pinker and others have talked about language as an adaptation for communication – a distinction that at some level is virtually impossible to resolve, because language is used for both functions, and the question of evolutionary origins is notoriously difficult, especially for such a complicated trait.

Let us return to the FLB–FLN distinction that I raised earlier on (see also Hauser et al. 2002) and that Cedric Boeckx picked up on in his paper here (see
Chapter 3
). Let us begin to think about the ways in which understanding of what goes into FLB vs. FLN can help us think about the nature of the evolutionary process vis-à-vis the internal computational system for thought and planning and its externalization in spoken language, or sign language. The hypothesis that Chomsky, Fitch, and I have been pinned with is what Pinker and Jackendoff (Pinker and Jackendoff 2005) have called “the recursion-only
hypothesis.” But that is not actually what we said (Hauser et al. 2002). What we said was that FLN – as an hypothesis – consists of the computations that enter into narrow syntax (we specifically spoke about recursion) and the interfaces to semantics and phonology. We think this is a useful way to frame the problem because it forces one to look not only at the evolution of the computational system alone, but also, how it interfaces with and is constrained by the other mind-internal systems. This move opens the door to interesting comparative issues, which I turn to next.

If we look at the songbirds, species that learn their vocalizations based on some innate structure that guides the process of acquisition with critical periods and windows of opportunity very much like language acquisition, what we see are exceptional capacities for vocal imitation. This is especially true in the open-ended song learners like starlings and lyrebirds and mockingbirds – very complex streaming together of sound patterns that looks like a rich combinatorial system; but no meaning. The variation that you see in the songbird system does not generate new meaning, it is simply “I'm-Fred-the-sparrow-from-New-York.” And we're finished, that's it. Change the variation a bit and it's “I'm-Joe-from-California. But I'm still a sparrow, and the meaning or function of my song is simple: I have a territory and I am looking for a mate.”
Fini
! This is equally true of humpback whale song – again, very complicated, but the variation yields no new meaning. Now, in my lab, we have recordings of a starling doing its own version of a goat and a chicken, as well as starling song material. Functionally, they use these songs in mate attraction, but they also use it as a sort of “No-vacancy” sign. They flood the habitat and say “You don't want to come here – there are goats and chickens and all sorts of other things around here. Don't bother!” Now monkeys and apes, in striking contrast, show no evidence for vocal imitation. There is no capacity (and it has been fifty years of intensive looking by primatologists), absolutely no evidence for vocal imitation. Primates typically do not string their calls or notes together: no combinatorics evident at all, and weak meaning if at all in their vocalizations.

Thus far, most of what I have focused on concerns the sensorimotor side. Now I want to come back quickly to animal concepts. People talk about the Galilean revolution; I like “Gallistelean” for talking about concepts. I think Randy Gallistel has done a great deal to help the cause in thinking about animal concepts, especially in terms of the notion of isomorphism. For now, I want to focus on intentionality, and in particular, the puzzle concerning the richness of animal mental life and the poverty of their communicative expressions. It is what I have often described as the metamorphosis problem after Kafka's story, in which Gregor Samsa,
qua
beetle, has profound thoughts about the world, but cannot convey them.

Let's start with physiology, and mirror neurons in particular, and then build up to thought and behavior. In the mid-1990s, Giacomo Rizzolatti was recording from neurons in the pre-motor cortex of a macaque monkey when he noticed that cells firing in response to observing an experimenter grasp an object also fired when this same monkey grasped the same object, in the same way (Rizzolatti et al. 1996). Further recordings, from other cells in the premotor area, revealed a kind of gestural repertoire: cells firing when the action itself is in the repertoire of the animal, and the animal either performs the action or observes another individual performing the same action. There is a linking or coupling between perception and action. Now what I want to show you is how you can take these physiological findings and look at how they may be instantiated in real world behavior in animals trying to make decisions about goals in the world and what they pay attention to when they perceive somebody acting in certain ways.

I am going to explain a study on the island of Cayo Santiago that was carried out by one of my terrific graduate students, Justin Wood (Hauser et al. 2007; Wood et al. 2007). The star of the show, besides Justin, is the rhesus monkey again, and here is the experimental paradigm. You find an animal alone on the island. The monkeys on Cayo Santiago love coconut. Unfortunately, in over eighty years of living on this island, no single individual has ever figured out how to open one. Now this is a problem, because it is the most preferred food. If I crack open a coconut, they all come running. They call, they are very excited and so forth, but they can't work out how to open one on their own. However, they seem to understand that
we
can figure it out, so whenever we move towards some coconuts, they know that something interesting may happen, and it may be for them. This sets up a simple experiment. You show a subject two half coconuts face down. Because they are face down, but already cracked open, they can't see what is inside, but it is possible that one or both coconut halves has some flesh. For each experimental condition, the experimenter places these two half coconuts on the ground, face down, while the animal watches, and then approaches one of the coconut halves and interacts with it in some particular way before walking away. The psychologically relevant question is: does the particular form of interaction or action by the experimenter on the half coconut influence where the subject searches? The results of this study reveal the proportion of subjects selecting the coconut acted upon by the experimenter. For each condition, we use one subject per trial, but multiple animals (between twenty and twenty-four) per condition.

In the first condition, the experimenter simply grasped the top of the coconut but didn't lift it, and then walked away. Here, approximately 90 percent of the subjects approached the coconut that we grasped. Similarly,
grasping the coconut with a pincer grip (i.e., index finger and thumb) also resulted in a selective approach to this coconut – approximately 85 percent of subjects. Interestingly, there are cells in the mirror neuron system that distinguish between a full hand grasp and a pincer grip; that is, cells that fire to a pincer grip do not fire to a full hand grasp, and vice versa. In a third condition, we grasped one coconut with a bare foot. Though the rhesus have never seen humans grasp in this particular way, rhesus will use their feet to pick up food, especially when they are hungry and attempt to carry as much food away as possible, using both hands, feet, and their cheek pouches. In parallel with the first two conditions, 90 percent of the subjects go to the foot-grasped coconut. In the final condition in this series, we asked whether rhesus need to see the target goal in order to infer the subject's intentions or whether they can draw this inference when the goal is occluded or out of view. Mirror neurons will fire when an agent reaches for and makes contact with a visible goal as well as when the goal is occluded. Similarly, rhesus selectively approach the occluded coconut when the experimenter reaches for it behind an occluder. Here, therefore, is a class of behaviors or actions that result in selective approaching behavior by rhesus. But there is a simple, and rather trivial explanation for all of these results, which would be that rhesus approach anything that an experimenter touches. If this is the rule they are following, then it is rather uninteresting, explained by simple associative mechanisms. If this is the proper interpretation, then any contact, intentional or not, with one coconut, should lead to selective approach. I turn next to conditions that directly explore the nature of the contact between experimenter and coconut.