Of Minds and Language (18 page)

All is subjective, not just free will: it cannot be otherwise given the bizarre paths taken by evolved heaps of life, with their re-usable and promiscuous modular units.

P
IATTELLI
-P
ALMARINI
: It is certainly refreshing to see a geneticist saying that there is no difference between
innate
and
acquired
. In the world of language, I always receive this with grave concern. You know, some of my colleagues say the same; in linguistics a couple of people at MIT say the same, that we should abolish the
innate/acquired
distinction. I usually receive this with great concern because I can see where that's leading.

G
ELMAN
: I can think of no worse or more unacceptable message to take back to developmental psychologists. This is that it's all right to continue thinking that the mind is a blank slate. Your reason: just because you said so. But many in my field do not understand the fundamental problem, which is that we are dealing with epigenesis and hence the interaction with mental structures and a very complicated environment that has the potential to nurture the nascent available structures. The notion of what is given has to be stated differently, in a way that does not pit innate against learned. If we buy into the standard learning account offered by various empiricists, then we are once again assuming a blank slate: that is, no innate ideas, just the capacity to form associations between sensations and do so according to the laws of association. In this case you don't need any biology. For me there is no reason to pit innate against learned. To do so is to accept the widespread idea that there is but one theory of learning. Put differently, it allows empiricists to commandeer the learning account. This is not acceptable. Our task is to delimit the theory of learning that is able to deal both with the fact that domains like language, sociality, and natural number are learned early, on the fly, and without formal tutoring and that domains like chess, computer science, art history, sushi making, etc. require lengthy efforts and organized instruction.

D
OVER
: I said it tongue-in-cheek, slightly, because I've been reading Pinker's book
The Blank Slate
(2003) and I don't have another term for it, basically. I would welcome one. But actually, what I'm saying is that the genes, all those individual little units – all 30,000 of them in humans – have to get their act together all over again after each moment of fertilization. And it's not just a question of epigenetic influences that are coming in from maternal cytoplasm or maternal mitochondria or parental differences in DNA methylation patterns – all that stuff. It has little to do with that, in the first instance. As I said, the genes have to start renegotiating one with another in the sequential order of interactions expected of the human genome if a human phenotype is to emerge. And there is no-one there telling them what to do. There is no-one at home saying, “Gene A, you'd better start interacting with B, and then hold hands
with F, and then hold hands with X”. It will naturally, inevitably unfold that way, even though you start off with the genes all blankly spread out on the slates of the two parental genomes. We mustn't misunderstand what most biologists mean by genetic regulation “programs” – programs and blueprints and recipes are metaphors that are highly misleading.
Why
there are no programs, and
why
, nevertheless, reconstruction proceeds along species-specific lines, is a matter for evolution – all those billions of steps from the origin of life onwards that led to the human genome behaving as it does during development – literally giving life from a genetic blank slate – from a completely novel, post-sex, combination of genes.

G
ELMAN
: I totally understood what you said, I'm very sympathetic to it; it's consistent. But you asked for the return, at the beginning, to the notion of blank slate. And that's what I object to.

D
OVER
: Well, the genetic blank slate is this. This is the genome of a frog [holds up a piece of blank paper]. There's nothing written on it; there are no dotted lines indicating how we are going to turn that into this [holds up a paper frog]. This is a frog, a squashed frog! So how do we get from that [the blank paper] to this [the frog], when there are no instructions of any sort on this piece of paper as to how the folding should proceed? Nor are there any extraneous hands of cooks following a misconceived idea of a recipe, or anything of that sort. So that is the genetic blank slate. If we have to use a different term, that's fine by me, because it is bound to be misunderstood given the history of usage of the term. We need a term to cover the process of
total nurturing
during the highly personalized reconstruction of a phenotype and all its networks, involving novel combinations of genes, novel epigenetics, and novel environments – and all starting from the “blank slate” of a unique fertilized egg, the first diploid cell.

R
IZZI
: I had a comment on your puzzlement about different views of parameters. I'm not sure it is exactly the same thing, but there is a debate in linguistics between two views of parameters. This to some extent emerged in our discussions here, and probably the thing is important so I think the debate should be more lively than it actually is. There is an interesting paper by Mark Baker on that. It is between a view that considers parameters as simple gaps in universal grammar (UG), so there are certain things on which UG says nothing, and then the role of experience is to fill these gaps – this is a kind of underspecification view – and then there is an overspecification view that says that UG contains specific statements for certain choices, which must be fixed by experience, but it is an overspecified view of UG somehow.

The argument for the underspecification view, of course, is simplicity. It is a simpler concept of UG. The argument for overspecification is restrictiveness, essentially. That is to say, those who argue for the second view observe that the underspecification view is not sufficiently restrictive in that it predicts possibilities that you actually do not find. Just to take the case offered by Cedric Boeckx in this conference,
⁷
those who argue for a headedness parameter, something that says explicitly that the head precedes the complement or follows the complement, seek to account for what actually is found across languages. If you did not have a statement in UG about that, the effect would simply be a consequence of the fact that you have to linearize the elements, that you have to pronounce words one after the other, so you do not get what you actually find. That is to say, in one language, for instance, you could sometimes produce VO structures, and some other times OV structures, because as the only goal is linearization, there is nothing that tells you that you must always go consistently. So there are these two views, overspecification and underspecification, which somehow transpired in our discussions here. That may be a source of your puzzlement about different conceptions.

Donata Vercelli and Massimo Piattelli-Palmarini

I have to tell you a story and the story is that the reason I am here is that I can't say no to my friends. Juan Uriagereka was both very insistent and very eloquent in inviting me, so here I am, presenting something that Massimo and I have been thinking about. I have to tell you that the division of labor is such that Massimo takes all the credit and I take all the blame. So this, by way of disclaimer, that I think we acknowledge that there is a little element of absurdity in what we may be saying, but we hope that we also have something that may be relevant to you.

Today we would like you to think about a biological trait, and for reasons I hope will become clear to you, let us call it biological trait
L. L
has certain features. It is species-specific, and in particular is unique to humans. It has a common core that is very robust but allows for inter-individual and inter-group variation. It has both heritable and non-heritable components. It goes through critical developmental windows of opportunity: that is, its developmental patterns are time-dependent. It is very plastic, particularly in response to environmental cues. It has multiple and discrete final states, it is partially irreversible, and it is robust and stable over a lifetime.

The question we are trying to answer is, what kind of biology may underlie a trait such as
L
, or, how is a trait such as
L
implemented in our genome. Classical genetics (which I will define in a minute) can certainly account for some features of
L
: species specificity, uniqueness to humans, and a very robust common core that allows for variation. The problem is that classical genetics, we maintain, would not buy us the other features that
L
has. And this is where we think we need to go a little bit further. Let us qualify why.

Fig. 7.1. Aspects of biological trait L

1953 is the year in which DNA, as we know it today, and classical molecular genetics were born. It is the year in which Watson and Crick published their rightly famous paper stating that the structure they proposed for DNA, the double helix, could be very effective to replicate, faithfully copy, and transmit information. The success of classical molecular genetics has been spectacular. In their labs, molecular biologists apply the paradigms of classical genetics every day. The notion that a DNA sequence is transcribed into an RNA sequence which is in turn translated into a protein is something we use, for instance, to make proteins
in vitro
starting with a sequence of DNA. This successful notion of genetics emphasizes the amount of information that is encoded and carried by the DNA sequence. What this genetics can give us is great fidelity and specificity in the transmission of information. What this genetics does
not
buy us is a fast,
plastic
response as well as environmental effects and memory of a functional state – nor does it buy us cell fate decisions. In essence, classical genetics is necessary, but not sufficient. This is where epigenetics comes in.

We are stressing the importance of plasticity, because we think plasticity is probably one of the defining features of our trait
L
. From a biological point of view, here is the puzzle. Let us consider the different stages our blood cells go through to become the mature cells circulating in our bloodstream. We have red cells and white cells, and they have quite different tasks. Red cells transport oxygen, some white cells fight infection from bacteria, some white cells fight infection from parasites. Therefore, all these cells do very different things, but they all derive from an initial common precursor cell – that is, they are
genetically identical, but they are structurally and functionally heterogeneous because they have different patterns of gene expression that arise during development. Such differences are epigenetically implemented.

To talk about epigenetics, we need to introduce a difficult but fascinating concept.
¹
The DNA double helix is not linear in space. It is a very long structure, if you unfold it, but it is actually very tightly packaged, to the extent that in the cell it becomes 50,000 times shorter than it is in its extended length. Packaging is a stepwise process during which the double helix initially forms nucleosomes, that is, spools in which the DNA wraps around a core of proteins (the histones). In turn, each of these beads-on-a-string is packaged in a fiber that is even more complex, and the fiber is further packaged and condensed until it becomes a chromosome. All this packaging and unpackaging, winding and unwinding, provides a way to assemble a huge amount of information within a very small space, but also makes it possible to
regulate
what happens to the information encoded in the DNA.

This is the subject of epigenetics. Epigenetics is the study of changes in gene expression and function that are heritable, but occur without altering the sequence of DNA. What changes is the functional state of the complex aggregate formed by DNA and proteins. These changes – extremely dynamic, plastic, potentially reversible – occur in response to developmental and/or environmental cues that modify either the DNA itself (by appending chemical groups to the sequence, which remains unaltered) or by changing the proteins around which the DNA is wrapped (i.e., the histones). By modifying the core proteins around which the DNA is assembled, or the chemical tags appended to the DNA, the functional state of a gene is also modified (Vercelli 2004).

Deciphering these modifications is quite complex. For DNA to become active, to release the information it carries, the molecule needs to unwind, to become accessible to the machinery that will transcribe it and turn it into a protein. This cannot happen if the DNA is very compressed and condensed, if all the nucleosomes, all the beads-on-a-string, are so close to one another that nothing can gain access to a particular region. Such a state is
silenced chromatin
, as we call it – chromatin being the complex (which is more than the sum of the parts) of DNA and proteins. When nucleosomes are very close and condensed, chromatin is silenced. That happens when methyl groups are added to the DNA or the histones bear certain chemical tags. On the other hand, when other tags are added to the histones or the DNA is no longer methylated,
the nucleosomes are remodeled and open up, the distance between them becomes greater, and the machinery in charge of transcription can get in. Now, transcription can occur. Hence, active chromatin is marked by accessibility.