bohm2 said:
They basically argue that this uniquely human part of our language faculty (FLN-see links for details) having the properties of recursion (also found in our mathematical abilities) emerged in human brains for “physical” reasons yet to be fully comprehended;
OK, bear in mind that I consistently explained systems in terms of three principles.
1) They self-organise hierarchically in that you have local constructive degrees of freedom finding some mutual equilibrium balance with downwards acting global constraints.
2) This is thus a dynamical or process view where things develop, starting out as the vaguely potential and becoming crisply realized (as a hierarchical situation of local degrees of freedom~global constraints).
3) Complexity of a higher order (like life and mind) can arise if a system discovers the trick of
constructing its own constraints. This requires semiotic mechanism, the epistemic cut, memory structures, etc. A system is no longer helplessly self-organising, it can harness dynamics for a purpose.
This is a view of nature now standard in theoretical biology. It shows through in "new models" of evolutionary theory like evo-devo - the realisation that natural selection can only harness self-organisational potentials, growth gradients. Or what Pattee called rate dependent dynamics.
The shift in biological thought is of course based on the deeper breakthroughs in physics and maths when it comes to modelling dynamics and self-organisation. All the work on criticality, dissipative structure, condensed matter physics, fractal geometry, chaos theory, etc.
So nature does have its fundamental patterns. There is a mathematics of development and self-organisation that we have only quite recently been discovering. And it is anti-reductionist or holistic in spirit because it highlights the causality of downwards-acting constraints.
Reductionist thought views causality only in terms of upward acts of material construction. Complicated things do not develop, they get built. Which then usually leads to the problem of who did the building? It is clear that too much information has gone into the construction of any natural system for something like Darwinian selection to be the designer, for genes to be a simple blueprint.
But the systems view, the evo-devo view, the epistemic cut/semiotic view, says what actually happens is that only a little bit of rate independent information (constraints that a system can construct) is needed to harness or control a whole lot of "order for free" in the shape of naturally-occuring developmental forms. The genes, the hand of natural selection, etc, only have to do a bit of judicious tweaking at the right moments to channel the kind of growth that will happen anyway of its own accord.
You could call this an organic logic, as opposed to the mechanical logic beloved of those whose main interest in life is building machines (systems which actually are over-determined, constrained to the point where they have no free dynamics, and so operate purely from local effective cause). A gardener will understand though the difference between judiciously nudging nature and deterministically trying to control nature. Likewise a parent with a child.
Anyway, to address your point, you can see that the Piattelli-Palmarini/Uriagereka paper is reinventing this particular wheel. Surprise, surprise, nature is mathematical in that developmental processes show general (ie: universal) patterns like log/log growth. When you get hierarchical expansion - with both the local and the global growing at equal rate - the inevitable outcome is powerlaw, scalefree, fractal, etc.
So naked self-organisation is constrained to fall into certain self-consistent (ie: self-constraining!) patterns. Then these natural patterns are what epistemic/semiotic mechanism can harness, so as to channel growth towards goals. And surprise, surprise, even this controlling machinery has certain inevitable characteristics. It looks to be ruled by certain universal mathematical principles (as only certain forms can achieve the resulting effects).
Recursion in grammar is this kind of story.
Nested hierarchical form is a natural outcome of self-organising development. Fractal or Fibonacci patterns are the "order for free" that result from free (ie: "random") local constructive actions within a stable set of global constraints. The animal brain is functionally structured as a nested hierarchy. Take the macaque visual hierarchy as the standard example. (
http://www.cse.yorku.ca/~billk/billkPres1b.html)
It is all about differentiation~integration - two mutual processes, one localising, one globalising, going on with equal energy. The result is a tremendously dynamic hierarchical organisation. Whether you are talking about sensory input or motor output, there is a hierarchical process of composition and decomposition going on. Which can look "modular" once you frame it as higher cortical areas doing the broad planning and oversight (setting the general constraints) and the lower cortical areas doing the detailed grunt work (constructing the particular actions to fit within these constraints).
So the animal brain is good at "recursion" in this sense. To move a paw requires the hierarchical composition of the intent coupled to the hierarchical decomposition of its execution.
But there is something still limited about the animal mind. It's thinking is holistic, entirely patterned according to the demands and possibilities of the moment, not able to step back and think self-consciously, reflectively, objectively, about the moment. The animal cannot
construct a recursive structure of such moments where this moment is seen within a dynamically/deliberately selected context of such moments. There is no secondary life story concerning a "me" who is having "this experience".
The reason is the animal mind/brain has the semiotic machinery (genes, synapses, other dynamic-tweaking memory devices) to develop hierarchical states of experience. But it is missing the mechanism for constructing the constraints necessary to go up that further level of recursion.
This is what language brings. And again, in terms of systems causality, it is only in fact a small difference that can produce a very big effect.
The key to language evolution (in my view, after Leiberman, etc) is the development of serial utterance - a palate and throat which broke vocalisation into a syntactic stream. As Piattelli-Palmarini/Uriagereka point out, Fibonacci type recursive growth needs the emergence of digital codes - the 0/1 of onset and coda in their "F-game". The human vocal tract does that, breaking the flow of sound into vowels and cosonants. Once that simple trick emerges, then unlimited recursion - an endlessly growing nested hierarchy - can be constructed.
So the animal brain was naturally a hierarchy, able to compose~decompose. But only in response to the very immediate world. There was no way to
construct a back-story of thought and meaning. The recursive capacities of their brains could not be exploited to do this higher level recursion because there was not the further machinery to harness it.
Humans stumbled into such a machinery with simple changes to the vocal tract. Yes, undoubtedly all this was connected to changes also going on in the brain - selective pressure for machavellian sociality, theory of mind, tool-use - which were making Homo capable of more elaborate recursion anyway. So it is a complex paleoanthropological tale.
But essentially the problem was finding a tool to unlock the potential. The animal brain was very good at self-organising mental states of anticipation, intent, etc. But only as a response to constraints largely given by the world around them. Langauge was a way to code for constraints - to construct them "at will" (or rather, by social learning and habit) - and so harness the moment-to-moment personal development of states of mind. I speak therefore I can control what I think. (and having found I spoke too hastily, I can then correct that)
So there is recursion here on the grand scale. Development has a systems logic. It is the natural and inevitable patterns that result from dynamical self-organisation. It is both physical (what happens in reality) and mathematical (how we model what happens in terms of formal causality). And it looks to have the form of a nested hierarchy when it is optimal or otherwise extremal - the max ent principle, the reason powerlaws are ubiquitous in natural structures.
But evo can harness devo. Syntax can harness semantics. Give a system a memory, some kind of epistemic mechanism to construct its own constraints, and freely-occuring dynamics can be chanelled to achieve localised purpose. You get life and mind. Genes were the first level great breakthrough, words the second.
The complex structure of a sentence was inevitable just as soon as some mechanism arose to break a flow of sound into discrete chunks. Syntax is the construction of constraints on semantics. A cat is a general idea. Which can be made hierarchically more constrained by adding qualifiers such as [cat [sat [mat]]]. Or [cat [kitten [cute]]]. Or [cat [cartoon [Krazy Kat]]].
What matters in the mind is that an idea, an anticipatory image, forms. This is going to happen every moment anyway because that is how brains are structured. But states of mind can be constructed via acts of verbally-base, recursively nested, constraint.
Sure, you can get obsessed about the hunt for a brain module which does "syntactic constraint". Stick a person's head in a scanner, ask them to generate regular and irregular verbs, and you will find different parts of the brain working hard. Perhaps the left dorsolateral prefrontal for regular verbs, and the left superior frontal gyrus for the irregular (until the next study using german rather than english speakers contradicts you).
But this is missing the point. Brains are already hierarchically organised. And the cause that is making such a "neuroscientific" difference in humans is not the addition of some novel brain module/genetic modification but another level of the epistemic cut. That is what needs to be the focus of attention.
