Monday, February 2, 2015

Linguistics and philosophy

There are many paths into linguistics. Some meander in from various languages, some from math, CS, psych, and bio. I came from philosophy. In fact, when I was a young lad, linguistics was pregnant almost to bursting with philosophical implications and lessons. It was the modern venue for the great debate between Empiricism and Rationalism, a major challenge for neuro-reductionism (of the connectionist variety), and the poster child for the computational theory of mind. In other words, discoveries in linguistics seemed to have implications for the great questions of the ages and this was (and still is, IMO) what made (and makes) it so exciting.   

If your tastes run in the same direction that mine do, here is a recent little paper by Chomsky that might interest you. It touches on a host of themes, including: “the creative aspect of language use,” the mind-body problem, the limits of scientific inquiry, scientific intelligibility, the reduction of chemistry to physics as a model for the reduction of cognition to neuro-science, among others. These are themes that Chomsky has developed over his linguistic/philosophical career, and he brings them together here in a very accessible form. Let me say a word or two about some of the issues he touches on.

He starts the paper with a discussion of how the creative aspect of language use (i.e. the fact that the “ordinary use of language is typically innovative without bounds, appropriate to circumstances but not caused by them –a crucial distinction- and can engender thoughts in others that they recognize they could have expressed themselves” (1)) motivated Descartes’ dualistic distinction between mind and body.[1] Chomsky’s makes two interesting points. First, that Descartes’ mind/body dualism was a reasonable scientific theory given the prevalent (and by prevalent I mean the entire whose who of 17th century scientific revolution) contact mechanical doctrine of physical interaction (viz. the position, that implied, for example, that there could be no action at a distance).  And second, that Newton’s discovery of gravitational action at a distance effectively exploded Descartes’ dualistic position not by exorcising the mental ghost in the bodily machine, but by dispensing with the machine. Newton’s work led to the collapse of contact mechanics as the overarching criterion of scientific legitimacy and so exploded the notion of body on which Descartes’ position relied. No body, no well posed mind/body problem. 

Chomsky observes that this consequence of Newton’s very successful physical theory was widely appreciated both in his time and later. Indeed, it was understood by Locke, Hume, Newton, Lange, Koyre (indeed everyone really until the modern period) that Newton’s work rendered materialism (the body part of the dualism) an insignificant (actually contentless) doctrine for after Newton there was no principled way of distinguishing the material from the non-material forces save sociologically (viz. materialism being whatever physicists decided to accept as part of physics).  

Chomsky further notes some of the consequences that this had for the relation between minds and brains. It was quickly assumed that mentation was what brains secrete, though how this was possible (i.e. how mental activity could supervene on brain activity) was somewhat mysterious, as remains more or less true to this day (see especially Chomsky’s quote of Vernon Mountcastle on p. 5).[2]

All of this has consequences for the reduction of cognition to neuro-biology. Chomsky touches on this, and helpfully illuminates the issues at stake by reviewing the history of the “reduction” of chemistry to physics. Yes, those are scare quotes. For as Chomsky notes reduction was quite tortuous, with the principles of chemistry more or less preserved in a radically reconfigured physics. Chomsky points to relevant morals for the current concerns of how to reduce minds to brains and points to something that IMO is very important: that the best way to proceed is not to worry too much about evident “explanatory gaps” (6). In fact, the best thing to do is emulate the earlier history of chemistry and physics which means “developing a body of doctrine” (Joseph Blacks’ phrase) in each discipline before worrying too much about how to unify them.  Right now, as Chomsky notes, neither cognition (and even linguistics!) nor neuro-science are as advanced as chemistry and physics were in the 19th century so there is likely still to be a lot of doctrine worth developing in both domains independently of any issues of unification or reduction. And, just as happened in the chemistry/physics unification, there is no reason to think that the right way to proceed will be via reduction of cognition to neuro-science. In fact, given how little we know about the brain it is just as likely that any unification awaits deeper understanding of the latter (a point that Gallistel has also frequently made).[3]

The last part of the paper addresses the common view that there is nothing opaque to scientific understanding. The opposite is called “mysterianism.” Chomsky dismantles this view, and linguists should read it for it touches on a concern that non-professionals often have with Generative Grammar (GG). What is it?

GGers argue that humans come with a lot of specific built in structure that allows humans to learn Natural Languages (NL). Many object to this kind of nativism for they see it as a reactionary doctrine that treats human characteristics as immutable, thereby limiting human potential. This is incorrect, but it is a pervasive reaction. Chomsky’s discussion provides a useful contrary picture. As he notes, our prized powers are the flip sides of our necessary limits. No limits, no scope. If you prize human creativity and originality, then you must also prize the structures that support these. And any structures will impose limits, for that’s what structure does.

There is one last overarching theme of this little paper: intellectual modesty. It seems that the great Enlightenment figures were less prone to scientism than modern scientists are. They seemed to appreciate not only what they knew but what they didn’t know and didn’t disparage the latter because it resisted our efforts at explanation. Chomsky proposes some modesty regarding these matters. As he notes: “The quest for better explanations may well be infinite, but infinite is of course not the same as limitless. English is limitless infinite (thx MJ), but doesn’t include Greek” (9). Similarly, there is no reason to doubt that some questions might forever be beyond human understanding (at least of the scientific kind), as Hume and Locke and Descartes (and most other Enlightenment luminaries) thought. There is no reason to think that among the animals only humans have no cognitive limits. Indeed, the contrary view comes close to being a modern form of Dualism.

In sum, there is a lot packed into this paper’s 10 pages. Enjoy. I did.

[1] This creative aspect of language use is different from the recursive hierarchy that natural languages display though it is related to it (Note the “innovative without bounds” above). This creative aspect is a feature of action, not a property of knowledge, which is why Descartes saw this sort of activity as indicative of something like free will and Hume saw it as a manifestation of imagination. Neither philosopher believed that these powers would ever be fully comprehensible to us.
[2] There is still a hot debate about this mysterious connection that flies under the banner of qualia, e.g. why does firing of fiber X result in my perception of red?
[3] Gallistel likes to discusse another similar case: the reduction of genetics to biochemistry. As he notes, after Watson and Crick, classical genetics remained largely the same. However, earlier biochemistry was largely discarded. See here for a great history.


  1. On a very related theme. The present issue of the New Yorker has a profile of Yitang Zhang, the mathematician of the twin primes fame. Not a particularly good piece but one of the themes is the beauty of mathematics. A beautiful proof "should resemble a simple and clear-cut constellation, not a scattered cluster in the Milky Way." Zhang's proof has "a renaissance beauty, meaning that though it is deeply complex, its outlines are easily apprehended. The pursuit of beauty in pure mathematics is a tenet."

    Now of course. "Last year, neuroscientists in Great Britain discovered that the same part of the brain that is activated by art and music was activated in the brains of mathematicians when they looked at math they regarded as beautiful".

    Do they have to ruin everything?

    1. yes, I think aesthetics plays a crucial role in science. I worked on automatic theorem proving a decade or more ago (when I was working on parsing as deduction). The big problem there is not so much generating correct logical deductions (our computers can generate millions of them a second) as figuring out when you've found something interesting or useful. Every deduction is correct, but most aren't insightful.

      I occasionally wonder whether there could be other intelligences ("space aliens") with the same laws of logic but with very different mathematics because their aesthetic sense is so different to ours.

  2. Largely off topic, but I’m intrigued how the agent nominalization of ‘generative grammar’ is ‘generative grammarian’ because of ‘grammar’, but Norbert’s agent nominalization of ‘GG’ is ‘GGer’ and not *‘GGian’. Phonotactics of -ian? Regularity of -er?

  3. I think you got the quote wrong -- English is infinite but not limitless.

    1. You are, as usual, correct. I will try and remedy this soon.

  4. I'm not sure I'm correctly understanding C's point about Newton destroying Cartesian dualism by destroying a purely contact-based mechanistic world; we all now believe in non-contact based forces like gravity. Is the point that we should stop worrying about how mental phenomena are instantiated in the brain, and just accept them as mysterious primitives?

    I think Chomsky elsewhere remarks that our understanding of the mind is not far beyond Aristotle's, but our physics would be incomprehensible to him, and suggests that perhaps this is because we humans are capable of comprehending physics but not the mind. That might be the case, but it could be wrong. After all, pretty much all the development of physics took place in the last few centuries; for millenia Aristotle ruled supreme in physics too.

    1. You ask a good question. Here's how I understand things. The most important consequence of the dissolution of the classical mind/body problem is what it says about how unification is to proceed. In the classic version, we KNEW what physical forces HAD to be: contact mechanics. Thus, unification meant reduction to physics as we know the limits of physical explanation. Once we give this up, we don't actually know the "limits" of the physical; after all, if God can link gravity, an occult property to matter, then she can link anything to matter. Thus unification can go either in the direction of reduction or in a more roundabout way by changing the "reducing" science and leaving the "reduced" one more or less as is. This is exemplified by Chomsky's discussion of chemistry and physics. The classical mind body conception privileged physics and required that every other science march to its tune. With tis demise there is no preferred direction of unification: either the physics changes or the chemistry or both.

      So, no there is what has been called Broca's Question in language: how do brains realize minds. But with the demise of the classical mind/body conception there is no reason to believe that the neurologists have it right and the cognitivists have it wrong when the two clash. THe way I see it, what C is saying is that after Newton, the scientific playing field has been conceptually leveled.

      This also has a practical implication: it is critical to develop "bodies of doctrine" within a discipline as well as figuring out how to unify across them. And these bodies of doctrine must be taken seriously. So, if you want to exlain how brains do language, then you need to explain how brains to the kinds of things that linguists have discovered over the last 60 years. This stuff has an integrity and counts intellectually even if there is no Nobel Prize in cognition. In reading much of the neuro literature there is this sense that the important big shot science is neuro and mental matters must just toe the line when the two appear to be in conflict. This is a residue of the dualism that Newton exploded. C's point is that this posturing has no scientific merit anymore, as we just don't know the limits of the physical/neurological.

      That's how I see it.

    2. Of course I agree that very often it is the base science that needs to be expanded; another example is how the gas laws relating temperature and pressure etc. were reduced to physics -- that only happened when an entirely new branch of physics (statistical thermodynamics) was invented. (Inconsistencies in the classical version of this theory led to quantum mechanics).

      I suspect we have much to learn about how the brain represents hierarchical structure (e.g., trees). The evidence for hierarchical structure in language is very strong, and while we might be mistaken about the details, any explanation of language that doesn't explain how and why it has words and phrases is missing something crucial.

      Thinking more about Chomsky's remarks on Cartesian mind / body dualism, I can easily imagine Chomsky agreeing with a twinkle in his eye that yes, dualism is scientifically untenable, but it's not because the mind is gone (which is what most of the audience would expect) but because Newton showed purely a mechanistic conception of the body has to be abandoned.

      Thanks for posting the Chomsky articles. They are always intriguing.

      It would be interesting to see what the field looks like in a century or two. It could be that neuroscience has advanced enough that we understand how minimalist computations are implemented in the brain. But it could be that some other empirical phenomenon (e.g., phonology, discourse?) is the wedge that lets us figure out the mind / brain connection. After all, those funny bands in the hydrogen spectrum played a crucial role in figuring out quantum mechanics. (Speaking of which, if I had to guess what kind of knowledge would be likely to be beyond the capacity of the human mind, quantum mechanics would be high on my list. Still, I guess we'll never know what facts are out there that we simply can't conceive of).