LINGUAFORM PHILOSOPHY AND WHY LANGUAGE IS SUPPOSEDLY NECESSARY FOR THOUGHT
My friend Matt Bush when criticising analytic metaphysics plausibly accuses a lot of them of being too deeply wedded to interpreting reality in terms of language. Matt’s criticisms are in my view entirely to the point. Since the birth of analytic philosophy the analysis of language as a way of understanding philosophical problems has been a central foundation. Philosophers like Frege and Russell by constructing artificial languages to help them speak more accurately about the foundations of mathematics set the ball rolling with analytic philosophy’s obsession with language. Wittgenstein in his ‘Tractatus’ using the insights of Russell and Frege tried to construct a theory of meaning for language which he used to demarcate between statements with sense and nonsensical statements. His later ‘Philosophical Investigations’ took the facts of ordinary language more seriously and focused more on the conditions of sensible use of language as opposed to the more abstract logical view of language set out in his ‘Tractatus’. Nonetheless, in both the earlier and later phase of his philosophical development Wittgenstein thought that an analysis of language was the way to solve philosophical problems. Non-naturalistic philosophers like Austin, Ryle and Hacker have followed him in this approach to philosophy.
I have discussed ordinary language philosophy in detail in other blogs so I will not go into it in this blog. More naturalistic inclined philosophers have followed Frege, Russell and Wittgenstein in making language central to philosophy. They have used language as ways of trying to solve philosophical issues in both ontology and epistemology. Matt has criticised non-naturalistic analytic metaphysicians attempts to understand metaphysics through linguistic analysis. I share his misgivings when it comes to analytic metaphysics, though that will not be my focus here, I will instead focus on a misplaced emphasis on language in naturalistic philosophy of mind. The emphasis on language which has been with analytic philosophy from the start I will argue has had a distorting influence on how naturalistic philosophers have thought about the mind.
Before proceeding I should add that I think that the linguistic turn while excessive has lead to some real improvements in the way that philosophy is done. So my aim here is not to slate linguistic philosophy, merely to curb some of its excesses in the philosophy of the mind. In the second part of this blog I will argue that naturalistic analytic philosophers have vastly overestimated the degree to which cognition is tied to linguistic abilities. However prior to doing this I will firstly try to summarise the best arguments in favour of the view that thought requires language.
Contemporary(ish) philosophers who make language central to cognition such as Davidson, Dennett, Rorty, and Brandom have two primary influences Wilfrid Sellars and W.V. Quine. I will now try to outline some of the best arguments from the preceding six thinkers that language is central to thought.
In his 1956 ‘Empiricism and The Philosophy of Mind’, Sellars argued that if we want to interpret a creature to being a thinker we need to do so in terms of language. Here is Sellars on the topic:
“All awareness of sorts, resemblances, facts, etc., in short all awareness of abstract entities-indeed, all awareness even of particulars-is a linguistic affair. According to it, not even the awareness of such sorts, resemblances, and facts as pertain to immediate experience is presupposed by the process of acquiring the use of language” (Science, Perception, and Reality, p, 160)
Sellars notes that we can distinguish between two different kinds of awareness. The first type of awareness is the ability to reliably distinguish between aspects of ones environment. The second type of awareness is awareness that comes from being in the logical space of reasons and having the ability to justify what one claims or believes about the world.
Obviously when one considers Sellars first two types of awareness the first does not require language while the second does. Sellars argues that the first sense of awareness is so permissive that it includes virtually everything as aware. In his (1995) ‘The Problem of Objectivity’ Davidson agrees with Sellars that awareness 1 (classification) is too permissive to be considered a criterion for saying what creatures are aware:
“My reason for resisting this usage is that if we were to accept it we would be committed to holding that the simplest animals have concepts: even an earthworm, which has so little brain that, if cut in two, each part behaves as the undivided whole did, would have the concepts of dry and moist, of the edible and inedible. Indeed, we would have to credit tomato plants or sunflowers with the concepts of day and night.” (Davidson: 1995 p. 8)
Robert Brandom in his (1995) ‘Making it Explicit’ makes a similar point:
“Reliable differential responsive dispositions are only a necessary condition for observational knowledge. Parrots and thermometers can have such dispositions and so can be used by us in effect as measuring instruments to acquire knowledge. But what they have is not knowledge. For they do not understand the significance of their responses; they do not take these responses as reasons for further claims; and they do not understand claims as potentially in need of reasons” (Brandom: 1995 p. 215)
So philosophers like Rorty, Davidson and Brandom amongst many many others follow Sellars in thinking that a creature who we discover through behavioural tests can reliably distinguish between an aspect of their environment is not necessarily aware. For true awareness to occur a creature needs to have the ability to place themselves in the space of reasons.
Sellars argument that if we use the criterion of being able reliably distinguish between aspects of their environment as the criterion of awareness then we will be committed to the view that thermometers and plants are aware, has convinced a lot of philosophers that justification is the best criterion. So Sellars argument is one of the key reasons that philosophers have for thinking that language is central to thought.
However Donald Davidson merges the views of Quine with those of Sellars to provide further arguments that language is necessary for thought. Quine argued that prior to mastering the syntax of quantification we have no justification for imputing an ontology to a child. A child may learn to mouth a sound e.g. ‘MAMA’ in certain intersubjectively salient circumstances, the child may even internalise recursive abilities. However these abilities do not prove that the child has an ontology. When the child says ‘MAMA’ we have no behavioural evidence which indicates whether the child is using it as a term, or as an observation sentence, as a count noun, a mass term, etc. Davidson following Quine argues that:
“What calls for ontology is the apparatus of pronouns and cross reference in natural languages, what we represent by the symbolism of quantifiers and variables in elementary logic. These devices provide the resources for constructing complex predicates, and at this point semantics must map names and predicates on to objects.” (Davidson: 2001 p. 140)
While Sellars argument above has a lot to offer, Quine’s argument is manifestly false and has been shown to be so by the experimental work of Soja et al. 1991 which I will review in the next blog.
A further argument against attributing propositional attitudes to non-linguistic creatures is the difference between the logic of referential opacity in linguistic and non-linguistic creatures. So, for example, consider the statement (1) John believes that ‘Batman caught the Joker’, and (2) John believes that ‘Bruce Wayne caught the Joker’. Sentence (1) can be true while sentence (2) can be false. This is because even though Bruce Wayne and Batman refer to the same person John may not know this. We can see from the preceding examples that when quantifying into belief contexts referential opacity obtains (the same is true of quantifying into modal contexts). When dealing with ordinary sentences referential transparency obtains. Thus consider (3) Batman caught the Joker. And (4) Bruce Wayne caught the Joker. If Batman and Bruce Wayne refer to the same object then (3) and (4) must have the same truth value.
Davidson notes that the logic of belief ascription changes when we consider non-linguistic creatures. So he asks us to consider the case of a dog chasing a cat the cat runs up tree x and the dog who doesn’t see this starts barking up the tree at the cat. Davidson notes that a person viewing the scene could say that the dog is barking up the wrong tree. So, for example, I could say (5) The dog believes ‘that the cat is up the oak tree’, or (6) The dog believes that ‘the cat is up the tallest tree in the forest’. Now in the case of a linguistic creature, for example me, I may not know that tree x is both the tallest tree in the forest and an oak tree, so one could say that (5) and (6) are referentially opaque. However things are different with a non-linguistic creature. (5) and (6) do not behave in the same logical way as they do for linguistic creatures. We have no way of saying that (5) and (6) change their truth value in referentially opaque circumstances because there is no fact of the matter as to whether the dog thinks that ‘the tallest tree in the forest’ and ‘the oak tree’ refer to the same object. Davidson argues that since referentially opaque sentences changing their truth value in belief contexts does not apply to non-linguistic creatures then this shows that we MAY not be justified in applying propositional attitude explanations to such creatures.
So the three primary arguments that philosophers use for the claim that only language using creatures can think are (1) It is over inclusive to claim that creatures who can reliably distinguish between aspects of their environment have thoughts (thermometer objection) (2) Only creatures who can have mastered the syntax of quantification have an ontology (3) Belief ascriptions to non-linguistic creatures behave in logically anomalous ways which imply that creatures are not really believers.
In my next blog I will critique the above three arguments and show that they do not actually demonstrate that non-linguistic creatures cannot think. In my final blog of this series I will consider Dennett’s more subtle linguaform views as they apply to consciousness and will show that his views also do not work.
I’ve always wondered about Quines radical translation argument in this respect. Dennett has a Quinian crossword–just a four by four grid with eight clues. It has two solutions–both equally good–and he uses this as an intuition pump to suggest that there is no answer to the question of the real answer to question one (say). Ok. Its very ingenious. Buts its very ingenuity makes me suspicious. As Dennett himself has emphasised in other contexts–practical problems can be philosophically interesting. The idea of constructing a 10X10 (let alone a 50000 X 50000–about the size of a natural language vocab) crossword should boggle the mind. But, such a thing should be possible if the Quinian hypothesis is true. Heres my problem with it–something like convergent evolution. There probably just arent that many ways to be a successful living thing occupying a particualr niche. Sure, we have 6000 languages here on earth but the idea that each one could systematcially mistranslate onto another one–by comparison with the outside world–and yer never be detectable…well…Dennett wants to preserve the “in principle possible” bit. I guess we have to give him that–but I have to wonder what can be derived from it!
Just off the top of my head.
I think Putnam’s brain in a vat argument uses the same point you ascribe to Brandom about the adequacy of differential response as a indicator for consciousness, to critique Turing machine models of thought and make the point about programmed patterns are not purposive. In this manner he promotes his own views of intentionality. Putnam does not conclude that animals are incapable of thought, but that surely Turing machines are….
I can’t see how Brandom would be able to insist that this is evidence that animals can’t think, just that we can’t use differential response as a good indicator of the possession of “propositionally structured cognitive states” (whatever that comes to). I think this is a decent argument, in so far as it also applies to you and me too, and as such cannot be used to draw a distinction between human and animal cognitions.