Kripke’s First Argument against Descriptivism

1. ◊ Benjamin Franklin was not the first postmaster-general of the US.
   
2. Benjamin Franklin ◊ was not the first postmaster-general of the US.
   

The modal operator’s scope in ii) is about Franklin himself; or the possibility that he might have done such and such. The first example is a de dicto possibility – about the possibility of the statement being true. ii) is de re possibility.

The point about i) and ii) is that we can be clear about the modal application in that its position tells us its scope. Despite the placement of the modal operator, the semantic value of the whole sentence remains the same. What's more important - from this Kripkean perspective - is that the semantic value of the name ‘Franklin’ remains unchanged in both i) and ii). This is, of course, going to affect the central argument about Kripkean proper names. However, this doesn't mean that the name must be somehow ‘empty’ if it allows us to change the scope of the modal operator.

What content does the name have if not, say, a description? Its semantic value is the named object or person itself. Clearly this conclusion has a de re feel – it's about an object, not a statement. Indeed we'll also see that it has an essentialist feel. In essentialist terms, we see that in both i) and ii) the object named must retain its semantic value no matter which world we place it in. In other words, the name’s semantic value remains fixed right across possible worlds. And if its semantic value is the object that is its referent, then the name holds fast to the object-referent it names across possible worlds. No matter which world the object is placed in, it will be held by its name despite the world’s differences and the different counterparts of the object itself. In order for this to be the case, that object must retain its essence at all possible worlds at which it is placed.

Now in terms of the statement of possibility which includes a description rather than a proper name, the case is different. For example, it may only be at our world that the description "the inventor of bifocals" is correctly assigned to Franklin. This, according to Kripke, makes this aspect of Franklin a contingent - not essential - property of the man. It's shown to be so by the introduction of possible-world scenarios. In other words, the descriptions of Franklin won't pick out Franklin - our Franklin - across all possible worlds. So

◊ The man who invented bifocals was not the first postmaster-general of the US.

doesn't make the same claim as the statement that uses a proper name. Whereas prima facie we think "the inventor of bifocals" and "Franklin" pick out the same person and that they do so contingently. At our world they do so; but at other possible worlds they may – or do - not. The above, rather counter-intuitively, isn't about our Franklin because it's a possibility-statement that doesn't use a proper name. It would need to use a proper name in order to secure stability and necessity of reference. The statement above tells us that it could be – or is – true because Franklin - the barer of the name ‘Franklin’ - does (or may) not exist.

There's a possible argument against Kripke’s conclusion. We can read the above as if it has wide scope. Strangely enough, it's the case that wide scope refers to a modal operator embedded within a statement; not one about a whole statement. In that case, wide scope is about de re necessity and possibility, not de dicto necessity and possibility. So a re-reading of the above as a narrow scope de dicto statement can now be:

The inventor of bifocals ◊ was not the first postmaster-general of the US.

Now we have a statement with both a singular description and a de re claim about the inventor of bifocals. In other words, we have freed the description, ‘the inventor of bifocals’, from the possibility operator. What does this modal operator movement attempt to achieve? Firstly, we take the description to be about the inventor of bifocals at the actual world, but then we hold that semantic value as ‘constant’ at other possible worlds – worlds at which he is not the first postmaster-general of the US. The status of being the inventor of bifocals is retained, but not its necessary connection with being the first postmaster-general (which is the case at the actual world). Although the statement above uses a description, because that description lies outside the modal operator it has wide scope. We must therefore initially be talking about our inventor of bifocals, not a possible one at another world. Kripke argues that it is our Franklin himself who is the semantic value of the description – a res, however strange that is - it gives the description its semantic value. We can now say that this rephrased wide scope de re statement that uses a description now makes Franklin’s property being the inventor of bifocals a necessary one, whereas being first postmaster-general is merely contingent. This, prima facie, appears to be a purely arbitrary distinction on the descriptivist’s part.

But names do not have a narrow scope reading. That is, if the modal operator is placed to the left on the entire statement, the name will not have an equivalent narrow scope reading in that no matter where the name is placed in the statement, or how it is used, it will always have the same semantic value – the object-referent at the actual world as he appears across other possible worlds. This property of names holding fast to their referents is encapsulated by Kripke by giving them the title ‘rigid designators’. They rigidly designate because they always refer to the same object at every possible world at which that object exists, even if these worlds are different to ours and the object itself has different contingent properties.

The rigid designator idea is quite simple. It is about the stability of reference across possible worlds. However, unlike theories of reference, this has nothing to do with naming processes such as ‘causal reference’ or ‘direct reference’. We can say, now, that it is not a ‘theory of reference’ at all. It is about objects keeping their names; not about them getting their names. The ‘rigidity’ of names tell us nothing about how names get attached, as it were, to objects and persons. It has nothing to do, for example, with Kripkean ‘naming baptisms’ or direct causal contact between names and named objects. However, it does apply to the relation between names and named objects; but only once such objects have been named. From then on, a name and its object remain together across possible worlds. The primary purpose of Kripke’s rigidity thesis is to enable us to identify individuals across possible worlds.

So now we can say more about the following:

◊ Franklin was not the first postmaster-general of the USA.

The claim above is not only true about a possible Franklin or a Franklin ‘counterpart, it is also true of our own Franklin. These possible worlds are always taken within the context of our world. The above is true, but it has nothing to do with the possibility that someone at a possible world merely has the name ‘Franklin’. It is not about a name, or someone else named ‘Franklin’. It is not about finding a world at which a ‘Franklin’ was not the first postmaster-general of the US either. It is about the possibility that our Franklin might not have been the first postmaster-general of the US. And we investigate other possible worlds to establish this possibility. Again, it is not about the possibility about another Mr and Mrs Franklin who also named their child ‘Benjamin’ and that this child did not become the first postmaster-general of the US. It is not about names at all; it is about an object/person. More than that, it is about our person – the person named ‘Franklin’ and how he may (essentially) be at other possible worlds.

All this is an attempt to show that we cannot rely on any descriptive content for the name ‘Franklin’ to establish its necessary semantic value. At other possible worlds we could not rely on the description ‘the inventor of bifocals’, or any other description, to fix the name ‘Franklin’ or pick out the object Franklin. No description can fix what it is about Franklin that must remain constant across all possible worlds. Therefore, according to Kripke, names must not have content. What a name must pick out at every possible world is Franklin’s essence. This means that all the descriptions we have used of Franklin must only pick out his contingent properties (in our world). And, therefore, they may not pick out Franklin at all at other possible worlds at which he exists because he may not exemplify any of these contingent properties (which are the source of our descriptions).

(A quick digression. What are the essential properties of Franklin according to Kripke? It is here that philosophers start to talk about ‘origins’ [see Forbes, 1997]. For example, Forbes must be indirectly talking about a named man’s beginning in the womb at the moment the zygote is formed from a sperm and an egg.)

One conclusion to all this is that names are indexed to the actual world. No matter what we say about the name ‘Franklin’ and the person Franklin at another possible world, it will still be determined by our Franklin and therefore our name ‘Franklin’. That is because no matter what a possibility-statement says about Franklin, if it uses the proper name, rather than a description, within that statement it will be saying:

The object/person who is actually Franklin might not have been the first postmaster-general of the US.

Despite all the possible-world scenarios, they are still about our person – the actual Franklin. With descriptions within a possibility-statement, we will need to know which possible world is actually being spoken about. A proper name, on the other hand, is always indexed to the actual world. And a name is thus indexed because it only depends on the referent or object of a name, not a Fregean ‘sense’ or a description. It is, therefore, de re rather than de dicto; essentialist rather than non-essentialist; ontological rather than conceptual, etc.

Now we can take on board a possible counter-example to Kripke – referential uses of descriptions as offered by, amongst others, Donnellan [1966]. However, Kripke would argue that such descriptive references are seen to belong to the ‘pragmatics of use’ and have nothing to do with the semantics of descriptions. Such a pragmatic reference must be so because it is indeed the case that particular persons will use particular descriptions in order to pick out - and then fix - a particular named object. Of course they will. However, this has nothing to do with the semantics of proper names or descriptions. The semantic value of ‘Franklin’ must be the same for all users and at all times. This could not be the case with any single description, or even Searle’s ‘cluster’ of descriptions [Searle, 1967]. This universality and a-historical reality of a proper name is determined by the fact that its only semantic value is the named object-referent itself - and nothing more! It is, therefore, de re. Descriptions, on the other hand, are conceptual or de dicto in nature. Descriptions belong to the pragmatics of use in the sense that they are used practically by particular persons at particular times, not by all persons at all times. We can therefore say that semantics, or Kripkean semantics, is essentialist in nature, whereas pragmatics is non-essentialist or even anti-essentialist.

The thesis of rigid designators tells us that proper names are given their semantic value at our world and must retain that value at all possible worlds at which the named object exists. This means that a possibility-statement with a proper name has a fixed semantic value because that value is fixed at the actual world by an actual object or person. Therefore it is the person himself, as it were, who is found at different worlds (or his counterparts), not a mere similar.

However, Kripke has argued that rigidity is not just about modal contexts. In the following

Aristotle was fond of dogs.

the name is rigid without a mention of modality or possible worlds. And yet Kripke goes on to write something which does highlight the modal nature of proper names:

".... the thesis that names are rigid in simple sentences is, however, equivalent to the thesis that if a modal operator governs a simple sentence containing a name, the two readings, with large and small scope, are equivalent [1972/1980)."

To conclude and repeat: the whole thesis of the rigidity of proper names is not actually a theory about reference in the sense of our giving an object a name, etc. However, if one accepts the rigidity thesis, then it automatically commits us, or so it has been argued, to what has been called a ‘direct theory of reference’. And Kripke, of course, also offers us a direct - causal - theory of reference to follow his account of rigidity. The Kripkean conclusion must be, then, that reference must be either direct or descriptive.

References

Forbes, G – (1997) ‘Essentialism’, in A Companion to the Philosophy of Language, Blackwell Publishers
Kripke, S - (1972/1980) Naming and Necessity, Oxford: Blackwell
Searle, J – (1967) ‘Proper Names’, in Philosophical Logic, ed. P. F. Strawson, Oxford University Press

The Basics of Logical Identity

Firstly we can say that logical identity is reflexive. In other words, everything (or every thing) is identical to itself. In symbols:

∀x (x = x)

To translate:

For every x (or for every thing), x must equal x (or everything must be identical to itself). That is, everything has the (quasi)-relation of self-identity..

The logic of identity is also symmetrical. In symbols:

If a = b, then b = a

This can also be expressed in quantificational logic:

∀x ∀y ((x = y) ⊃ (y = x))

To translate:

For every x, and for every y, if x equals — or is identical to — y, then it must also be the case that y is identical to x.

Logical identity is transitive:

If a = b, & b = c, then a = c

The above means that something is “passed on” all the way from a to c. That’s why it’s a transitive relation (or expression). Thus if a is identical to b, and b is identical to c, then a must also be identical to c - by what’s called transitivity. After all, all the symbols above simply refer to one and the same object. This too can be expressed by in quantificational terms:

∀x ∀y ∀z ((x = y) ∧ (y = z) ⊃ (x = z))

To translate:

For every x, for every y, and for every z, if x is identical to y, and y is identical to z, then x must also be identical to z.

All these types of identity are satisfied by every equivalence relation (such as “is the same colour as”). Until now we’ve only talked about relations in the abstract, not specific and concrete kinds of relation. In addition, we haven’t assigned values to the symbols x, y, a and b. So with a = b we can say that the David Jones (i.e., a) and the The Thin White Duke (i.e., b) are in the relation of standing for David Bowie.

Or with a = b, and b = c, then a = c, we can say:

If Mary is the same height as John. And John is the same height as Tony. Then Mary is also the same height as Tony (as well as being the same height as John).

Mary (or a) has the relation of being the same height as (in this case) both b and c (or two different people).

Finally, the logic of identity satisfies Leibniz’s law (or the identity of indiscernibles). This law can be expressed in different ways, such as:

If a is identical to b, then everything true of a is also true of b.

In this version, Leibniz’s law is expressed with reference to the semantic property of truth. We can, instead, express it in terms only of the properties of identical objects which were perhaps initially taken to be two different objects. This, then, would be a de re (as opposed to de dicto) take on logical identity.

We can also say that Leibniz’s law entails the symmetrical notion (mentioned earlier) of logical identity, as expressed thus:

If a = b, then b = a

That is:

((a = b) ⊃ (b = a))

All this can be expressed quantificationally:

∀x ∀y (F) ((x = y) ⊃ F ((x) ≡ F (y)))

To translate:

For every x. For every y. And also for every property (i.e., F). If x equals y, then it is the case that property F belongs to both x and y. (In this case, that property is true of.)

The identity of thing x and thing y is established by the truths applicable to x and the truths applicable to y. We can see if the two sets of truths correspond with each other. Therefore we’ve established that object x is in fact identical to object y.

The symbol ≡ can be seen in the quantificational expression above. This is the symbol for the biconditional.

The biconditional symbol is defined truth-functionally (just as with the constants and connectives). The symbol truth-functionally operates on (or is applied to) the symbols (or expressions) which surround it (or to which it applies). That is, the symbol ≡ operates on, for example, F(x) in the example above; as well as on the F(y) which follows it. More precisely, it’s truth-functional because both F(x) and F(y) both have the truth-value of either true or false. Therefore the symbol ≡ operates on their truth or falsehood in order to come out with something with a different (or the same) truth-value.

In addition, we can say that biconditional symbol ≡ is equivalent to:

((p ⊃ q) ∧ (q ⊃ p))

In other words, the biconditional is symmetrical in nature, unlike the simple logical conditional. That is, if we accept that p entails q, then we must also accept that q entails p. This is unlike this simple conditional:

(p ⊃ q) or (q ⊃ p)

This means: If p then q or if q then p. However, the symbols p ⊃ q alone don’t entail their converse: q ⊃ p. The conditional is therefore asymmetrical.

We can also say that in

p ≡ q or A ≡ B

the q (i.e., a proposition) and B (a state of affairs or an object) above offer us both the necessary and sufficient conditions for p’s and A’s truth.

Now to cite a philosophical example:

m ≡ b

This is a logical expression of mental supervenience in that for every mental change (m) there must be a physical change in the brain (b). However, clearly m and b aren’t identical otherwise we’d have m = b, not m ≡ b. This conditional is also bi because it works in two directions. Alternatively put, when anything happens in the case of m, something must also happen when it comes to b, even though m and b aren’t identical.

The conditional (i.e., not the biconditional) is a different case. That is, in m ⊃ b it’s not the case that every mental change must entail a parallel (though not identical) change in the brain. Thus:

(m ⊃ b) ¬(m ≡ b)

In terms of propositions alone (i.e., not of mental and physical changes), another way of explaining the symbol ≡ is by stating the following:

p ≡ q is true if and only if both p and q are true; otherwise it is false.

In the above, p and q stand for propositions which can be taken as being either true or false. If p is true, then q is also true; even though they aren’t identical propositions. (Similarly, if p is false.) However, if p is true and q is false (or vice versa), then the overall biconditional itself (i.e., p ≡ q) would also be false. Its falsehood would be entailed by the asymmetrical truth-values of both p and q.

Eliminative Materialism

Historically, or I should say in the fairly recent past, there have been many arguments offered against materialism in the mind and matter debate. Philosophers have posited mental phenomena that were supposed to be irreducible to matter (or the brain). Churchland himself offers us four examples: emotions, qualia, raw feels and now - Churchland was writing in 1980 - propositional attitudes. Such things were/are proposed to be both ineliminable and irreducible. Churchland was saying that the goal posts kept on shifting to make way for an irreducible or ineliminable mind (or aspects thereof).

Churchland said that the resistance eliminative materialism encounters surprises him. “After all,” he said, “common sense has yielded up many theories.” Here the reader can fill in his own examples rather than relying on the many that Churchland offers.

Propositional Attitudes

Why are we so absolutely certain that our minds contain propositional attitudes (e.g., beliefs and desires)? Could we possibly make a mistake about the workings of our own minds? Churchland thinks that we can. Many people think that although we can make mistakes, sometimes big ones, about the external world, the same is not true about the internal world (the mind). But the reasons for making mistakes about the external world are pretty similar to those that can be applied to the mind or the mental. An “introspective judgment” is a “conceptual response to one’s internal states”. So just as we rely on contingent and possibly false concepts to acquire a picture of the external world, similar conceptual schemes are brought to bear on our introspective judgments. The mind is not transparent (as Descartes and others have thought). So the judgments we make about the workings of our own minds are “always contingent on the integrity of the acquired conceptual framework (theory) in which the response is framed”. We look through, as it were, contingent and possibly false concepts when we introspect. But what has all this to do with the existence or non-existence of propositional attitudes (which is Churchland’s primary concern)? The answer to this is simple: we may be wrong about our minds being the “seat of beliefs and desires”. That is, beliefs and desires as we know them may not exist. Indeed, a belief in beliefs and desires may be as misplaced, according to Churchland, as the ancient belief that the “star-flecked sphere of the heavens turns daily”.

But let’s be clear about the folk psychologist’s position is on the reality of propositional attitudes. The things we believe and desire, or, more correctly, the propositional content of our beliefs and desires, are effectively quantified over by the folk psychologist. Not only that, but he sees that the relations (e.g., entailment, equivalence and mutual inconsistency) between beliefs and desires and other beliefs and desires are “lawlike”. So all this sounds very much like a scientific theory, despite the fact that we are talking about the mental. And that’s why folk psychology is a theory. Churchland goes into more detail about the theoretical status of folk psychology.

If folk psychology is a theory, it wasn’t thought to be one until the second half of the twentieth century. Churchland thinks that it’s a mystery why the theoretical nature of folk psychology was never recognised, especially bearing in mind that he thinks that it is “so obviously a theory”. Churchland then goes onto compare folk psychology with the theories of mathematical physics. He finds very interesting parallels. But whereas mathematical physics has a domain of numbers to quantify over, folk psychology quantifies over the domain of propositional attitudes.

There are many options on reduction and ineliminativity to be taken in the philosophy of mind. Apart from his own, Churchland cites three alternatives: the identity theory, dualism and functionalism. He summarises them thus:

The Identity Theory

Identity theorists believe that folk psychology “will be smoothly reduced by completed neuroscience, and its ontology preserved by dint of transtheoretic identities”.

Dualism

The dualist “expects that [the mind] will prove irreducible to completed neuroscience”.

Functionalism

The functionalist also “expects that [the mind] will prove irreducible… [because] the internal economy characterised by folk psychology is not…a law-governed economy of natural states, but an abstract organisation of functional states, an organisation instantiable in a variety of quite different material substrates”.

Churchland happily concedes that FP enjoys a “substantial amount of explanatory and predictive success”. And then concludes: “And what better grounds than this for confidence in the integrity of its categories?” Of course, Churchland is being rhetorical here. He doesn’t believe that its explanatory and predictive successes are in fact, after 2,000 years of its hegemony, that great. He then gives a list of its notable failures. He says:

“…consider the nature and dynamics of mental illness, the faculty of creative imagination…Consider our utter ignorance of the nature and psychological functions of sleep….Or consider the miracle of memory, with its lightning capacity for relevant retrieval.”

Folk psychology offers virtually no insights on all the above. It is explanatorily and predictively more or less useless. And the main reason for this, as Churchland perceives it, is how the folk psychologist explains the mechanisms of thought. The folk psychologist sees learning “as the manipulation and storage of propositional attitudes”. That is, within the mind-brain there are sentences or propositions of some kind, or possible analogues of sentences or propositions. It is a thoroughly sentential or propositional approach to learning and thought. But, Churchland argues, something must have predated the storage and manipulation of propositional attitudes. Propositional attitudes couldn’t have shown us how to manipulate propositional attitudes before such they were on the scene. So certain non-propositional mechanisms must have predated the storage and manipulation of propositional attitudes. He says that it “is only one among many acquired cognitive skills” and therefore FP faces “special difficulties”.

Beyond Folk Psychology

So what precisely is Churchland alternative to FP? What does he believe are the mechanism of thought, learning, etc? For a start, his descriptions could be called “scientistic” (I am using that term in a non-judgemental way). He relies heavily on the findings of neuroscience and rejects any a priori philosophising. So I will describe in full his very non-linguistic and non-logical (possibly non-philosophical!) account of mind-brain activity:

“[The neuroscientific theory] ascribes to us, at any given time, a set or configuration of complex states, which are specified…as figurative ‘solids’ within a four- or five-dimensional phase space.”

Clearly there is no mention of logical relations, propositions, beliefs, desires and all the rest. This is essentially the language of neuroscience rather than the language of philosophy. Having said that, these descriptions of brain activity are not necessarily the result, or entirely the result, of research into the nature of the brain. It is a theory after all. So there is as much speculation here as one would find in any philosophical theory. The difference being that the landscape described and terms used are those of neuroscience and not of philosophy. The brain, according to this theory, has read the book of science, and not the book of language and logic.

The mental states that FP postulates are simply not law-governed. It is this that Churchland has a major problem with. It essentially takes FP outside the realm of science and into Kantian world of noumenol freedom. More precisely, propositional attitudes and their relations are not law-governed. Kinematical states and configurations, on the other hand, are law-governed.

There is an alternative to propositional or sentential language. Just as we are familiar with the language of propositional attitudes and their relations, we could become familiar with kinematical states and their relations and interactions. We could “acquire a vocabulary” that could “characterise our kinematical states”. And, of course, these law-governed relations and interactions would be scientifically bona fide. We could also learn how these kinematical states cause behaviour. We would therefore be able to predict behaviour to a higher degree than we do now. More to the point, if we could literally read the brain, we would have first-person access to other mind-brains!

Churchland clearly thinks that philosophers of mind, and philosophers generally, have over-stressed the importance of language when it comes to mentality (and have also, incidentally, lingui-fied and logi-fied the mind itself). (See quote from dynamics paper.) Churchland claims that natural languages “exploit only a very elementary portion of the available machinery, the bulk of which serves far more complex activities beyond the ken of the propositional conceptions of FP”. It is of course very hard to accept that natural languages have been over-stressed, considering the importance of them in the lives of virtually all human beings. But there are other mental activities that are just or more important, it’s just that philosophers, because of their linguistic or logical bias, haven’t really registered them. These alternatives can even be deemed forms of language, according to Churchland. Though their makeup is very different. Their syntactic and semantic structures could be “decidedly alien”. However, these non-natural languages, as it were, “could also be learned and used by our innate systems”. This new language, or these new languages, would have a “new and more powerful combinatorial grammar over novel elements forming novel combinations with exotic properties”. But because they are not

propositional, or even statemental, they could not be evaluated as true or false. Nor could the relations between these elements be “remotely analogous to the relations of entailment, etc. that hold between sentence”.

In order to demonstrate his position Churchland basically says that such a non-natural language already exists in the brains of human beings. He tells us that the left hemisphere of the brain does communicate with the right hemisphere (and vice versa). And, of course, this communication is non-propositional. So if two parts of the same brain can communicate so effectively without propositional forms, then why can’t two brains? Between the two hemispheres of a single brain is what is called the “commissure”. The commissure carries the messages from one hemisphere to the other. It is a kind of bridge between the two. So in order for two separate brains to communicate non-propositionally with each other, we would need to construct an artificial commissure. But we would need more than an artificial commissure to ensure communication between two brains. This is how it could work out, according to Churchland. We would need to implant a transducer in both brains. This transducer would “convert a symphony of neural activity into (say) microwaves radiated from an aerial in the forehead”. This would run through the artificial commissure and enter the recipient brain. And, alternatively, rather than neural activity being converted into microwaves, we would also need to convert microwaves back into neural activity. That is, the information-receiving brain would require this.

And if two brains can communicate in such a manner, why not three or even more? Such are group of artificially connected people could “learn to exchange information and coordinate their behaviour [like their] own cerebral hemispheres”. One result of this, that is, of brains communicating directly with brains, is that “spoken language of any kind might well disappear completely”. Also, in library books we wouldn’t find words and sentences but “long recordings of exemplary bouts of neural activity”. In essence, we would be reading the neurophysiological goings on of other people’s brains. Of course, we would initially need a translation manual from the brain states to our understanding of the brain states. However, if we don’t need a translation manual to understand our own cerebral activities, perhaps we wouldn’t need one to understand other people’s brain workings.

People will of course ask: “How will such people understand and conceive of other individuals?” And Churchland answers his own question by saying: “In much the same fashion that your right hemisphere ‘understands’ and ‘conceives of’ your left hemisphere – intimately and efficiently, but not propositionally!”

Is Eliminative Materialism Self-Contradictory?

There is a well-known argument offered against elimitivist materialism. It revolves around its self-contradictory or incoherent nature. The argument is this. Because eliminative materialists are not meant to believe in propositional attitudes (like belief, intention and knowledge), then their statements in favour of eliminative materialism are “just a meaningless string of marks or noises”. Why is this so? Because they believe in what they say. And they have an intention to communicate. And they have knowledge “of the grammar of the language” and knowledge of the ‘truths’ of their own findings. But belief, intention and knowledge are deemed to be propositional attitudes, and eliminative materialists don’t believe in these things. Here’s the self-contradictory bit. If the statement of eliminative materialism is true, then it is false. It is false because there are no beliefs or knowledge to express, according to its own doctrine. According to itself, the primary statement of eliminative materialism is a “meaningless string of marks or noises” because it cannot be a belief (true or false) and it can’t be knowledge because these things are propositional attitudes. It therefore can’t be true by its own standards. As Churchland himself says: “Therefore it is not true. Q. E. D.”

Patricia Churchland offered a riposte to these ‘refutations’ of eliminative materialism, which Paul Churchland quotes in full. Firstly he offers a psychological rather than philosophical argument. That is, EM is just so radical, revolutionary and, perhaps, in initially counterintuitive, that it is understandable that people react fiercely to it. Churchland finds an historical parallel with eliminative materialism. It revolves around the reaction against vitalism. It was once held that when a vital spirit inhabited inanimate matter, it would become animate (alive). This belief in vital spirits was shared by just about everyone at certain points in history. It was “integrated with many of our conceptions”. So if anyone were to reject vitalism, which they did, the “magnitude of the revisions any serious alternative conception would require” would have encouraged just about everyone, at first, to fiercely reject anti-vitalism (this in fact happened). So, according to Patricia Churchland, the vitalist could have, and perhaps did, offer arguments the anti-vitalist that would be very much like the arguments the anti-eliminativist offers against the eliminativist. This is how it goes:

“The anti-vitalist [eliminativist] says that there is no such thing as vital spirit [propositional attitudes]. But this claim is self-refuting. The speaker can expect to be taken seriously only if his claim cannot. For if the claim is true, then the speaker does not have vital spirit [propositional attitudes] and must be dead [contradicting himself]. But if he is dead [or lacks propositional attitudes], then his statement is a meaningless string of noises, devoid of reason and truth.”

So the antagonist says that the eliminative materialist can only be saying something true if he agrees with him. This effectively means that the eliminativist can only speak the truth about propositional attitudes if he doesn’t believe what he says is the case.