14.       Eliminative Materialism

Given the difficulties so far, perhaps we should defend physicalism about the mind by denying that there are such things as mental states. This view is eliminativism. Let’s consider their views about propositional attitudes. (Note that it is not a self-refuting view.)

 

The Case for Eliminativism

 

Turns on two claims: one concerning how we should think of PAs, and the other concerning the history of science.

 

Propositional Attitudes as posits

 

We should think of PAs as posited internal causes in a theory about behaviour – a common-sense theory, not a properly scientific one. This gives us BDA psychology.

(By contrast an Instrumentalist thinks of them as tools for prediction only.)

 

The role of the propositional objects

 

For an E-ist, the propositional objects of the PAs describe how the inner causes work. PAs get their objects by what they do – assigning objects to PAs describes what they do.

E-ists call the common-sense theory a folk psychology, to associate it with folk theories of other kinds.

 

The lesson of the history of science

 

The History of Science is full of overthrown folk theories – in astronomy, medicine, nutrition, physics, etc. We no longer explain in terms of epicycles, horor vacui, phlogiston, vital spirits, humours, aether, etc. The E-ists suppose B&D will disappear for similar reasons.

E-ists do not say we can do w/o folk psychology now – we still need to use it as there is no replacement theory (yet.)

They do say that there are similarities between flok psychology and those other folk theories.

It is ancient, not much changed (we’ve added subconcious desires [and removed gut, heart, etc. as mental sites.]) It can’t handle a lot of things; like dreams, sweating, shivering, etc. It can make bad predictions. Neuroscience looks like it might be able to do better.

 

Degenerating research program

 

It looks like what Lakatos calls a degenerating research program. Ancient, stagnent, lacking explanatory resources for obviously related facts, the lack seems irreparable, there’s a better theory in the offing. Compare all the other defunct fok theories.

 

The quick dismissal

 

The E-ists can say that we must talk about B&D (and their own project is a B&D) until we can talk in terms of B*&D* - where B*&D* talk is the replacement for B&D talk.

[Just so, doctors talked in terms of a balance of bilious and sanguine humours etc., until they could talk in terms of germs and neurology etc.]

 

The Functionalist Reply to Eliminativism

 

We shall claim that the posits in some kinds of theories are immune to elimination by new information. The posits of F-ism are like this. They are functionally defined and say nothing about any intrinsic properties of the posits.

Consider an example

 

Imprinting

 

Ducklings imprint on the first large thing they see. We ascribe imprinting per observed behaviours, but what we know goes beyond that. E.g. the sighting lays down a trace in the duckling’s head, and it is connected to leg movements, etc.

We don’t accept that the idea of imprinting, even though it has the faults mentioned for folk psychology, is ripe for replacement. It is possible, but we don’t think likely, because its behavioural evidence is too great. What we expect will be revealed is not the nonexistence of imprinting but the mechanism.

 

Modesty in explanation

 

Same thing holds for B & D. B & D are posits, but very plausible and well-supported ones.

We can distinguish modest from immodest theories.

Both make posits to explain phenomena, but modest ones are quiet on the intrinsic nature of the posits, while immodest theories are not. Modest ones describe roles for the posits, but not how they play them.

E.g. Observation makes it almost certain that a fridge’s temperature is regulated by a thermostat – but in hypothesizing a thermostat modestly we don’t have to say anything about how it works.

Common-sense F-ism is like that. It’s logically possible that it’s wrong – but not really possible.

 

Is common-sense functionalism modest?

 

CSF is probably modest because:  

1.   it says nothing about intrinsic nature

2.   it is restricted to claims about causal roles

 

Explanation by discrete beliefs

 

One can dispute this claim. It might be that CSF-ism makes enough claims about the mechanism of B&D to count as immodest. E.g. For X believing both P and Q, if we think that P and not Q causes A, then we must think that P and Q are distinct states. We may be committed to a LoT by our CSF; but this is really a matter in doubt – to be decided by neuroscience.

 

Immodesty and what goes on inside

 

The argument for immodesty may confuse two possibilities.

Divide platitudes about mind into two groups: one is neutral about brain processes, the other is not. Blockhead is one way a brain’s constitution might make it impossible to have a mind according to CSF-ism.

If CSF-ism is immodest it might be because

  1. the 2nd group of platitudes demand it. E.g. we might require a LoT or something else more plausible than Blockheadedness. We [authors] don’t think this is likely.
  2. the 1st platitudes demand it. E.g. the functional roles posited may tightly constrain types of brain architecture. But we know that roles are occupied, so this just means we can know more about the brain from knowing about behaviour than we first thought.

An argument for E-ism along these lines has to say there are implementation commitments in F-ism’s platitudes (as per 2) and the implementation commitments are empirically risky (as per 1.) But the claims can’t be mixed in that way, so E-ism doesn’t get support from that.

 

Does functionlism make it too easy to save folk psychology?

 

Is functionalism too cheap a way of avoiding elimination?

 

Chuchland thinks that sort of F-ist reply is too easy to make. It could have been applied in just the same way to save all the old defunct theories we listed above – phlogiston, aether, humours, etc.

   

Functionalism and phlogiston

 

Phlogiston was understood as whatever it was that disposed its container to combustion correlatively with its concentration, and it was expelled in combustion. But it was eliminated when combustion was understood as oxidation and as a ‘taking in’ not a ‘giving off.’ There was nothing to play the functional role of ‘being given off.’ There was no such role to be played.

 

Avoiding elimination by redefinition

 

Phlogiston could be kept if its F roles were redefined appropriately. But this is a criticism of radical redefinition – we’re no longer even pretending to talk about the same thing in the same words.

Some redefinition is OK. Consider ‘atom’ for example. The redefinition can only be limited however. [Where are the limits?] ‘Atom’ still works because enough of the former defnintion still works, and CSF-ists think that in any likely scenario, enough of the folk roles of B&D will persist to count as B&D.

 

Empirical functionalism: how to have strong internal constraints without inviting scepticism

 

Recall (ch. 5) that EF-ists think B&D have essences distinct from their folk roles. The folk roles fix references on internal roles that make a B a B and a D a D. Once the states filling those roles are found they are the essences of the B&D.

 

Eliminativism for Martians

 

EF-ism puts strong constraints on intentionality.  No creature very different from the exemplar of an intentional creature can be an intentional creature. Martians, that behave and look much like us, may fail to be intentional because of internal architecture. (This was the chauvinism critique of EF-ism.) E-ism would be true of Martians in that case. But, of course, it couldn’t be true of us, or true in general.

 

Natural Kinds and Scientific Reductions

 

Another argument against folk psychology appeals to types of reductions in science.

 

Two famous reductions

 

Genes were posited to explain Mendel’s observations and folk knowledge about inheritance. Gnetics became molecular biology when these patterns could be explained in terms of DNA molecules etc. So genes were identified with strings of DNA because they played the right roles.

The thermodynamic theory of gases became the kinetic theory when explanations in terms of pressure, heat, etc. could be replaced by explanations in terms of mass molecular dynamics. This showed what heat and pressure actually were.

 

Smooth reductions

 

Those examples show that new theories can incorporate the posits of older theories by ‘explaining’ them. We may anticipate that the same sort of thing will happen with neuroscience. But the wrong lessons may be being learnt here.

These reductions have relatively simple bridge laws that relate objects in the ontology of the old theory with those of the new. This gives us a smooth reduction (though that can be exaggerated.)

 

Smoothness and elimination

 

Smoothness is not essential for a reduction not to be an elimination. If it were then E-ism would be more likely since reductions might be quite rough. Imprinting in chickens might be differently implemented from the same in ducks; so there couldn’t be a smooth reduction from imprinting theory to neuroscience. But that wouldn’t eliminate imprinting.

 

Failures of smooth reduction without elimination

 

Note that the relationship between genes and DNA is not that smooth really.

Also note that the kinetic theory of heat can be reduced from mass molecular to individual molecular terms; but there are no interesting biconditionals linking such terms. There is no smooth reduction there, but that doesn’t mean we have to eliminate mass molecular terms or thermodynamic terms.

Finally note there is no prospect of a smooth reduction of meteorology to any lower level science, and yet tornadoes are not fictions and cannot be so eliminated.

 

Psychological kinds, natural kinds, and serious science

 

There are arguments that the states that play the CSF-ism roles have to be tokens of natural kinds.

There is a crude argument that, since only natural kinds exist, the states that play the CSF-ism roles have to be tokens of natural kinds, but lots of things – like famines, ways to get to Paris, etc. – exist without being natural kinds.

 

The demand that psychology traffic only in natural kinds

 

An interesting argument to the same point is that only natural kinds deserve to be included in a serious science. B&D may exist as CSF-ism describes them, but unless they are natural kinds they shouldn’t be in a serious psychology.

E.g. constellations exist, but they don’t play a respectable explanatory role, [and aren’t natural kinds?] so they shouldn’t be in any serious astronomy.

B&D may be like constellations.

So E-ism1 says B&D are like phlogiston, while E-ism2 says they’re like Orion.

 

Why beliefs and desires are better than constellations

 

B&D do still play an explanatory role. If to be a natural kind is to play an explanatory role in a serious science then there are plenty of functional kinds that will qualify – like vitamins, typhoons, chairs. Constellations won’t.

If natural kinds are more than just explanatory then B&D may not quaify as NKs, but then there are plenty of non-NKs in serious sciences.