Chmess and twaddle

by Chris Bertram on June 26, 2004

Brian Leiter has a couple of interesting posts reflecting on the state of analytical philosophy, and also links to Dan Dennett’s The Higher-Order Truths of Chmess , which I hadn’t read before. Dennett cites Donald Hebb’s dictuum “If it isn’t worth doing, it isn’t worth doing well,” and remarks

Each of us can readily think of an ongoing controversy in philosophy whose participants would be out of work if Hebb’s dictum were ruthlessly applied.

I confess to succumbing to feeling of utter despair whenever I have to listen to people talking twaddle about twater on twin-earth, so that would be my candidate even though I have dear colleagues who care passionately about the topic. But the twaddlers themselves would, no doubt, want to consign some of my pet interests to the bin. Commenters are invited to nominate the disputes that drive them crazy, and those who care about the tw-topic are invited to explain to the rest of us why we should think it matters.



Daniel Nolan 06.26.04 at 2:28 pm

I don’t talk about Twin Earth cases much myself, but I think the people who do talk about them are often trying to make progress on some quite important questions – and that talk about Twin Earth and similar cases _has_ made progress on important issues. So I thought I might try to say something about how Twin Earth cases relate to some general issues that seem important to me, and I hope might show Crooked Timber readers why this work seems “worth doing”, even if they don’t find it personally interesting.

First, a brief recap for people who haven’t heard of Twin Earth. Twin Earth cases are thought experiments, where we are invited to consider someone who is just like us, but lives in an environment where one of the features of the environment is superficially the same but is different at some more basic level: the original example was someone who is just like you in all obvious respects, and lives in an environment that appears just like yours, but lives on a planet where the substance that fills the seas, comes out of taps, falls as rain etc. is a colourless, odourless liquid made up of XYZ rather than H2O, (where XYZ is some fictional alternative chemical that has the observational behaviour of water). Arguably, our word “water” picks out H2O, and their word “water” instead refers to XYZ. Many people think that our word “water” doesn’t pick out the colourless, odourless stuff on Twin Earth, and likewise the Twin Earthians don’t refer to the stuff that we drink, bathe in etc. with their word “water”. Hilary Putnam, who pioneered the thought experiment, thought that this was a way of demonstrating that “meaning ain’t in the head” – what our words mean depends not just on our internal processes, but also on what those processes typically respond to in our immediate environment. (To avoid worries that the two worlds will not seem different because chemists say different things in the two worlds, you may wish to consider an Earthling of 1700 and her Twin Earthling counterpart – since neither know anything much about the chemistry of the liquids around them, their two environments might be the same in all observational respects, though the Earthling refers to the stuff we now know is made from H2O and the Twin Earthling refers to a quite different substance.)

I don’t want to get stuck into any arguments about whether this sort of case does successfully show Putnam’s claim is right, but I do want to talk about three issues that Twin Earth cases are supposed to help with. The first is an issue about what the natural sciences can tell us about the world. A traditional view of the sciences was that they were a way of discovering essential truths about the world – not just how things happened to be, but how they must necessarily be. The consensus in the early and mid twentieth century was that this was wrong – science could only tell us how things happened to be, not how they must necessarily be. Perhaps logic and mathematics delivered necessary truths, and maybe there were other ways to discover them, but disciplines like physics and chemistry couldn’t lead us to discover necessary truths about the physical or chemical natures of objects or the world.

Twin Earth cases, along with other work by Putnam and Saul Kripke, brought back the idea that science could allow us to discover necessary truths about scientific matters. After all, if our word “water” always serves to pick out H2O, then “water is H2O” is plausibly a necessary truth. (Just as the Twin Earthling’s sentence “water is XYZ” expresses a necessary truth – though not about the stuff we call water, but rather about XYZ). But we can only find out “water is H2O” through doing chemistry – so it looks like chemistry is telling us about the necessary nature of water – water’s “essence”.

Naturally enough, philosophers have wanted to better understand how science might be telling us necessary truths after all. We argue about whether it does (i.e. whether the Twin Earth cases show what they are supposed to), and argue about whether, if it is a necessary truth that water is H2O, whether this necessity might be “factored” into something contingent that science tells us and something necessary that we know through some other means (e.g. through the grasp of our concept of water). We wonder whether this sort of case refutes the view that necessity is somehow a product of our linguistic conventions – whether it is an artifact of how we represent the world, or whether there are neccessary facts that have nothing to do with representation that we can discover. As you can guess, not much of this is uncontroversial.

What is at stake is the kind of insight into the world that science gives us (e.g. the insight into chemical composition that chemistry gives us). Is it just how things happen to be made up chemically, or is there some range of chemical facts with the same sort of inevitability that the truths of logic and mathematics have? If there are such facts about essences, over and above contingent facts about how chemicals happen to combine, how do we fit those facts into our picture of the world? If there is a difference between discoveries about essences and discoveries about contingent or accidental truths of chemistry, where is the boundary?

Now, I presume some people aren’t interested in questions about the kind of insight into the world that science gives us. Remember, though, that the questions of the previous paragraph generalise – at stake is not just our understanding of chemistry, but physics, biology, psychology, sociology, economics, ethics… “twin earth” sorts of questions, along with questions about what discoveries are discoveries of necessary truths, can be asked about just about any area of enquiry, though the arguments might go differently in different domains. I hope that the people who aren’t personally interested in how to understand what science reveals about the world might be able to recognize that there’s some point in having people who are interested in that question consider it. And hopefully they’ll be able to see why it might be worthwhile to work out how much, if any, of science reveals how the world must be of necessity, and how much is just telling us how things are in fact (though we might be able to put such information with other information we have about the workings of language to discover things we might have thought were possible, in some generous sense, aren’t, in the end).

So that’s one use, in the philosophy of science/metaphysics area. I happen to think that the use twin-earth cases are to questions in philosophy of mind and language reflects on issues of even greater general interest: the nature of psychological explanation, the understanding of the contribution of context to the meaning of utterances, the relationship between a speaker or community’s understanding of their words and the meanings of words in public languages. I suspect it’s even easier to make the case that they’re relevant and indeed important to those questions. But I might leave the attempt to explain why twin-earth cases get discussed in those cases to someone else, since I’ve gone on long enough for one comment.

I should make one thing clear. It’s not that I claim we can solve any of these broad questions, about our understanding of what science does, or what we can discover about how the world has to be rather than just how it happens to be, or about the other issues I touched on like the nature of mental representation or the function of language, just by waving a wand and uttering the magic expression “twin earth”. The debate that swirls about twin-earth cases is only a piece in those bigger jigsaws. And philosophers might disagree about the relative importance of the issues here and other issues in philosophy of science, metaphysics, philosophy of mind and philosophy of langauge. But it’s certainly not true that philosophers are just considering an outlandish example for its own sake when they discuss twin earth: many of them think that they’re gaining some important insight into some very important questions, questions which it’s very hard to make progress on.


Kieran Healy 06.26.04 at 2:44 pm

Hi Daniel! Hope you’re doing well in St Andrew’s these days. I think you just won a prize for the longest _bona fide_ informed comment ever left on CT.

NB Chris: This is what happens when you needle the metaphysics people. You’re asking for trouble.


Matt 06.26.04 at 2:44 pm

Something that I think is important for thinking about twin-earth type cases, and rarely mentioned by those who play that game, is that Putnam himself soon abandoned the approach. He still promoted externalism about the mind and content, but from a quite different way. And, he quite clearly changed his mind about whether water is necessarily H20. I think his later possition is much more reasonable, and that his arguments against the old straight _Meaning of Meaning_ argument are very good. You’d not know this, though, if you read most of the twin earth papers.


Kieran Healy 06.26.04 at 2:46 pm

Putnam himself soon abandoned the approach.

But isn’t this true for all x?


Matt 06.26.04 at 3:30 pm

_But isn’t this true for all x?_

No. He still takes much the same approach to much of his earlier work on logic, including the analytic-synthetic distinction, and to his work in philosophy of science proper. And, as I noted, he still supports externalism about the mental and about content, which was the real core of the twin-earth argument, but just dropped the particular argument since people tended to think it was showing more than he would (now) want to take it to show. That is, he doesn’t think we can do the sort of arm-chair science that many try to do w/ the twin-earth arguments (the “water is necessarily H20” claim). When he later argued against that claim, he looked to the work of actual scientists for a starting point. It’s the a priori element that Putnam (rightly, I think) wanted to distance himself from, and that is, to my mind, the equivilant of chmess.


DonBoy 06.26.04 at 3:53 pm

If you read Dennett’s piece and substitute the word “blog” for “philosophy”, much of it still applies.


Chris Bertram 06.26.04 at 4:25 pm

Daniel, thanks ever so much for responding with such an enlightening comment! I guess I’m less bugged by the original Putnam piece than I am by the amazingly intensive cottage industry that has grown up around this particular thought experiment. But I think you’ve said enough that next time I’ll attend more rather than thinking “oh God, not another paper on TE….”


Matt McGrattan 06.26.04 at 5:32 pm

I think the Twin Earth thought experiments are useful but find myself convinced by Alan Sidelle’s work [see Necessity, Essence, and Individuation (1989)]* which argues that what’s really going on in these experiments is a teasing out of certain linguistic (and therefore analytic) principles of individuation.

An important consequence of his work is that the necessity and the a posteriority in Twin Earth cases separate out. The necessity on Sidelle’s account comes from the fact that in scientifically-informed English-usage being the same chemical substance as x simply means having the same chemical structure as x. There’s nothing metaphysically wierd about this necessity — it’s just common or garden analyticity.

The a posteriority comes from the fact that we need to head off and do some empirical investigation to find out what that structure is.

On the ‘Sidelle’-type account of what’s going on in these cases no substantive metaphysical conclusions fall out of these thought experiments. Nevertheless, as a way of exploring or drawing attention to certain relationships between concepts they can be pretty handy.

* Sidelle’s book builds quite heavily on Nathan Salmon’s ‘Reference and Essence’.


Matt Weiner 06.26.04 at 5:35 pm

I think the twin-earth examples also helped inspire the philosophers like Dretske, Fodor, and Millikan who began to give analyses of concepts such as (very roughly) “The concept of X is the thing in the brain that is ceterus paribus caused by X.” This analysis has the consequence that it’s not always transparent what our beliefs are–we can’t always tell whether a belief expressed using the word X and a belief expressed using the word Y actually incorporate the same concept–which is a very interesting result. I’m not an expert in this sort of stuff, though, so don’t take me that seriously.

What I really want to say is–I wish they would stop calling it ‘twater’.


Matt McGrattan 06.26.04 at 6:41 pm

Re: ‘twater’ – yeah, for Brits like myself that’s particularly amusing. Although more so on the written page…

I’ve been trying to think of ‘chmess’ -like areas of philosophy and apart from pretty much the whole of post-Kantian ‘continental’ philosophy :-) I’m finding it hard to come up with one.

Work on perception, and on colour in particular, often seems to be woefully under-informed about the science of colour perception. I’ve lost track of how often I hear people confidently pontificating away and making claims that are straightforwardly contradicted by things like Land’s experiments on vision from the 50s and 60s.

Similarly, I do sometimes get frustrated by the general trend (post-‘linguistic turn’) to make substantive inferences from language about metaphysical claims. [Of which the Twin Earth cases are one type of example]

It’s irritating to sometimes hear such claims made on the basis of things that are local to the particular grammar and/or semantics of English, for example.


Matt Weiner 06.26.04 at 9:46 pm

Oh, is “twater” not pronounced like the obscenity in Britspeak? I pronounce them the same, when I pronounce them at all. (Modulo the “er” of course.)


Michael Kremer 06.26.04 at 11:28 pm

Warning: off-topic remark.

I once heard Alasdair McIntyre say something like this: “if it’s worth doing, it’s worth doing poorly.”

I like that as much as I do Hebb’s (or maybe more).


Matt McGrattan 06.26.04 at 11:35 pm

Well, they aren’t pronounced the same in my accent.

Twater is pronounced like ‘water’ only with a ‘t’ at the start. Same vowel sound as ‘caught’. i.e. backwards-c in IPA notation.

The obscenity is pronounced with the same vowel sound as ‘cat’. i.e. /a/ in IPA notation.

This might vary slightly from accent to accent. Some ‘traditional’/rural Scots accents, to my certain knowledge, would pronounce them the same. There may be others.


foxforce5 06.27.04 at 1:53 am

Where I’m from “twat” rhymes with “hot”, but “twater” — alike “water” — rhymes with “caught her” (rather than “got her”).


intrigued guest 06.28.04 at 2:40 am

‘Work on perception, and on colour in particular, often seems to be woefully under-informed about the science of colour perception. I’ve lost track of how often I hear people confidently pontificating away and making claims that are straightforwardly contradicted by things like Land’s experiments on vision from the 50s and 60s.’

Sorry to backtrack, but I think I might be exactly the sort of person that Matt has in mind here (I certainly know nothing of Land’s experiments, and I’m trying to write a thesis on colour perception).

Could I therefore ask Matt to say, in a nutshell, what sort of pontifications he has in mind?


Matt McGrattan 06.28.04 at 7:46 pm

Well, it might be worth while to take a look at Land’s work. I believe the Hardin book ‘Color for Philosophers’ has a summary. The original papers are in Scientific American … I don’t have the exact citations to hand.

Basically, Land carried out a range of experiments which taken together show that colour has little to do with surface spectral reflectance and is only indirectly connected to the wave length of light. [I’m probably over-simplifying here]

For example, he showed how one could project an image using nothing but red and white light and yet create the impression in viewers of a full colour image. He also carried out a range of experiments using so-called colour ‘Mondrians’ in which he was able to take patches of colour and manipulate them so that a single patch’s surface spectral reflectance remains constant and yet its perceived colour alters.

There’s a number of other experiments all of which taken together suggest that perceived colour is the product of a complex interplay of wavelengths across the visual field where the relative rather than absolute lengths of the wavelengths give rise to perceived colour variation.

I’ve lost track of how often people appeal to things like wavelengths of light, surface spectral reflectance and so on when talking about colour. Of course they are usually hedged by ceteris paribus clauses and references to ‘standard conditions’ and so on. But they still seem then to proceed more or less as if colour was nothing more than surface spectral reflectance. [And I’m not just talking about objectivists about colour here… dispositionalists also often proceed as if the ground of the disposition was straightforwardly something like surface spectral reflectance.]

Of course there’s loads of philosophical work on colour where the Land experiments aren’t strictly relevant and lots of other work that DOES take that kind of data into account. But I do often attend seminar presentations or other talks which do proceed in ignorance of Land’s work when it is relevant.

[Sorry for the slightly rambling nature of this reply….]


intrigued guest 06.28.04 at 11:40 pm

O.K. – thanks Matt. So the moral is: don’t talk about surface spectral reflectance and wavelengths without reading up on the Land experiments first. Got it!


Anders Weinstein 06.29.04 at 5:00 pm

I thought Land’s experiments were taken by him to support the “retinex theory”, according to which the stimulus for color *is* reflectance (actually, a triplet of reflectances). The point of Land’s experiments is that the stimulus for color is *not* intensity of light of various wavelengths alone. But the retinex theory does nevertheless give us some fairly natural physical property, one which is in fact more invariant under changing illumination, as the basis for color experience.

In artificially contrived situations, illusions can of course result if the way the visual system (in effect) estimates reflectances in the normal environment under which it evolved is no longer reliable under these circumstances.

I haven’t read Hardin myself, and I have also heard that the neat retinex theory itself has its problems in turn. But Land’s reflectance-based theory has been taken to support a physicalist theory of color (e.g. by Paul Churchland in, I think, _Scientific Realism and the Plasticity of Mind_).


Matt Weiner 06.29.04 at 10:44 pm

Where I’m from “twat” rhymes with “hot”, but “twater” — alike “water” — rhymes with “caught her” (rather than “got her”).
Same with me, except that “caught her” and “got her” may also rhyme, especially if you say them fast.

Next question: How should you pronounce “Kant”?


dsquared 07.01.04 at 12:29 pm

with a short German “a” and a steely gaze.

Comments on this entry are closed.