Become a Fan

« Joshua Miller guest post: Proposal for a First Year Seminar | Main | Expressivism: A Misunderstood Meta-Ethical View »

27 September 2012


TrackBack URL for this entry:

Listed below are links to weblogs that reference How can philosophical theories be empirically tested?:


Feed You can follow this conversation by subscribing to the comment feed for this post.

Sara L. Uckelman

• The waiter who just grinds the pepper will not get any tips.

• The waiter who now ground the pepper will not get any tips.

Perhaps the Dutch versions which she investigated are illustrative of the points you're discussing than these English translations. The first is not ungrammatical in English, it just requires a different reading of "just". Given the present-tensed verb "grinds", "just" has to be interpreted as "only", that is "The waiter who only grinds the pepper and does nothing else will not get any tips". It is only in the context of a past-tensed verbs such as "ground" that "just" has a temporal sense of "immediately in the past".

I'm not sure I'd go so far as to say that the second is ungrammatical either, though it would certainly sound more natural as "has now ground".

Catarina Dutilh Novaes

The 'just' is my own translation; the translation she was using is 'just now', which I think is problematic because it uses 'now' (which is exactly what is under discussion!) The Dutch expression I was translating as 'just' is 'zonet', which has a clear temporal dimension, unlike 'just'. Do you have suggestions on an alternative to 'just' as a translation to 'zonet'?

Sebastian Lutz

Two questions and one comment:

First, you write

This does not mean that all philosophical questions can or should be reduced to empirical questions (I’m not a positivist!)

What kind of positivism are you thinking of here? I figure that it's not logical positivism, but who held this view?

Second, it seems that Brederoo's research relied on the assumption that the philosophical theories about 'now' have to capture (more or less?) the speakers' behavior or intuitions. But that seems to make them into comparably straightforward empirical theories (like those of, presumably, linguistics). How would one go about empirically testing perdurantism or nominalism, say? It seems that this example of empirically informed philosophy does not easily generalize.

Finally, the results on the example sentences seem to me rather difficult to interpret, since they contain two verbs with different tenses ('ground'/'grinds' and 'will get'). The sentence 'The waiter who now ground the pepper was particularly attractive' seems to me very nicely grammatical, for example (it could appear in a story told in the past tense). The original 'now'-sentence accordingly seemed grammatical to me until I reached 'will'. That alone might explain the difference in the P600-effect (think of a really long sentence that is clearly ungrammatical once you have spent considerable time analyzing it).

Antonio Negro

Maybe "recently" can do the trick as well, ideally in other contexts.

Catarina Dutilh Novaes

I was referring to Comte-style positivism (and it was a joke anyway...).

While it may not generalize to all areas and topics that philosophy is interested in, I take this approach to be at least the right approach to pretty much everything that is done under the heading 'philosophy of language'. What I found particularly interesting is that this particular analysis of the results was only made possible because of the background philosophical theory that Brederoo had found in Recanati. The people who conducted the experiment themselves were not able to draw the same conclusions. So it seems to be a nice case of true division of labor.

Catarina Dutilh Novaes

Yes, something like "The waiter who recently grinds the pepper will not get any tips" should work for the contrast in question.

Sebastian Lutz

Ah, humor. Not my strength.

I wonder whether, in a sense, you are underselling the generalizability of the example: There is a passage in Van Fraassen's /Laws and Symmetry/ (I think) in which he argues that some philosophical position entails that there is one eternal electron (I think). That philosophical position can thus be empirically refuted. Similarly, if a position in the philosophy of language entails that 'now' is in ordinary language only used for the present tense, that position can be refuted as well.

In another sense, your claim may be too strong, however: If I have two theories in the philosophy of language that are empirically equivalent (say, realism and anti-realism about propositions), I doubt that one could devise an experiment that speaks in favor of one or the other theory.

Jeff Ketland

Catarina, any physical theory T in a reasonable language L (not containing the predicate "x is a god") is conservatively extended by adding the metaphysical sentence
(G) "there is at least one god",
if quantifiers are relativized. So, if A is an L-sentence, let A* be the relativization of quantifiers in A to "x is not a god". Then you get:

If T* + "there is a god" implies A*, then T implies A.

To prove this, suppose T does not imply A. So, we have a model which M satisfies T but not A.
Now let a new model M' be obtained by adding one new element g to the domain of M, keeping the extension of all non-logical predicates in L fixed (i.e., except =, so, the extension of = is extended by the ordered pair (g,g)).
Let the extension of "x is a god" be this element g.
Then M' satisfies T* + "there is a god". Furthermore, M' does not satisfy A* either, since its quantifiers are restricted.
So, T* + "there is at least one god" does not imply A*. QED.

The theories T and T* + "there is a god" are equivalent in what they assert about every x that is not a god; assuming that no god is observable, they are empirically equivalent.

The key here is that the posit is an "idle posit", and does not logically interact with the entities postulated by the theory. This will not happen if one has mixed predicates. If one adds "there is a god who modifies the local quantum field at certain regions in spacetime", then the theories may become distinguishable.

Roger Koppl

Seems like your student is taking X-phi in a somewhat new direction, which seems entirely positive to me.


Not really a new direction, and not really X-phi. Stuff in this general vein has been going on for a while now in philosophy. The clearest examples that I know show up in the work by Paul Pietroski and his collaborators on 'most'. For example:

Reinhard Muskens

Hi Caterina, interesting stuff but I'm puzzled. As far as I can see this is about the semantics of the word 'now'. There are several theories and there are data against which we can test those theories---native speaker judgements about syntax, native speaker judgements about entailments and truth conditions, and P-600 effects. Why isn't this just plain vanilla linguistic semantics? There is of course a continuity between semantics and philosophy (especially logic), semantics uses logic, but is this really a case of 'a philosophical theory that can be empirically tested'? To me it just seems a linguistic theory that is being tested.

I'm curious about the Dutch versions of the example sentences! Could you give them, please?

Catarina Dutilh Novaes

I don't mean to say that this kind of approach applies only to phil language, nor that it applies to every and all questions within phil language. I think I'll write another post to give another example, this time from my own research.

Catarina Dutilh Novaes

Hi Jeff, sure. But even in less toy-like examples there will be lots of cases where two competing philosophical theories are empirically indistinguishable (one of the reasons why I I emphasize that I am *not* saying that all of philosophy can be reduced to empirical questions). But empirical data can in many cases at least reduce the logical space of available theories. Suppose you have theories T, T* and T** about a given phenomenon. Now, it may well be that the currently available data is equally compatible with T and T*, but not with T**. In such a case, eliminating T** from the pool of plausible theories is an informative move, even if you still have T and T* as equally plausible.

Catarina Dutilh Novaes

Hi Reinhard, I'd say your comment speaks precisely in favor of my plea for more fluidity between the different fields! It doesn't really matter much under what heading the research is done, but the theoretical backbone for the thesis was provided by discussions which took place within philosophy, among philosophers. So the semantics of 'now' appears to be a legitimate philosophical topic too.

The sentences in Dutch are:

*De kelner die zonet de peper maalt krijgt geen fooi.

*De kelner die nu de peper maalde krijgt geen fooi.

Catarina Dutilh Novaes

Thanks for the pointer!

Reinhard Muskens

Thanks for the Dutch originals, Catarina! I think that the second sentence is in fact grammatical. I wouldn't think twice about it if someone uttered it and such things are probably said all the time.

I'm not an expert on now, but isn't it true that now sometimes refers to the reference time? Here is an example from Alice in Wonderland where that seems to happen:

Just then her head struck against the roof of the hall: in fact she was now more than nine feet high, and she at once took up the little golden key and hurried off to the garden door.

I'm all for fluidity between fields and as a matter of fact I think research questions are all that matters. What field a research question belongs to is not in itself interesting. So I guess we agree. As you know there is now in fact a large community of people (logicians, linguists, and computer scientists) working on questions related to the meaning of language. This basically started in the late 1960s when Barbara Partee exported Richard Montague's groundbreaking work on formal semantics to linguistics and the cooperation still continues. I think of Kamp, Kaplan, Recanati and Predelli and others as people that are all to some extent part of this larger cooperation. (Two others are, for example, Irene Heim and Angelika Kratzer---both trained as philosophers, but now very influential in linguistic semantics.)

Catarina Dutilh Novaes

Sure, as an ILLC child, this is really the kind of thing that I was fed on from the very beginning :)

But there is one important proviso, which is that only recently has formal semantics become interested in empirical data. My former colleagues Maria Aloni and Katrin Schulz now have an important empirical component in their work, and if I remember correctly so does your project, but these are recent developments. In this sense, I'd say that this large field of language studies where logicians, philosophers and linguists collaborate can still become more empirically informed.

Jeff Ketland

Hi Catarina, thanks,

"But empirical data can in many cases at least reduce the logical space of available theories."

Yes, I understand, but I am sceptical. Is there an example of this for philosophical theories? Can one identify:
(i) a metaphysical claim C, and
(ii) an empirical claim E that contradicts C?
There may be some; I can't be sure. But for a large class of metaphysical claims, I tend to think the pervasive situation is the reverse. E.g., consider:

(C) There are at least aleph_0 possible worlds.

Alone, this doesn't imply any empirical statements, except logical truths.

If one adds mixed/bridge principles, this can change. E.g., "the structure of each possible world is encoded in the DNA of some actual virus on Earth" - which might make for a nice sci-fi story involving a modal metaphysician ....


Sebastian Lutz

Hi Jeff (if I may), I think Swinburne's translation of "There is a God" ("There exists necessarily a person without a body (i.e., a spirit) who necessarily is eternal, perfectly free, omnipotent, omniscient, perfectly good, and the creator of all things") could be seen as a metaphysical thesis that is empirically testable and has in fact been disconfirmed repeatedly. I think Anthony Flew was right in claiming that theism had suffered the "death by a thousand qualifications"; when all occurring terms are understood in the usual sense, theism is thoroughly disconfirmed, and it is only by redefining 'good' and 'omnipotent' that one can still stick with the claim. In short: C="There is a God", E="Some earthquakes have killed newborns". The result is not a contradiction, but only a disconfirmation. However, to demand a logical contradiction would make it hard to find even empirical theories that can be tested. I gather that Catarina would also be o.k. with empirical data reducing the class of plausible rather than possible theories.

Of course, one could always say that theism consists of one empirical theory—the deductive and inductive inferences that theism makes possible—and a metaphysical claim—whatever remains of theism after its empirical component has been removed. But I doubt many would agree with this meaning postulate for 'metaphysical' (incidentally, I might).

Catarina Dutilh Novaes

I usually try to stay away from metaphysics, and one of the reasons is precisely because it's hard to articulate how metaphysical theories could be tested empirically :) Actually, I tend to stick to things that *people* do (language, cognition, logic).

But anyway, at least some people (Katherine Hawley, Barry Loewer) seem to be doing great work on the interface between physics and metaphysics. This leads me to believe that for at least some metaphysical questions, empirical data from physics is of some relevance. (This is a bit of an 'argument from authority', thus not very compelling, but I guess it's the best I can do...)

Jeff Ketland

Hi Catarina, "...precisely because it's hard to articulate how metaphysical theories could be tested empirically".

Yes, & here's my explanation: when we look closely, mathematics, modality, semantics and mereology (maybe other bits of metaphysics - e.g., moral claims; which sets count as extensions of genuine properties) generate conservative extensions, at least on their own (no bridge principles). Their empirical content is, in a sense, nil. It seems an extremely pervasive phenomenon. In a sense, all of the epistemological/logical action is in the bridge principles.

On the other hand, a metaphysical theory may be part of, or implied by, a scientific theory T, which is testable in a more Quinian/holistic way, as you say ...

"for at least some metaphysical questions, empirical data from physics is of some relevance."

I think that's right.


Jeff Ketland

Hi Sebastian, by "there is a god", I meant, "for some x, x is a god", where "x is a god" is primitive.
When we have bridge principles and/or mixed predicates, the situation changes, and we can have logical links.
So, if "x is a god" is defined in the base language, over governed by axioms, the logical situation changes.
(A) if x is a god, then x is perfectly good
and a bridge principle, like
(B) if x is perfectly good, then x would stop bad stuff happening.
Then we infer: really bad stuff happened; ergo, there is no god.
I assume that theists reject (B).

Didn't one of these famous atheists - can't recall who, Flew maybe? - change their mind?

The one electron theory you mentioned earlier is associated with Wheeler and Feynman.


Sebastian Lutz

Hi Jeff, I completely agree with your example. I just think that according to many, metaphysical theories can have bridge principles. Swinburne's God-sentence is one such example, where 'good' and 'omnipotent', for example, are not newly introduced by the metaphysical claim. That theists reject (B) is indicative of them having changed the meaning of these two terms.

And it was indeed Flew who changed his mind (from atheist to deist, if I recall correctly), although only when his mind was already deteriorating (this is not meant in a glib way—I remember reading an article on the whole thing in which that was argued at some length).

By the way, if metaphysics is a conservative extension of empirical theories, it is simply non-falsifiable relative to these theories, right? (Allowing any set of empirical sentences, including quantified ones, to do the falsifying.) That would leave open its verifiability and of course its inductive confirmation and disconfirmation. (Niiniluoto, Tuomela, and recently Raatikainen have argued for the disconfirmability of a theory's Carnap sentence, for example.)

Thanks for the reference to the Wheeler-Feynman discussion. The one electron theory I was thinking of was something much less recherché, if I recall correctly. It was some run-of-the-mill metaphysical theory (about probabilities maybe?) for which van Fraassen showed that it entailed that there is at least one electron that never decays (or something like it). I really have to look this up though, lest I end up with my foot in my mouth (which might already have happened, I realize).

Jeff Ketland

Hi Sebastian,

By the way, if metaphysics is a conservative extension of empirical theories, it is simply non-falsifiable relative to these theories, right? ... That would leave open its verifiability and of course its inductive confirmation and disconfirmation

Right, and, as you say, one may go on to consider inductive/confirmation relations, as in the papers you mention. And with appropriate bridge principles, interesting consequences might follow from the metaphysical assumptions.

By the way, I read your thesis several months ago - well done!


Sebastian Lutz

Hi Jeff, thanks for reading my thesis! By the way, I found the article on Flew I was thinking of:

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Your comment could not be posted. Error type:
Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.


Post a comment

Comments are moderated, and will not appear until the author has approved them.