A decisive experiment is a single experiment intended to determine, between two theories, which one is true and which one is false.
According to Kuhn, there is no such thing as a decisive experiment. This is because he thinks of competing theories as being incommensurable. Incommensurability amounts to the notion that competing theories do not even speak the same language, use none of the same theoretical terms, and so no decisive experiment will be possible. This is because the meanings of each of the theoretical terms in each of the theories acquires its meaning uniquely from its place or role within its respective theory. A theoretical term, then, is only meaningful in the context of a theory. So any proposition of T1 must be saying something different than any proposition of T2 — they share none of the same concepts, and so there is nothing between the two theories that can be directly compared. And a direct comparison is necessary for a decisive experiment (since it is a single, deciding experiment).
But from a scientific realist perspective, this doesn’t seem quite right. Kuhn’s rejection of a decisive experiment is predicated on the notion that T1 and T2 have nothing in common — they talk about none of the same things (except for, maybe, the direct observable phenomena [sense-data]). But this is a mistaken assumption. T1 and T2 can talk about the same things. There is no reason to suppose that when T1 makes a claim about electrons and when T2 makes a claim about electrons, that T1 and T2 are not making (different) claims about the very same thing.
This rests on a realist attitude (at least toward theoretical entities). Here’s why I’m a realist about theoretical entities. And given this, I also think that Bohr, Stoney, and Thomson were all talking about the very same natural kind despite their different theories (and theoretical frameworks). Each of their stereotypes about the electron may have differed (indeed, some of them may have even been incompatible with each other), but this does not entail that the extension of each term was different. As Putnam argues, meaning just ain’t in the head. The fact that Bohr, Stoney, and Thomson had different stereotypes associated with the electron does not mean that they were referring to different things — meaning is not determined by one’s mental/psychological state. Natural kinds, like ‘water’ or ‘electron,’ are (to an extent) indexicals. That is, they directly refer to their corresponding object without the aid of any definite description. Here’s how ‘water’ gets its meaning: I point to water and say, ‘Now this is water.’ By using the demonstrative ‘this’ I am directly referring to the thing that I am pointing to — I am not generating any kind of definite description. Here is how ‘water’ does not get its meaning: ‘I define water to be whatever is that odorless, clear, tasteless fluid of such-and-such density.’ The same is true for the electron. The electron is whatever we mean to refer to that causes those streaks in the cloud chamber and say, ‘that thing (is an electron).’ We do not have in mind any necessary and sufficient conditions for the use of the term, and do not mean to supply and specify them.
If we accept that Bohr, Stoney, and Thomson were all talking about the very same natural kind, then we might think that decisive experiments are possible (insofar as we can decide, between two competing theories, which one must be false). Suppose T1 makes a claim p about a natural kind, E. And T2 makes a claim q also about natural kind E. Further suppose that p and q are incompatible claims. The important thing here is that, whatever the differences between T1 and T2 may be, T1 and T2 are still making claims about the very same, real entities.
We should be able to take E, put it in the appropriate circumstances, and observe whether p or q holds. If neither hold, then neither theory can be true (as per modus tollens). If p holds, but not q, then we reject q as false (and vice versa). If both hold, then there must be some inconsistency in our background theories (to which both T1 and T2 are a part). But suppose that p, in fact, holds and q does not. Because they were both making observable predictions about the very same real, physical thing, the experiment (which must contain the real, physical thing in question) falsifies q and does not falsify p. In this way, we ‘decide’ to accept p over q. But this does not entail that p is true — only that it has not been proven false.
But suppose of you’re of the antirealist, rather than realist, persuasion. Are decisive experiments still possible? Generally, with an antirealist view, you really will see two competing theories as genuinely incommensurable, as really speaking two different languages, and all the theoretical entities thereof cannot be identical (for their meaning is derived from their role within the theory). T1 and T2 cannot assert p and q (respectively) about (something like) E. For the antirealist, there is no E, each T1 and T2 has its own E1 and E2 (respectively) — and E1 and E2 are not identical. So it follows that there is no common element to link the two for direct comparison. So, prima facie, it seems like decisive experiments will not be possible.
But I’d like to note that while there may be no such thing as a decisive experiment between T1 and T2, this does not entail that T1 and T2 are incomparable. There may still be a method for deciding between the two. If T1 and T2 are both equally empirically adequate, then they adequately predict and organize all the phenomena we have been acquainted with. Kuhn, if anything, was a keen observer of the history of science. And one important observation was this. Within any scientific paradigm, there will eventually (and inevitably) be some phenomena that the concepts, methods, and knowledge of the paradigm cannot account for. If the paradigm cannot provide an adequate solution, then a revolution or gestalt-switch will occur and a new paradigm, with a new conceptual apparatus, will emerge. Eventually T1 or T2 (or both) will encounter an anomaly. If neither can account for their respective anomalies, then T1 and T2 will be abandoned and T3 (or T3 and T4) will emerge. But suppose T1 can account for its anomaly while T2 cannot (and, moreover, this is an anomaly we really care about). Then we may well feel just fine about abandoning T2 and pursuing whichever program is suggested by T1.
Crucially, this cannot amount to anything like direct comparison. And if you think that science progresses toward some kind of truth, then it is possible that we may be led astray. (Though I do not think we’ll wander too far.1) But this does seem to be a method for deciding between two equally empirically adequate incommensurable theories, if, albeit, an imperfect one. There is no one decisive experiment, but a series of experiments within each respectively may eventually lead us to prefer T1 over T2.
- Perhaps I will write a post discussing this. ↩