| Author: | |
| Website: | |
| Page title: | |
| URL: | |
| Published: | |
| Last revised: | |
| Accessed: |
The eighth assignment requires us to look at the concept of inference to the best explanation (IBE), the basic idea of which is to look at a dataset or a body of evidence and, given a number of possible hypotheses, choose the one that best explains it.1 To this end, we are asked to refer to the work of James Ladyman2, Carl Hempel3 and John Worrall4, as well as that of Sebastian van Fraassen5. The assignment questions, together with my answers, are shown below.
State the basic scheme of the DN-model of explanation. Of the various problems that Ladyman discusses for this account, which do you think is the worst? Explain why.
The deductive-nomological (DN) model of explanation is one of several that form part of the general covering law model of explanation which says (according to Hempel) that an explanation of some phenomenon or occurrence can be shown to follow from the application of some law or set of laws, given the existence of some set of initial conditions. He goes so far as to say that all explanations, not just in science, refer to some covering law (or laws), even if it is not always immediately apparent that this is the case.
The terminology used in defining the covering law model includes the terms explanandum (which refers to the thing that is being explained) and explanans (which refers to whatever is used to do the explaining). The term deductive is fairly self-explanatory. The term nomological refers to the use of natural / physical laws to explain something. The explanations themselves state that whatever happened did so as the consequence of some natural law or laws.
According to the DN model of explanation, the explanandum has to be deduced from one or more laws of nature, plus background facts, and a set of initial conditions. Ladyman expresses this schematically as a series of laws and conditions that can be enumerated, which consequently entail a number of explananda that can also be enumerated. The model also imposes the following logical conditions:
A further condition says that the argument must be true (i.e. not use false premises/laws – which seems rather obvious to me; isn't this a bit like saying "no cheating"?). The problems outlined by Ladyman include the inclusion of irrelevant arguments (they offer an explanation, but it is essentially meaningless), pre-emption (something does indeed happen, but for reasons other than predicted), and overdetermination (where two or more sets of causal conditions are in place, any one of which might account for some outcome).
In general terms, I would say that the worst problem is that a number of theories that are very successful at prediction (e.g. Newton's laws of physics used to predict the return of Halley's comet) they don't actually offer an explanation. This potentially undermines the use of the DN model of explanation as a mainstay of realism, because if there is no meaningful explanation to be had, a lot of otherwise very useful theories might have to be rejected.
What role does IBE play in the 'local' defense of scientific realism? How might Anti-realists attack this argument?
IBE is used in the global defence of scientific realism, where the explanandum (that which is being explained) is the success of science as a whole, and the most frequently used and powerful explanans (that which does the explaining) is the "no miracles" argument first articulated by Putnam6; simply, that that realism is the only approach to science that does not require a scientific achievement, or the success of science in general, to be looked upon as miraculous.
The local defence, in contrast, is applied to a particular theory. The example Ladyman uses is that of the theory of molecular structure and how it "unifies a vast array of experimentally determined facts about chemical phenomena". The realists argue that we should accept the theory because it not only predicts the phenomena that we observe, but also explains them.
The realist accepts the existence of atoms consisting of nuclei and the electrons that orbit them, as well as the physical laws that govern their behaviour. The existence of such things explains why chemicals behave in the way that they do. A competing theory that simply says that the world behaves as if there are such things is able to predict the same observable phenomena, but not explain them.
It is this explanatory power that (according to the realists) overcomes the underdetermination problem created by having two competing theories that are both empirically adequate. The argument goes that the theory that can best explain a phenomenon should be chosen over any competing theory or theories. If the theory in question refers to unobservable entities, we should accept the existence of these entities, and (as Ladyman puts it) "therefore the practice of IBE in science commits us to realism".
In considering how anti-realists might attack this argument I would simply comment that it sounds somewhat circular, in the same way that an inductive defence of inductivism is circular. If we use the inference to the best explanation as our criteria for choosing a theory, we are presumably already realists. Furthermore, my understanding of the anti-realist position is that they don't actually require a theory to have explanatory power, merely that it should be consistent with the observational data.
Then again, according to Ladyman, IBE can't be defended using a non-circular argument, but then neither can deduction. He outlines the counter (realist) argument that "although they cannot force the non-realist to accept IBE, they can show that its use is consistent and then argue that it forms part of a comprehensive and adequate philosophy of science". He goes on to say, however, that van Fraassen4 will attempt to undermine this claim as well.
Explain, in your own words, van Fraassen's example of meeting his friend at the market. What does he take this example to show?
Van Fraassen describes an occurrence in which both he and his friend go (separately) to the market and subsequently meet each other there. Van Fraasen can explain why he (Van Fraasen) has gone to the market, and his friend can also explain why he (the friend) has gone to the market. There was no pre-existing intention on the part of either of the two friends to meet at the market; it is simply a coincidence. Indeed, one assumes that neither friend knew of the other's planned trip to the market until the occurrence of their (purely coincidental) meeting.
Van Fraassen uses the example to show that, whereas a realist would claim that because a particular occurrence of some phenomenon was simply accidental or coincidental it could have no explanation, there is in fact an explanation of sorts; just not one that could have predicted the phenomenon in question. Both he and his friend are acting in accordance with predetermined, though unrelated, plans when they make their respective visits to the market, therefore each can explain his presence there at the time of the meeting. Neither could have predicted, however, that such a meeting would occur.
I am not entirely sure what point van Fraassen is attempting to make in respect of the realist's use of IBE as a rule of inference, but I assume that he is saying that choosing the theory that best explains a particular phenomenon is not always going to be possible, since some phenomena simply cannot be explained in terms of cause and effect, or through the application of some law (or set of laws), or by a combination of the two.
However, as Ladyman points out, this does rather seem to miss the point of Putnam's "no miracles" argument, which centres on the repeated predictive success of scientific theories not being due to coincidence or luck or, indeed, miracles. It also conveniently ignores the fact that, had each friend been aware of the other's plans, then the meeting could have been predicted (or at least seen as a probable occurrence), so I don't really see it as a valid objection.
What is the 'selective scepticism' objection? Who's views does it target?
According to the constructive empiricists, if two or more competing theories are both supported by the observational data, then according to the non-realists (led by van Fraassen, of course), they are empirically equivalent. This makes them subject to the underdetermination problem which, in a nutshell, states that there are no empirical grounds for choosing between them.
If it is believed that this problem may be resolved in the future through the acquisition of new observational data, any preferred theory is said to be weakly underdetermined. If, on the other hand, it is believed that further observational data is highly unlikely to resolve the deadlock, the preferred theory is said to be strongly underdetermined. Realists argue for inference to the best explanation (IBE) as the solution to this problem.
The idea of "selective scepticism" as practised by the constructive empiricists (according to the realists) is that they exercise what Ladyman calls an epistemic discrimination against unobservables. Frankly, I don't entirely understand the argument, but it seems to hinge on the fact that, if the constructive empiricists believe two theories to be empirically equivalent, then the only way they can choose between them is by using some non-empirical criterion.
Van Fraassen (according to the realists) appears, then, not only to have no grounds for making such a choice, but also no grounds for considering a particular theory to be empirically adequate in the first place (that's the bit I don't quite get, by the way, since I thought the only requirement was that a theory should "save the phenomena", i.e. agree with all the observational data). If we are saying that only one of the competing theories can be empirically adequate, it makes more sense.
Since the only grounds for choosing between empirically equivalent theories must (from a constructive empiricist point of view) be non-empirical, then realists can argue that either van Fraassen is an outright sceptic and therefore unable to break the underdetermination of which theory is empirically adequate, or he must accept some non-empirical criterion or other in order to do so, in which case he is exercising a selective scepticism. Does he, for example, exercise scepticism about IBE in the case of unobservables, and yet endorse IBE for theories concerning observables?
On the other hand, it is probably worth bearing in mind that van Fraassen does not object to IBE per se, merely to the idea of it being used as a rule of inference. In other words, to believe a theory on the basis that it has been found to be explanatory is perfectly acceptable, but to insist that a theory must be believed because it is explanatory (or more explanatory than any alternative theory) is not acceptable.
Worrall characterizes a potential [anti-?]realist response to van Fraassen's criticisms (detailed in Ladyman pp. 225-7). Do you find this response adequate? Why or why not?
Worrall responds to the critique of van Fraassen by simply, in essence stating that IBE is the most reasonable position to adopt. At the same time, he accepts, as no doubt he must, that nothing can force someone to adopt a realist attitude towards theories, and even appears to accept that the endorsement of scientific realism is not strictly necessary. He seems to be comparing this standpoint to the defence of metaphysical realism (the view that there is an external world that exists independently of our minds, and that we can refer to and say something about the nature of the objects and events therein) insofar as it, too, is the most reasonable position to adopt.
We are not compelled to assume that the external world exists just because our perceptions tell us so, but it seems to be a common-sense assumption to make. Following on from that, it is surely not too great a leap to accept the existence of unobservable phenomena, if those phenomena successfully explain the things that we can observe – what Ladyman calls "the analogue of commonsense metaphysical realism in the unobservable domain".
Whether or not this is an adequate response to the criticism of van Fraassen and other anti-realists probably depends on who it is intended to convince. For those already leaning towards the realist point of view, it will probably push them further in the direction of scientific realism. For those sitting on the philosophical fence it may also have some merit.
One suspects, however, that hard-core anti-realists may be more resistant to this argument. It does not appear to offer anything more concrete than an appeal to "common sense" and "reasonableness", both of which virtues are somewhat open to interpretation. I would imagine that a hardened anti-realist would claim that common sense and reasonableness were irrelevant in the context of scientific inquiry, particularly since so much of modern scientific theory deals with (unobservable) phenomena that are vastly different in nature to anything we can directly experience.
Notes: