Thursday, October 6, 2016

A Gentle Introduction to Scientific Realism

Here are some notes for a discussion that I led yesterday at Ohio State's Philosophy Club. There is nothing really new here, but these notes might be helpful for students who want a short, basic introduction to some aspects of the scientific realism debate. More thorough treatments can be found via Chakravartty's "Scientific Realism" entry on the Stanford Encyclopedia of Philosophy.

Does Science Tell the Truth? Notes for Ohio State Philosophy Club (Oct. 5, 2016)

Modern science presents us with many claims: the universe is more than 10 billion years old. The human species arose via evolution around 6 million years ago. Material objects are composed of very small molecules and atoms, built up out of even more fundamental particles.

Are these claims true? If they are true, how do we know they are true? The scientific realist argues that science aims at the truth and that many of the claims found in modern science actually are known to be true. However, many reject scientific realism: it is said to be too optimistic concerning our abilities. On this view, we may never know the truth about many scientific claims, and so we should adjust our aim to something more tractable.

What are the alternatives to scientific realism? One option is simple skepticism. The skeptic argues that we can never know any claim whose subject-matter goes beyond our personal, present experiences. In particular, we can never know about the past or the future. Some attribute this skeptical position to David Hume (1711-1776). It strikes many people as too pessimistic. Surely, there is something wrong with a philosophical argument if it reaches this pessimistic conclusion. I am more certain that I know that I have hands, to use G. E. Moore’s example (1873-1958, "Proof of an External World" (1939)), than I am in any philosophical premises of a skeptical argument. If this is right, then we do know certain claims, and the truth of these claims involves the past and the future.

We can draw on another example that Moore deploys in his lectures, Some Main Problems in Philosophy (1910-1911): "the sun and moon and all the immense number of visible stars, are each of them great masses of matter, and most of them many times larger than the earth" (p. 3). Here is an example of a kind of scientific common sense that most of us accept, and this shows we not only reject skepticism, but come some ways closer to the scientific realist.

There is an important intermediate position, though, that is best defended in our own time by Bas van Fraassen (b. 1941). He calls his view "constructive empiricism": it is based on a distinction between observable and unobservable entities. An observable entity is one that can be detected by an ordinary human being, unaided by instruments. So, a tree is observable because when it is there, and a human is appropriately close to it, the human can rightly come to believe that the tree is there simply by looking. But bacteria are unobservable because even when the bacteria are present, a human needs an instrument like a microscope to reliably detect it.

Clearly, it is easier to know the truth about observable entities. It is not trivial, though. The far side of the moon is observable in van Fraassen’s sense because if a human stands there, they can directly see its features (with a flashlight). But it is practically very difficult to get to the right position. Van Fraassen is not focused on these practical difficulties. He argues that there is a deeper kind of obstacle to knowing the truth regarding unobservable entities. As a result, he concludes, science should aim only at the truth regarding observable entities. He invented a special term for a collection of claims that get things right about observable entities: this collection or theory is empirically adequate. So, the constructive empiricist aims at empirical adequacy, and not truth. And much of our best modern science is, for the constructive empiricist, empirically adequate, even though we have no basis to conclude that it is true.

What is the difference, really, between truth and empirical adequacy? Consider the case of bacteria. If you are a scientific realist, then you believe in the existence of bacteria and their role in causing illnesses, e.g., from eating certain foods. However, if you are a constructive empiricist, you may use the bacteria theory, but you do not think that all of its claims are true. You accept only what it says about observable entities. So, the theory supports our practice of pasteurizing milk. Milk is observable, and it is observable that some people get sick drinking milk that has come directly from a cow. Heating is also observable, and we find that when we heat the milk, fewer people get sick from drinking the milk. All of this the constructive empiricist can accept. They can even use the word “bacteria”, but they do not think the claims about bacteria living in the milk, or being eliminated by the heat, are known to be true.

The scientific realist claims that the entire theory is true. Why would they add the truth of these claims to the empirical adequacy of the theory? One influential motivation is tied to explanation. The existence of bacteria is a crucial part of a good explanation for why pasteurization limits these illnesses tied to drinking milk. As realists put it, this is in fact the best explanation: the illnesses drop off because the bacteria are eliminated. But this explanation requires that the claims about unobservable entities be true. Our commitment to the bacteria explanation requires scientific realism. The constructive empiricist cannot offer this explanation.

Why should that matter? It seems a kind of wishful thinking: we want to have explanations, and so we adopt theories that allow us to explain what we observe. Often those explanations will appeal to unobservable entities. So our desire for explanations leads us to adopt scientific realism. Is this tie to explanation anything more than wishful thinking?

The realist responds that this form of reasoning is widespread and accepted by everyone who believes in substantial knowledge, i.e. everyone who is not a Humean skeptic. Why, for example, should we believe that the observable regularities that we find extend into the past and the future? Consider the very regularity that the constructive empiricist adopted for the case of pasteurization: when you heat milk, it is less likely to cause a certain kind of illness. This is what we have found in the past, but why accept that this pattern will continue into the future? One explanation of the past instances of the pattern is that we have a genuine regularity that is based somehow on the features of milk, heating and humans (the observable entities). This is a better explanation than the proposal that what we have found so far is just a massive coincidence.

Typically we accept the best explanation available, and believe its claims primarily because their truth does explain what we have found. This is inference to the best explanation (IBE). We employ it everyday life when (to borrow van Fraassen’s explanation) we conclude that there is a mouse in our house based on various sounds and visible signs. And the constructive empiricist uses it in a restricted way when they conclude that the bacteria theory is empirically adequate. And finally the scientific realist uses an unrestricted form of IBE when they conclude that the bacteria theory is true.

This brings us to the central issue that divides the constructive empiricist from the scientific realist. Is there a coherent way to restrict IBE to observable entities in a way that does not entail Humean skepticism? That is the realist challenge to the empiricist. Is there a convincing way to justify extending IBE from observable entities to all entities? That is the empiricist challenge to the realist. Let’s conclude by considering these two challenges in more detail.

Here is why it is difficult to restrict IBE and yet avoid skepticism. The arguments that try to show that one should not use IBE for unobservable entities seem to also show that one should not use IBE for observable entities. But if we don’t use IBE, then we seem forced to skepticism. An example of this problem recalls Descartes’ (1596-1650) method of doubt in the Meditations. He resolved to reject any claim if its truth could be doubted, even if that claim involved a fantastic scenario. So, I can doubt the existence of the past if I suppose that a powerful demon created me five minutes ago with all my memories intact. If we use the method of doubt to call into question IBE for unobservable entities, then it clearly extends to IBE for observable entities. And so we are forced to skepticism.

Constructive empiricists can respond by offering a different reason to worry about IBE for unobservable entities. Consider, they say, the history of science. The following pattern has played out many times in the history of science. A scientist uses IBE to justify their claim to the existence of a new sort of unobservable entity. That claim is then widely accepted, and leads to many additional scientific successes. However, after a period of time, a new scientific innovation is made, and the scientific community comes to reject that unobservable entity as an illusion. The worry, then, is that IBE for unobservable entities has a bad track-record. We should not use this method of forming beliefs because that method has been unreliable when arriving at the truth.

There are many examples that fit this pattern. One famous one concerns the "aether" that was proposed in the nineteenth century as the medium for light and then electro-magnetic radiation more generally. Here is how James Clark Maxwell (1831-1879) put it in 1878: "Whatever difficulties we may have in forming a consistent idea of the constitution of the aether, there can be no doubt that the interplanetary and interstellar spaces are not empty, but are occupied by a material substance or body, which is certainly the largest, and probably the most uniform body of which we have any knowledge" (1878). The same point applies to theories of disease: before bacteria and germs were blamed for disease, many blamed "bad air". The "miasma" theory, as it was called, had many successes, but is now dismissed as a massive error. Who, then, can be confident in our own realist commitments, given this poor track-record?

The advantage of this argument is that it is not fully general, and does not obviously support skepticism. For the empiricist can point out that there are fewer cases of these sort of errors for IBE when it is used only to draw conclusions about observable entities. For example, we have theories about how to build bridges so that they do not collapse. Here the theory is tested by its successes. Sometimes bridges still do collapse, but the focus on the observable seems to have helped us get these claims right.

Does this meet the original realist challenge? If IBE about unobservables really is so much more unreliable than IBE about observables, then the realist challenge has been met. However, it is not clear if the historical examples really support this interpretation. Perhaps IBE about unobservables as it is done now really is very reliable. Various realists have tried to pinpoint what is different about 21st century uses of IBE. This is an ongoing debate that combines historical and conceptual claims about how science has been, and is being, done.

Let us turn then to the difficulty in convincing someone who accepts IBE for observable entities to extend IBE to unobservable entities. What can the realist say to convince someone to be a scientific realist? The key move is the link between an explanation and the truth. If we have a genuine explanation for why something occurs, then the explanation is made up of true claims. Consider something that we have found throughout modern science in the 20th and 21st centuries: science has made enormous contributions to technology and has also made many novel predictions about experiments that we subsequently found to be correct. To take two examples: atomic physics led to the development of nuclear weapons, and biologists have mapped the human genome. What is the best explanation for all these successes, both practical and experimental? The best explanation is clearly that all of these theories developed by the scientists are true. So, on this second or "meta" level, the success of science supports scientific realism. This is just IBE applied to science itself.

The constructive empiricist has a powerful response. IBE is admitted to be appropriate for observable entities. But this argument uses IBE for unobservable entities: the truth of these scientific theories requires the existence of unobservable entities. So, this argument in fact presupposes that it is appropriate to use IBE for unobservable entities. It presupposes what is in fact at issue between the constructive empiricist and the scientific realist.

So, does science tell the truth? The two main positions in the philosophy of science respond with a qualified "yes". The scientific realist argues that most or all of what science says, when it has generated successes, is true. The constructive empiricist argues that a restricted part of what science says is true, namely the claims it makes about observables, past and future. Neither position argues that science tells the whole truth. For both, additional philosophical reflection is needed to figure out what is true in science, and so it seems that philosophy is needed to get at one kind of truth: the truth about science itself.

Friday, March 11, 2016

Special Journal Issue: Indispensability and Explanation

There is a great new special issue of Synthese that brings together a number of new papers on explanatory indispensability arguments. The editors have also included comments on some of the articles. Here is the lineup:

Indispensability and explanation: an overview and introduction
Daniele Molinini, Fabrice Pataut, Andrea Sereni Pages 317-332
Parsimony and inference to the best mathematical explanation
Alan Baker Pages 333-350
Comments on “Parsimony and inference to the best mathematical explanation”
Fabrice Pataut Pages 351-363
The explanatory dispensability of idealizations
Sam Baron Pages 365-386
Which explanatory role for mathematics in scientific models? Reply to “The Explanatory Dispensability of Idealizations”
Silvia De Bianchi Pages 387-401
Evidence, explanation and enhanced indispensability
Daniele Molinini Pages 403-422
Equivalent explanations and mathematical realism. Reply to “Evidence, Explanation, and Enhanced Indispensability”
Andrea Sereni Pages 423-434
Should scientific realists be platonists?
Jacob Busch, Joe Morrison Pages 435-449
Indispensability and the problem of compatible explanations
Josh Hunt Pages 451-467
The varieties of indispensability arguments
Marco Panza, Andrea Sereni Pages 469-516
Naturalizing indispensability: a rejoinder to ‘The varieties of indispensability arguments’
Henri Galinon Pages 517-530
Grounding and the indispensability argument
David Liggins Pages 531-548
Nominalistic content, grounding, and covering generalizations: Reply to ‘Grounding and the indispensability argument’
Matteo Plebani Pages 549-558

Saturday, March 28, 2015

Scientia Salon Discussion of Abstract Explanation Preprint

There is lively discussion of my preprint "Abstract Explanations in Science" (BJPS) over at Scientia Salon. Readers of this blog might be interested in checking it out, and supporting the site more generally.

Friday, November 14, 2014

Paperback of Mathematics and Scientific Representation is out!

Thanks to some good work at Oxford, the paperback edition of my 2012 book is now available. (It is listed on Amazon at least, and should be on the OUP USA website soon.) As much as I wanted to, I resisted the urge to make corrections and improvements.

Monday, November 3, 2014

My PSA 2014 talk (title, abstract and change in time)

This week is the 2014 edition of the Philosophy of Science Association conference. A great program has been assembled here.

Due to an oversight on my part, a conflict developed, and I had to request that the program chair move the time for my talk. The talk will now be presented on Friday Nov. 7th in the 4-6pm session on Explanation. I am grateful to the program chair for accommodating this last minute request.

Title: Newton, Laplace and Salmon on Explaining the Tides
Abstract: Salmon cites Newton's explanation of the tides in support of a causal account of scientific explanation. In this paper I reconsider the details of how Newton and his successors actually succeeded in explaining several key features of the tides. It turns out that these explanations depend on elements that are not easily interpreted in causal terms. I use the explanations offered after Newton to indicate two different ways that non-causal factors can be significant for scientific explanation. In Newton's equilibrium explanation, only a few special features of the tides can be explained. A later explanation deploys a kind of harmonic analysis to provide an informative classification of the tides at different locations. I consider the options for making sense of these explanations.

Monday, October 27, 2014

Two new papers on abstract (mathematical) explanation

There has not been much activity here lately, but I wanted to link to two new papers of mine that tackle the vexing issue of mathematical explanation in math and in science. I try to isolate a kind of "abstract" explanation using two cases, and explore their significance.

The Unsolvability of the Quintic: A Case Study in Abstract Mathematical Explanation
Philosophers' Imprint, forthcoming.
Abstract: This paper identifies one way that a mathematical proof can be more explanatory than another proof. This is by invoking a more abstract kind of entity than the topic of the theorem. These abstract mathematical explanations are identified via an investigation of a canonical instance of modern mathematics: the Galois theory proof that there is no general solution in radicals for fifth-degree polynomial equations. I claim that abstract explanations are best seen as describing a special sort of dependence relation between distinct mathematical domains. This case study highlights the importance of the conceptual, as opposed to computational, turn of much of modern mathematics, as recently emphasized by Tappenden and Avigad. The approach adopted here is contrasted with alternative proposals by Steiner and Kitcher.

Abstract Explanations in Science
British Journal for the Philosophy of Science, forthcoming.
A previous version of this paper is online here.
Abstract: This paper focuses on a case that expert practitioners count as an explanation: a mathematical account of Plateau's laws for soap films. I argue that this example falls into a class of explanations that I call abstract explanations. Abstract explanations involve an appeal to a more abstract entity than the state of affairs being explained. I show that the abstract entity need not be causally relevant to the explanandum for its features to be explanatorily relevant. However, it remains unclear how to unify abstract and causal explanations as instances of a single sort of thing. I conclude by examining the implications of the claim that explanations require objective dependence relations. If this claim is accepted, then there are several kinds of objective dependence relations.

It remains to be seen if this "ontic" approach is the best way to go, but I believe it is a promising avenue to explore.

Saturday, September 7, 2013

New Symposium on Glock's What is Analytic Philosophy?

The most recent issue of the Journal for the History of Analytical Philosophy has just appeared with a long-awaited symposium on Glock's book on the nature of analytic philosophy. Discussants include me, Leila Haaparanta, Panu Raatikainen and Graham Stevens. Glock also offers an extended and helpful reply. This issue marks the end of the term of our first editor in chief, Mark Textor. I would like to thank him for all his work in getting this new open-access journal going. I would also like to welcome our new editor in chief, Sandra Lapointe!