Bas Van Fraassen is a nifty philosopher of science. He received his PhD in Pittsburgh in 1966, under the guidance of Adolf Grünbaum, he taught at Yale University, the university of Toronto, the University of Southern California, he has been McCosh Professor of Philosophy in Princeton, and eventually joined the department of philosophy at San Francisco State University, where he has the title of Distinguished Professor of Philosophy.
He first gained attention with his book An Introduction to the Philosophy of Time and Space where he tried to develop a formal theory of space and time based on the notion of causality. The book had an enormous legacy, with experts of the likes of John Earman and David Malament joining the debate. However, he achieved V.I.P. status with his classic The Scientific Image, where he defends a combination of empiricism and antirealism towards unobservable entities based on a re-definition of what the scientific enterprise is. His last achievement is the tome Scientific Representation: Paradoxes of Perspectives, where he combines his scientific empiricism with the view that theories are best thought as models or structures, rather than sets of sentences.
In this interview, we talk about his philosophical influences and the birth of The Scientific Image during a journey through North-Africa, Turkey and Eastern Europe, we talk about saving the phenomena and suspending judgement over the existence of unobservable entities, living in world full of mysteries and leaving unanswerable questions unanswered, rationality and irrationality, living in a simulation, the historical interplay between theorizing and experimenting, the meaning of particle detectors for an empiricist, the unity of science and physicalism, the condemnation of Galilei by the Church, and the distinction between Appearance and Reality…
Iphilo: How did you become a philosopher?
Bas van Fraassen: Through books. When I was 17, working part-time in the Edmonton public library in Canada, I came across Plato’s Phaedo. That was so different from everything else I had been reading that it came as a revelation.
Until that year it had not been in my mind to go to university (not at all an obvious choice for an immigrant boy), but my teachers began to suggest it. So then I told them I wanted to study philosophy. Nobody seemed to think this was a rational decision, and perhaps it wasn’t under the circumstances, but somehow it worked out. A main factor was the support of my first philosophy teacher, Karel Lambert, a logician. He gave me special directed study on logic and probability, using books newly published. Quite a different fare from the Phaedo! But in some ways not so far since Lambert liked arguing with me in the same way Socrates argued with his students.
In my second year I was working part-time in the University library, and there I came upon Hans Reichenbach’s Philosophy of Space and Time. That is what led me to philosophy of science, it was my second transformative experience with a book.
IP: In The Scientific Image, you claim that from hearing the pattern of little feet we can infer the existence of a mouse, but from the traces in a cloud chamber we cannot infer the existence of muons or other invisible particles. It seems that the two inferences are formally identical. Why would you accept one and not the other?
BvF: I do think the two cases can be given a parallel treatment, but we have to be careful about what the conclusion is that is inferred. As I suggested in that passage, the thesis that we infer to the truth of the best explanation has a rival: the thesis that we infer to its empirical adequacy. As it happens, the hypothesis that there is a mouse in the wainscoting is empirically adequate if and only if it is true. So there the two rival theses have the same implications. That is not so in the case of the hypothesis that what we see in the cloud chamber is the track of an unobservable particle – that could be empirically adequate without being true.
I should add that I definitely did not suggest that rival thesis because I wanted to make it part of my epistemology! I offered it to embarrass the scientific realist with the question: which of these theses is the better explanation of our ordinary ‘inductive’ behavior? In fact, in the ordinary case of the mouse droppings etc. I already know that these indicate the presence of a mouse, with high probability, so no inductive reasoning is needed anyway.
IP: How did you come to your view on science?
BvF: Reichenbach’s Philosophy of Space and Time showed me a new world, the world created by the 20th century philosophers of science who took an active part in debates about the new revolutionary changes happening in physics. But Reichenbach had recently died. I found out though that another philosopher had donned his mantle, Adolf Grünbaum, and so I went to study with him in Pittsburgh. Reichenbach was a prominent logical empiricist, and his theory of space and time was of an empiricist sort, ‘relationalism’ rather than ‘absolutism’; Grünbaum continued in this vein. My first large project in philosophy of science was to continue the development of this sort of view of space, time, and space-time. I was wonderfully lucky again, for Grünbaum didn’t laugh at my naïve ambitions, he was tremendously supportive.
I had another teacher in Pittsburgh, Wilfrid Sellars, prominent among the philosophers who had developed the new scientific realism in opposition to logical positivism and logical empiricism. His seminars were the most popular, his lectures fabulous, his grasp of history amazing, his writings profound. I learned a great deal from him, but unlike some of my class mates did not become a disciple. On the contrary, I felt a strong tension between his scientific realism and the empiricism that had so attracted me in Reichenbach. But I was not in any position to express a coherent contrary view at that time.
The breakthrough for me came about ten years later. I had in the meanwhile worked out a relationalist, empiricist view of time, space, and space-time. On this view, there are on the one hand physical relations in the world that ground temporal and spatial judgements, and on the other hand mathematical structures that play the role of time, space, or space-time in physics. The crucial requirement is that the former, which may be scant in mathematical eyes, are embeddable in the latter. There need not be anything in reality that is precisely ‘mirrored’ by the mathematical structures, as long as there is a satisfactory fit to what there is.
Eventually I realized that this could be generalized to the relation between phenomena and their scientific models in general. I was very excited when that became clear, and I sent a long letter to Professor Grünbaum to outline this, from a campsite in Algeria. This was in my second full year’s leave from teaching. I was traveling with a car and tent on a large loop, starting from England and passing through North-Africa, Turkey, and Eastern Europe. I wrote the first versions of the main chapters of The Scientific Image in campsites along the way.
IP: Why is it important to save the phenomena, that is to devise theories that correctly predict the observable outcome of experiments? Is it only for technological purposes?
BvF: The empirical criteria of success in the sciences are exactly what need to be met for application, prediction, manipulation in their subject matter. Even in ancient times technology in that broad sense was of great value: think of the empirical success, despite theoretical error, of ancient astronomy.
But there is also an intellectual value in the acquisition of empirical knowledge: not just the value of finding our way around in the world but of knowing our way around in the world. According to Wilfrid Sellars, to be in that position requires that we can answer all relevant why-questions, ex cathedra so to speak. I disagree – empirical knowledge, and the sense of knowing our way around, can live comfortably in world full of mysteries, once we understand what modelling is.
IP: Fig-leaf realism is the view that unobservables exist but that this is the only thing we can know about them. Is your constructive empiricism a variety of fig-leaf realism or is it even a weaker theory, in the sense that it allows not to have any ontological commitment to unobservables?
BvF: The latter, certainly. Since the reality of the unobservable is irrelevant to the criteria of success operative in scientific practice, you might as well be agnostic about them.
What seems crucial to me in philosophy, and perhaps in much else in life, is a willingness to leave unanswerable questions unanswered.
IP: In your retrospective essay, “Constructive Empiricism Now”, you said that scientific realism is a rational attitude. But you defended the rationality of suspending belief in unobservables. What is it in your view to be rational? Can it be captured in standard ways such as the Bayesian framework?
BvF: Logic and probability theory, the conceptual tools developed in Bayesian (and more generally, probabilist) epistemology pertain to the bounds of rationality independent of context. While it seems to me that the models developed in formal epistemology are still very simplistic, they can be developed further, both to make better contact with what “evidence” means in the sciences and with such topics as voluntarism in traditional epistemology. But so far they pertain to context-independent criteria only, so do not relate at all closely to the mucky worlds of daily life and scientific practice.
If you and I were to look at a real case where we wanted to figure out what was or was not rational, it wouldn’t be a context-independent criterion we were looking for. We would work in a context of shared prior values, intentions, commitments, standards, and aims, and these would set the bar much higher than logical consistency. Is it rational for a young person today to stake his career on the possibility that string theory rather than quantum gravity will be dominant in tomorrow’s physics? Or to marry, or not to marry, or to become a priest?
IP: Are there cases in your view of rationality under which suspending judgments would be irrational?
BvF: Even if only context-independent criteria are to apply, the answer is Yes, since other opinions that are already in place may make suspending judgment impossible. If you do want to suspend judgment, and find it impossible because of your strongly entrenched prior opinions, you can try to dismantle those bit by bit, but this could require a lifetime.
Imagine for example that someone suggests that you are totally mistaken about recent history, in fact it was Nazi Germany that won World War II, and now rules Europe. You want to deny that, but she asks you to be open-minded and suspend judgment while she brings you the evidence. I think that if the evidence began to look convincing, it would make the idea that I was myself a patient in a lunatic asylum more probable than that she was right. But the intellectual effort of suspending judgment about whether I am sane or insane would involve the same task of dismantling an overwhelmingly large part of what I believe. Finally, to suspend judgment on something, while keeping other opinions that rule on it, would be logically incoherent.
IP: In the same essay, “Constructive Empiricism Now”, you say that each one of us must choose where to draw the line between what to believe and what merely to accept for practical purposes. You seem to treat scientific realism and constructive empiricism as giving us two acceptable ways to draw the line. In the mainstream media there has been a lot of discussion of a paper of Nick Bostrom claiming that we very probably live in a simulation. Is it rational to draw the line à la Bostrom and suspend judgement about ordinary physical objects?
BvF: I meant rather that there are different ways of drawing a line between what we’ll count as the deliverances of observation while maintaining an empiricist view of science. If someone started out as a constructive empiricist and then decided to let in much more to count as observable, she would not thereby become a scientific realist. That would only happen if she changed her mind about what is the aim, the basic criterion of success, of science. As to Bostrom’s argument and its kin, it seems to me naïve as long as it is not responsive to Hilary Putnam’s critique of what earlier used to be called the ‘brain in a vat’ hypothesis.
Constructive empiricism presumes a ‘common sense’ realism, with reference to rocks, trees, magnets, interferometers, and so forth as entirely unproblematic. The issue goes back to the old ideas about sense data or phenomenalism, which in my view are already in the dustbin of History. Notice that even when supposing that we are part of a simulation, Bostrom grants himself the liberty of referring to humans and computers, although if he were part of a simulation, his ‘words’ would not have any connection with anything that could establish reference to either!
IP: If the Higgs boson does not exist, do you still see a reason to construct a Large Hadron Collider just to see if we can predict the statistical patterns that arise when we turn on a gigantic magnet?
BvF: Referring to an experiment as “turning on a gigantic magnet” is a bit like referring to the Mona Lisa as a piece of painted wood. Literally, both may be accurate, but both ignore meaning.
However, your question is a good and provocative way of asking just what could be the meaning of such an experimental set-up on an empiricist view of science. The answer, briefly put, is that an experiment has meaning within the theoretical context in which it was designed. Experimentation is the continuation of theory by other means. As physics develops, there are open questions. Some of these are settled by theoretical calculation; in a sense then they were not really open. Others however are genuinely open, and experiments designed, with the help of the theory itself, has the deciding voice – for those who are developing the theory, about how they should go on. The outcome guides how the theory should be continued (or in some famous cases, drastically changed or replaced).
This is an empiricist answer, for it makes sense of the historical interplay of physical theory and experimentation, without implying that it is a story of a progressive revelation of an unobservable reality behind the investigated phenomena.
We can relate this to the example of the Higgs boson. Talk of particles in current physics doesn’t refer to little bits of matter rushing around, of course. The Higgs boson is an excitation of the Higgs field. The introduction of the Higgs field was initially controversial, but it became an established part of the so-called Standard Model before the experiments could be carried out. Eventually this could be done, and was done with a positive result, with the Large Hadron Collider. In this set-up, there is a model of the detector in which a certain ‘global signal strength’ parameter can have value 0 or 1, with the theory implying that 1 corresponds to an excitation of the Higgs field.
This example illustrates very well the procedure of empirical grounding: the data generated by the detector determine, via calculations by the theory itself, the value of the theoretical quantity.
Initially the question was: should the Higgs field be accepted in the development of the Standard Model? Theoretical reasons for doing so accumulated, as its introduction solved problems for the theoretician, but these are not enough. As long as the question is still open, there is a gap in the theory that needs to be filled in one way or another. The experiment had to dictate how it was to be filled in, whether as the advocates of the Higgs field thought it should be, or differently.
The role played by the theory itself in the design of the experiment and in the significance attached to its possible outcomes is crucial here: the experiment has meaning in this theoretical context, and thereby constrains the further development of the theory.
IP: Our best physical theories, the standard model of particle physics and general relativity, are able to save all the phenomena we know of. However, their mutual inconsistency motivates (realist) physicists to look for a theory of quantum gravity. Do you think that the search for a unified theory is justified or should we content ourselves with several yet mutually inconsistent theories?
BvF: Nancy Cartwright has argued convincingly that the sciences need unification neither in practice nor in principle, and I am convinced by her arguments that we must learn to live in a dappled world.
But this general insight cannot be applied too narrowly in specific areas. Contradictory implications of various ways of modeling a single substance such as water, or sand, or a single process like the frog population’s growth in our local lake, cannot be just taken in our stride without obstructing the very applications for which they are designed.
In the specific field of theoretical physics dealing with the fundamental theories, I would say that this caveat applies. The reasons are very well explained by Carlo Rovelli in his book, Reality is Not What It Seems.
These two different points taken together would be in trouble if we added the conviction that all of science, or even all of natural science, is in principle reducible to fundamental physics. But I see no warrant for any such conviction.
IP: Many have noticed a similarity between constructive empiricism and the position of the Church towards Galileo and his use of the telescope. Cardinal Bellarmino and other men of the Church insisted that the Copernican model be viewed only as a mathematical fiction to predict the phenomena and they refused to believe that the images given by the telescope corresponded to something in reality. What is the main difference between your view on unobservables and that of the Church on celestial bodies?
BvF: Well, there is quite a difference! The controversy over Copernican versus Ptolemaic astronomy concerned the motions of the planets, and no one raised any doubt about the reality of the planets, which are observable.
The long-standing problem in the science of the day was that Ptolemaic astronomy ‘saved the appearances’ but was incompatible with Aristotelian physics. This problem was not significantly changed by the advent of the Copernican model which equally saved the appearances and was also incompatible with the accepted physics. When Tycho Brahe came up with yet a third system,it was also one that saved the appearances and was incompatible with the physics. Bellarmini expressed one venerable reaction to this problem: that astronomy had no claim to anything more than saving the appearances, and there was no evidence to decide between the rival models.
What Bellarmini did not appreciate was that for Galileo, the terms of this discussion were too narrow. For Galileo, it was part of a much larger issue. For he was at the beginning of a revolution in physics, and was constructing a new kinematics, with what we now call Galilean relativity, and a new dynamics. Motion of the earth made no sense within the old physics but it would, he was convinced, make sense within the new.
In any case, the controversy over Copernicus had nothing to do with the unobservable. The other story that you mention, of Galileo’s telescope, comes closer to dealing with observability. The evidence that Galileo submitted by this means did favor the Copernican system (and Bellarmini ignored that). Galileo confronted the learned of his day with new appearances: the appearances in the images produced by a telescope of shadows on the moon and moons other than our own. Once again there was a question of how to save the appearances. Could, for example, epicycles be added to any system of astronomy as orbits of moons of other planets? But there was a prior question: were those appearances, produced by the telescope, artifacts of measurement, or genuine data to be accommodated? Questions of that sort arise for us in the sciences today, and it is only with hindsight that we can dismiss their significance in that historical episode.
I hope I have not just side-stepped your question. The learned doctors of the Church did not have the view that the criterion of success for physics was empirical adequacy. Quite the contrary, they would have heartily disagreed with me! It is because they regarded the task of physics to give us a true account of what is going on behind the phenomena that they had such a great problem with astronomy.
The position that astronomy should only be held to the criterion of saving the appearances was taken because they did not see how they could do better, and did not see how the physics could or should be changed. In this respect, they were I think typical of older generations of physicists resistant to new theoretical developments. What was startling, and disconcerting also to many Catholics, was that in 1616 the Church hierarchy suddenly took a stance quite at odds with its own scholarly traditions, and prohibited teaching that the earth moves. It has been suggested that this was a reaction to the vocal Protestants, who insisted that such teaching contradicts the Bible (see Thomas Kuhn’s account in Ch. 6 of his book The Copernican Revolution).
IP: For the readers at iPhilo that want to get further into the philosophical topics we discussed so far, are there three books you could recommend that would help them?
BvF: Certainly! I would recommend, not to express agreement, but because each of these can be inspiring in its own way:
- Pierre Duhem, The Aim and Structure of Physical Theory (1906)
- Hermann Weyl, Philosophy of Mathematics and Natural Science (1926/1949)
- Hilary Putnam, The Many Faces of Realism (1987)
IP: Last but not least: What is your favourite unobservable?
BvF: Light. At every revolutionary stage in the history of physics light has been in the avant-garde. We see things when they reflect light; light does not reflect light, so we don’t see it. Nor can we hear or touch or smell it. When we see a searchlight playing in the heavens above Hollywood, what we see is the dust particles in the air, which do reflect light.
But who would deny that light is real?
The question, if made precise, is just how our talk about light, in our descriptions of optical phenomena, can be understood. On the ancient atomists’ and Newton’s account a light ray is a stream of particles. On Aristotle’s and the 19th century account light is a changing condition of the aether. In the 20th century both these sorts of models for the optical phenomena ran into such serious difficulties that once again, light had to play a unique role in the development of physics. To say something more definite I would once again emphasize modeling: optical phenomena, puzzling and fascinating throughout the history of science, are represented in successive theories in ever further imagination-transgressing ways.
In conclusion, I want to thank you very much for this as well as for the other provocative questions! I have enjoyed thinking about them and about how to answer them, and that was a valuable experience for me.
Interview by Joshua Babic, Lorenzo Cocco and Michal Hladky (July 2017).