Almost  60 years ago, Ernest Nagel concluded his meticulous discussion of the cognitive status of theories (The Structure of Science, 1961, ch. 6) by suggesting that when apparently opposing views (realism and instrumentalism) “are each stated with some circumspection, each can assimilate into its formulations not only the facts concerning the primary subject matter explored by experimental inquiry but also all the relevant facts concerning the logic and procedure of science…. the opposition between these views is a conflict over preferred modes of speech” (p. 152). Not, then, a debate worth prolonging.

But this is philosophy, our debates go on for millennia! And these days there are pressures to publish or disappear. So here comes Rowbottom to revitalise what he sees as a currently unfashionable ism. He has provided a new logo: cognitive instrumentalism, and a new vision statement. He lacks Nagel’s meticulousness, so the main discussion of what realism amounts to occurs as an appendix at the end (understandably since it is a depressing series of quibbles directed at friend and foe alike) and one often wonders who would want to disagree with what he claims as distinctive of his position.

Everyone agrees, it seems, that the terms of debate are indeterminate. Before reading Rowbottom, I might have offered a rough distinction by contrasting Osiander’s take on Copernicus (that his theory gives us a better way of calculating but is not to be regarded as a literal claim about the actual movements of the solar system) with Galileo’s eppur si muove: there is a fact of the matter and we are on the track of it, however flagrantly it contradicts common sense and what we all observe. Prima facie, there is more going on here than different preferences for modes of speech.

Rowbottom introduces instrumentalism as “the view that science is primarily, and should primarily be, an instrument for furthering our practical ends” (p. 1). So is a telephone directory, so he goes on immediately to make instrumentalism “more precise” by associating it with “a complex cluster of related views in an empiricist philosophical tradition” (p. 1), nodding to Mach, Duhem, Bridgman, Bohr, and assorted 19^th century British scientists. But these views (which Nagel took to be instrumentalist too) suffered from various philosophical defects and misunderstandings. Rowbottom offers us a refurbished cognitive instrumentalism.

Cognitive instrumentalism has three core components (all on p. 1):

1.       an account of the progress/value of science: it increases our understanding of and ability to do things with the observable world around us;

2.       a criterion for when to take a bit of science literally: “when it is grounded in talk about observable things”;

3.       an account of what we can expect science to achieve: it doesn’t need to fixate on unobservable things.

He chooses the qualification cognitive for two reasons: science is a tool for understanding, not merely saving/predicting, the phenomena (so did Osiander simply get science in general wrong?); “talk of unobservable things within science is primarily a cognitive tool for comprehending how observable things behave” (p. 1).

The book is structured to offer detailed accounts of the three components in the first three chapters; these are followed by a historical review of atomic theory from 1885 to 1930 as an illustration of the fertility of non-literal thinking in physics; a chapter on understanding; a chapter that has two tasks (how we should draw the observable/unobservable distinction; how cognitive instrumentalism measures up to some recent alternatives); and finally a chapter detailing some psychological experimentation which suggests that people are overconfident about explanations they can offer that involve unobservables: this is said to explain why realisms have been found so attractive among philosophers in recent years. As noted already, the book concludes with an appendix, a reprint of an article on the nature(s) of scientific realism.

That is the plan. Its implementation is sometimes surprising, but let me stick mainly with the key issues.

To be distinctive we need to know how instrumentalism differs from our default realistic interpretation of what people, even theoretical scientists, say. To take an item from Rowbottom’s bibliography, Earnshaw’s 1842 paper is entitled ‘On the Nature of the Molecular Forces Which Regulate the Constitution of the Luminiferous Ether.’ So one takes Earnshaw to have assumed that there are molecules, and an ether for light to wiggle in. We still think there are molecules, even Rowbottom. We have discarded the ether. For all I know, Earnshaw might have said somewhere that he wasn’t committed to these beliefs but just wanted to see how they played out, but unless he did say something to block a literal interpretation, the natural default reading is the literal one. Many instrumentalists have suggested that they refuse the natural way of taking these claims because they think that all we can properly do here is make predictions about what we can check. The bounds of the world we can responsibly characterise are the bounds of our sensory experience.

 

It is not only politicians who steal the opposition’s policies. Rowbottom’s appeal to understanding versus prediction serves to undercut a large swathe of objections to the traditional empiricist instrumentalisms I have just mentioned. What remains distinctive? One issue is his cutting a link between understanding and truth (e.g. p. 15 — the chapter devoted to understanding did not grab me since it was concerned more about the nature of “eureka” moments than with offering an account of what such a response is typically a response to) but again this link is something I think all of us should abandon. Ptolemy’s system explained why we see some of what we see and so provided more understanding than the lack of any mechanism or an appeal to the whims of celestial divinities could give. As with knowledge, so with understanding —  we have to go with what we have got. “Real” understanding may demand the true story, but we have to make do with what our current theorising offers, and in most of our cognitive activities it is often quite sufficient to make do with what we recognise to be only approximations to the final, whole truth. Bricolage, as I have agued elsewhere, is to be found, not just in the French countryside, but in science and philosophy too.

 

The other way that Rowbottom differs from many realists is his continual insistence that what we understand are observables; we may use unobservables to provide that understanding and so we may allow that they exist and have properties etc. but what gives them their entry visa is what they do for observables. So we certainly need what Rowbottom promises as his second component: a defensible distinction between what is observable and what is not.

 

We also need a reason for thinking that any such distinction matters, that our dealings with the unobservable are somehow improper, not to be taken literally. But here one wonders what Rowbottom is up to. One of his recurrent examples is electron spin (pp. 48-54, 98-102). It is not till the fourth chapter (p. 100) that we learn why spin here cannot be taken literally: to produce what we observe electrons would have to be spinning faster than the speed of light, which our best theorising tells us is impossible. So what blocks the default interpretation is another bit of theory. But the prior issue here is that the proponents of this bit of electron theory don’t say we should take a literal view. They use ‘spin’ for a property that bears some analogy with ordinary everyday literal spin. Rowbottom asks (p. 54) whether we should take electron-talk literally; answer, no. “But talk of some tiny thing possessing discrete mass and charge may be taken seriously” and their ‘spin’ is a fiction for helping us understand how these quasi-electrons behave. But he only gets there by insisting on taking literally what the theorists have told him is not to be taken literally. (The supposed fictional status of the ‘spin’ derives from Rowbottom’s claim that Bohm’s variant of quantum mechanics can get by without it. Other views should say, I think, that electrons have a property we gesture at that looks a bit like intrinsic angular momentum.)

 

Let me turn to the fundamental distinction: observable/unobservable. My take on the discussion over recent decades is that this distinction has been shown to be much too indefinite and context-bound for any interesting philosophical thesis to be erected upon it. (With perhaps an exception for van Fraassen’s insistence that scientists are not usually aiming at what philosophers aim at, so that they rightly have an agnostic attitude to unobservables until their key properties can be measured somehow.)

 

In one sense Rowbottom agrees with this. The distinction is vague, new instruments change where it lies, …. He makes the unexceptionable points that we are stuck with the senses natural selection has given us (p. 131ff) and that we must rely on them to begin our acquaintance with things and their properties (p. 31ff). But he wants these, I hope, uncontroversial points to yield the thesis that it is “impossible to comprehend or discuss … any truth behind the appearances” (p. 31). So he wants what Charles Pigden has aptly called a “coercive theory of meaning” (v. the version of his 2010 Logique et Analyse paper at https://philpapers.org/rec/PIGCTO) which might show inter alia that he himself cannot understand the perfectly intelligible views he attributes to the realist opposition. We underestimate the constructive powers of language at our peril. His blinkered view of what we can do with language is what puts him in the empiricist camp, though he recognises clearly the value of non-literal uses of language (especially it seems for Anglophone scientists — the French are happy with pure abstractions). What he seems to discountenance is the move from ‘we can see atoms as non-literally little solar systems’ to ‘we can say literally that atoms are like solar systems in these respects and not in those.’

When he turns to the variability of the distinction due to technical changes in instruments, Rowbottom plumps for a very conventional Austinian preference for observations “laden on highly stable folk theories” (p. 141) as against the deliverances of sophisticated instruments that we can only unravel with the help of scientific theories and long chains of inference. But while experience does deliver itself without explicit inference, one thing that has been obvious since Plato remarked on sticks looking bent in water (and ought to have been a starting assumption since Locke remarked on the interpretations built into mature experience) is that there is a lot of theorising built into whatever mechanisms produce our experiences and the judgments we make on their basis. The logic of folk theory is no different from the logic of other theory. Why should we privilege the hypotheses natural selection and the tribe have given us?

I have presented realism as a consequence of our default interpretation of language, a default that we rightly block when inconsistency looms (or in story-telling, etc.). On such a view, while realism is the default attitude, it is not for that reason the attitude we should always adopt; we have to take cases by cases. If anyone had read Mendel’s original paper they might have used his story of factors in an instrumentalist fashion, to arrive at certain populations of pea plant, because there was no other reading available at that time. The partial success of the story might have prompted a deeper investigation to tie down these factors, but at that time taking a realist stance could only have been an act of faith.

Many realists suppose that the theories we have now are not only to be understood generally as true/false claims but actually are true claims. Rowbottom correctly disputes this last move. In chapter 3 he exploits Stanford’s appeal to unconceived alternatives (Exceeding Our Grasp, 2006) and extends the range beyond theories to instruments, observations, mathematical techniques, perhaps even to missing shades of blue. In chapter 1 he convincingly shows by reference to the behaviour of pendulums that stating the truth about the actual world and getting an understanding of how it works often push in opposite directions, though a later example (tides, p. 72) shows that one should be careful here. The success at one level of Newton’s theory led people to think that everything about tides would be explained by it, but as they went down to more detailed levels new factors were found to play a role. In both cases one has a simple overarching theory and a lot of niggling details. What matters for understanding depends on one’s interests: the broad picture or the local anomaly.

Rowbottom asks for charity in the interpretation of his text (pp. 3-4) so let me note that his use of unconceived alternatives is not so much to cast doubt on current theorising as to remind us that all our judgments, including those of the probabilities of evidence and hypotheses, are in flux and liable to change when a new idea is born. So overall scientific agnosticism (p. 77) is to be preferred to overweening confidence. He is right (but who would disagree?) that there are advances in science that do not involve probing ever deeper into the unobservable. But, remove the atavistic preference for medium-sized dry goods, and it certainly looks like the difference between his cognitive instrumentalism and a circumspect realism is little more than a superficial preference rather than a profound disagreement.

 

© 2019 Ed Brandon

 

Ed Brandon is, by training, a philosopher, and now is retired from working in a policy position in the University of the West Indies at its Cave Hill Campus in Barbados.