A hotly debated topic in the philosophy of science is whether we should consider our scientific and social scientific theories descriptions of reality, or if we should instead just consider them instruments for influencing the world. One of the main difficulties facing proponents of scientific anti-realism (hereby SARism) is distinguishing themselves from anti-realists more generally. P. Kyle Stanford, a SARist, devotes a substantial amount of the first chapter of his book, Exceeding Our Grasp: Science, History, and the Problem of Unconceived Alternatives (2006), to criticizing his colleagues for failing to thus distinguish themselves. He proposes that they fail to carve out a proper philosophical niche for SARism because their arguments are ultimately just variants of larger philosophical problems. Stanford therefore puts forward his own novel SARist argument, the problem of unconceived alternatives, which is, roughly, that the history of science provides us with inductive reason to believe that there are unconceived alternatives to our best scientific theories. I contest that Stanford does not succeed where others have failed. First, Stanford's argument seems to cut too deep, undermining the dichotomy that he is trying to create between that which we should consider real and that which we should not, and, second, Stanford fails to identify how SARism as an attitude is notably different from scientific realism.
Stanford's new induction and its motivation
To properly grasp the place of Stanford's problem of unconceived alternatives in the debate between scientific realists and anti-realist, it is important to understand Stanford's dissatisfaction with other SARist theories. The SARist agenda is not the same as the agenda of a skeptic, nor is it that of the anti-realist, because the SARist must reconcile their scientific anti-realism with a general belief in realism. The two central challenges posed in the name of SARism are: the pessimistic induction (our past theories have been false, therefore our present theories are likely false), and the challenge of underdetermination (there are alternative theoretical understandings that are empirically equivalent to our present theories). Historically, Stanford says, SARists have mostly focused on trying to universally demonstrate underdetermination (2006, 14). In doing so, he thinks they have “accepted… a devil's bargain” because their objection is then transformed into “traditional radical… scepticism” (10). Their argument just becomes a variant of Descartes's deceiving demon. Of equal concern for Stanford are those arguments that prove too little. The “local algorithmic strategy” seeks to demonstrate that we can modify the features of some theories to produce empirical equivalents (e.g., Newtonian physics assumes the universe is at rest, but we can produce an infinite number of equivalent theories by assuming that the universe is moving at different constant velocities (11)). This, Stanford says, is a variant of the “tacking problem”; we can tack on spurious claims to a theory and the combination remains equally supported (since the evidence has nothing, negative or positive, to say on the spurious claim). Stanford's criticism is therefore that SARists fail to make proper SARist arguments.
Stanford proposes a novel argument that utilizes aspects of both the pessimistic induction and underdetermination, which he calls the problem of unconceived alternatives (PUA). Stanford's innovation is that, instead of looking for empirically equivalent theories, he proposes that the history of science inductively suggests that there are unconceived alternatives to all of our modern theories that are equally supported by the evidence. Returning to physics, Stanford points out that the evidence that once led scientists to endorse Newtonian physics equally supports Einstein's general relativity. We can see many examples of the development of such alternatives in the history of science, Stanford claims, and it seems very likely that such alternatives exist for our current theories as well. These unconceived alternatives are “ordinary theoretical alternatives of the garden variety scientific sort” (14), rather than philosophical fantasies, because they are just those theories that scientists themselves have created or will create. By this argument Stanford thinks he has demonstrated that we have good reason to doubt that our current theories represent reality.
Though Stanford's induction is powerful, I do not think that he succeeds in making a proper SARist argument where his colleagues have failed. Stanford is not entirely explicit in defining what he means by realism, but it seems to me he is using the concept in two ways. He is primarily trying to argue that there is a categorical difference between our theoretical understanding of inaccessible domains and our knowledge of more mundane matters. He is also, however, making claims about proper scientific conduct. That is to say, he is claiming that realism is a kind of epistemic attitude of high certainty, and that adopting the SARist attitude is important for doing good science. Though it is easy to see how these two arguments come bundled together, I am going to pull them apart and address them separately.
What does Stanford think is real?
Stanford's primary goal is to use his induction to argue that claims about inaccessible domains are categorically different then other kinds of claims. I propose (drawing on Stathis Psillos, 2009) that Stanford has made a strong anti-realist argument, but has not succeeded in making a SARist argument. The problem has two parts: first, naive common sense claims are poor candidates for literal belief; and second, all theoretical claims seem vulnerable to Stanford's PUA.
Naive common sense claims do not seem like proper candidates for literal belief because they are either vague approximations or brute observations. To quote two of Psillos's examples: “Is the surface of the pool table flat? Well, it depends. Is the height of John 1.73? Close enough” (2009, 14). Our common sense claims are shorthand notions that allow us to function in our everyday lives. They are not vulnerable to Stanford's induction because they say very little: instead they persist as part of the observational bedrock upon which we continuously build and tear down theories. Despite their constitutive importance, they are not claims “towards which we can have a stance of strict and literal belief” (Psillos 2009, 14). Stanford gives an example of a claim that a dropped cannon ball will hit the ground in x seconds (Stanford 2009, 34), but this kind of claim seems hardly distinguishable from relating an observation. All that seems to have been said is “I believe what my eyes and ears tell me,” but without theoretical content it seems our observer is trapped in the most immediate sensory realism. I do not think that Stanford is interested in this kind of crude phenomenology, but the only way beyond it toward the actual substantive nature of things is theory. Admittedly, common sense does not stop at such observations; it also encompasses many rough-and-ready heuristics, such as our intuitive concepts of psychology. However, it seems quite unlikely that these are the kind of theories that Stanford wants to label as “real” either. The question, then, is whether there are sophisticated theories to which Stanford's induction does not apply.
Most, if not all, complex theorizing seem vulnerable to Stanford's induction. That is to say, it inductively seems to be the case that, historically, complex theories of all sorts seem vulnerable to the PUA. The more you say, the more likely it is that you will be wrong. The line that Stanford wants to draw in the sand, as a proponent of SARism, is a distinction between theories that pertain to inaccessible domains and those that do not. But what kind of theory does not pertain to inaccessible domains? One proposed candidate is our understanding of amoebas. Psillos wonders whether there is a principled difference between “being committed… to the reality of amoebas and being committed to the reality of atoms.” (2009, 16) To this Stanford responds:
virtually all that we think we know about atoms comes from the role they play in a highly elaborate… theory. But quite a lot of what we know of amoebae (how fast they move, what they eat…) does not come to us… this way, but by means we routinely gather knowledge about the world… (Stanford 2009, 35).
This response is problematic because our “routine” means of gathering knowledge just seems to be our common sense. I can describe how fast an amoeba moves, or what it eats, but only if I refrain from any substantive theorizing: my knowledge is therefore so impoverished that it hardly seems worthy of the name. Based on the preceding account, I therefore propose that, on this interpretation, Stanford's PUA succeeds as a strong anti-realist argument, but fails as a SARism argument.
Realism as an epistemic attitude
Stanford is not just trying to draw a distinction between those theories that likely represent reality and those that do not: he is also trying to define how we should conduct ourselves differently in relation to the two. Indeed, it would be hard to understand Stanford's motivation if he did not think that the distinction represented an important difference in scientific conduct. Stanford says that we can take some claims as true: “in the same straightforward sense that the scientific realist thinks the claims made by Newton’s radically false theory about the behavior of the rockets, moons, cannonballs, and tides of our everyday experience are more-or-less just plain true.” (2009, 34) Here he has, notably, shifted the terms of the debate from realism to truth. The question is no longer “Is it real?” but rather “How confident am I in my expectations?” Realism has therefore become a high state of confidence, in some sense. How does the realist attitude differ from that of the SARist? Stanford draws this distinction as follows: “The instrumentalism I advocate is strictly modelled on the realist’s own attitude towards a theory like Newtonian mechanics, which she thinks is fundamentally false but nonetheless serves under a wide range of conditions as a powerful and reliable instrument…” (33). To make an interpretive metaphor, the SARist seems to use theories as tools: they are always on the lookout for the next big upgrade. I think this metaphor also captures why Stanford thinks the SARist attitude is superior: one should not get too attached to one's tools.
Stanford's distinction does not seem entirely correct to me. For once we incorporate our modern scientific theories into instrumentalism, it seems we must immediately draw a distinction within instrumentalism along the same line that used to divide instrumentalism from realism, that is to say, the distinction between those theories that we currently endorse and those that are behind the times. The SARist must, to adopt a concept of anti-realist opponent Simon Blackburn, give an account of how being “animated” by a theory is different from believing in a theory (2002, 119). Animation is the term Blackburn proposes to represent the SAR alternative to realist belief: it entails “learning to speak the theory as a native language, and using it to structure one’s perceptions and expectations” (119). Stanford would likely argue that the SARist attitude is more restrained. Blackburn is sceptical of this claim. He points out that the SARist seems to substitute belief in a theory with belief in the empirical adequacy of a theory, and wonders why the first is obviously more prone to dogmatism then the second. He says, “If I conceive myself as knowing that a theory is totally empirically adequate, I might look on attempts to investigate further as a waste of time.” (121). Therefore, given that we must immediately cut a belief-shaped slice out of our instrumentalism, it does not seem that Stanford has yet proposed a substantive difference between the disposition of the scientific realist and that of the SARist.
Stanford aims to provide a proper SARist argument where his fellows have failed. I propose that he does not succeed. Accepting his induction as correct seems to attack all realism, and Stanford fails to demonstrate how the SARist attitude importantly differs from that of the realist.
 This is Blackburn's interpretation of a proposal by Van Fraassen.
Blackburn, Simon. 2002. “Realism: Deconstructing the Debate.” Ratio 15 (2): 111–33.
Stanford, P. Kyle. 2006. “Realism, Pessimism, and Underdetermination.” In Exceeding Our Grasp: Science, History, and the Problem of Unconceived Alternatives. New York: Oxford University Press.
Stanford, P. Kyle, Juha Saatsi, Stathis Psillos, and Rasmus Grønfeldt Winther. 2009. “Grasping at Realist Straws.” Metascience 18 (3).
–Ryan Workman recently completed his Masters in Philosophy and Public Policy at the London School of Economics and Political Science, and is now interning with the Police Foundation, a think tank dedicated to developing knowledge and understanding of policing and crime prevention.
daniel halverson says
Hi Ryan. Thanks for your thoughtful article. A couple of points in response:
According to SEP Descartes solved the demon problem by invoking God. God is good and wise and does not want us to be deceived, therefore we have good reason to trust our senses, and, further, we can believe that He has created us and the universe such that we can understand it. In short, the existence and benevolence of God authorizes belief in an intelligible universe, knowable through our senses. But if we do not allow ourselves this assumption, as most modern philosophers would not, then the relation of our senses to truth becomes problematic. Despite alot of complaining about “dualism” and how old-fashioned it is, I haven’t seen a substantive argument for doubting that it describes the situation as it really is. People just seem to *dislike* it. So if Dr. Stanford argues that it’s “just Descartes’ demon” and believes we can dismiss the insights of Kuhn etc. on those grounds, I can’t agree. As a recent article by the biologist David Hoffman points out, and as has long been recognized by pragmatist philosophers, the open-ended nature of Darwinian biology strongly undermines the claim that the we live in a fundamentally comprehensible universe – unless, that is, we hold with Alvin Plantinga that the evolution is not an open-ended process after all. In short, teleology and truth rise or fall together – hence anti-realism .
With respect to “the problem of unconceived alternatives,” these cannot be viewed simply as the theories that scientists decide on at a later date. Rather, it contains all potential theories consistent with the evidence at the time, including those that would later go on to be realized. From the perspective of the 18th century Relativity is indeed an unrealized alternative, but hardly the only one. The point here is much broader than “later theories will replace older ones.” It’s that the history of science, like history generally, has to be seen (on current metaphysical theories) as an open-ended, contingent process. Hence we can know what it did produce, but not what it could have produced, even though we can be reasonably certain that these potentials could in fact have been, and may yet be, realized. If we acknowledge the existence of this potential, we do indeed have good reason to doubt our present theories, as Larry Laudan pointed out. Hence we are led, again, to anti-realism.
I’m not sure that anti-realism is necessary for good science, but a stronger case can be made that it is necessary for good history. If we hold that current theories are accepted because they are true, then we are implicitly invoking an ethnocentric and teleological progress narrative. It’s objectionable for alot of reasons, and generally thought to be bad historical practice. In so much as history is supposed to furnish important evidence for the philosophy of science, I think its methodological standards have to be taken seriously.
I’d write more but I’m afraid my note is already quite long. Thanks again for your article.
Ryan Workman says
Hey Daniel, thanks for the response.
I think your first two paragraphs well capture the anti-realist motive. I am uncertain whether you are arguing against my central thesis that a meaningful distinction cannot be drawn between anti-realism and scientific anti-realism. Putting my article aside for a moment, I think my intuitions around truth and dualism is to argue against the line traditionally drawn between subject and object: an agent and their world are inseparably one, meaning an agent cannot be separated from their senses. I’m currently reading through John Dewey’s Experience and Nature, and find myself very sympathetic to his account. I am quite content to describe history and knowledge as a open-ended process: perhaps I therefore abandon realism in the traditional sense.
As to the last paragraph, my immediate response is that science’s understanding of itself seems important to the history of science, and so philosophy of science seems to fold into history. If you want to talk about the history of science, you can talk about the progression of how scientists have understood the profession, but I think you must also weigh in on the actual nature of the project (otherwise you cannot identify your subject matter).
The key claim in my paper on the matter is that the practice of science seems little affected by scientists believing in theories and being ‘animated’ by them. Indeed, I’m uncertain the two can be concretely defined as distinct. If that is the case, then it seems to me like it shouldn’t make a difference for the history of science either.
Thanks for your response, really happy to get a comment.