Are You a Scientific Realist or an Instrumentalist?

Are You a Scientific Realist or an Instrumentalist? September 5, 2018

Are scientific theories useful because they’re true? Or do we only consider them true if they’re useful?

A recent find at the Harvard Book Store’s warehouse sale was Physical Theory: Method and Interpretation, a compilation of essays by philosophers of science edited by Lawrence Sklar. One of the essays, by University of California at Irvine philosopher P. Kyle Stanford, concerns the different ways scientific theories, and theory itself, is interpreted. In “Reading Nature,” Stanford investigates the development of two opposing approaches to science: realism and instrumentalism.

The Real Thing

The realist position is probably the most familiar to science fans, the straight-ahead correspondence theory that states that scientific theories can be judged on how well they describe reality:

Having more or less settled that scientific inquiry really does proceed in large part by proposing theories or hypotheses about the entities, events, and dynamical principles at work in otherwise inaccessible domains of nature, the position known as scientific realism has often seemed the most natural or straightforward approach to interpreting these hypotheses. The scientific realist is a forthright and commonsensical soul who tells us that our best scientific theories simply offer (at least probably and/or approximately) true descriptions of how things stand in the world, including most contentiously those domains or aspects of the world about which it is extremely difficult to get information in any other way because the entities or events in them are extremely small, unavoidably remote, occurred in the distant past, or are otherwise inconveniently situated. But the realist also typically holds a distinctive position concerning how the truth of those theories is to be understood. Her “correspondence theory” of truth holds that the claims of scientific hypotheses are true when the propositions they express “correspond to” or mirror the way the world or the natural order really is.

Though Stanford ultimately seems persuaded of the validity of realism, he points out various problems with the approach. First and foremost, there’s no conceivable way to verify to what extent our theories represent reality, since our only contact with reality is through the methods of inquiry that depend on the theories we’re trying to verify. If we’re opposed to unverifiable claims, we should acknowledge that scientific realism seems predicated on suspiciously circular logic.

Also threatening realism is the concept of pessimistic induction, that points out that even our best theories throughout the history of scientific inquiry have proved inadequate, and have needed to be replaced by constructs unimaginable to proponents of the old theory. Do we have any reason to believe that our theories are immune to such revision, apart from our intellectual investment in them? Stanford mentions that problem of underdetermination too: there’s more than one theory that can conceivably explain any set of data, so we’re never in a position to single out one theory as the true one.

Instrumentalism and Pragmatic Acceptance of Scientific Theories

The instrumentalist critique can be either semantic or epistemic in nature:

The instrumentalist doubts whether simply reporting the facts about otherwise inaccessible domains or aspects of nature is what even our most successful scientific theories actually do. But such doubt can originate in more than one way, and the grounds for the instrumentalist’s caution are as diverse as the challenges facing the realist’s position itself…One traditional strand of such instrumentalist thinking has challenged whether scientific theories should be understood as making the claims that the scientific realist sees them as making in the first place. Such semantic instrumentalists suggest that either truth itself should not be conceived in the realist’s preferred manner (whether for the claims of theoretical science or quite generally) or that scientific claims require some special semantic understanding or analysis. The other influential strand of instrumentalist thinking has concentrated its dissent instead on the realist’s view of our epistemic entitlements:  it does not fault the realist interpretation of the claims of our scientific theories but argues instead that we are not justified in actually believing those claims to be true. Such epistemic instrumentalists have sought to articulate an alternative cognitive attitude toward successful scientific theories to which they think we can be epistemically entitled. What unifies these two strands of instrumentalist thinking is their common commitment to an alternative positive construal of scientific theories as mere conceptual tools or “instruments” for achieving our practical goals and objectives rather than descriptions of nature that are either probably or approximately true in the realist’s correspondence sense.

The instrumentalist, then, just questions whether the degree of confirmation a scientific theory enjoys is justification for considering it true. If a theory makes testable predictions and serves as the basis for useful research, there’s no need to assume that it corresponds to some objective reality.

What’s The Difference?

To me, at least, an instrumentalist position best demonstrates skepticism and parsimony. In addition, it seems to describe exactly the position we take with scientific theories in general. One can accept a theory as useful and still be agnostic about whether it “corresponds” to reality. Even schoolkids use Newtonian mechanics in their science labs, and its predictive power is demonstrable. We realize that Newton’s theory is radically false, and yet we can employ it in a wide array of experimental and real-world situations. If a theory doesn’t need to be true to be useful, it seems pointless to assume that utility correlates with truth.

Not to mention that the realist position seems to me a recipe for dogmatism, and the history of science bears out this suspicion. If a realist believes a theory to be true, there’s no incentive to investigate alternate constructs even when his or her theory faces competition from a scientifically valid rival. Instrumentalism would appear to be the more prudent approach in terms of scientific progress.

Which position best describes your approach to science? Do you think scientific inquiry is getting us closer and closer to the truth, or do you think science is just generating more useful models to guide research? Could it be the realist-instrumentalist contrast a distinction without a difference? 

"Books on feminism and science, both written by women : one saying there had been ..."

The Social Construction of Reality
"I wasreading a book by Richard Rorty on this and was nodding along to all ..."

The Social Construction of Reality
""Reality" implies "the one truth" when a "multiplicity of possibilities" is the actual concept that ..."

The Social Construction of Reality
"Somewhere along the line, someone decided that "social construction" explanations should include the word "reality".This ..."

The Social Construction of Reality

Browse Our Archives

Follow Us!


TRENDING AT PATHEOS Nonreligious
What Are Your Thoughts?leave a comment
  • Anthrotheist

    Well apparently I am more of an instrumentalist. I guess I see scientific realism as a natural tendency toward narrative and certainty. That is, humans like to tell stories, it is how we arrange our understanding of events (and thus our world). Realism tells the story of what is real. Also, we like to be certain about what we believe, and realism appears to essentially say, “This is the truth about the real world we live in.” That certainty is surely reassuring.

    Ultimately, for me, it is a matter of pragmatism. Quantum field theory may claim that everything we experience is fluctuations in energy fields interacting with one another, that mass and solidity are just how we perceive those interactions and therefore aren’t really “real.” That narrative doesn’t mean that I can walk through a wall, but then again it might inspire someone’s imagination and thinking to the point where they develop a technology that could allow such a thing.

    So we live by what we can use, but we dream of what we can imagine to be true. One is useful, the other inspiring.

  • So we live by what we can use, but we dream of what we can imagine to be true.

    Very well said!

    I’m pragmatic about the matter too. I just wonder how open-minded scientific realists can be if they declare that their pet theories aren’t just useful and robust, but true. You and I can look at it in narrative terms, but I’ll bet most science fans can’t.

  • Anne Fenwick

    I would technically be an instrumentalist, according to the rather tight definition you give, especially noting the proviso: ‘If a theory makes testable (and presumably confirmed) predictions…)

    Coming from the humanities, I would be quite concerned about what can happen if people lose sight of that very big ‘if’, and that’s perhaps why I thought twice about this question.

    I think I’m going to call myself a limited realist in so far as I think the usefulness of our theories is directly related to their accurate correspondence with reality. But I don’t think the correspondence is likely to be complete or perfect. Do we even know for sure what we would mean by ‘true’ in this context? (Genuine question, but to be taken at a philosophical level).

  • Coming from the humanities, I would be quite concerned about what can happen if people lose sight of that very big ‘if’, and that’s perhaps why I thought twice about this question.

    I fully agree. Stanford quotes philosopher Simon Blackburn in the essay as ridiculing the false modesty of the instrumentalist: “I don’t believe trilobites existed, but I structure all my thoughts about the fossil record by accepting that they existed.” However, I think it’s just as disingenuous of realists to make it sound as if they only provisionally think our currently accepted theories are true: “I think scientific theories are true because they directly correspond to reality, but I’m willing to completely change my mind.”

  • Sastra

    I’d like to split the difference and say I ascribe to both ( in limited fashion.) Instrumentalism, however, worries me for its potential to minimize the importance of not just seeking, but discovering truth ( in limited fashion.)

    I recently got into a discussion with a Catholic friend who tried to segue from the reasonableness of instrumentalism to the reasonableness of faith. If all views are acknowledged unprovable and we can’t be sure of any truth, then believing in a science theory is like believing in God. Objective reality is ungraspable and above us, we can only know what works for us. Thus, science and faith are not only compatible, but basically equivalent.

    An interesting, but long, stretch.

  • As I explained in the OP, I don’t think instrumentalism in any way compromises the scientific program. If anything, it leaves scientists more open to new ideas or methods, because they’re not so invested in a construct they’ve already decided corresponds directly to the objective truth.

    Objective reality is ungraspable and above us, we can only know what works for us.

    I have no problem with this. I’m not a conspiracist and I don’t dispute any mainstream scientific theory.

    However, in many ways scientific inquiry and religion both try to limit our imaginations by presenting something eternal and objective, a set of unquestionable answers. Whether it’s to God or objective reality, were supposed to submit to some all-encompassing authority rather than understand our shared responsibility in making sense of the chaos.

  • Anthrotheist

    I agree, and share the same concern regarding science. It’s good though, forums like this are important to me; it is part of reminding myself that you don’t need religion to have dogmatism.

  • Sastra

    I don’t see that particular correspondence between scientific inquiry and religion, because the answers in science are supposed to be checked by reality, not scientific authority. If a test fails, it fails. A beautiful theory may have to be revised or discarded, limiting our imagination by forcing us to accept our responsibility in ‘making sense of the chaos.’

    In religion, however, as I see it the imagination has no natural limits. If you believe something wrong about “God” there’s nothing which could happen to correct you. There are no tests. Unlike changes in scientific explanations, any revisions in personal theology come about through fancies, dreams, preferences, and choice.

  • You’ll notice that I wasn’t trying to compare science and religion. Yes, we all realize that religious belief is groundless and resistant to correction. What I was trying to make clear is that even the certainty that science promises is illusory; at the amateur level, it’s characterized by too much credulity to qualify as free inquiry.

  • winmeer

    The word “truth” is often misused as “absolute i.e. unchanging” which pertains more to religion. In science truth by correspondence theory is “provisional i.e. a work in progress” ( K Popper). As Lakatos points out the original question a theory/paradigm seeks to address changes with more research into the problem; one “gets progressive problem shifts” until the new theory that emerges is/can be regarded as a “scientific revolution” (to use Kuhn). The instrumentalist approach is simply one that uses scientific ideas/formulas(ae) to solve practical problems as for example in engineering.

  • Edward Silha

    Shem wrote: “We realize that Newton’s theory is radically false”
    I feel that this is an overstatement.

    The equations of motion developed from the theory of relativity account for the effects of mass and velocity on time and space. However, when the variables in those equations are replaced by actual values that do not approach the speed of light or relatively extreme masses, they devolve into Newton’s equations. From the outset, it was observed that Newton’s equations were not totally adequate to describe reality because they could not account for the obit of Mercury. The equations developed from the theory of relativity are sufficiently correct to describe the orbit of Mercury and the adjustment of time necessary for the Global Positioning System to produce accurate results. However, they are still not sufficient to describe what happens near a black hole.

    I am a retired engineer. I assumed that well tested scientific theories are good enough approximations to reality that I could use them in designs with confidence that the design would be adequate for the application. Whether well tested theories are approximate descriptions of reality (my preference) or simply tools to predict outcomes has no practical consequences except for members of philosophy departments. I believe that my view of this is common among scientists and engineers, and does no dissuade researchers from probing the frontiers in an attempt to refine the theories. I do not know of any well tested scientific theories that have been completely overturned in the last seventy or eighty years.

  • Teto85
  • aвѕolυт clancy

    Fascinating post, Shem, thanks!

    Could it be the realist-instrumentalist contrast a distinction without a difference?

    I’m inclined to say no, at least after a first read. The realist approach seems optimistic to me, and a bit of a misnomer because the instrumentalist seems more realistic. And I agree that instrumentalism is the more skeptical approach.

    But I think it also matters whether we’re talking about basic or applied research. One is directed toward learning for its own sake; whether the knowledge is immediately applicable isn’t really that important. The other is trying to answer a specific question. So basic research seems more instrumentalist by nature and expectation. I’d also note that with the frequency of serendipitous findings even in applied work, instrumentalism seems the more accurate descriptor here too.

    As far as science reflecting reality, one trend I’ve noticed in my own industry is a growing emphasis on orthogonal methods. Fifteen or so years ago it was common to identify a drug by one test or technique – a UV detector for example. Now FDA is frequently asking for additional and different methodologies: run a wet chemistry test too, to prove a certain moiety is there; use another instrumental procedure like IR or mass spec. I think this approach does lend increased likelihood that the experimental is giving a better picture of reality, but the difficulty is with the fields we’re postulating about often being way ahead of our detection capabilities.

    I think it might be typical of human inquiry to ask questions way before we have advanced means of finding out. We were wondering about communicable disease long before we had a microscope.

    Which position best describes your approach to science?

    These terms are all new to me, but I’d argue that pessimistic induction is the default position of a good scientist or any type of investigator. I think we’re all constructing approximations. I can’t tell a patient their drug is “pure” – I can only give a level above which I’ve detected no impurities using the methods known today. But I expect the detection levels to drop, and that today’s anomalous data will be interpreted someday in the future.

  • Great to see you again!

    I tend to think there’s a lot more common ground between the two approaches than it seems at first glance. But that hinges on whether the realist can be pragmatic enough to admit that the “truth” our research describes is constantly in flux, and whether the instrumentalist’s skepticism isn’t just denial.

    As far as science reflecting reality, one trend I’ve noticed in my own industry is a growing emphasis on orthogonal methods. Fifteen or so years ago it was common to identify a drug by one test or technique – a UV detector for example. Now FDA is frequently asking for additional and different methodologies: run a wet chemistry test too, to prove a certain moiety is there; use another instrumental procedure like IR or mass spec. I think this approach does lend increased likelihood that the experimental is giving a better picture of reality, but the difficulty is with the fields we’re postulating about often being way ahead of our detection capabilities.

    Thanks for providing the real-world context for these questions. It seems obvious that the results of testing give us confidence that our hypotheses are in line with some aspect of reality. As you say, it depends on how specific our question is. We have to be prudent about how many stops the “inference ticket” can take us.

  • Edward Silha

    Shem wrote: “However, in many ways scientific inquiry and religion both try to limit our imaginations by presenting something eternal and objective, a set of unquestionable answers.”

    I feel you are conflating popular science descriptions commonly found in news media (which sensationalize articles to attract readers) with how science and reliable science reporting works. Some media hype the results of a single study but often omit context that would limit the impact of the article. Media like Science and Science News seem to do a good job of identifying the limitations of new research.

  • I think you’re focusing on science’s research function, and I admit that the media has a habit of sensationalizing the results of research for a credulous public.

    But that’s not what I’m talking about here. This is about our philosophical approach to scientific inquiry itself, and our beliefs about what it’s possible for science to study and reveal. Treating science not like a collective human endeavor that provides useful information but rather as a portal to the eternal and unchanging Truth is just way too religious for my liking.

  • TinnyWhistler

    I’m not convinced that the realist must believe that their current theory is true, only that it’s better than anything else at describing reality. That doesn’t close the door to the theory changing in the future. In general, I tend to see “true” as a shortcut for “best explanation so far for what we observe” when discussing scientific theories. I don’t know if that prevents me from being a realist.

    I do agree that there’s a problem inherent in assuming that scientific consensus provides some sort of Truth, but I really don’t ever bump up into that among actual researchers as I do among fedora-having “philosophers” online who like to claim that they’ve constructed the Ultimate Worldview based on Science.

    “What unifies these two strands of instrumentalist thinking is their common commitment to an alternative positive construal of scientific theories as mere conceptual tools or “instruments” for achieving our practical goals and objectives rather than descriptions of nature that are either probably or approximately true in the realist’s correspondence sense.”
    I hope this isn’t an argument for ignoring scientific research that doesn’t seem to have any immediate practical use. Much of science has been formulated when the particular area of study was still considered “useless” and practical applications often didn’t come for decades. Consider the study of charge and electricity, for example. That certainly started as an effort to describe things that were being observed, rather than to achieve any practical goal.

    I take issue with this paragraph:
    “Not to mention that the realist position seems to me a recipe for dogmatism, and the history of science bears out this suspicion. If a realist believes a theory to be true, there’s no incentive to investigate alternate constructs even when his or her theory faces competition from a scientifically valid rival. Instrumentalism would appear to be the more prudent approach in terms of scientific progress.”
    This seems to be ignoring the possibility of instrumentalism’s lack of incentive to investigate further simply because someone’s found a theory that seems useful enough. If utility is the purpose of scientific inquiry, why would you ever kick the tires on an established theory?

    I’m probably misrepresenting both schools of thought. It’s been a long day.

  • TinnyWhistler

    Newton also couldn’t account for Uranus until Neptune was discovered. As far as I know, people were actively looking for another planet near Mercury when Einstein published.

    Here’s a fun one for you: An engineer and a mathematician were each faced with an attractive partner. They were told they couldn’t walk up to the person but could only ever decrease the distance between them by half. The mathematician threw up their arms because they realized they’d never actually reach the person, but the engineer didn’t despair because they could get “close enough”

  • Instrumentalism appears to generally be a more valid way to treat models when we are in a poorly evidenced state of speculative thinking. In those circumstances, paradigm shifts occur frequently, and paradigm shifts make little sense at all when treated with a realism model.

    However, realism is far more useful to apply in many other situations. One example — science often relies upon people dedicating their lives to it. From biographies, this dedication is to a search for truth — IE this committmetn relies upon assumig realism. Few people would dedicte theirlives to developing an abstract and meaningless puzzle.

    A second example — we LIVE based on the assumption of realism. The reality of the external world is assumed through indirect realism supported by massive empirical confirmation. As is the reality of other minds. Take the intersection of these two hypotheticals, in the form of a friend. A nominalist I read recently described the external world in nominalist terms as basically saying: “this fact makes it advisable to accept the thing language.” where “thing language” means external world model. So — if the “fact” is the recent death of ones friend — is addressing this death to just “accept the thing language”???? Anyone who addresses the world in those terms — would be considered insane. (the author was Carnap)

    A third example is from mathematics rather than empiricism, but the realist/conceptualist split is the same there. Conceptualism in math has been shown to not support degrees of infinity, yet degrees of infinity are needed to construct numerous proofs. Realism once more is pragamtically more useful/powerful.

    Fundamentally, we are constucting models, and are uncertain of their validity. But treating the models as uncertian, while also assuming there is a reality — works best over both realms, the immature and the mature.