The Nature of Science:
The "Problem of Unconceived Alternatives" and its Significance

The 25 August 2006 issue of Science carried a review of a new book, Exceeding Our Grasp: Science, History, and the Problem of Unconceived Alternatives, by Kyle Stanford, a philosopher of science at the University of California, Irvine. The reviewer, Tim Lewens, who is himself a philosopher of science at the University of Cambridge, sets an important context "The question is ... whether we should think our best theories - in chemistry, physics, biology, and elsewhere - are even close to the truth. So-called scientific realists say yes. Stanford says no." summarizes Stanford's contention "Time and time again, [scientific] theories that enjoyed impressive predictive and practical successes, and that were regarded as beyond doubt, have later been rejected as fundamentally mistaken. The argument concludes that the theories we now hold to be true will eventually go the same way." and expresses concern about an extension of that claim "... if a scientific community's repertoire of concepts expands over time as it interacts with the world, then that community cannot generate theoretical knowledge effectively. I do not know whether Stanford would embrace such a sweeping conclusion." Paul Grobstein, a biologist and neurobiologist who also co-teaches a course in philosophy of science, responded to the challenge in an email to both Lewens and Stanford "Stanford's argument from history for "the problem of unconceived alternatives" is paralleled by an argument from contemporary understandings of brain function (Getting It Less Wrong: The Brain's Way). And both indeed point to a conception of science and scientific progress quite different from those preferred by "scientific realists" (Revisiting Science in Culture). The lack of certainty about "theoretical knowledge" is neither a peculiarity of history nor due to any idiosyncratic failing of the brain. It is instead an inherent characteristic of inquiry itself, one about which to be pleased rather than concerned (Writing Descartes)." and noted that the philosophical issues under discussion have quite significant practical implications in a variety of contexts including education (cf Science as story telling in action and A conversation about science education (and science)).

Continuing conversation based on this starting point is documented here. To join the conversation, email your thoughts.


Kyle Stanford to Paul Grobstein - 7 September 2006

I thought I would pass along a part of my reply to a message from another colleague who asked what I think about the final issue Tim raises in the review. I think it is connected to the point you are making in your letter, but please let me know if I have misunderstood you. Here is the relevant chunk of that message:

Re: Tim's generous review, I think it's natural enough to ask whether my argument doesn't prove too much and (in a related vein) what sort of cognitive abilities or history of inquiry would leave us in a position to endorse realism--and, in fact, I think I can answer each of these questions effectively. But I don't feel the pull of Tim's worry that by my lights any scientific community that reliably expands its conceptual horizons will, ipso facto, automatically be precluded from realism about its theories. We must first answer the further question of whether the historical evidence shows the conceptual limitations of our earlier scientific communities to have indeed resulted in systematic failures to conceive of well-confirmed alternative hypotheses that would later turn out to be important--after all, it could have turned out that our conceptual blinders, although real, were simply not of a sort that systematically prevented us from formulating or considering theoretical possibilities that would later be accepted or treated as important. If they did, however, then (barring some specific reason to think we are situated differently now) it seems perfectly reasonable to me to infer that we are likely in the same spot. Perhaps we are supposed to find it implausible that such conceptual limits could fail to prevent us from conceiving of well-confirmed alternative possibilities, but if we assume that it seems to me quite right to conclude that this should lead us to reject realism. That is, it does not seem to me to get there by setting the evidential bar so impossibly or unfairly high that recognizing any sort of conceptual progress is incompatible with the embrace of realism--perhaps any cognitive agents whose history of inquiry is systematically populated with suitably fundamental and consequential cognitive blind spots really are precluded from realism, but (1) that doesn't sound quite so sweeping or implausible, and (2) that's just how it is--it is not the job of the world to make sure that realism remains an option for us. In addition, Paul, perhaps I should mention that I am deeply interested in the question of what (revised) positive picture of scientific inquiry we should embrace if we accept that even the best contemporary theories are probably false. Indeed, in the last chapter of the book I try to characterize the version of an "instrumentalist" alternative to a realist understanding and practice of science that I think recognizing the problem of unconceived alternatives should lead us to adopt. One last point on which I would insist, however: lack of certainty isn't the issue--even realists should be fallibilists, prepared to admit that our theories might be wrong. What my argument must show to be important or interesting is that it is at least as likely or not (or some such) that our best scientific theories are fundamentally mistaken in their central claims.


Paul Grobstein to Kyle Stanford - 17 September 2006

Let me amplify a bit. "Getting It Less Wrong: The Brain's Way" argues that the existence, at any given time, of multiple different viable interpretations of a given set of observations is not only an historically demonstrable feature of scientific practice but "is an inevitable and inescapable characteristic of all human inquiry into material things - because it is a fundamental aspect of the organization of the brain, which is itself the "inquirer." That organization consists of an information gathering process and a "story telling" process that has the capability to conceive multiple "theories" to account for any given set of observations. The relation between observations and theories (or stories) has always been and, for a variety of reasons, is always likely to be one to many instead of one to one. To put it differently,

"External "reality" (ambiguous or otherwise) is a good hypothesis, supported by an overwhelming number of observations over a very long period of time, but it is not the starting point for how the brain (and hence the inquirer, at the deepest level) works. What is "real" to the brain is the signals it receives (and it itself generates), signals which are always ambiguous in the sense of having multiple interpretations. In the face of this, what the brain has evolved to do is not to lessen its imperfections in painting pictures of "reality", but rather to make of the ambiguous information it has to create candidate unambiguous paintings, not one but many, which it can then test by additional observations. The brain is not designed to have a single picture of "reality" as an outcome, but rather to explore an infinite variety of candidate pictures. Ambiguity and uncertainty are not ... the ripples of the imperfect glass through which the brain tries to perceive reality. They are instead the fundamental "reality", both the grist and the tool by which the brain (and, hence, all humans, you and I among them) creates all of its paintings" Notice that the issue here is not "conceptual limitations". Both your argument from history and mine from the brain are subject to the challenge by philosophers (and scientists) of a "realist" persuasion that the only problem is some sort of "blind spot" that gets in the way (temporarily or permanently) of actually seeing "reality" and hence that the argument is, at least in principle, answerable from a realist position, ie there really IS a "reality out there" (a "fact of the matter") whether we can see it or not. I gather you may be comfortable accepting this, ie acknowledging that the problem may in fact be a "blind spot" but insisting that one ought in practice to abandon realism because of it? I'm certainly with you this far. If we can't, for whatever reason, use proximity to "reality" as a measure (and motivation) for progress, then we need some other way of thinking about what science is/does.

I think, though, that this doesn't quite get to Tim's point, and its here where the brain argument more than the historical argument may encourage one (productively, I think) to "embrace such a sweeping conclusion." I would, I think, be inclined to argue that if the problem WERE a "blind spot" we would eventually discover and transcend it (as both science and the brain have repeatedly done in the past). In fact the brain argument provides a more direct reply to the "realist" position. There is not only no "reality" by which one can decide how close the "theory/story" actually is to it nor which would allow us to characterize all conceivable "blind spots," there is no "reality" at all except as an ingredient of some forms of theory/story. And so I would be inclined to accept without concern Tim's conclusion that science "cannot generate a theoretical knowledge" effectively, if by "theoretical knowledge" one means "foundational understandings that are not themselves subject to challenge and change at some future time."

Along either track, of course, the question is how to achieve a "(revised) positive picture of scientific inquiry", what can one can use, other than a correspondence theory of truth, to motivate, evaluate, and explicate scientific "progress"? Along these lines, I look forward very much to hearing (and reading) more about your "instrumentalist" alternative. And to exploring its similarities and differences to what I have started calling "empirical non-foundationalism". There are sketches along these lines in several of the links I sent you.

From Getting It Less Wrong: The Brain's Way

""Progress" in science is not measured by increasing closeness to "truth" or to the "real"'. It can't be, because neither "truth" nor the "real" is a known location against which proximity can be measured. Progress in science has instead always been (and can't but be) measured in terms of distance from ignorance. Science proceeds not by proving "truth" or "reality" but rather by disproving falsity, not by painting the "right" picture but by painting a picture "less wrong" than prior pictures."

From Writing Descartes

"Maybe at this point in human history we've finished cataloguing all the possible things that one MIGHT have used as a solid starting point for continuing inquiry and we can conclude (for the moment at least?) that NONE of them are in fact a solid starting point, in the sense that none can be taken as a given not subject to further skepticism and exploration. Maybe its time to seriously entertain the possibility that looking for a single solid starting point just isn't the right way to go, that one has to find another, different way to proceed."

"The big thing, of course, is that by fully and completely following through on a posture of profound skepticism one very much expands the space for exploration and inquiry. While it may be a little uncomfortable to give up the security not only of authority and logic and sense data and thinking but also the "self", one achieves along this path the freedom to become, and, in becoming, to be oneself the agent of new territory to explore and inquire into."

From Revisiting Science in Culture

"the story of science as story telling and story revising may provide a foundation for a less wrong view of science, one that usefully engages a wider array of experiences and perspectives. That though is for others to decide. As with all scientific stories, the ultimate test of the value of this one is not in the past but in the future, not in whether it is right given the observations, but in what new things happen, what new observations are made, and what new stories develop because of it."
And in an on-line forum for the philosophy of science course I co-teach "Here I want to argue that in some cryptic way they (Popper and contemporary realists) continue to accept the idea that inquiry is about "reality", and how to deal with/uncover/expose it. And that it is time to recognize that inquiry isn't about "reality" at all; it is instead about finding/creating new things, things that have never existed before. It is exploration of what might be given what has been rather than discovery of what is" I trust you can hear in these excerpts the "instrumentalist" (or "pragmatic") conception of science, the idea that it doesn't NEED the concept of "reality" on which I think we're in full agreement, on both historical and philosophical grounds. What the brain perspective may (or may not?) usefully add is the notion that science is as much about creating new "realities" (within the constraints of empiricism and skepticism) as it is about uncovering existing ones. And hence that the concept of a fixed (and in principle discoverable) "reality" is not only unnecessary but actually gets in the way of a process of inquiry that is intimately and bi-directionally interacting with its own subject.

And that's why I am perfectly happy to embrace Tim's "sweeping conclusion", perhaps more so than you?. If one starts with an empirical perspective, "theoretical knowledge" has no better claim to permanence than any other kind of knowledge (there is no metaphysics independent of epistemology). And so a scientific community reflecting a thorough going empirical/skeptical perspective can be no more (and no less) "effective" at generating theoretical knowledge than knowledge of any other kind. Is this a direction your historical argument would take you, or am I here off on my own?


Kyle Stanford to Paul Grobstein - 22 September 2006

It sounds to me like you are worried about realism both as a metaphysical and an epistemological issue--that is, you are worried both about the nature/character/existence of an external world and about whether science puts us in a position to know anything about such a world (and/or what else it puts us in a position to know). The argument of my book is much more about the second of these questions than the first, but let me say at least a little something about each.

On the metaphysical side, I think you might be giving up a little too easily on the idea of an "external world". Philosophers have always been worried about how we acquire knowledge, but at least since Kant one line of thinking about this issue has been that when we talk about the external world we really mean something like the-external-world-as-we-experience-it (or our experiences of it as we organize and systematize them into a coherent unified framework, or something in this ballpark)--after all, it is sometimes asked, what sense of "the external world" beyond this could we coherently grasp, refer to, or be worried about? More recently, Pragmatist philosophers argued that the truth about the external world was simply constituted by whatever we will (or would) come to believe about our surroundings in the fullness of time or in the limit of inquiry. (Note that this is far removed from idealism, which holds instead that mental items are all that exist in the first place.) Your brain argument may make this kind of metaphysical picture attractive to you, in which case I'm tempted to say that you are sympathetic to a Pragmatic conception of truth, or a Kantian conception of the external world, even if you think that this process of revision and replacement of our fundamental pictures of the world really will go on forever (perhaps this is what you mean in suggesting that there can be no metaphysics independent of epistemology).

Now it so happens that I am sympathetic to both a Pragmatic conception of truth and a (broadly!) Kantian conception of the external world, but this turns out not to matter to the argument I make in the book. One of the points I make there is that retreating to a weaker conception of external reality or the truth about it that science seeks won't help avoid the problem of unconceived alternatives: as long as we have reason to believe that we are neglecting well-confirmed and serious alternatives to even our best scientific theories, we have reason to think that we will continue to find future scientific communities embracing beliefs that are among those alternatives, even if we (and they) reconceive the object of their inquiry to be reality-as-we-experience-it, or the beliefs that would meet all further experience in a fully satisfying way, or some other alternative to the more robust and reassuring kind of external world envisioned by the metaphysical realist.

Also, you are right that I am arguing from "blind spots" in a certain sense, but an important part of the argument is that we have no reason to think that we will or can ever "transcend" (as you describe it above) the having of such blind spots (as opposed to any particular blind spot). Perhaps more carefully, even if our inference procedures do lead us to the true theory of the external world (in whatever sense), the process and history of our inquiry itself ensures that we will not be in a position to justifiably believe that it is indeed the true theory (as it leaves us with every reason to believe in the existence of serious and well-confirmed unconceived alternatives to that theory). Thus, the realism I am suggesting we abandon is an epistemological position, regarding what we can take ourselves to know about the world around us, not a metaphysical position, regarding what the world around us is or isn't like. And this epistemological conclusion follows, I think, no matter what picture you have of the external world or what you would like to put in place of the metaphysical realist's conception of the matter--this issue is simply bracketed for the purposes of the argument I pursue.


Paul Grobstein to Kyle Stanford - 7 October 2006

I understand and share your interest in challenging "epistemological" realism, and agree that an argument for doing so can be made on historical grounds while leaving the issue of "metaphysical" realism open. But I am indeed, as you suggest, concerned about realism "both as a metaphysical and an epistemological issue" or, more accurately, as an issue that arises from and needs to be dealt with in the absence of any clear distinction between epistemology and metaphysics. That distinction, useful as it has been for several hundred years of scientific inquiry, is becoming increasingly difficult to sustain as science probes a variety of subjects where the act of inquiry itself produces significant changes in what is being inquired into, such as the brain itself. In such circumstances, it becomes steadily more obvious that there is in general no "fact of the matter" that exists independently of the particular form of inquiry being used, and hence of the perspective and approach of the inquirer. That, in turn, highlights what has (and continues to be) sometimes convenient to ignore but has always been and will always be so: the products of scientific inquiry are integrally related to the methods of inquiry.

This, I would argue, is not a "blind spot" but rather is of the very essence of scientific inquiry (and of inquiry in general). One makes observations, creates "stories" about them (which are always one of many ways of accounting for those observations, and necessarily involve a "subjective" element), and then makes new observations in an effort to falsify the story, again and again and again. What emerges from this is not only all past and current understandings (both practical and theoretical) about things being inquired into but also understandings of the nature of inquiry itself, including the concept of their being a world independent of our stories about it, ie metaphysics. One may study the comparative strengths and limitations of various known ways of inquiring and call that "epistemology". Alternatively, one may look for commonalities among the products of various ways of inquiring and term that "metaphysics", but the two cannot and should not be regarded as parallel and independent studies. There can be no meaningful metaphysics without multiple forms of inquiring. The upshot is that metaphysics grows out of epistemology. Moreover, it is always subject to change as forms of inquiry evolve, and so can't be used as a standard against which to measure forms of inquiry. In particular, there is no way to use metaphysics itself (or a description of "reality" associated with it) to say that an inquirer has a "blind spot". A "blind spot" exists if and only if there exists some alternate form of inquiry that reveals it.

Given that one can, as you do, make a strong argument against epistemological realism while leaving the issue of metaphysical realism "bracketed", why force the issue as I'm inclined to try and do? Yes, I am indeed, like you, inclined to "a Pragmatic conception of truth" and so the question boils down to "Are there practical differences between the narrower rejection of epistemological realism and the broader rejection of realism in its entirety?" I think there are, and that among them are differing implications of the two positions for trying to characterize a "(revised) positive picture of scientific inquiry."

On the route to that, let me gloss a little bit what you characterize as the Kantian understanding of the "external world" as "the-external-world-as-we-conceive-it." I hadn't run across that particular move and am indeed comfortable with it, as you suggested I might be. It does, though, differ from the more standard realist position in ways that are important in the present context.

The standard realist position, as I understand it, is something along the lines of "There is out there a reality with a uniquely appropriate description that I may not yet have achieved but have a reasonable aspiration of achieving using familiar procedures in which I have confidence." Denying epistemological realism would then require a modification along the lines of "There is a reality out there with a uniquely appropriate description but I am unable to provide it now or ever because of inherent limitations in procedures for uncovering it." And a reasonable "more positive picture of scientific inquiry" might have it that what matters isn't in fact achieving the final description but instead continuing improvements in the picture one has at any given time.

The "external-world-as-we-conceive-it" position is a little different. Because one is always working with the external world as conceived, the possibility is opened that there is no "uniquely appropriate description", that there have been and always will multiple "external worlds", ie different worlds conceived by the brains of different individuals or different groups of individuals. Notice that this too isn't "idealism" in the sense that "mental items are all that exist". Other things exist (probably) but because "worlds" are "mental" there need be no "uniquely appropriate description." In this case too one might promote a "more positive picture of scientific inquiry" that focuses on continuing improvements but one has the problem of disagreements about which "external world" to take as the basis for moving forward. Moreover, one is, I think, missing an opportunity to develop a still more "positive picture of scientific inquiry".

The potential for pluralism inherent in the "external-world-as-we-conceive-it" position puts it very close to a rejection of not only epistemological but also metaphysical realism, so why not just go all the way? Everything we know about the brain says the existence of multiple different worlds is not only a potential but an actuality. Might one perhaps develop a "more positive picture of scientific inquiry" starting from there? By taking the existence of multiple "external-worlds-as-we-conceive-them" not as a problem (as it is for a "realist") but rather as a virtue? Perhaps along the following lines ...

  1. Any set of observations is amenable to more than one summary/story/theory
  2. Any new set of observations can falsify some summaries/stories/theories but there will always be several summaries/stories/theories that are adequate summaries/stories/theories of observations to date
  3. Some summaries/stories/theories will prove to motivate new observations more effectively than others but there is no way to be certain in advance which these are and so scientific inquiry always includes an element of subjective choice
  4. It is possible to make use of several different existing summaries/stories/theories to generate new summaries/stories/theories that motivate additional observations
  5. By doing so, one expands the range of "external-worlds-as-we-conceive-them"
  6. Progress in scientific inquiry reflects not only the numbers and scope of observations incorporated in adequate summaries/stories/theories but also the number and scope of "external-worlds-as-we-conceive-them"
  7. Scientific inquiry is as much about creation as it is about discovery
  8. Scientific inquiry depends as much on seeing new possiblities because of multiple summaries/stories/theories as it does on falsifying particular summaries/stories/theories
My guess is that items 1-3 on this list are more or less common ground for both your rejection of epistemological realism while suspending judgement on metaphysical realism, and mine of rejecting what I see as unavoidably intertwined epistemological and metaphysical realism. And they provide fully adequate grounds to encourage a more Pragmatic ("instrumentalist"?) approach to scientific inquiry, one that I would certainly endorse. As a scientist, though, my feeling is that one can do still better, by recognizing that "unconceived alternatives" is not actually a "problem" but rather an opportunity. As per my items 4-7, scientific inquiry actually proceeds most effectively by looking for and trying to create as yet "unconceived alternatives". The realist posture, whether epistemological or metaphysical, gets in the way of that for both scientists and others, and so ought to be abandoned in its entirety. Metaphysical agnosticism doesn't, I think, go quite far enough along these lines.






| Science and Culture Forum | Science in Culture | Serendip Home |

Send us your comments at Serendip
© by Serendip 1994- - Last Modified: Saturday, 07-Oct-2006 15:42:55 EDT