What I think we’re converging to here is the agreement that the progress of science benefits from scientists engaging in philosophy, and in a different context often society does as well. It’s been my experience that the best scientists do tend to have an appreciation of philosophy, which is really just a subset of the larger observation that the best scientists tend to have multidisciplinary awareness, and sometimes even formal multidisciplinary engagement. But this is not the same as philosophers per se usefully contributing to science. The way I’d sum it up by example is that Lawrence Krauss, brilliant physicist though he may be, is a narrow-minded blowhard who would be a greater credit to science if he understood philosophy. But on the second point, a philosopher like Dreyfus should stick to his phenomenology and not try to venture into machine intelligence or cognitive science, which he’s been doing with a great deal of pointless futility for about four decades.
No one made you read it.
No one made you respond.
Stop being a jerk.
[ /Moderating ]
I have enjoyed this thread more than any I have read here, not sure why.
The ‘philosophy is useful to science’-issue is really a bit of a sidetrack, and I’m starting to wish I hadn’t gone down that road. My main point is that those that criticize philosophy on the basis of a comparison to science, or that claim that science will replace philosophy, are comparing eggs to apples; even if philosophy had over the years exactly zilch to say that helped science, it would still be a worthwhile endeavour, because science doesn’t, and constitutionally can’t, encompass all knowledge-generation—it is only appropriate when there is an objective fact of the matter that it can approximate. But there’s knowledge to be gained through conceptual analysis, through discourse, that isn’t reducible to the empirical (in any sense other than perhaps a purely behaviourist or instrumentalist one).
Well, it’s debatable whether throwing more math at the problem will help; it’s quite possible for physicists to produce mathematically sophisticated, but conceptually incoherent works.
My example regarding Lawrence Krauss was facetious, but only slightly so. In fact, I think it’s exactly the uncertain status of modern cosmology that makes philosophy a possible boon to it: deep down, philosophy is nothing but the systematic attempt to get our muddled thinking clear. As such, it can’t tell you whether A or B applies to the world, but it can tell you that if you believe A, you also have to believe C, and can’t simultaneously believe D. In other words, it can clarify our commitments. If what we are committed to by advocating a particular position is nonsensical or even contradictory, then that’s a good reason to re-examine that position.
Krauss, due to his philosophical innocence, simply fails to heed this advice. One thing is that philosophical discourse has shown us that we can’t treat existence as a property, because then, negative existence statements become contradictory: if I say, ‘the ball is green’, then I can parse this as ‘there exists x such that x is a ball and x is green’—thus, any property ascription includes an implied existence claim. So what about ‘pegasus does not exist’? I obviously couldn’t parse this as ‘there exists x such that x is pegasus and x doesn’t exist’.
Related to this is the impossibility to ascribe properties to nothing. If I say, ‘nothing is p’, then I’m first of all faced with the linguistic ambiguity of deciding whether I mean ‘there is no x such that x is p’ or ‘there is an x such that x is nothing and x is p’. The first one is a perfectly valid statement, but it’s the second one we’re interested in—and that one’s flatly contradictory, basically claiming that there is something that is nothing. But that’s exactly what Krauss does: his ‘nothing’ is, e.g., a closed pseudo-Riemannian 4-manifold of zero radius (or the quantum vacuum, or something else—I don’t recall his exact definition). But that’s just not nothing—it’s something. So, his book should have been called ‘A Universe from Something’—only then nobody would have bought it. And when faced with this criticism, all he had to say for himself was basically that the notion of nothing has changed, and that’s what physicists mean when they talk about nothing.
There’s a lot of philosophers who are quite capable mathematically (Sunny Auyang’s book How is Quantum Field Theory Possible? was severly taxing on my math skills when I read it, and could easily double as an introduction into fibre bundle geometry). True, often they do have training in math or the sciences, but it becomes really too restrictive if you define as ‘true philosophers’ only those without scientific training. (I know physicists who work in finance. Are they not bankers, because they have scientific training?)
One thing that philosophers have helped scientist appreciate is the importance of the Kochen-Specker theorem (a kind of generalization of Bell’s theorem)—while it languished in obscurity in the physics community, philosophers were even putting the central construction of its proof on the covers of their books. In recent years, it’s also become an object of intense study by physicists (being now suspected, for instance, to be the engine behind he quantum-computational speedup).
I don’t understand your reasoning here. What do you think is being invalidated, and how?
Well, as I said, my main issue is not the applicability of philosophy to science; this is, at best, a side-effect. I think attempting to shoehorn the evaluation of philosophy into ‘but is it useful for science?’ starts the discussion on the wrong foot by assuming that science is the only game in town, and thus, to be valid philosophy must be useful to it. I’m arguing that precisely that is not the case.
And it’s also a good answer to Procrustus’ earlier question regarding who changes their worldview or life in response to philosophical arguments: millions of vegetarians and vegans do.
Bell himself considered what he was doing to be ‘quantum philosophy’—at least that’s what he named his book which contains a reprint of his original paper on EPR.
Bell’s theorem is the statement that if a certain inequality (or an inequality from a certain set of inequalities) is experimentally violated, then a certain ontology (local realism) doesn’t work. This is not experimentally falsifiable: if the inequality were not violated, then the last sentence would still be just as true; just the antecedent of the claim would be false. The experimental part is just a straightforward claim of quantum mechanics, which could have been made without tying it to metaphysics; but tying it to metaphysics is exactly what Bell did.
I think that’s really all I’ve been arguing for.
The name Hubert Dreyfus has been mentioned a couple of times as an example of ignorant philosophical meddling in science. Now, I’ll readily confess to not knowing much of the whole affair, and Dreyfus may indeed be full of shit when it comes to AI—I’ve nowhere claimed that every pronouncement by every philosopher is golden. Like usual, I think that Sturgeon’s law is a good guideline: 90% of everything is garbage, including philosophy.
But the wiki article seems to paint a quite different picture:
[QUOTE=wiki]
When Dreyfus’ ideas were first introduced in the mid-1960s, they were met with ridicule and outright hostility. By the 1980s, however, many of his perspectives were rediscovered by researchers working in robotics and the new field of connectionism—approaches now called “sub-symbolic” because they eschew early AI research’s emphasis on high level symbols. Historian and AI researcher Daniel Crevier writes: “time has proven the accuracy and perceptiveness of some of Dreyfus’s comments.”
[/QUOTE]
And further down:
[QUOTE=wiki]
By the early 1990s several of Dreyfus’ radical opinions had become mainstream.
[/QUOTE]
Basically, it seems that Dreyfus critiqued the early, overly optimistic AI enthusiasm (in what seems to have been a somewhat abrasive manner), in a way that puts him in line with many of today’s views on AI, like embodied cognition, subsymbolic processing, etc. And of course, we all know that those early predictions of human-level machine intelligence by the 70s, etc., never quite materialised—they couldn’t have, since there were too many unforeseen challenges, things like Minsky’s paradox, that what seems hard to us is easy for machines, while what we accomplish with little difficulty, such as facial recognition or navigation of changing, complex environments, are formidable challenges for AI.
The following passage is interesting:
[QUOTE=wiki]
Part of the problem was the kind of philosophy that Dreyfus used in his critique. Dreyfus was an expert in modern European philosophers (like Heidegger and Merleau-Ponty). AI researchers of the 1960s, by contrast, based their understanding of the human mind on engineering principles and efficient problem solving techniques related to management science. On a fundamental level, they spoke a different language.
[/QUOTE]
So, what seems to have happened in this case is that a failure to communicate between science and philosophy preempted a valuable contribution that philosophy might have made:
[QUOTE=wiki]
McCorduck writes “His derisiveness has been so provoking that he has estranged anyone he might have enlightened. And that’s a pity.” Daniel Crevier writes that “time has proven the accuracy and perceptiveness of some of Dreyfus’s comments. Had he formulated them less aggressively, constructive actions they suggested might have been taken much earlier.”
[/QUOTE]
But I really don’t have anything more than that wiki article to go on, so if that grossly misrepresents things, I’d like to know.
I don’t think the Wiki article “grossly” misrepresents things, it’s just that Dreyfus got a lot more things wrong than he got right, and the advances in AI came from evolutionary developments within the field itself rather than any guidance from Dreyfus. Indeed there were other philosophers – the late John Haugeland, once a student of Dreyfus, being a notable example – who had a much more accurate vision of the field. Dreyfus for example harped on his phenomenological thesis that machine intelligence doesn’t fundamentally model how people do things, because if they did they should be able to reach human performance levels, they would be generalized and not embody specialized heuristics, and they would, in contrast, embody recognizable cognitive processes. The first is just nonsense, the second is an area where Dreyfus makes some valid points but is demonstrably not always true, and the third is a controversial area in which his misunderstandings abound.
Dreyfus insists that experiential human cognitive qualities are fundamentally different than the theoretical ones of formal systems, that AI will always fail because it “simulates”, and to be successful it must not just “simulate” but “represent” the human cognitive process, and in his early missives was even insisting that the mere fact that computer processes were digital was a precursor to failure. Indeed he goes on about this at length in his first book, going so far as to claim that a digital computer cannot do what an analog system can do, citing as an example that “the only way a digital machine could ‘represent’ the division function of a slide rule would by assigning colinear spatial coordinates to the mantissas of two log tables, and effecting a ‘translation’ by subtracting.”
This is not only a ridiculous assertion of what a model is, it’s also a basic misunderstanding not only of the goals of AI but of the fundamental goal of science: not to duplicate phenomena but to make them accessible to the intellect, and to recognize the fundamental uniformities of nature even when they’re not obviously apparent. The consequence of conducting this science in this particular area is the constant advancement of machine intelligence beyond the bounds Dreyfus keeps setting for it, much as one of the first really successful chess-playing programs surprised him by beating him badly, and much as we fly in jets and space ships far faster and far above the birds yet without flapping our wings, which the ancients imagined to be synonymous with flight.
Vaguely amusing, as there are slide-rule emulators out there that exceed the accuracy of real physical slide rules – let’s see you read one where the degree of accuracy is on the order of 1/100th of a screen-display pixel!
If one’s emulator exceeds the quality of the “real” thing being emulated, which one is really the more limited?
(In the same way, people have calculated pi to what – billions of digits? Let’s see him do that with a compass and straightedge!)
Thanks for your input. Reading around somewhat, it seems that the story is that he got a few things right (mostly about the workings of the human mind) and a few things wrong (about the workings of computers), and the current view from at least some in the AI camp is that he could have offered some valuable suggestions for the field’s development.
Basically, this sort of thing is the reason why I started this thread: imagine how things might have developed if, instead of chest-thumping stay-off-my-turf displays, they’d listened to each other, and worked out their misunderstandings. Dreyfus might have learned that analog and digital computers can compute exactly the same functions, while AI might have learned (sooner) that human reasoning is not well captured by the kind of giant if-then-else tables used by expert systems. But instead, Dreyfus felt the need to brashly rebuke those AI newcomers daring to encroach on topics that philosophers had been mulling over for two and a half thousand years, and the AI folk in turn spat ritualistically upon the mentioning of dirty words such as ‘phenomenology’.
Of course, considering the larger picture, it seems that things are changing somewhat, with both philosophy getting new impetus from AI developments, and AI realizing that there’s people around who’ve thought about the problems they’re facing for millennia, so maybe there’s something they have to tell; so it’s not all that uncommon anymore to see both philosophers and computer scientists engaged together in the same endeavour, as for example with Daniel Dennett’s participation in MIT’s Cog project.
In any case, I think this supports my thesis that there’s ample room for a fruitful cooperation of science and philosophy, but unfortunately, this still is hampered by the anti-philosophical bias of leading authorities in science these days.
Philosophy is fun. I treat it as any pasttime or hobby.
I dropped a note about this to someone I know who’s one of the leading researchers in cognitive science. Part of his reply:
I think the key point here is that the question of whether phenomenology can play a useful role in cognitive science is entirely different from the question of whether Dreyfus can. This may interest you [PDF file]:
Thanks for your effort. I’ve found a video of Dreyfus presenting his critique online, and I’ll have a look once I get the time in order to be able to better comment. Basically, it’s hard for me to reconcile what I’ve read so far with pronouncements that he’s ‘blinded by some combination of ideology and maybe some childhood trauma involving computers’ against the possibility of AI. For one, he seems to agree that modern sub-symbolic approaches do capture the way human minds work, as opposed to the good old-fashioned approach, so it seems he’s not opposed to computers themselves, but merely to the assumptions that went into using them originally as a model of the mind. And those four assumptions he criticizes have indeed, at least in part, been rebuked in many modern AI approaches. But I recognize that I’m speaking from a position of too much ignorance to make any sort of definite pronouncement here (and it’s not the main object of this thread, anyway; not that there seems to be much interest in discussing its main points anymore).
Thanks, I’ll have a look.
Quite the contrary, the Brooks paper is directly relevant to the topic, and supports your thesis – that philosophy has much to contribute to science – as well as mine, which is that Dreyfus does not. My previous point was that Dreyfus seems to be as much reviled in the cognitive science community as he is in the AI community.
You’re right, thank you. But my comment was (for the first part) aimed more specifically at the whole issue of whether Dreyfus was right or wrong regarding AI, and (for the second part) at the fact that everyone else seems to have lost interest in debating the points they’ve brought up.
I’ve read the paper by now, and it’s at least made me want to check out the book by Gallagher and Zahavi, and prompted me to get a clearer idea of what, actually, phenomenology is about—so either way, I’m getting out of this with a net plus.
Actually, I thought that we had pretty much reached a consensus on this. That being that science would be well served if more scientists took a more philosophical view of their field of study and that many of the best scientists in fact do so. The converse to that seems to be that it’s unclear that actual philosophers per se very often contribute to science, which is a different claim altogether. Maybe the Brooks and G&Z papers will persuade you that sometimes they do, but it can’t be often.
The problem with the latter claim being that useful contributions to modern science typically require a tremendous amount of in-depth subject matter knowledge and specialization. This may seem to contradict what I said earlier, that some of the best scientists have cross-disciplinary awareness and expertise, but this only works when the areas have a strong connection – for instance, psychology, cognitive science, and AI, or paleoclimatology and specific areas of math, chemistry and physics.
I think one could make a much stronger case that pure mathematicians contribute much more to science than philosophers because math is so pervasive in science, but even there, I would argue only within the context of the math itself.
It occurs to me that there’s an analogy I can cite that’s a pretty good math parallel with the Dreyfus case. When a climatologist named Michael Mann produced a millennial-scale global temperature reconstruction showing the dramatic uptick in post-industrial temperatures, a fellow named Steve McIntyre decided to look into his statistical methods. McIntyre wasn’t a mathematician but he did have a strong background in statistics and approached it from that angle. In fact he found some minor quibbles (which Mann subsequently fixed) and that should have been the end of that. But it wasn’t, not by a long shot.
To paraphrase an earlier comment, McIntyre might have been said to have had “some early childhood trauma regarding global warming” :p, though actually his long professional association with the oil and gas industry had more to do with it. Emboldened by having made a couple of criticisms that turned out to be correct, he launched an all-out attack on everything Mann had done – starting out by claiming that Mann’s use of principal component analysis utterly distorted the data and created a fake “hockey stick” shape (it didn’t) and moving on to claiming that some of the important temperature proxies Mann had used were spurious and non-representative. He’s now basically made a career out of accusing Mann (and other climatologists) of scientific fraud, whereas in fact Mann is one of the most respected and recognized climatologists working today – a pioneer whose results have now been validated countless times by countless different investigators – while McIntyre himself, now generally shunned by the scientific community and the respectable journals, is reduced mostly to ranting in his personal blog.
Such are the perils of someone with an axe to grind, be it a philosopher or a mathematician or anyone else, venturing far outside their field of expertise.
Well, the problem is really that I don’t want to see philosophy evaluated by its usefulness to science, because that basically assumes that science presents the gold standard that anything must approach in order to be considered worthwhile. But I think this is a somewhat fallacious way of viewing things that depends on questionable assumptions, most notably that there is an objective fact of the matter accessible to empirical methods regarding every possible question. But questions like ‘how should I live my life’ and similar ones don’t seem to.
Nevertheless, I believe that there is meaningful discourse that can be had on such questions, which does produce concrete advances. It’s just not discourse in the sense of approximating the objective truth of the matter ever closer, as it is in science, but rather, something that’s at its base more like simply clearing up muddle-headed thinking, exposing our assumptions and commitments, and giving us grounds to question them.
There are edge cases in which this latter dialogue produces results that can be used to further the scientific process, but these are, by and large, incidental; they’re not the reason why we undertake this dialogue. In a certain sense, philosophy and science are incommensurable—they’re not primarily after the same sort of knowledge, and thus, use different methods, and produce qualitatively different results. But criticism of philosophy is almost always provided with a ‘science-centric’ viewpoint in mind: it’s precisely those ways in which philosophy differs from science—not producing quantitative results, not being falsifiable, and so on—that make it somehow less worthwhile. Science, it is alleged, by having all of these qualities has the power to replace philosophy with an inherently better way to get at ‘the truth’.
To me, this is wrongheaded, since there simply isn’t always a ‘the truth’ to be got at—things like ‘what is the best form of government’, ‘what is the morally right way to act’, etc., don’t ask for facts that are written into the laws of the universe, but which depend on the social context in which they are asked. Many fear the dangers of relativism in this sort of stance, and I think that’s an important factor in the reaction to philosophy (it’s all arbitrary, just words, etc.), but I don’t think this is true—fundamentally, the question is just, how do we best cope with the world we find ourselves in, and of course, this will depend on the state of the world. And in answering this question, or discussing it, we change that world, which then will prompt a new reevaluation, and so on.
It’s ultimately like evolution: there is no ‘true’ or ‘best’ form of life that all evolutionary lines will eventually converge to, no target state or objectively ‘most evolved’ being, but an ever ongoing process of change and adaptation which leads to further change, etc. Science, in a sense, then produces boundary conditions to this process, or rather, discloses them (not that that’s a conception that all philosophers would agree to, but to enter into this discussion as well here would exceed even my willingness to write walls of text). (Another point one could raise, but which would take us too far afield, would be that science really only produces structural knowledge, but we don’t seem to live in a purely structural world, but rather, one with content and meaning; this is in my view tied to the problem of consciousness and its scientific appraisal—it’s ultimately not a phenomenon itself, but the base of all phenomena, and hence, in a certain sense, all science has to take it for granted, as a brute fact not admitting further analysis; thus, it’s hard to see how a scientific study of consciousness could come about. And then this takes us right back to phenomenology…)
For the record, I agree with just about everything you’ve said, the bolded parts being my emphasis. It’s just that your OP comments, headed as they were with II. The Relationship of Science and Philosophy and III. Philosophy aiding Science seemed to steer the discussion to (a) a specific argument about the role of philosophy in science, and (b) even more specifically to the role of philosophers in science. I think the value of (a) is significant in many contexts, the value of (b) can be very limited (as we’ve seen), but the value of philosophy in helping us govern our lives and understand our place in the universe is, in my view, indisputable. Empiricist ideologues like Lawrence Krauss are pretty short-sighted, IMHO, no more helpful to that understanding than religious dogmatists on the other extreme who cling to Biblical literalism.
And you’re quite astute in pointing out that questions like “what is the morally right way to act … depend on the social context in which they are asked”. In fact, that has become the entire foundation of the deontological moral philosophy founded by Immanuel Kant called “the categorical imperative”. In its very simplest interpretation it states that first you have to agree on basic societal values, and then your actions can be rationally governed by whether or not they conform to those goals, as tested by the hypothetical for any proposed action: “what would happen to those values if everyone acted as I propose to act?”
Fair enough. I intended the OP as a response to a certain sort of criticism which to me seems to come mainly from the direction of those who think that science has already replaced philosophy, or will do so in the foreseeable future, and to show that science may be hurting itself by essentially declaring philosophy a forbidden subject. The contribution of philosophers to science was never my main aim, though I see how one can view my post in this way (in fact, I’d say that the more significant contribution is regarding the form, rather than content, of scientific inquiry—Hume’s problem of induction, Popper’s falsifiability, Kuhn’s paradigm shifts, and so on—which is not surprising: both science and philosophy are incredibly vast and specialised subjects, and only a few exceptional individuals are truly capable of making substantive contributions to both; that’s part of my reason for advocating a greater interdisciplinary communication and lamenting the tendency to downplay the contributions of the other discipline, even if that’s just the natural outcome of wanting the world to be surveyable within one’s own horizon).
I don’t think every deontologist would necessarily agree to that; there’s certainly deontologists that would consider themselves to be morally absolutist, i.e. believing that there is a unique fact of the matter regarding what’s morally right, and I think a good argument can be made to the effect that Kant himself falls into this category in so far as he considers the possibility of a truly ‘universal law’ that should govern each individual’s actions, which can be determined by ideal rationality—in other words, if we all were ideally rational beings, there would then be no question about what actions were morally permissible/forbidden/mandatory, and we would reach a static state of ideal morality.
But again, this is a different thread… 
Just as a brief aside, having refreshed my memory about this stuff, it appears that what I was describing was the first formulation of Kant’s categorical imperative, the “universal law of nature” formulation and probably the best known. You’re quite right that here, as in the other formulations, Kant did try to argue for moral absolutism. I’m not sure where “…first you have to agree on basic societal values…” came from except from my own head, but I probably said it because I emphatically believe it, as I don’t see how one can judge morality except with reference to what is invariably some framework of values. But that, as you say, is a discussion for another thread.
I don’t think it’s comprehensive enough to say science proponents have a problem with philosophy. I know that’s what they say, but nuances often get broken down into talking points. Most scientists don’t have a problem with ethics or political and social philosophy.
Science doesn’t just come from, but is completely based on the philosophy of Empiricism: The epistemological philosophy that states that knowledge only comes from physical evidence and experimentation interpreted by our senses. Not every scientist claims all knowledge is exclusive to science, but the ones you’re referring to do. Through the lens of Empiricism; All other knowledge claims are just unsubstantiated conjecture, and are therefore, unjustifiable assumptions.
Empiricism also relies on at least 2 ontological assumptions in addition to the assumption that empiricism is true. They assume they exist and so does the universe. Empiricists have no problem agreeing with these assumptions (and if they’re intellectually honest they’ll even admit they’re assumptions). It’s only when you challenge any of these assumptions that it undermines the scientific method and all we’ve achieved with it.
The more pedantic the empiricist; The more likely they are to reject anything unempirical.
I think most scientists accept that mathematics is a form of knowledge, albeit a non-empirical one.