Foundationalism, Skepticism, Coherentism
There are few issues in philosophy that run deeper or have broader ramifications than that of foundationalism versus coherentism. I object to foundationalism that it is incoherent; by parity, the foundationalist objects that coherentism is unfounded. Such exchanges may be cute, but they don't really help to clarify things. To that end, I want to say something about what foundationalism is and why people -- myself at one time included -- accept it. With that kind of background, the coherentist critique and alternative have a better chance of being understood and appreciated.
Foundationalism is a view about the structure of our knowledge. The foundationalist envisages solid foundations driven into bedrock and a sound structure erected thereon ranging over the most fundamental facts and principles, the most detailed understanding and the most embracing generalizations. It says, in essence, that certain beliefs or truths or bits of knowledge are specially privileged starting points for any further development of our knowledge. These are where we have to start in order to "get off the ground" in finding out about the world. If we start anywhere else or deny any of these, the whole structure can come crashing down -- many of our beliefs will be completely unreliable because they will depend on mistakes we've made about the foundations. For any of our higher-level beliefs -- those that are one or more layers of evidence or argument removed from the foundations -- justifying them or showing them to be correct or (perhaps) probabilifying them or showing them to be acceptable means to show that they have the right kind of connection to the foundations.
This is an attractive picture for any number of reasons, not least of which is its appeal to familiar metaphors such as the structure of a multi-story building, the links of a chain or the steps of a ladder. It is attractive as well because it seems to reflect our practice in criticizing or questioning knowledge claims. We ask, "How do you know?" -- and expect in answer something easier to know than the conclusion we have questioned. In imagination, we project a series of such questions and answers and infer that we would finally reach foundations - the most easily and readily knowable things of all. The unstated idea behind this is that we just can't keep on asking questions and giving answers forever; the unstated wish is that our stopping point be proper - that we stop where we do because no further relevant or sensible questions can be asked. There is further appeal in the analogy with logic where we know that false premises can lead to false conclusions with complete validity; the foundationalist views himself as making sure that the "premises" aren't false.
The deepest appeal of foundationalism, I think, derives from an attempt to achieve certainty. As long as there have been philosophers and scientists and ordinary reflective human beings, there have also been skeptics who claimed that we don't or can't know anything. They were the ones who kept pushing the `How do you know?' question and refusing to take easy answers. It didn't seem sufficiently satisfying just to say that the skeptic's own beliefs and assumptions -- for example, that people don't or can't know anything -- presupposed that he knew something about human beings and their cognitive capacities or lack thereof. Foundationalism offered the promise of a more ambitious response: By tracing our knowledge back to its foundations, we would get something the skeptic couldn't deny or call into question. We would be able, finally, to make the skeptic shut up.
Foundationalism comes in different flavors depending on what particular type of belief or unit of knowledge or whatnot is taken to be foundational. The foundations may be self-evident truths, synthetic a priori principles, presentations of sense-experience -- and the list goes on.
For present purposes, thankfully, it isn't necessary to disentangle and deal with each of these separately. Something quite general has gone wrong and the best way to see what is by taking the foundationalist model seriously -- that is, by trying to trace some particular knowledge claim to its "foundations." (I invite you to think of your own example -- the kind of thing that you would confidently and without hesitation assert and, if pressed, would claim to know.) Here's an example based on an actual conversation:
A: Is Gingrich a Republican?
A: How do you know?
B: It says so on the back of this book he wrote. Besides, he's Speaker of the House. Stuff about his party is in the news all the time.
A: Well, maybe reporters are just careless.
B: Hmph . . . somebody would have noticed.
A: Besides, he wants to close the national parks and sell them to big business.
B: Maybe, but so what?
A: Well, maybe it was a guilt-by-association ploy. The publisher and reporters could be trying to discredit the party.
B: Why would they all do that?
A: Maybe they have a hidden agenda ....
And so it goes, on and on and on ....
There are a couple of points to notice here. One is that one of the people involved is pretty plainly being more reasonable than the other and that we can tell that even though neither of them has said anything about foundations. The second is that, however reasonable or unreasonable these particular conversants are, what they're doing is an example of a very common pattern in argument which I'll call horizontal justificatory spread. When some belief is questioned, we answer by appeal to other beliefs or facts on the same general level. In practice, people often find something they can agree upon fairly early and the dispute ends, but, in principle, there's no need for either the questions or the answers to run out -- though the participants are pretty likely to eventually end up repeating themselves or reformulating and rearranging material already covered. What doesn't happen -- or doesn't have to happen -- is any kind of approach to "the foundations."
Now consider another common pattern -- the appeal to principles or generalizations. (Ex.: "Why don't you think there will ever be an intergalactic civilization?" "Because nothing can travel faster than light." Or: "How can you object to capital punishment?" "Because it risks killing innocent people.") This adds more to the picture we're building up -- another dimension at least. Even more is added if we include higher-level or broader principles and generalizations that themselves cover or explain lower-level or narrower principles and generalizations.
Does "thickening" the model in this way get us moving in the direction of "foundations"? Well, no, for at least a couple of reasons. First, it really isn't clear until after we know what the foundations are whether principles and generalizations are closer to or further away from them. It might be, for example, that we accept a general principle expressing opposition to killing innocent people because of lots of particular judgments that killing innocents hadn't worked out well. Second, there's a different model of the way in which all these knowledge claims fit together and connect with one another -- of which more shortly.
So what does move us toward foundations? What motivates foundational descent? Only one thing, I think. That one thing is the skeptic. He won't accept answers based on other facts "of the same general level." He calls whole ranges of alleged reasons into question at once. He wouldn't have been interested in asking whether Gingrich was a Republican; he would have asked how I knew there was such a person, how I knew there were any other persons, why I counted on perception or memory -- and so on.
This is where the foundationalist comes in. He takes on the project of answering the skeptic. If the skeptic doubts sense-perception, the foundationalist will defend it; if memory is doubted, he'll defend that. Wherever the skeptic raises questions, the foundationalist is ready or working on getting ready or hoping that someone somewhere is ready or that someone somewhere is working on getting ready or at least, in principle, could be ready -- to give answers.
Whatever we may think of the strategic wisdom of letting one's opponent set the terms of conflict, I think we have to admire the drama and heroism inherent in the foundationalist's self-portrait. He is an intellectual guardian of civilization, fending off the encirclements, encroachments and seductions of the forces of subjectivity, relativism and blind willfulness. He is, to borrow from Yeats, "the best" who does not "lack all conviction, while the worst are full of passionate intensity."
Having been there, having thought -- for
too long -- that the project of answering the skeptic was central to
philosophy, we also have to admit, I think, that the
foundationalist's self-portrait is not quite realistic. It is a bit
too dramatic and self-important.
The foundationalist thinks there is available, in principle, an answer to all of the skeptic's doubts. That "in principle," however, covers a good bit of hand-waving. Hardly any foundationalist thinks that all those answers have been worked out in satisfactory detail.
Suppose our imaginary foundationalist is one who hopes to trace our knowledge to "self-evident truths." For the time being, we'll skip all the intermediate steps. Suppose he boils down our knowledge of some sort to the self-evident truth that all triangles have three sides. (The example was chosen to be as innocuous and uncontentious as possible. Real-world foundationalists, of course, try to come up with self-evident truths that carry more content than that.) Surely, here he's got something that's really self-evident -- something that can't be denied by anyone who understands it. But, if he thinks that, he's wrong. Those who would doubt that every triangle is three-sided may all be or belong in mental institutions -- but that doesn't mean they don't exist. The "self-evident truth" has to be qualified: No sane or rational person who understands it can deny it. But that gives away the whole case. If the foundationalist makes his favorite set of "self-evident truths" defining criteria for sanity and rationality, his argument will be circular: "All the people who agree with me ... agree with me." But otherwise, the foundationalist has to appeal to notions of sanity and rationality that don't depend on accepting his favorite "self-evident truths."
Consider a different foundationalist -- or maybe the same one at a different time or in a different mood. Suppose he starts by defending the reliability of sensory perception. It is all but inevitable that somewhere in his defense he will appeal to some remembered fact or argument or distinction. What if the skeptic questions that? Then the foundationalist embarks on a defense of memory. (It is best to leave this argument conveniently off-stage so as not to be embarrassed by questions about how the defense will work without relying on memory.) Does the skeptic then give up and go home as a good foundationalist? Of course not. He just raises a different question -- for instance, whether time is real or not. And if time is somehow unreal, of course, then the reliability of memory is again in question. The foundationalist may have an answer to this as well. He may if you allow him enough time and also assume that he can reliably remember the phases and connections of his argument from beginning to end. But let's not be petty. Suppose he does have an answer. He thinks that time is real. The skeptic will point out that any serious position on that subject is going to have to face questions about the relationship of time to what clocks measure. And how do we know anything about what clocks measure (or what clocks are!) without counting on sensory perception? And isn't that where this whole argument started?
This is what I mean by hand-waving. The foundationalist is invincibly persuaded that somehow all the holes can be plugged and all the questions answered. In practice, this amounts to faith that he is cleverer and faster at coming up with answers than the skeptic is at coming up with questions. Lacking that faith, it just strikes me as an undignified game and hardly a suitable pastime for adults.
It may seem that the foregoing must either be a caricature of foundationalism or a prelude to an endorsement of skepticism. It is intended to be neither. There may be no foundationalist who exactly matches my description, but the portrait aims to be representative. It is intended to sketch general outlines, to point to typical motivations and rationales and to indicate as well the generic problems foundationalism faces.
If you doubt that, there is in the end only one test. Take it seriously. Adopt the project as your own. Taste for yourself the intoxicating prospects and the tantalizing whiffs of a final victory that seems never to be quite final -- to be constantly, chronically, within your grasp but never quite yours. Accustom yourself to a regular diet of small frustrations as you notice -over and over -- "just one more issue," "just one more question" where the corrosive acids of skepticism can slowly burn their way into your foundational structure. You can try to hold on by covering the wasting structure with massive plastered acts of faith. Or you can embark on the truly difficult task of reinforcing it, making it corrosion-resistant and acid-proof. I do not expect you to succeed.
As to skepticism, I have assumed throughout that it was mistaken. I think that, properly understood, the argument that the wholesale skeptic's doubts presuppose knowledge is a satisfactory answer.
Let's take a brief detour: Almost everything that people do or can do (with the possible exception of some automatic bodily functions) are also things that they can imagine doing, fantasize doing, pretend to be doing, feign doing. The reason the skeptic consistently wins the running debate with the foundationalist is that he has had a monopoly on doubt. What if we open up doubting to fair competition? What if we doubt his doubts? Are the skeptic's doubts real -- or is he just pretending? If he's just pretending, then the right response might be to pretend that we have an answer -- and pay him no more attention.
The skeptic predictably will complain
that his doubts are real and cannot be dismissed as the fantasies or
pathologies of a disordered mind. "Fine," I agree, "why
not?" And here the skeptic has major problems. The only way to
show that his doubts are real is to show that he has reasons for them
-- but having a reason for a doubt assumes all sorts of background
Suppose, for example, that he doubts the existence of an external world on the grounds that (a) it is non-deductively inferred from immediate experience, (b) that being so inferred, it could be mistaken, and (c) that we shouldn't claim to know something if we could be mistaken. It may seem as if the skeptic is only committed to the apparently trivial claim that non-deductive inference is less than certain. It may seem that way to the skeptic and presumably does to anyone who finds his argument persuasive. But it shouldn't seem that way.
In the first place, the apparently trivial claim is actually a rather bold speculation about the class of all non-deductive inferences -- to wit, that none of them are certain. This back-fires rather dramatically. For if some such inferences were certain, then a non-deductive inference to the existence of the external world might be one of them. On the other hand, since the skeptic doesn't have a complete list of all non-deductive inferences, he could only know that none of them are certain in two ways: First, he could find something that's common to all non-deductive inferences and argue that that feature rules out their being certain. But that just raises the same problem over again. How could he figure out something that every non-deductive inference has to have without looking at them all? Second, he might be able to do it by engaging in a non-deductive inference of his own -- the result of which would either be certain or not. And if it were certain it would furnish a counter-example to his thesis while, if it were uncertain, then it would tend to undermine his second premise.
Second, where did the skeptic get his premise (a)? Is it really obvious that our belief in an external world is derived from inference? Suppose I said, "Oh no, it's not inference at all; it's a matter of direct awareness. I don't `infer' the external world - I see it." Whatever the skeptic says here leads to trouble. He might succeed in undermining that claim of direct awareness, but only by making all sorts of empirical claims about how perception works, how it is related to inference and controversial interpretations of these facts- none of which he would be entitled to unless he already believes that he knows quite a bit about the supposedly doubtful external world.
Third, whether I should claim to know something when I could be mistaken may depend on how I could be mistaken and what the available alternatives are. What if the chance were very small - say, one in a trillion trillions? What if there were no real chance at all, as when I "could" be mistaken in thinking that eleven is a prime number? What if, given some basic intellectual equipment, one cognitive or belief-management policy generated lots of knowledge claims that were, for the most part, but not invariably, reliable, and the only alternative policies that generated fewer errors did so by being so fastidious about the evidence they'd accept and/or so painstaking in examination of reasoning that they also produced far fewer (or no!) correct knowledge claims? As these examples illustrate, the skeptic is really trading on the claim that "knowing something" occurs only if it's impossible to be mistaken. And he could be entitled to that claim only if he knew a great deal about the role that knowledge claims play in our belief- economies and a great deal about the different kinds of error that are possible in some sense or other.
Why have I gone into so much detail here? To show that the skeptic's doubts, if they are real, rest on all sorts of knowledge claims, just as pervasive, just as precarious, and just as liable to error as the ones to which he directs his critical fire. When the skeptic's doubts are challenged, they turn out to be either not real doubts -- and therefore not in need of real answers -- or they rest on quite large assumptions about what he knows -- in which case, skepticism turns out, more elaborately, to be a pretense.
But if skepticism is a pretense, a matter of pretended or feigned or imagined or fantasized doubt, then the skeptic's questions can't motivate foundational descent. When skepticism evaporates, foundationalism goes with it.
Suppose foundationalism is out. What kind of alternative is there? What does the coherentist have to offer? How is our knowledge structured if we don't have privileged foundations as starting points?
Before saying much on this, it may help to say something about what coherentism doesn't mean. It doesn't mean that there can't be linear arguments that proceed, perhaps through many intermediate steps, from premises to conclusions. It doesn't mean that everything is up in the air or that anything goes. It doesn't mean that we can't have convictions or that nothing is certain. It does not even mean that our starting points can't turn out to be certain.
So where do we start -- without privileged starting points? There's a prior question here: Do we start? For, in at least one sense, we don't. If we set relatively high standards for what is to count as a belief so that it has to have linguistic or verbal components or aspects, then, of course, we have first beliefs -- since linguistic competence is acquired, not inborn. Such high standards are unpersuasive, though, since we often ascribe beliefs to creatures like dogs or newborn babies entirely lacking in linguistic competence. If, on the other hand, we allow certain response-dispositions or interpretive tendencies to count as beliefs, then believing, though not the particular content of our beliefs, is as old as we are.
"Where do we start?" may be misleading. By the time we're able to consider such questions, we already have beliefs, and there's no guaranteed-to-be-reliable sorting strategy that will tell us which to keep and which to discard. What we do is the only thing that we can do: We start where we are with a vast array of already formed beliefs, patterns of reasoning and habits of inference. These beliefs, patterns and habits may vary greatly in quality. Some will be true or correct or reliable and others will not. Some will seem true or correct or reliable on the best evidence we have but will in fact be false or incorrect or unreliable. We will be able to give reasons for some but not for others. We will feel certain about some and not about others. When we feel less than certain, we will feel more confident of some than of others -- and our rank- ordering may be mistaken.
We may, in certain moods, regret this mixed state, our lack of omniscience or infallibility -- but there can be little doubt that it is our state. Given this, what are we to do? Abstractly, there are only a few basic possibilities:
We could reject all these beliefs unconditionally and without respect to content. If this could be done, it would be, quite literally, an arbitrary act with no justification whatsoever. I don't think this really can be done. We are believing animals and can no more stop believing by an act of will than we can stop metabolizing by an act of will.
We could reject them all conditionally -- that is, until and unless they are separately justified. This is the strategic position staked out by the skeptic and is untenable for reasons indicated above. He wants to preserve the notion of justification while denying that it applies to anything. His position can be saved from the charge of pretense only at the price of admitting knowledge claims -- and thus giving up skepticism.
We could accept them all unconditionally and unrevisably. This is approximately the mirror-image of the skeptic's position: we would keep the beliefs and give up justification. Apart from the fact, obvious to any other parties, that this would amount to enshrining and sanctifying a great deal of error, it is fully as untenable as skepticism, even in its own terms. For the beliefs themselves, even on quite rudimentary levels, make reference to justification. Without accepting the notion -- and the relevance -- of justification, we could neither understand nor, therefore, believe a remark such as "I took my umbrella because I thought it was going to rain," or "It was sheer luck that they hired me."
We could try to sort beliefs in terms of truth or reliability. This is, more or less, the foundationalist's approach. It's excellent advice but not very practical -- unless you have a reliable sorting principle or strategy. And how do you get that unless you have a reliable sorting principle for selecting among allegedly reliable sorting principles? And how do you get that ....?!
Last is the option I'd recommend. We can accept them all provisionally. We recognize that our beliefs are of varied quality, so we don't hold them immune to criticism nor do we assume that they must succumb. We are entitled to continue to believe whatever we find ourselves believing -- until and unless it generates a problem. Then, some adjustment has to be made. Since problems come in many shapes, ranging from clashes between or among our beliefs to felt dissatisfactions to disappointed expectations to unfulfilled predictions, the needed adjustments also come in many shapes -- ranging from rejection of some belief or set of beliefs to alteration of one's aims or objectives to the creation of new theories to explain what was previously problematic to reconstrual or reconceptualization of a problem situation. How do we decide what adjustment to make? -- in light of other beliefs which, for the time being, aren't problematic.
Does this mean that our beliefs and
knowledge claims are hostage to the possibility of undiscovered
errors? In a sense, yes, because errors later may be the result of
errors earlier. But that's just as much true on any theory of
knowledge and should probably be a more serious worry for the
foundationalist. Coherentism, at least, does not begin by setting
apart a special class of beliefs and refusing to entertain (except
perhaps in great extremity) the possibility that they are in error.
No belief or knowledge claim is, in principle, immune to criticism
and revision; any error may, in principle, be discovered and
There is another sense in which the coherentist may escape suffering, cognitive or otherwise, for early mistakes. Our beliefs are not to be envisioned as hard pellets with "meanings" fixed for all time, unaffected by what goes on around them. New discoveries, additional information and arguments going on elsewhere in a belief system may leave a statement of belief unchanged but alter the allowable inferences from it. We can still speak of the rising and setting of the sun though we no longer think that the events referred to involve the sun's motion. Especially clear examples of this kind of change are often buried in etymologies. We can talk of lunacy without thinking it caused by the moon and study economics without even considering household management. How far can such change go? Occasionally, even to the extent of direct self-contradiction: It is surprising to be told that a manufactured good is handmade though, etymologically, it must be. A mistake may be, not so much eliminated as erroneous but swamped by the implications of other beliefs -- thereby ceasing to be mistaken.
It remains to say something more general about the structure of our knowledge if we give up foundationalism. Remarks about horizontal justificatory spread and about "thickening" the model with the inclusion of principles and generalizations may have suggested some kind of three- dimensional network of knowledge claims. I think that's right in spirit -- but probably not nearly enough dimensions.
We can approach an answer by asking what the ideal completion of our system of knowledge would be. Of course, we are nowhere near such an ideal completion and never shall be. But in it there would be no unsupported claims nor any dogmatism. It would be comprehensive and inclusive, in complex interface with the external world, open to upset but would not in fact be upset. Our knowledge, ideally, is a network of claims that mutually support, reinforce and qualify one another; none is completely beyond revision in the light of changes elsewhere. Perhaps, in the limiting case, any single claim excised from the system could be reproduced -- on the basis of the implications of its nearer and further neighbors.
There is room for certainty here but it
must be reconceived. It becomes humanized and this-worldly. It
becomes less like a privileged starting-point and more like a center
of gravity -- maintained and stably preserved because of the complex
relations in which it stands to the things around it. No longer is it
the province of super-human powers or god-like detachment.
Certainties are not the foundations from which we begin but the ideal
end-points of investigation.