The Motivation for Classical Foundationalism

I’d be interested to hear people’s thoughts on the following question. Why does classical foundationalism claim that basic beliefs are epistemically privileged in the sense that they are indubitable, infallible, indefeasible, and so on?

Roughly speaking, foundationalism is the view that some beliefs are basic in the sense that they are non-inferentially justified, while all other beliefs are non-basic in the sense that they are inferentially justified by their relations to basic beliefs. It’s a further commitment that basic beliefs are epistemically privileged in the sense that they are indubitable, infallible, indefeasible, etc. Moreover, contemporary foundationalists tend to distance themselves from further commitments of this kind: see, for example, Pryor’s discussion of foundationalism in “Highlights of Recent Epistemology”. So, what is the motivation for classical foundationalism to incur these further commitments in the first place? Here are three suggestions.

The first suggestion appeals to the contents of basic beliefs. Classical foundationalism tends to claim that basic beliefs are restricted to beliefs about one’s own experience and perhaps a priori justified beliefs, whereas contemporary foundationalism tends to include a posteriori beliefs about the external world among one’s basic beliefs. To explain why classical foundationalism is committed to the epistemically privileged properties of basic beliefs, we need to add the further claim that a priori justified beliefs and beliefs about one’s own experience are epistemically privileged in the relevant sense, whereas a posteriori justified beliefs about the external world are not. I think much of the opposition to classical foundationalism is motivated by the rejection of these sorts of epistemic asymmetries. However, I’m not satisfied that this suggestion answers the original question about motivations, since I think it get things backwards. My suspicion is that classical foundationalism restricts basic beliefs to beliefs about experience and a priori matters because of the prior commitment to the claim that basic beliefs are epistemically privileged, not the other way around. So the question remains: what’s the motivation for this commitment?

The second suggestion appeals to skepticism. I’m not sure exactly how to flesh this out, but I suppose the rough idea is that an adequate response to skepticism requires establishing that our basic beliefs are epistemically privileged in the sense that they are indubitable, infallible, indefeasible, etc. The problem with this suggestion is that it remains quite unclear why this should be a criterion for an adequate response to skepticism. Indeed, contemporary foundationalists (e.g. Pryor) are quite explicit in rejecting these sorts of criteria of adequacy. So, I think we need to look deeper in order to find an answer to the original question about motivations.

The third suggestion, which seems to me roughly on the right track, is that classical foundationalism is motivated by epistemic internalism, i.e. roughly the idea that epistemic facts about which propositions one has reason or justification to believe are epistemically privileged in the sense that they are certain, infallible, indefeasible, etc. Moreover, it seems plausible that these epistemic facts are epistemically privileged only if they are grounded in non-epistemic facts that are epistemically privileged too – the best candidates being facts about one’s own experience and certain a priori facts. As it seems to me, the commitments of classical foundationalism cannot be explained without taking into account its commitment to epistemic internalism.

Of course, there are further questions about the rationale for epistemic internalism, which I won’t take up here, although I’d be interested to hear people’s thoughts. I’d also like to hear whether people think that this is a plausible diagnosis of the motivations for classical foundationalism and whether there are better alternatives that I haven’t considered. If anyone has references to useful discussions of these issues in the literature, those would be very much appreciated too.


The Motivation for Classical Foundationalism — 17 Comments

  1. Declan, first, I’d want to be careful about how “iinferentials understood, but that’s tangential to your main point.

    As you saw at the last MEW, my motives fit into the first category. I don’t think it gets the order reversed (I’m pretty confident of this regarding my own case). For me, it goes like this (train of thought, not argument):

    *Some of my beliefs are based on other beliefs
    *What’s rock bottom, I wonder?
    *[Thinks about it.]
    *Experiences are rock bottom
    *So if basic evidence is propositional (which I’m quite unsure about), it consists in propositions which describe our experiences/state that we have had those experiences
    *What interesting epistemic properties do these items of basic evidence have, I wonder?
    *[Thinks about it.]
    *They are privileged in this way: They are certain or nearly certain (for latter, see the work of Jeffrey and Conee’s “Comforts of Home” and maybe the end of Chisholm 1942)

    That’s how it works for me anyway, when I’m not a moderate foundationalist or coherentist (if there is any difference there).

    So, and I talk about this in the Intro to, the third option is the one that has it wrong way round. I’m an internalist *because* I’m a foundationalist. It “just so happens” that what is *in fact* at the foundations of my evidence (experiential propositions) are (nearly)indubitalbe items about what is internal. This is the feeling I get from Chisholm, too, though that’s quite compelex.

  2. My 2nd year at Rochester, I read through all three editions of Chisholm’s _Theory of Knowledge_ consecutively. It was a very rich and rewarding experience. In the 1st edition (1966), he starts out with the Theatetus and wants to know–without doubting the possibility of knowledge–what’s the extra bit that turns true belief into knowledge. He more or less assumes–though provides some motivation for–the idea that the extra bit is sufficient justification, i.e. evidence. He then proceeds to give an account of evidence via his epistemic principles (the material ones take you from a descriptive state to a normative state, the formal ones from one normative state to another, roughly). He pursues a line of what he calls “Socratic questions” which are not unlike Descartes’ questions about what cannot be doubted. So it seems to me, that Chisholm’s reasoning is as I described mine: we want to know the foundations, what is most evident. It turns out to be stuff about our psychology. But that was not necessarily foreseen, it comes (logically) at the end of the dialectic, not the beginning.

    In another way, though, the quest for foundations *just is* the quest for the (as) indubitable (as possible). Chisholm’s Cartesian use of Socrates seems to bear this out.

    In 2nd edition, the historical stuff (very sadly) drops out and he begins with definitions of terms and goes straight to the theory of evidence. Like the 1st edition, there is no chapter on skepticism. The book ends with a consideration of the problem of the criterioin.

    The 3rd edition (1989) opens with “The Skeptic’s Challenge” and ends with “What is Knowledge?” There is no chapter on knowledge in the first edition.

    The sad thing about all the history dropping out after the 1st edition (Chisholm’s knowledge of the history of philosophy is breathtaking!) is that one no longer gets to see how the internalist project has really been a consistent stream throughout the history of philosophy as a search for a stable foundation for belief. I highly, highly recommend reading the first three chapters of the 1st edition.

  3. hey man,

    This is really interesting, but – at least as a historical matter – it seems to me that there’s almost certainly no single motivation for Classical Foundationalism in its various forms.

    Still, maybe we can avoid this issue by looking at a particular example – e.g. Descartes – and asking what his motivation for accepting something like this position was. (Of course, the attribution of this view to Descartes isn’t entirely uncontroversial, but there are no uncontroversial examples. And please note that I’m certainly no Descartes scholar!)

    As we all know, one of Descartes’ goals is the discovery a method for forming beliefs whose products will be epistemically privileged in something like your sense. But his interest in this project seems to me to be largely independent of his foundationalism. In fact, Descartes seems to be interested in foundationalism largely because it provides him with a means for discovering what this method will involve. So his interest in discovering a method for generating these sorts of beliefs is independent of and priori to his foundationalism – contrary, I guess, to your first suggestion.

    But why is Descartes interested in discovering such a method? Contrary to your second suggestion, I don’t think that his interest in it is motivated primarily by the idea that such a discovery is required in order to avoid skepticism in ordinary situations. Rather, Descartes wants to find such a method because doing so would allow him to form a set of beliefs that has certain other epistemic virtues that he takes to be important – i.e. it would put him on the path to something like real scientia (real systematic knowledge).

    So what are these virtues and how do they relate to your third suggestion? This is, I think, a really interesting and contested topic. Scientia of P seems to require that one clearly and distinctly perceives both the truth of P and the truth of the general principle that what one clearly and distinctly perceives is true. So it requires not just knowledge that P but also knowledge of the principles that (in some sense) support that knowledge. That does sound a bit like what you’re describing, of course, since it involves a kind of knowledge of the grounds of your beliefs. But it’s not really the same as your story as far as I can tell. And scientia also has some other special features for Descartes – it’s perfectly safe, it’s universal, it’s systematic, and so on.

    In the end, then, I’m not sure that we need to see any of these features as the unique source of Descartes’ interest in scientia – after all, a lot of them have obvious philosophical interest, whether or not scientia is required in order to respond to skepticism. Or, to put the point more forcefully, the discovery of the sort of method Descartes was after would be uncontroversially awesome for all sorts of reasons. So I’m not sure we need to dig very deep to discover the motivations behind his search for this method. (I think this is similar in spirit to Trent’s comment?)


  4. Trent,
    Not to derail the thread, but why propositions that describe your experience and not propositions that represent what you experience and capture experience’s content? Otherwise, it sounds as if introspection is the foundation for all knowledge of an externalnworldnwhichininvitesnthenquestion as to why introspection can do what perception cannot (I.e., provide the foundations).

  5. Many thanks for the interesting comments everyone.

    Trent, I had a similar question to Clayton’s: couldn’t you hold that experience is rock bottom in the sense that experience is a source of foundational justification for belief, while denying that perceptual justification for beliefs about the external world depends epistemically on introspective justification for beliefs about one’s mental states?

    On the order of explanation point: my worry is that without some further theoretical motivation, it can look a bit shaky to claim that one’s justification for beliefs about experience is infallible, indubitable, indefeasible, and so on. For instance, these claims are pretty unpopular among philosophers of mind who work on introspection outside the framework of traditional concerns about foundationalism, internalism, etc. If I remember correctly, even Conee, whom you mention, argues against the luminosity of mental states by arguing for the defeasibility of introspective justification. My sense is that without some independent theoretical motivation from arguments for epistemic internalism, these traditional claims about the epistemic privilege of introspective justification can seem a bit arbitrary and unmotivated.

    I’ll be interested to read what you’ve written on this. I couldn’t find the introduction to the Evidentialism volume on your website, but I’d like to see it if you have a copy you could send me. As for the Chisholm, I’ve read one of the editions, but I can’t remember which, and I must admit I didn’t realize the various editions were so different from one another! Perhaps I should follow your example and tackle the trilogy at some point.

  6. Karl, you’re quite right of course: classical foundationalism is not a monolith. Ultimately, I suppose, the question I’m interested in is whether Pryor and other contemporary foundationalists are right to abandon classical foundationalism in favour of more moderate versions or whether there are good motivations for the classical version. No doubt, thinking through the history can help to cast some light on this.

    Is Descartes’ idea that if all of our beliefs are to be indubitable, infallible, indefeasible, and so on, then our basic beliefs had better have these properties too? That’s a bit different from the kind of classical foundationalism that I had in mind. I’m imagining a version on which it is granted that our a posteriori beliefs about the external world are justified in a way that is fallible, dubitable, defeasible, and so on, but which depend epistemically on certain basic beliefs that are justified in a way that is infallible, indubitable, indefeasible, and so on. If you have this kind of epistemic asymmetry between basic beliefs and non-basic beliefs, then it’s perhaps a bit more difficult to see why the basic beliefs need to be epistemically privileged in the relevant way.

  7. Hi Declan,

    I started typing out a comment earlier, but I’m still learning to use the ipad and just gave up. I think I agree that there’s a plurality of considerations that explain the motivation for classical foundationalism. Among the pressures that I thought should be mentioned builds on a point Trent made. Once we turn to experience to serve as our foundation, the arguments that are supposed to show that the objects of awareness are mental intermediaries, it makes some sense that our foundational beliefs will be about ourselves and our experiences rather than external states of affairs. If independently you think that the kinds of arguments from error that push people away from direct realism do not create a divide between use and the immediate objects of awareness, perhaps we’re supposed to get indubitability and the rest for free. I don’t think I’d want to say that this is a dominant motivation, but I suspect that this is part of what pressures people of a certain era in the move from foundationalism to classical foundationalism.

  8. Right, Declan. What I really meant to do in my (far too long) comment was to suggest that the historical concern for locating beliefs which have the relevant privileged status often works quite differently than your setup suggests.

    As you present things, we begin with some intuitive notion of the sort of justification ordinary beliefs often possess, and then want to discover what needs to be the case in order for a belief to be justified in this sense – which, in turn, leads us to the idea that such beliefs must be based (directly or indirectly) on basic beliefs which have the relevant privileged status.

    But for someone like Descartes, things seem to work somewhat differently. Descartes begins with the desire to locate a way of forming beliefs that will have the relevant privileged status – which he takes to be required for some sort of appropriately heavy-weight conception of knowledge (scientia) – a kind of knowledge that is of epistemic interest whether or not it is required for the sort of justification that ordinary people’s beliefs have.

    I think something like this is actually quite common. If so, then Classical Foundationalism in your sense – insofar as this claims that “everyday justification” requires that basic beliefs have the relevant privileged status – might well be the product of a certain sort of natural misunderstanding of what Descartes (et al) are up to.

    But none of this, of course, directly impacts your substantive point.


  9. Clayton, good point about the connection between classical foundationalism and traditional arguments against direct realism: these issues do seem to be tied together, e.g., in Russell’s philosophy. It may be that the arguments against direct realism are driven in part by epistemological assumptions, rather than the other way around, although I’m not an expert on the history. In any case, I’d like to find some motivation for classical foundationalism that doesn’t depend on the rejection of direct realism, since I doubt many contemporary foundationalists will be much moved by these sorts of considerations.

    (Apparently, Tim McGrew has a probabilistic argument for classical foundationalism inspired by C. I. Lewis that relies on the principle that all probabilities must be grounded in certainties. The rationale for that principle is not immediately evident to me, but it might be worth thinking about.)

  10. Clayton, since when do you have scruples against hijacking a thread? 😛

    1. I’m not sure what you have in mind by “propositions that represent what you experience and capture experience’s content.” Could you give me an example?

    2. The reason I think introspection can do what perception cannot is that our mental states are immediate in a way that external world objects are not. As Leibniz says

    Our direct awareness of our own existence and of our own thoughts provides u with the primary truths a posteriori, the primary truths of fact, or, in other words, our primary experiences; just as identical propositions constitute the primary truths a priori, the primary truths of reason, or, in other words, our primary insights. neither the one nor the other is capable of being demonstrated and both can be called immediate–the former, because there is no mediation between the understanding and its objects, and the latter because there is no mediation between the subject and the predicate. (New Essays Concerning Human Understanding, Book IV., Chap. 9).


    3a. I didn’t say anything about *beliefs* about one’s mental states. I’ve become partly convinced of tacit beliefs by Tim McGrew, but I’m not all the way there yet. Here is a discussion between Tim and I when I was still defending Modest Foundationalism.

    3b. The reason I think propositional evidence can’t stop before propositions describing our experiences (e.g. ) is that for any other kind of proposition, it makes sense to ask for further justification. My biggest concern here has been a kind of Hume’s regress worry (which Chisholm seems not to have shared). But I’ve assuaged those doubts in three ways: 1. A formal result which Alex Pruss and I have, 2. Tim’s tacit belief route, 3. A linguistic-grounding account which I haven’t shared with very many people now. Suffice it to say, I’ve migrated a long way from modest foundationalism (though, you’ll see in a moment, not all the way).

    4a. I can give an explanation as to why the current fad in Phil Mind is what it is, but that won’t make me any friends.

    4b. Note that I’ve hedged considerably on the “indubitability” or “infallibility” claims. Like Earl, I think that we likely can’t have those properties (which differ from luminosity), but that we can have something just as good. He calls the property “epistemic comfort” and it is a property such that “we cannot have without our being in a position to know it by acquaintance and rely on it as evidence.” After he considers all the arguments for and against, he concludes

    “Nothing in the arguments that we have considered jeopardizes the fact that phenomenal qualities are always available to be known to us by acquaintance. Likewise, the arguments do not jeopardize the fact that conscious characteristics are always among our ultimate evidential resources.” (p. 450)

    I like to call this “privileged *enough*” status. Whether it’s indubitable or infallible depends on the definitions of those terms. But introspective evidence is privileged in the most important sense: it, and not justified beliefs about the external world, is suited for being the foundations of our epistemic system.

    5. Chisholm was completely convinced by his mentor C.I. Lewis that “nothing is probable unless something is certain.” Richard Jeffrey despised this and attached it vigorously. His doctrine is “radical probabilism” which entails that it can be probabilities “all the way down.” This is my position. However, like Chisholm and unlike Jeffrey, I do think our basic evidence is of the kind Lewis thought. So my position is one tick to classical from Jeffrey. It’s in between Chisholm and Jeffrey at any rate. Here is a discussion between Johnny Blamey and I on whether there needs to be certainty for probabilitiy.

    Clayton and Declan,

    6. I take my position to be completely compatible with Direct Realism. Chisholm was a direct realist–it was largely the motivation for his adverbial theory of sense, and Michael Huemer–another “experience first” epistemologys–is also a direct realist (See _Skepticism and the Veil of Perception_, Chapter IV., esp. $$ 5-6). Fumerton–the arch-direct realist–has a very interesting paper on the nature of direct realism on his website.

  11. Trent, I’m trying to understand the relationship between your view and classical foundationalism as I’m defining it, i.e. (i) there are some basic beliefs that are non-inferentially justified; and (ii) all basic beliefs are epistemically privileged in the sense they are infallible, indubitable, indefeasible, etc. You raise an interesting set of issues about whether we can weaken (ii) in ways that are nevertheless stronger than countenanced by standard moderate forms of foundationalism. But it’s not entirely clear to me that your view counts as a form of classical foundationalism, at least as I’ve defined it, since your claim is about which propositions comprise our evidence, i.e. propositions about experience, rather than which beliefs are non-inferentially justified, i.e. beliefs about experience. Do you have some other way of defining classical foundationalism in terms of evidence? Or are you assuming some principle linking evidence and justification, e.g. one has non-inferential justification to believe p iff p is comprised in one’s evidence?

  12. Trent, sorry for the barrage of questions, but here’s just one more. Why do you think that our evidence is comprised of propositions that describe epistemically privileged conditions, e.g. conditions that one is in a position to know by acquaintance whenever they obtain? My impression from your first comment was that this is just a happy accident for you, i.e. there are independent arguments that our evidence is comprised of propositions about experience and independent arguments that experiences are epistemically privileged conditions. But my impression from the more recent comment is that you think there are more principled reasons why the foundations of our epistemic system must be epistemically privileged in the relevant sense. That’s what I think too, although perhaps we disagree about what those principled reasons are…

  13. Hi Trent,

    So, I might have not understood your view, but you said, “if basic evidence is propositional (which I’m quite unsure [but completely right] about), it consists in propositions which describe our experiences/state that we have had those experiences”

    Consider five propositions:
    (1) The glasses are on the piles of paper.
    (2) The glasses are near me to the right.
    (3) What is near me to the right has the look of a pair of glasses/looks to be a pair of glasses.
    (4) It seems as if I see a pair of glasses off to the right.
    (5) I’m appeared to glassesly. (Is that how you say it? I never learned to speak that language.)

    (1)-(3) are the sorts of things that I thought would capture what I experience without being about my experiences whereas (4) and (5) seem to be about experiences rather than what I experience. So, while I’d think that I can know (4) and (5) via introspection, I don’t think I could know (1)-(3) via introspection but could know them via perceptual experience. I took you to be saying that your evidence ultimately consists of propositions like (4) and (5) which could be true even if you were a Cartesian spirit disembodied in the dark deceived by a demon. They don’t seem to represent what I experience because they don’t tell us what it would take for my experience to be veridical.

  14. Hi, Declan and all,

    I’m both amazed by the vastness of the ground covered in this excellent thread and surprised at what strikes me as the omission of a key element to a satisfactory answer to Declan’s main question here. That question is: What is the motivation for the classical-foundationalist characterization of basic beliefs as infallible, indubitable, etc.?

    (But, first, let me note that I’ve just finished reading this thread for the first time. So, maybe this is an instance of *irresponsible posting* on my part. Maybe the surprise would wane with more reflection on the thread. Sorry if I’m just adding noise to the thread!)

    That key element is the mention of an *epistemic goal*. Every epistemology is based on an assumption about what the primary epistemic goal is. The default assumption in contemporary epistemology – one that can, and has been, intelligently challenged, of course – is that truth is the primary epistemic goal. (As a cognitive agent, you’re presumably guided by the goal of having only true beliefs.) From that perspective, the interest in finding infallible beliefs seems to follow. (A belief is infallible iff you cannot have it while it is false.) And from that, arguably, on a JTB+ conception of knowledge, we soon face the issue of the truth-connection for justification: If justification must somehow be truth-conducive, let’s look for infallible justification first. So, now we would seem to have the desire for infallibly justified beliefs (their truth being implied by the nature of their justification). And, if a (non-inferential) belief is infallibly justified, it cannot be Gettierized (the contemporary epistemologist would add). Just call it “knowledge”. So, find us some basic beliefs and we’ll have some knowledge.

    Is this off the rails? If not, is it obviously implied by the previous posts? (I don’t doubt that it is non-obviously implied by the discussion.)


  15. Thanks, Claudio. It’s possible that this thread was a little too vast; maybe next time I’ll just try to prove a theorem or something!

    A quick comment on your suggestion: it’s not obvious to me why acceptance of truth as the primary epistemic goal should push one in the direction of foundationalism, as opposed to, say, coherentism; and even if one accepts foundationalism, it’s not clear to me why it should motivate a classical infallibilist version of foundationalism as opposed to the moderate versions that are more common in contemporary discussions. At least it’s striking how many coherentists (e.g. BonJour 1985) and non-classical foundationalists (e.g. Goldman 1986) seem to endorse the claim that truth is the, or at least one of the, primary epistemic goals. Of course, you’re right that infallibility would be awesome from the point of a truth-based epistemology, but I think the majority are persuaded that it’s not a “sine qua non”.

  16. Thank you, Declan!

    Well, I definitely didn’t mean to suggest that the thread was “a little too vast”. But, sure, proving a theorem every now and then won’t do any harm either! 🙂

    Some quick, hopefully not useless, replies:

    1. If I’m not mistaken, coherentism is not a good fit for a truth-linked notion of justification. Yes, BonJour did say, circa 1985, that he was interested in a truth-linked notion of justification. And look where he ended up.

    2. And yes, if you want both foundationalism and a truth-linked notion of justification (by any name), externalism *is* an obvious choice. So, you’ll be a moderate foundationalist. (I find it hard to understand how one can succeed in having it all: moderate foundationalism, truth-conducive justification and internalism. But it’s easy for me to understand how an internalist foundationalist turns into a BonJour or a Fumerton.) But we were talking about the motivation for *classical* foundationalism.

  17. Just tripped over this thread …

    It strikes me that the motive for the classical position is that in order for something to be /evidence/ it needs to be a candidate basis for /intersective learning/: one needs to throw out all possibilities incompatible with the evidence as no longer to be taken seriously as candidates for actuality. That’s a pretty bold move: once thrown out, the possibility can’t come back again (since learning is intersective).

    In order to incur no risk of throwing out the actual world (and therefore plunging irrevocably into error), the content of any particular piece of evidence would have to be true. If we think of the receipt of evidence along the lines of the acceptance of a sentence, the sentences would have to be /analytic/: true whenever accepted.

    Two familiar sorts of analytic sentence are unsuitable for the role of evidence: the ‘2+2=4’ type because its content is the necessary proposition (so that it is uninformative); the ‘I am here now’ type because its content, though contingent, is opaque (so that what it says is not really /evident/).

    My paper ‘There it is’ talks about a third sort: sentences which are ‘Lagadonian’ in the sense that their referent is part of their orthography. So in ‘this is thus [looking at a red thing]’ for instance, the ‘thus’ has redness itself as part of its orthography. In accepting this sentence, one would be assured of the presence of something red.

    Now of course since the language we think in is not Lagadonian (perhaps a sense in which perception is ‘nonconceptual’), there will need to be some translation of the Lagadonian language into this language. There we get the risk of ‘coding errors’: Hesperus/Phosphorus type confusions. My view is that this is what happens when we are taken in by hallucination.

Leave a Reply

Your email address will not be published. Required fields are marked *