Coming out soon, J. Adam Carter, Emma C. Gordon and Benjamin Jarvis have an anthology on the “knowledge first” approach to epistemology and mind (based on the work of Timothy Williamson in his Knowledge and Its Limits). (Maybe the volume is already is out, but I could not find it on the interwebs.) Their introductory essay contains some clear and insightful explanatory summary of various knowledge-first theses (including what they take to be the central one) and discusses a central motivation for the knowledge-first approach (and what they take to be its central thesis). Here is that essay:
And here are some excerpts/summary from this text and some commentary from me (in bold).
(If there is a theme to my recent philosophical commentary here at PoT, it is the importance, in many cases, of understanding how things function (or what their function is) in order to understand better what they are.)
- (p. 1, para. 1) “Knowledge-first epistemology is (in short) the idea that knowledge per se is an epistemic kind with theoretical importance that is not derivative from its relationship to other epistemic kinds such as rationality [belief, justification, truth]. Knowledge-first epistemology is rightly associated with Timothy Williamson (2000) in light of his influential book, Knowledge and Its Limits (KAIL). In KAIL, Williamson suggests that meeting the conditions for knowing is not constitutively explained by meeting the conditions for anything else, e.g., justified true belief 1 . Accordingly, knowledge is conceptually and metaphysically prior to other cognitive and epistemic kinds. In this way, the concept know is a theoretical primitive. The status of know as a theoretical primitive makes it particularly suitable for use in making substantive constitutive and causal explanations of a number of other phenomena, including the nature of belief, the nature of evidence, and the success of intentional actions 2.”
The core idea here is simply that (a) knowledge (not anything that might in some sense constitute or compose it) has its own distinctive explanatory work to do in the epistemic realm (just what this constitutive/explanatory work is would need to be filled in). But this is consistent with knowledge being constituted (and constitutively explained) by justification (or rationality), belief, truth (along with and some final element that rules out the merely-lucky confluence of the other elements – the ‘jtb+’ approach). What it is not consistent with is the identity-reduction of knowledge to these other elements leaving knowledge itself with no explanatory work to do. If so then why endorse the more dramatic claims that are more typical of the knowledge-first approach to the metaphysics and epistemology of belief and knowledge?
- (p. 1, para. 2 & p. 2, para 1) “As just indicated, Williamson takes the view in KAIL that knowledge—considered as a kind or type—has no constituents. (This should not be confused with the view that instances of knowledge aren’t at bottom physically constituted—Williamson is, in fact, a physicalist 3 .) This negative idea seems to be that there are no further kinds that constitute knowledge when collectively instanced; there is no correct theory that identifies the kind ‘knowledge’ with some mix of distinct epistemic and cognitive kinds meeting specifiable conditions. Nevertheless, in KAIL, Williamson also offers a positive characterization of knowledge as the most general factive mental state 4 … Williamson takes factive mental states to be at least on a par with non-factive ones 7 . Moreover, with respect to the (allegedly) central mental factive state—knowing—and the central cognitive non-factive correlate—believing—Williamson is clear that the former is no less explanatory than the latter. Even if it is possible to understand knowing as a kind of “apt” believing 8 , it is also possible to understand believing as a kind of “botched” knowing 9.”
So here we have some of the stronger knowledge-first claims. Viz., (b) knowledge is metaphysically primitive (relative to other epistemic or cognitive kinds) and (c) belief can be explained (and understood) in terms of knowledge (as well as vice versa). Presumably, the mode of explanation here is not precisely constitutive. Perhaps the idea is that (c.1) both belief and knowledge are constitutively primitive (relative to all other epistemic or cognitive elements) but nevertheless are necessarily related to each other in ways (that philosophers might characterize as a metaphysical entailment relationship) such that each partially explains the other. Regarding the epistemic analogue to explanation – understanding – the idea here might be that (c.2) we can understand the concept of belief through understanding the concept of knowledge as well as vice versa. (I’m not sure, but it may be typical of knowledge-firsters, and perhaps Williamson in other moods or in other work, to claim that (c.2*) we come to have and understand the concept of belief through having and understanding the concept of knowledge.)
For the record: I’m sympathetic to (a) and (c.2*), but not so sympathetic to (b) and (c.1). What interests me more are what the authors give as the main explanatory motivation for the knowledge-first program (whether just (a) or the other claims as well). That motivation is providing a (constitutive) broadly functionalist explanation of what beliefs are in terms of what they (the, or the various, not-necessarily-belief-constituting mental or physical elements) tend to do and how they tend to do it. So on to that.
- (p. 2, paras. 2 & 3; p. 3, para. 1) “A central project within epistemology is to understand the proper assessment of belief. A central project within the philosophy of mind is to understand what a belief is. A not wholly implausible idea is that these central projects are, in fact, related. To understand better what a belief is, one needs to think about what happens when belief goes right, and to understand better what happens when belief goes right, one needs to think about what beliefs are… For some time, the dominant approach to the theory of belief has been functionalism (at least broadly construed)—so that, to a first approximation, beliefs are what they do, i.e. believing any particular proposition is largely a matter of occupying a certain role 10. Arguably, belief plays a number of roles—assertions express them, actions are based on them, topical understanding consists in them, and so forth. Consequently, the approach of understanding belief by understanding its proper assessment might begin by considering what it is for belief to go right in each of these roles… A natural suggestion… is that going right for belief is a matter of knowing. Williamson defends individual theses about the explanatory primacy of knowledge in KAIL—e.g., that it is the standard for proper assertion 11 , that it is central to the explanation of action 12 —and others have defended further theses—e.g. that it is required for topical understanding 13 . Arguably, a unifying feature of these individual theses is that the phenomena at issue are closely associated with belief—so that belief might even plausibly be at least partly constituted by its role in each case. A knowledge-first addition to this last plausible idea is that the role that beliefs play generally is parasitic on the role that they play when things go right so that the belief qualifies as knowledge: there are surfeit of ways for a belief to fail in assertion, in action, in understanding etc., but we understand how beliefs can fail in these ways by considering what happens when they don’t fail—because the subject doesn’t merely believe but rather knows.”
Okay, there is a lot here. First, there is the idea that (d) belief, as a kind, is functionally constituted (by doing a multiplicity of things, hence the different functional roles that constitute belief). Second, there is the idea that, (e) when things have functional roles they have success-conditions (with respect to both the outcomes that they tend to produce and with respect to instantiating, in the right ways, the characteristic procedures that tend to produce those outcomes) and hence are subject to broadly evaluative/normative standards. (In this sort of case, having a function might come to nothing more than physical structures in an organism being finely-calibrated in such a way that, in the type of environment that they are suited or adapted to, they tend to achieve certain results in specific, characteristic ways. Also ‘normative’ here does not mean normative in the action-guiding sense that is relevant to what we morally, not-morally, or generically should or have reason to do.) I’m quite sympathetic to both (d) and (e).
I find the suggestion that things “going right” for belief is constituted by belief being knowledge puzzling. Functional-role evaluation has two distinct elements – appropriate means/procedure and the achievement of relevant outcomes (that tend to be produced by the means/procedures). But, at least on an initial, naive reading, the suggestion seems to be that knowledge is something basic in the cognitive or epistemic realm, the achievement of which constitutes the sole sort of general success in believing. Setting aside the metaphysical claim, it is plausible that knowledge constitutes success in both ways at once and that it is this robust sort of success-for-belief that is most explanatorily important (e.g., in explaining the systematic – not merely episodic – success in the agential pursuit of aims or carrying out of plans). And this fits well with what the authors take to be the central or most important claim of the knowledge-first approach, which is pretty much just the idea that knowledge is a natural kind. But given that there are two different sorts of general success in believing – constituted by achieving truth and coming to and maintaining one’s beliefs rationally or with sufficient justification – there is no sin in splitting them up analytically and seeing what work they do. And it makes sense that one would need to do this, if not for everyday explanatory purposes, then for a more sophisticated, detailed account of what success in belief is. Of course, if truth and rationality do their own work and are necessarily correlated with knowledge, this cuts against the metaphysical thesis that knowledge is not (in part) constituted by truth and rationality/justification.
The next paragraph in the essay (not quoted above) suggests that the authors are somewhat alive to this issue. What they say may even indicate that, true to their initial characterization of what is essential and important in the knowledge-first approach (at least as an account of what belief and knowledge are, regardless of the order of concepts or understanding) and in line with my expressed sympathies, the defensible metaphysical claim is simply that knowledge is a natural kind (or that it, not belief or true belief or rational belief, is the most explanatorily important natural kind).
- The authors close the substantive part of the essay by examining two objections to the idea that knowledge constitutes the explanatorily important standard for success in belief. The first is that belief does all sorts of things and it is not clear that it is always via the same route. So perhaps there is no one thing that constitutes functional-cum-normative success in believing (and the one thing certainly is not knowledge). The authors note both that this kind of radical difference in functional role might not be true and that, if it is true, we would do well to treat the outliers as being states that do not count as beliefs (at least not in the full-blooded sense that is correlated with good explanatory individuation of types). So this objection to the explanatory thesis is not so powerful. More powerful, perhaps, is the idea that on-off all-or-nothing belief – and hence knowledge – is not explanatorily important; rather, degrees of credence are. The authors say a bit about this in relation to how some of Williamson’s theses might bear on the problem, but do not offer anything much of their own by way of a response. The threat here to the knowledge-first program is put in terms of belief, and hence knowledge, turning out to be epiphenomenal (in some sense not fully real or not as real as credence and degrees of credence).
My initial response is this: degrees of credence, like truth in belief and rationality in belief, probably do important explanatory work, relative to some things and at a certain pretty-fine-grained level of explanation. But this is consistent with belief, relative to different explanatory work that is not quite so explanatorily ambitious at the level of detail, doing this work when credence cannot (at least not for creatures like us, with the kinds of concepts we are capable of having and kinds of propositions we are capable of entertaining). The explanatory power of belief (true belief, rational belief) is consistent with the distinct explanatory and important causal/explanatory work that knowledge does. Of course, this is at best an implied and schematic defense of each of the explanatory/natural-kind theses about belief (as against credences) and knowledge (as against belief, true belief, rational belief). And of course the metaphysical issues about the priority relationships between properties/kinds must be carefully distinguished from the priority relationships between concepts (or the order of understanding). I suspect that, with regard to the central theses here concerning the properties/kinds (and the relation to doing important explanatory work), the authors would be in broad sympathy with my answer to the worry that the reality and fundamentality of degrees of credence rendering belief, and hence knowledge, epiphenomenal. If “reality” (or degrees of such) is determined by or correlated with there being good explanatory work that gets done, then there is a clear strategy, and a plausible one, for maintaining that both on-off all-or-nothing belief and knowledge are real elements that do important work in explaining such things as truth or accurate representation, achieving scientific understanding, systematic success in achieving agential aims (and probably much more).