Book Review

Essays in Philosophy
A Biannual Journal
Vol. 2 No. 1

The Opacity of Knowledge

Abstract Here is a common ‘intuition’ that you’ll often find expressed regarding the epistemological externalism/internalism distinction. It is the thought that epistemological internalism, whatever its other faults, at least leaves the possession of knowledge a transparent matter; whereas epistemological externalism, whatever its other merits, at least makes the possession of knowledge opaque. It is the status of this view of the externalism/internalism contrast that I wish to evaluate in this paper. In particular, I argue that on the most credible interpretation of this ‘transparency’ thesis it is in fact inconsistent with even a minimal version of epistemological internalism. I conclude that knowledge is opaque on any plausible construal of knowledge, and consider some implications that this result has for the contemporary epistemological debate.

It is now part of the conventional wisdom in epistemology to argue that epistemological externalism, whatever its other virtues, has the huge drawback of leaving the possession of knowledge an opaque affair. In contrast, it is often cited as the main virtue of the internalist position, despite its other vices, to claim that it at least leaves the possession of knowledge a transparent affair.¹ It is this comfortable conception of the externalism/internalism debate in epistemology that I wish to challenge here. Before attempting to spell-out what this esoteric notion of 'transparency' is that so eludes an externalist treatment, I shall begin by describing, in the least contentious terms as possible, what the externalism/internalism distinction at least minimally consists in.

Despite being an integral part of the contemporary epistemological debate, the exact nature of the epistemic externalism/internalism distinction has always eluded a sharp characterisation. Nevertheless, a consensus of sorts has emerged regarding how, in essence, this contrast should be formulated, and this minimal description should suffice for our purposes here. One finds a neat (and recent) expression of this view - what I shall henceforth call the 'Standard View' - in the following quotation from John Greco where he characterises the contrast insofar as it applies to justification:

Internalism is the position that the conditions for justification must be appropriately internal to the knower's perspective. Roughly, something is internal to S's perspective so long as S is aware of it or could be aware of it merely by reflecting. Externalism is simply the denial of internalism, holding that the conditions for justification need not be within the knower's perspective. [Greco 2000, p. 181]

This conception of the internalist position is also echoed by James Pryor in a recent survey article of contemporary epistemology, where he characterises the core internalist thesis as follows:

Usefully, Pryor goes on to note what is involved in such "reflection":

a priori

In what follows I shall adopt the same reading.

The key thought underlying both quotations is that an account of justification is internalist if, and only if, that account demands that the conditions that allow one to possess such justification must themselves be reflectively accessible, with externalist accounts being all those theories that are non-internalist. We thus have an initial minimal characterisation of the Standard View of internalism regarding justification as follows:

The Standard View (about Justification)

An account of justification is internalist if, and only if, it demands that whether or not one is justified is determined by facts that one is in a position to know by reflective access alone.

Accordingly, externalism about justification shall be understood (again, at least minimally), as the thesis that a subject's belief could be justified without it being the case that the facts which determine that justification are reflectively accessible to that subject.

This formulation of epistemological internalism - (and thus, given that the negation of internalism is defined as externalism, of the internalism/externalism contrast itself) - ought to be widely accepted. It is, after all, implicit in the writings of a number of prominent epistemologists. Roderick Chisholm [1989, p. 7] has argued, for example, that "[t]he concept of epistemic justification [...] is internal [...] in that one can find out directly, by reflection, what one is justified in believing at any one time." In a similar fashion, Robert Audi [1998, pp. 231-2] has maintained that internalism about justification suggests "that justification is grounded entirely in what is internal to the mind, in a sense implying that it is accessible to introspection or reflection by the subject." We thus have good grounds for thinking that this minimal conception of the Standard View ought to be, at least relatively, uncontentious. Whilst different exponents of the internalist doctrine will spell-out the details of their position in distinct ways, this core conception of the position ought to remain as a common denominator.

In order for this characterisation to be of dialectical interest to the current discussion, however, it had better be the case that justification is a substantive epistemic property, for otherwise one could accept an internalist account of justification whilst simultaneously arguing that it is not justification that it central to knowledge but rather a different notion entirely that is more in line with an externalist account. Accordingly, if we are to mark a definitive epistemic contrast here, then we had better extend our understanding of the Standard View so that it applies to knowledge as well.

A natural way to do this would just be to stipulate that justification is whatever it is that turns true belief into knowledge, so that the Standard View about knowledge would simply be that an internalist account of knowledge demands that an internalist conception of justification is both necessary and sufficient (with true belief) for knowledge. The problem with this 'tripartite' suggestion, however, is that it would fall foul of the famous Gettier [1963] counterexamples to knowledge, which indicate that knowledge cannot just be (internalistically) justified true belief. A more plausible suggestion would therefore be to construe the internalist as demanding that the possession of internalist justification is at least necessary, along with true belief, for knowledge, with externalists demurring from that claim. In this way, we ensure that the debate between externalists and internalists is at least centred upon a substantive epistemic notion.

We thus get the following minimal formulation of the Standard View of internalism about knowledge:

The Standard View (about Knowledge)

An account of knowledge is internalist if, and only if, it demands that the possession of justification, internalistically construed, is necessary for knowledge.

Accordingly, we shall understand the externalist position regarding knowledge to minimally consist in the rejection of the thesis that the internalist conception of justification is necessary for knowledge. This position is equivalent, in effect, to the claim that an agent could know a proposition, p, whilst lacking internalist justification for his belief in p.

There are a number of advantages to characterising this distinction in this way. On the one hand, we can account for the fact that even internalist epistemologies can allow external conditions (over and above the truth condition) to play a significant role in the determination of the possession of knowledge. Since, as just noted, Gettier-type counterexamples reveal that mere (internalist) justification is not enough to distinguish knowledge from mere true belief, so this external component of the view is essential (I shall say more about this below).

On the other side of the equation, this formulation of internalism is also able to account for the fact that an externalist epistemology can allow a role for internal notions to play without thereby undermining its externalist credentials. Although only an epistemic internalist would regard the possession of such a property as being essential to knowledge possession, it does not thereby follow that the externalist must deny that such a property is at least pivotal to knowledge possession in general.² In order to see this, one need only recognise that externalists about knowledge can, and often do, allow 'internal' epistemic notions, like the internalist conception of justification, to play an important function in their epistemology. Robert Brandom, for example, whilst endorsing what is clearly (and self-consciously) an externalist position, holds that the "Founding Insight" of externalist epistemologies is that

internalist

Note the caveat here: "at least in some cases". The claim is thus not that an internalist notion of justification is inconsistent with an externalist epistemology, but rather that there exist instances in which an agent has knowledge even though the internalist rubric in this respect has not been met. In this way, Brandom can allow an internalist epistemic notion like justification to play a central role in his epistemology whilst simultaneously eschewing any commitment to the idea that the satisfaction of such an internalist condition is a precondition for knowledge.³

He is not the only one to view the debate in this way. For example, Alvin Plantinga's [1993] opposition to epistemological internalism does not consist in a rejection of the internalist notion of justification per se, but rather of the claim that such a notion could perform the sort of 'threshold' role that a fully-fledged epistemic property (which he terms, generically, 'warrant'), could play. That is, the internalist notion of justification may well be important to epistemology, but it is neither necessary nor sufficient for knowledge. In a related fashion, a number of externalists, among them Alvin Goldman [e.g. Goldman 1988], have found room in their epistemology for an internalist notion of justification.⁴

Given that this is the case, we can therefore distinguish between two sorts of knowledge operator. First, take the simple 'K' operator to be indifferent to the internalism/externalism dispute, so that it does not pick out the sort of knowledge that is at issue (the motivation for this will become apparent in a moment). We can then distinguish the simple 'K' operator from a different knowledge operator that specifically concerns knowledge as understood by the internalist where it essentially incorporates the internalist conception of justification. We shall symbolise this operator as 'K_I [ų]', where this says that an agent has internalist knowledge of the proposition, ų.⁵ Of course, given the preceding discussion, it follows that the internalist will think that all knowledge is internalist knowledge, and thus will affirm the following thesis:

Internalism
("ų) (K [ų] ® K_I [ų])⁶

In contrast, the externalist will demur from this thesis, arguing instead that there exists some knowledge that does not meet the internalist rubric:

Externalism
($ų) (K [ų] & ¬ K_I [ų])

Such, then, is the battleground upon which internalists and externalists meet - regarding the epistemic status of an agent's beliefs in cases where, although he lacks internalist knowledge, the externalist maintains that he knows nonetheless.

Consider, for instance, the famous example of the chicken-sexers.⁷ These are people who, simply by being raised around chickens, are able to reliably determine the sex of a chick. Typically, the sexers tend to assume that they must be seeing or touching something distinctive, but tests have shown that their actual means of distinguishing male from female chicks is olfactory. So even though these sexers have a highly reliable ability, they nevertheless tend to have false beliefs about how that ability functions. Now take the example of the 'naļve' chicken-sexer. This is a person who has this highly reliable ability but not only has false beliefs about how he does it but also fails to have sufficient reflectively accessible grounds for believing that he is reliable in this respect. Let us say, for example, that he has just never ascertained what his success ratio is.

The question we now need to ask is whether or not the 'naļve' chicken-sexer knows the sex of the chick, since it is on this point that the externalist/internalist divide should open up, with externalists tending to answer in the affirmative and internalists in the negative. On the one hand, the externalist will tend to maintain that this is a stable reliable ability that exhibits a sensitivity to the truth not only in the actual world but also in counterfactual circumstances, and thus ought to qualify as an instance of knowledge. The internalist, in contrast, will be inclined to maintain that the subject's lack of reflectively accessible grounds is fatal to the epistemic status of his belief. Sure, he has a handy skill, he might say, but he does not know the sex of the chicks as a result of it.

In effect, the dispute is over the necessity of the internalist conception of justification to knowledge, just as our formulation of epistemological internalism above suggests. We thus have what ought to be an uncontroversial account of the heart of the internalism/externalism distinction in terms of the necessity, or otherwise, of the internalist justification condition for knowledge, where this condition, in turn, demands reflective access to the factors which make it such that one is justified. With this account in mind, let us return to the issue of the putative transparency of knowledge on the internalist theory.

The question we now need to grapple with is how to characterise this enigmatic notion of transparency that is meant to be an integral feature of internalist epistemologies whilst notably absent from externalist accounts. Perhaps the most natural way of understanding transparency is in terms of the claim that the subject is always in a position to determine, by reflective access alone, that he possesses the knowledge that he takes himself to have. After all, one intuitive way of conceiving of the idea of transparency is as the claim that one is always in a position to tell, without requiring any further empirical investigation for example, what the nature of one's epistemic position is.

Moreover, one can see that the complaint against epistemological externalism is precisely that by making mere external conditions sufficient for knowledge (at least in certain cases), it therefore makes knowledge opaque in the sense that one is not always in a position to reflectively determine that one has it. Take the naļve chicken-sexer for example. He may well know (let us say), but he does not have adequate reasons for believing that he knows and thus his possession of knowledge in this respect is not something that is reflectively available to him.⁸

We shall call this the 'Transparency' thesis and characterise it as follows:

Transparency

An account of knowledge is transparent if, and only if, it characterises knowledge such that if an agent has knowledge then he is in a position to know via reflective access alone those facts that determine that he knows (rather than just truly believes).

As I shall now show, however, this thesis is unable to capture the idea that internalist knowledge is transparent because no plausible internalist account of knowledge could be consistent with it. In order to see why, we must look at the Gettier counterexamples to the classical tripartite account in more detail.

Consider the following standard Gettier counterexample to the tripartite account. Every morning our hero John comes down to breakfast around about the same time, 8.25 am, and forms his belief about what time it is by looking at the large grandfather clock he has in the hall. This clock has been working perfectly for over 100 years, let us say, and John has excellent grounds for believing that this is so, and also for believing that it is working on this particular morning. Nothing seems amiss with it and, indeed, it seems to be telling the right time, 8.25 am (let us stipulate that he has independent reasons for thinking that this is roughly the right time). He thus has a belief that the time is 8.25 am that possesses impeccable justification. Moreover, the belief is true - it is 8.25 am. Nevertheless, however, John does not know what the time is because, unbeknownst to him, the clock has in fact stopped - it stopped at 8.25 the previous morning. John thus lacks knowledge despite having a justified true belief. The reason for his lack of knowledge is that his belief, whilst being true, is far too lucky to count as knowledge. If he had come downstairs either a minute later or a minute earlier, for example, then he would have formed a false belief instead of a true one.

The moral of the story is that mere internal conditions are unable to rule-out the possibility that one's true belief was gained accidentally, and thus are insufficient to turn true belief into knowledge. Moreover, the reason why mere internal conditions are insufficient in this way is that they do not entail the truth of what is believed. Indeed, given that they are only internal conditions, it follows that they cannot be plausibly thought to entail the truth of what is believed (except perhaps in certain cases where the facts known are themselves 'internal' in the relevant respects). Accordingly, the possession of merely an internalist conception of justification will always be consistent with scenarios in which the truth condition only happens to obtain - that is, where the satisfaction of the justification condition is not appropriately related to the obtaining of the truth condition. And these are the very sort of scenarios that the exponent of Gettier-type counterexamples can exploit. As Linda Zagzebski has put the point:

of knowledge

At the very least, then, what is needed to evade such counterexamples is thus an extra external condition on knowledge that makes it such that the satisfaction of the epistemic conditions entails the truth of what is believed. One possible candidate might be, for example, that the method by which the agent forms the belief must be itself reliable, so ruling-out the stopped-clock example.

Of course, the mere addition of such an external component to knowledge does not itself prejudice the issue of where one stands on the externalism/internalism divide. After all, as was noted above, so long as one holds that the justification condition is a necessary component of knowledge, then one will still count as an internalist. Moreover, the simple fact that an internalist account must incorporate such an extra external condition is, in itself, entirely consistent with the Transparency thesis since, plausibly, the agent will be in a position to know, via reflective access alone, that the relevant external condition has obtained. In the example above, for instance, is it really plausible to imagine that a subject might (internalistically) know that the time is 8.25 am on the basis described whilst not being in a position to determine, via reflective access alone, that the clock from which he formed the belief that it was this time was reliable?

Nevertheless, beneath this point about Gettier counterexamples and the need for an external condition lies a deeper moral that internalism is unable to evade. Consider the following, more sceptical, Gettier counterexample to the tripartite account. Imagine that our hero John has a range of experiences that are perfectly consonant with him being in the green pastures of Harvard, and that on this basis he forms the belief that he is indeed currently in Harvard. With such experiential support we might legitimately suppose that his belief in this respect is sufficiently justified. Moreover, let us also grant that his belief is true - that John is indeed currently in Harvard. Nevertheless, John does not know that he is at present in Harvard because he is in fact a brain-in-a-vat, albeit a brain-in-a-vat stored at the research laboratories at Harvard University that is being 'fed' experiences of being in Harvard. He thus has a belief which, whilst true, is true in an accidental fashion. If he had been 'fed' experiences that indicated that he was in Oxford, for example, then he would have formed the (false) belief that he was there instead. Equally, had he been a brain-in-a-vat stored at the research laboratories in Oxford University, but still 'fed' the same experiences of Harvard, then he would have continued to believe that he was at Harvard even though, again, such a belief is false. As before, then, we have an example of an accidentally true belief, and accidentally true beliefs cannot constitute knowledge.

In order to eliminate cases like this (and thereby avoid radical scepticism), the internalist needs to demand some sort of extra external condition on knowledge that, however it is to be defined, at least ensures that the subject is not the victim of a subjectively undetectable cognitive illusion in this way. In so doing, the internalist can explain why the agent in this case does not know the proposition in question because such an external condition has not been met. Moreover, it is only given that this external condition excludes sceptical scenarios that the satisfaction of the epistemic conditions could possibly be such as to entail the truth of what is believed. If this were not the case then, as we have just seen, one could simply formulate a Gettier counterexample to undermine the view.

It is with this demand, however, that the crux of the problem lies. Unlike in the case of the stopped clock it is entirely implausible to suppose that the sort of anti-sceptical external condition at issue here, however it is to be ultimately cashed-out, could ever be such that the agent could be in a position to reflectively determine that it had obtained. For precisely the point about sceptical scenarios such as these is that there is (or at least need be) no subjectively detectable difference between the sceptical scenario and the corresponding non-sceptical scenario (such as between the 'real' experiences of being at Harvard, and the merely 'virtual' experiences of being at Harvard produced by neuroscientists in a research laboratory). As a result, it is impossible for the agent to have the requisite reflective access to the obtaining of this external condition. Consequently, there is a condition on knowledge which, even on the internalist account, the agent must, perforce, lack reflective access to, thus contravening the Transparency thesis.

It is important to note at this juncture that this way of understanding Gettier counterexamples is entirely consistent with an internalist epistemology, as described above. That there are external conditions on knowledge which an agent cannot (even in principle) reflectively determine to have obtained, whilst inconsistent with the Transparency thesis, is not incompatible with internalism because it does not prejudice the issue of whether or not the internalist conception of justification is necessary for knowledge. Accordingly, an internalist can perfectly consistently accept this result and therefore endorse the need for the external condition as characterised here.

Furthermore, it is also important to dispel any suspicion one might have that this point about the internalist's failure to meet the Transparency thesis only concerns a restricted class of cases that explicitly deal with sceptical scenarios. Since almost any item of putative knowledge can be subject to a 'sceptical' Gettier in this way, it follows that nearly all instances of knowledge will require such an external condition, and thus that the Transparency thesis is not met across a wide range of cases. Consequently, given that meeting this thesis is essential to transparency, we can conclude that internalist knowledge, just like externalist knowledge, is nearly always opaque.

It therefore follows from an uncontentious reading of the Gettier counterexamples that any non-sceptical version of epistemological internalism - or, for that matter, externalism - must allow that there are external conditions which are essential to knowledge but which an agent is never in a position to reflectively access. Accordingly, since the Transparency thesis therefore fails on both internalist and externalist accounts, so it follows that the possession of knowledge is opaque on either construal of knowledge.

III

Although we have here defined the Transparency thesis in terms of reflective access, another popular way of understanding this thesis is in terms of the 'Iterativity' principle. This principle demands, roughly, that knowledge always 'iterates' in the sense that if one knows, then one is at least in a position thereby to know that one knows. One can see the intuitive force of such a principle. If one's knowledge is transparent then, seemingly, it must follow that one is always in a position to tell that one has it and thus know that one has it. Accordingly, or so the thought runs, the transparency of knowledge on the internalist account is illustrated by the fact that it is an a priori truth that one's knowledge iterates; whereas the opacity of knowledge on the externalist account is illustrated by the fact that, as Richard Fumerton [1995, p. 174] has put it, it is "merely a contingent question as to whether we have knowledge of knowledge". An investigation into the putative transparency of internalist knowledge would thus be incomplete without an account of how the Iterativity principle figures in this debate.

Of course, there are those who argue that the principle of Iterativity is incoherent anyhow, and thus that no epistemological view could be regarded as entailing it. Consider, for example, the following passage from Greco, where he argues that the Iterativity principle

[...] cannot be right, since it leads directly to total skepticism; if the principle were correct, then I could not know even that I exist, and no matter what theory of knowledge is correct. This is because, according to the principle, I know that I exist only if I can know that I know I exist. But by a second application of that principle, I know that I know that I exist only if I know that I know that I know that I exist. Kp implies KKp, which in turn implies KKKp, which implies KKKKp, which implies KKKKKp, and so forth. But sooner or later I will reach propositions which I cannot even grasp, much less know. But of course I do know that I exist, and so the principle does not state an actual condition on knowledge. [Greco 2000, p. 183]

Although there certainly is a prima facie worry here, I think we can safely side-step this concern in this context for two reasons.

First, because it is not essential to the argument that I present below that Iterativity should be understood without any restriction on its application. Indeed, all that is required is that Iterativity be construed such that first-order knowledge always entails second-order knowledge, not also that knowledge should iterate to the nth degree, which is the feature of Iterativity that Greco's objection depends upon.

Second, because there is, in any case, cause to wonder whether the validity of rules of epistemic logic - such as, putatively, Iterativity - should be subject to censure on purely contingent psychological grounds. As Crispin Wright has expressed the point:

It is [...] not a relevant objection to Iterativity that the knowing subject might simply lack the appropriate second-order belief. [...] The acceptability of proof-theoretic rules for knowledge, and indeed for epistemic operators in general, does not hold out hostages to psychological contingency. [Wright 1991, p. 92n]

With these points in mind I shall thus take it that it is at least possible for the Iterativity principle to form part of a well-formed internalist epistemology.⁹

Since we are currently concerned with the properties of an internalist theory of knowledge, it ought to be the case that the knowledge which iterates on such a theory is itself internalist knowledge. We can thus express the particular Iterativity principle that we are interested in more formally as follows:

Internalist Iterativity

I_KI: ("ų) (K [ų] ® K_I [K_I [ų]])¹⁰

There certainly does seem to be a very close connection between Iterativity and the externalism/internalism distinction. For example, a number of prominent internalists in the theory of knowledge, such as H. A. Prichard [1950] and Chisholm [1982], have endorsed this principle.¹¹ Moreover, it has been a common externalist doctrine to hold that Iterativity - when not indexed throughout to internalist knowledge of course - should fail. One finds such a position advocated in the writings of such leading externalists as, for example, Fred Dretske [1971] and D. M. Armstrong [1973].

Indeed, some have argued that the epistemological distinction between internalism and externalism is entirely recoverable in terms of the relation that these two doctrines bear to this principle. As Michael Williams has expressed the point:

in contrast to internalist theories of knowledge

According to this view, whereas internalism entails Iterativity (or, more specifically, I_KI) - and thus, we might, think, the transparency of knowledge - externalism entails the negation of Iterativity (at least in its non-internalist guise) - and thus the opacity of knowledge.¹²

As I shall now show, however, this conception of the internalism/externalism debate is faced with a dilemma. On the one hand, insofar as I_KI can be understood such that it captures the idea that knowledge is transparent, then it must be invalid. On the other, insofar as I_KI is understood such that it evades this particular objection, then it is unable to capture the thought that knowledge is transparent. In order to see this, we need to consider the repercussions that our previous discussion of the nature of internalist knowledge at first-order has to the nature of internalist knowledge at second-order.

Presumably, the thought underlying the thesis that transparency is encapsulated in Iterativity is that if one knows then one knows that one knows in the sense that one is able to tell, by reflective means alone, that one knows. The problem is that, as we saw in the last section, insofar as we have a plausible internalist account of knowledge, then there must be cases in which there is a condition on knowledge at first-order that the subject is unable to know has obtained via reflective means alone. Given that this is so, however, then that a subject has iterated second-order internalist knowledge will not guarantee that the subject has reflective access to the conditions under which he (internalistically) knows at first-order. Accordingly, the presence of second-order knowledge, even on the internalist account, cannot ensure the transparency of knowledge. And if that is so, then the prospects for I_KI capturing the idea that internalist knowledge is transparent are non-existent.

Of course, one could just stipulate that I_KI should be understood such that the second-order knowledge in question incorporates reflective access to all the conditions under which first-order knowledge is achieved, thereby guaranteeing transparency. (This is, presumably, how a defender of the view that transparency is encapsulated in I_KI understands the principle). On this construal, however, I_KI is now completely implausible. Since, as we have seen, internalist first-order knowledge must incorporate a non-reflectively accessible external condition, it follows that any construal of Iterativity that has it demanding such reflective access must be inconsistent with the internalist view.

So either I_KI is possibly true, in which case internalist knowledge is no more guaranteed to be transparent at second-order than it is at first-order. Or second-order internalist knowledge is understood such that it is guaranteed to be transparent, in which case I_KI must, on the internalist view, be false. Either way, then, one cannot capture the thesis that internalist knowledge is transparent via I_KI, and thus one cannot argue for the transparency of internalist knowledge on this basis. So though it might be true that one could recover the externalism/internalism distinction in terms of this principle, that this is so would have nothing essentially to do with the transparency of knowledge. Moreover, this result emphasises the fact that we need to be careful when dealing with Iterativity principles not to read more into the 'K' operator at second-order than we would at first-order.¹³

That knowledge is opaque on even the internalist account also affects the way another pivotal epistemic principle should be understood. This is the so-called 'Closure' principle that knowledge is 'closed' under known entailment, which states, roughly, that if a subject knows a proposition, ų, and this proposition entails a second proposition, µ, then that subject also knows the second proposition, µ. Of course, since we are trying to capture this principle in terms of an internalist framework, it needs to be understood such that knowledge is indexed throughout to internalist standards. We can thus express the variant of the Closure principle that we are particularly interested in more formally as follows:

Internalist Closure

C_KI: ("ų) {(K_I [ų] & K_I [ų ® µ]) ® K_I [µ]}

The status of the Closure principle has elicited a great deal of discussion ever since it was explicitly denied (in its non-internalised form) by Dretske [1970] and Robert Nozick [1981]. Given that both of the accounts of knowledge put forward by these philosophers are externalist accounts, it is perhaps no surprise to find that a close correlation has formed in the philosophical imagination between epistemological externalism and the denial of Closure. Indeed, there is good reason for thinking that the two theses are closely linked. After all, a common externalist motif is that externalism allows an agent to know a proposition, because his belief is appropriately reliable, even whilst lacking any knowledge that the method via which he knows is itself reliable. (Indeed, one finds one expression of this particular construal of externalism in the quotation from Williams that we saw above, where he writes that externalist theories allow us to know things "even when we do not know that the appropriate reliability conditions are met"). Accordingly, one might know a proposition, know that it entails the obtaining of a certain 'reliability' condition, and yet fail to know that the reliability condition has obtained. And since this is itself a counterexample to Closure - at least on a non-internalised construal - so this externalist motif directly counts against the adoption of this principle.¹⁴

So just as many find it intuitive to suppose that one can recover the externalism/internalism distinction in terms of the status of Iterativity, similarly it is also common to think that one could do the same with Closure. If you accept it, then you must be an internalist; whereas if you reject it, then you must be an externalist. Moreover, the status of this principle is closely tied to the issue of transparency because, intuitively, if one is unable to tell that the known consequences of what one knows have obtained, then that one knows cannot be an entirely pellucid affair. Our previous discussion should, however, give us cause to doubt this neat view of the contrast.

One of the morals of the preceding discussion is that any plausible account of internalism must be such that it allows for the obtaining of certain conditions to be essential to knowledge possession even though the subject is unable, in principle, to reflectively determine that those conditions have been met. Given that this is so, however, then one should expect it to be the case that there are known consequences of the truth of what an agent (internalistically) knows that that agent is unable, in principle, to (internalistically) know.

As a case in point, consider a standard sceptical scenario. An agent may well (internalistically) know that he is currently seated at his computer, because his belief is not only true but he also has sufficient (internalistic) justification and whatever other external condition which needs to obtain has obtained. An entailment of the truth of this known proposition, however - an entailment which, moreover, we might legitimately suppose that the agent knows - is that he is not currently a brain-in-a-vat (since a brain-in-a-vat does not sit anywhere). But since the agent could not have, even in principle, sufficient reflectively accessible grounds for believing that he is not a brain-in-a-vat, it follows that this known consequence must be itself (internalistically) unknown.

On any plausible account of internalist knowledge, therefore, C_KI must founder, and thus any attempt to represent the transparency of internalist knowledge in terms of C_KI must be doomed to failure. As with Iterativity, then, whilst it might be the case that one could recover the externalism/internalism distinction in terms of the Closure principle, it would not thereby follow that this result would hold out any moral regarding the transparency, or otherwise, of knowledge.

Contrary to a great deal of received wisdom on this topic, we have seen that it is actually the case that knowledge is opaque no matter what epistemological theory one endorses. A consequence of this result is that it straightforwardly invalidates those objections to externalist accounts of knowledge that rest upon that claim that they make knowledge opaque, whether these objections are directly stated in terms of the Transparency thesis or indirectly in terms of the principles of Iterativity or Closure.

Consider, for example, the following passages from Fumerton:

tempting

level

may

... have knowledge of

constitute knowledge

¹⁵

In effect, the complaint that Fumerton is giving expression to here is that externalism allows that there are certain conditions on knowledge that we are unable to reflectively determine have obtained, or, in other words, that externalism leaves the possession of knowledge opaque. Indeed, Fumerton is more explicit about the focus of his objection when he goes on to write that

[...] the main problem with externalist accounts, it seems to me, just is the fact that such accounts [...] develop concepts of knowledge that are irrelevant. [...] The philosopher doesn't just want true beliefs, or even reliably produced beliefs, or beliefs caused by the facts that make them true. The philosopher wants to have the relevant features of the world directly before consciousness. [Fumerton 1990, p. 64]

Presumably, to argue that externalist accounts of knowledge are problematic because they fail to demand that the relevant facts should be "directly before consciousness" is simply to complain that such theories deal in external conditions on knowledge that are not reflectively accessible by the subject. Insofar as this is a problem for externalist accounts of knowledge, however, then, as we have seen, it is a problem that is equally applicable to internalist theories as well. Accordingly, if there is a legitimate worry about the inscrutability of knowledge being expressed here at all, then it is a worry that it is incumbent upon all epistemologists to share.

Fumerton is not the only one to put forward objections to externalism that run along these lines, though he is perhaps the most explicit about what the complaint that he is giving voice to amounts to.¹⁶ For example, a similar argument against externalism seems to be implicit in the following passages from Barry Stroud:

¹⁷

It is difficult to understand Stroud's objection here if it is not to be construed along similar lines to that found in the passages from Fumerton cited above.

For example, the particular problem with externalist approaches that Stroud is giving a lyrical expression to cannot be that they do not give us a "philosophical" account of our knowledge of the world (they manifestly do). Moreover, neither can it be that such theories do not allow us to know the "truths" that he writes of (which are, in effect, reliability conditions), since this is clearly an open question that is not settled on purely a priori grounds. The focus of the objection must thus be on the lack of "reasons" we have for thinking that such reliability conditions obtain. Again, however, taken literally what Stroud says is false - externalist accounts can indeed offer us reasons for thinking that such conditions obtain. As noted above, externalism is perfectly consistent with the idea that internalist notions, such as the internalist conception of justification, are fundamental (though not essential) to the possession of knowledge.

Accordingly, the objection that Stroud is proposing here must be that the reasons which externalist accounts offer us for thinking that the reliability conditions in question have obtained are inconclusive. And, of course, Stroud is perfectly right about this. Since, as we have seen, these external conditions must be such as to preclude phenomenologically indistinguishable sceptical scenarios, we could not, even in principle, have sufficient reflectively accessible grounds to conclusively indicate that they have obtained. Stroud's objection thus comes down to the same old complaint against externalist accounts (expressed by Fumerton above) that they leave the possession of knowledge an opaque matter. But since this is a difficulty that is not unique to externalist theories of knowledge, one can hardly censure externalist epistemologies on this basis.

Indeed, this way of understanding Stroud in this respect would also explain why he argues that externalism does not permit us to know that the reliability conditions have obtained. If one interprets the 'know' in question here so that it demands reflective access to all the facts which determine one's knowledge, then such knowledge is clearly lacking on the externalist account, just as Stroud says. But, as argued above, to understand knowledge in this way is to do far more than merely give it an internalist reading. Rather, what Stroud is doing here is adopting a construal of the operator that no epistemologist, of either persuasion, could consistently endorse. Stroud is thus assessing externalism relative to unreachable standards that no epistemology could fulfil.

It thus appears that a certain conception of the externalism/internalism debate - and, relatedly, a certain line of critique regarding epistemological externalism - is based, either explicitly or implicitly, upon a faulty view of what an internalist epistemology can reasonably be thought to offer. This is not to say, of course, that internalism should be abandoned for lacking this one merit, important though the presence of such a property seems to be for certain commentators. Rather, the point is that a mature debate between epistemological internalists and externalists needs to recognise that there are certain theoretical virtues that no plausible account of knowledge can be thought to provide, and hence be willing to proceed on this chastened basis.¹⁸

Duncan Pritchard
University of Stirling

Bibliography

Alston, William P., 'Levels Confusion in Epistemology', Midwest Studies in Philosophy, 5, pp. 135-50, (1980).
----------, 'Internalism and Externalism in Epistemology', Philosophical Topics, 14, pp. 179-221, (1986).
Armstrong, D. M., Belief, Truth and Knowledge, (Cambridge: Cambridge University Press, 1973).
Audi, Robert, Epistemology: A Contemporary Introduction to the Theory of Knowledge, (London, Routledge: 1998).
Bird, Alexander Philosophy of Science, (London: University College London Press, 2000).
Bonjour, Laurence, 'Externalist Theories of Justification', Midwest Studies in Philosophy, 5, pp. 27-52, (1980).
----------, The Structure of Empirical Knowledge, (Cambridge, MA: Harvard University Press, 1985).
Brandom, Robert, Making It Explicit, (Cambridge, MA: Harvard University Press,1994).
----------, 'Knowledge and the Social Articulation of the Space of Reasons', Philosophy and Phenomenological Research, 55, pp. 895-908, (1995).
----------, 'Insights and Blindspots of Reliabilism', The Monist, 81, pp. 371-92, (1998).
Brueckner, Anthony, 'Unfair to Nozick', Analysis, 51, pp. 61-4, (1991).
Chisholm, R. M., 'Knowing that One Knows', reprinted in his The Foundations of Knowing, (Minneapolis: University of Minnesota Press, 1982).
----------, Theory of Knowledge, (3rd Edition), (Englewood Cliffs, NJ: Prentice Hall, 1989).
Craig, Edward, 'Nozick and the Sceptic: The Thumbnail Version', Analysis, 49, pp. 161-2, (1989).
----------, Knowledge and the State of Nature: an Essay in Conceptual Synthesis, (Oxford: Clarendon Press, 1990).
Dretske, Fred, 'Epistemic Operators', Journal of Philosophy, 67, pp. 1007-23, (1970).
----------'Conclusive Reasons', Australasian Journal of Philosophy, 49, pp. 1-22, (1971).
Feldman, Richard, 'Fallibilism and Knowing that One Knows', Philosophical Review, 90, pp. 266-82, (1981).
Fumerton, Richard, 'The Internalism/Externalism Controversy', Philosophical Perspectives, 2, pp. 443-59, (1988).
----------, 'Metaepistemology and Skepticism', Doubting: Contemporary Perspectives on Skepticism, (eds.) M. D. Roth & G. Ross, pp. 57-68, (Dordrecht: Kluwer Academic Publishers, 1990).
----------, Metaepistemology and Skepticism, (Lanham, Maryland: Rowman & Littlefield, 1995).
Gettier, Edmund, 'Is Justified True Belief Knowledge?', Analysis, 23, pp. 121-3, (1963).
Ginet, Carl, Knowledge, Perception and Memory, (Dordrecht: Reidel, 1975).
Goldman, Alvin, 'The Internalist Conception of Justification', Midwest Studies in Philosophy, 5, (1980).
----------, 'Strong and Weak Justification', Philosophical Perspectives, 2, pp. 51-69, (1988).
Greco, John, 'Internalism and Epistemologically Responsible Belief', Synthese, 85, pp. 245-77, (1990).
----------, Putting Skeptics in their Place: The Nature of Skeptical Arguments and their Role in Philosophical Inquiry, (Cambridge: Cambridge University Press, 2000).
Hintikka, J., Knowledge and Belief: An Introduction to the Logic of the Two Notions, (Ithaca, NY: Cornell University Press, 1962).
Lewis, David, 'Elusive Knowledge', Australasian Journal of Philosophy, 74, pp. 549-67, (1996).
Nozick, Robert, Philosophical Explanations, (Oxford: Oxford University Press,1981).
Plantinga, Alvin, Warrant: The Current Debate, (New York: Oxford University Press, 1993).
Prichard, H. A., Knowledge and Perception, (Oxford: Clarendon Press, 1950).
Pryor, James, 'Epistemology c. 1988-2000', The British Journal for the Philosophy of Science, 51, (2000).
Sainsbury, R. M., 'Crispin Wright: Truth and Objectivity', Philosophy and Phenomenological Research, 56, pp. 899-904, (1996)
Sosa, Ernest, 'Philosophical Scepticism and Epistemic Circularity', Proceedings of the Aristotelian Society, (supplementary vol.) 68, pp. 263-90, (1994).
Stroud, Barry, The Significance of Philosophical Scepticism, (Oxford: Oxford University Press, 1984).
----------, 'Understanding Human Knowledge in General', Knowledge and Scepticism, (eds.) M. Clay & K. Lehrer, (Boulder: Westview, 1989).
----------, 'Scepticism, 'Externalism', and the Goal of Epistemology', Proceedings of the Aristotelian Society, (supplementary vol.) 68, pp. 290-307, (1994).
Williams, Michael, Unnatural Doubts: Epistemological Realism and the Basis of Scepticism, (Oxford: Blackwell, 1991).
Wright, Crispin, 'Scepticism and Dreaming: Imploding the Demon', Mind, 397, pp. 87-115, (1991).
Zagzebski, Linda, 'What is Knowledge?', Epistemology, (eds.) J. Greco & E. Sosa, pp. 92-116, (Oxford: Blackwell, 1999).