2. TWO GETTIER CASES FOR JUSTIFICATION

I will now describe two (hypothetical) cases in which a subject is in a Gettier condition with respect to justification. Both cases involve veridical hallucination. The first case is a case of a true belief, the second is a case of a false belief. Therefore, the second case is not a Gettier case for knowledge.

To begin with, in a case of veridical hallucination, as I understand it, a subject undergoes a perceptual experience whose perceptual representational content is correct (satisfied). We can assume that it is correct by mere luck. It is a mere coincidence that the way things appear to the hallucinating subject is indeed a way things are. I will assume, quite uncontroversially, that a hallucination is not a perception, but involves a perceptual experience. So things appear to be a certain way to the subject. We can also suppose that the subject is not aware of undergoing a hallucination, but thinks that everything is normal or ordinary. So the two cases are cases of unrecognized veridical hallucination.

Consider the first case. Call it the ‘tomato case’. Suppose that Daniel has taken some drug and is hallucinating. His hallucination involves a perceptual experience as of a red, tomato-shaped object in front of him. So there appears to be a red, tomato-shaped object to him. In fact, and by mere coincidence, there is such a red, tomato-shaped object in front of him. So the representational content of his perceptual experience is correct (satisfied). But he is not perceiving. He is undergoing a veridical hallucination. The right kind of (causal, presumably) connection between the red tomato and his perceptual experience is missing.Footnote ¹⁴

We can suppose that in normal conditions for visual perception, Daniel is quite good at perceiving the shapes and colours of objects like tomatoes correctly – he possesses ordinary, normal visual perceptual capacities with respect to these kinds of properties and objects. Now, however, he is not in normal conditions, since he has (unknowingly) taken a drug which induces hallucinations in him. Nevertheless, he is exercising his normal abilities to form perceptual beliefs on the basis of perceptual experiences. Therefore, by exercising these abilities, he forms the belief that there is a tomato in front of him. And the belief is correct (true). The belief is thus not just a shot in the dark. Rather, it is formed in a rational way. It is a rational belief.Footnote ¹⁵ Only because the conditions for exercising the relevant perceptual abilities are so bad is it the case that his perceptual experience is merely luckily correct. (In normal conditions, we can assume, his perceptual experiences are not merely luckily correct but very reliably correct.) But because the fact that there is a red tomato-shaped in front of Daniel is evidence for the presence of a tomato in front of him, Daniel's belief has supporting evidence. It is supported by real evidence, it is ‘evidenced’, as I will call it. And Daniel is in possession of this evidence, since he represents it perceptually. Things appear to him to be a certain way in his perceptual experience, and things really are that way. So the way things are has entered into his mind. And this way things are is really evidence, or a reason, for his belief that there is a tomato in front of him.Footnote ¹⁶ Nevertheless, I submit, Daniel's belief is not justified. It is rational and evidenced (and true), but not justified. Therefore, in the tomato case Daniel's belief is a kind of ‘Gettier case for justification’.

Of course, one could object at this point by saying that unless Daniel genuinely perceives the relevant fact he is not really in possession of the evidence fact. And veridical hallucination does not amount to genuine perception. One could hold that only knowledge or perception of facts counts as possession of evidence.

I will come back to this objection in the next section. Suffice it to say for the moment that according to the objection, one loses the ability to distinguish between lucky possession of evidence (as in the case of veridical hallucination) and non-lucky possession of evidence (as in a case of genuine perception) and to carve out the commonality. And one would have to introduce a technical notion of ‘quasi-possession of evidence’ in order to adequately deal with all cases. But this will then be a mere reshuffling of terminology.

We could leave it at that and let the case rest on the intuition that Daniel's belief is not justified. But additional, supplementary intuitions can be evinced in order to further support the claim. First of all, there is the further intuition that a quite intimate relation holds between justification and reasons (or evidence). It is often said that a justified belief must be ‘based on a reason’. The ‘being based on a reason’ here must be understood not merely as being in accordance with evidence (or being supported by evidence), however. Rather, a justified belief must be held because of the evidence supporting it. But it seems intuitively quite clear that Daniel's belief is not held because of the evidence, since the relation to the evidence is too accidental; it is merely in accordance with the evidence. And this is not good enough for being justified. Secondly, and related to the first point, there is an intuition that Daniel's condition is not as good as it could be, epistemically. His belief formation would be clearly better if he really perceived the evidence. Veridical hallucination connects him to the reason, but genuine perception would be an epistemically better condition. So intuitively, Daniel's belief ranks second-best, epistemically. Then, however, it seems appropriate to reserve the title of justification to the higher epistemic status. Thus we can conclude, again, that Daniel's belief lacks justification. (It still deserves the title of being rational.) These considerations may stand in need of further elaboration, but initially at least, they seem to provide some further reason for thinking that Daniel's belief is not justified.Footnote ¹⁷

Let us construct a second case, the ‘wax case’, as we can call it. In this case, everything is just like before, with the exception that the real tomato is replaced by a wax tomato which has the same appearance as the real tomato. Perceptually, the wax object is not distinguishable from the real tomato. Daniel's perceptual experience, again, is hallucinatory, but veridical. There really is a red, tomato-shaped object in front of him – and that is how things appear to Daniel perceptually. He forms the same belief, namely, that there is a tomato in front of him. This time the belief is false.

Again, his belief is formed in a rational way, by exercising normal and quite good perceptual abilities. (Daniel still has and exercises normal, good perceptual abilities. Only the conditions for exercising them are bad, since Daniel has taken the drug.) And again, there is evidence for the presence of a tomato – misleading evidence, unfortunately. So Daniel's belief in the wax case is rational and evidenced. Nevertheless, it is not justified. Therefore, in the wax case Daniel's belief is again a kind of ‘Gettier case for justification’.

In the wax case, Daniel's belief is false. Thus, it cannot be a Gettier case for knowledge. The second case, therefore, shows the independence of Gettier cases for justification from Gettier cases for knowledge. The two phenomena are really different.

3. POSSESSION OF EVIDENCE AND JUSTIFICATION

In the following, I shall point out an assumption that I have been relying on, and I shall try to defend it to some extent. As already mentioned, possession of evidence is to be understood in a factive way. So if Daniel possesses evidence e, e has to obtain. What then is the relation of possession? – It seems to me, and I have been assuming in constructing the Gettier cases for justification, that correctly perceptually representing evidence is good enough for possession of evidence. Possession of evidence consist in something like ‘registering’, or ‘representing’, or ‘being aware of’ the evidence. A subject which correctly perceptually represents there being a red tomato-shaped object is in possession of evidence – namely, the fact that there is a red tomato-shaped object which is evidence for there being a real tomato. Call these evidential facts ‘perceptual facts’ (since they can directly be represented perceptually). Then, perceptual facts can serve as evidence for non-perceptual facts, such as facts about natural kinds or functional types. And by undergoing a correct perceptual experience as of a certain perceptual fact one is in fact in possession of it, in the relevant sense. One is aware of it. Whether this allows one to form a justified belief depends on whether the possession of evidence is due to the exercise of some reason tracking ability. If you are merely luckily in possession of the evidence (as in the cases of veridical hallucination) and your possession of evidence is not due to your exercise of a reason tracking ability, then you are not in a position to form a justified belief – even if you exercise a reason tracking ability. You can still form the ‘right’ belief – the belief which is supported by the evidence you possess. This belief then will be a virtuously, competently formed belief, and we can call it ‘rational belief’. And it is an evidenced belief since there really is evidence in favour of it. This then is, again, a Gettier case for justification.

Why should we think that correct perceptual representation is good enough for possession of evidence? Do we not need genuine perception (so that veridical hallucination does not count)? – One important argument is from the analogy to the doxastic case of beliefs about evidence. Intuitively, if one correctly believes that p, and the fact that p is evidence for the further fact that q, then one is in possession of evidence for q. One may luckily possess the evidence in this doxastic way, by simply luckily (correctly) believing that p. Or one may believe it in virtue of exercising a reason tracking ability. But one has the evidence, one possesses it, in either case. If one goes on and forms the belief that q on the basis of the belief that p, one can acquire a rational, evidenced belief. So correct doxastic representation is good enough for possession of evidence. Analogously, correct perceptual representation should be good enough, too.Footnote ¹⁸

For further support, one could appeal to considerations of what one is under an obligation to take into account in one's belief formation. Intuitively, if one believes that p and understands that it is evidence for q, then one is under the obligation of taking it into account in one's doxastic deliberation about whether to believe q. It does not directly matter how one came to believe that p and whether one exercised any reason tracking ability. The mere fact that one believes it puts one under an obligation.Footnote ¹⁹ If evidence is supposed to play this role, we must allow for the possibility of lucky possession of evidence.Footnote ²⁰

One qualification can be added, however. Perhaps, correct perceptual representation is not quite good enough for possession of evidence, simply because phenomenal consciousness might be missing. And there is a quite powerful intuition that only conscious (i.e. phenomenally conscious) perceptual representation counts as possession of evidence, and unconscious perceptual representation does not (such as in cases of blindsight). We can add this qualification. And still, we do not need genuine perception. Undergoing a conscious (correct) perceptual appearance is good enough for possession of evidence.Footnote ²¹

By saying that correct (phenomenally conscious) perceptual representation, as occurring in cases of veridical hallucination, is good enough for possession of evidence, I oppose certain accounts of possessing evidence, such as the knowledge account of Timothy Williamson. (This in itself does not mean that I oppose Williamson's account of justification. We will come back to this issue soon.) The requirement that one know a fact in order to possess it as evidence is too demanding. I cannot go into a discussion of Williamson's arguments here. Suffice it to say that by requiring knowledge of evidence, one excludes the Gettier cases for justification described above. The price is that possessing evidence becomes a quite demanding status. In addition, the strategy of putting higher demands on possession of evidence leaves us with some questions concerning the cases of accidentally having evidence in mind without knowing it. Being aware of the evidence in such a way does not prevent one from using it appropriately, or responding to it appropriately. (How could one use the evidence without having it?) And, plausibly, it puts one under some obligations to use, or react to, the evidence appropriately. So, it seems preferable to allow for accidental possession of evidence. The distinction between accidental and non-accidental (competent) possession of evidence can be further illuminated by the following consideration. One of our goals in belief formation is to have beliefs which have some evidence in their favour, i.e. to have evidenced beliefs. (This goal is especially salient if we do not have at hand any competence for tracking the truth directly.) But this goal is to be distinguished from the way in which we can reach it, namely, by exercising abilities to track evidence. Possession of evidence can play a role in how we arrive at evidenced belief. One important way to arrive at evidenced belief is by becoming aware of the facts which are the evidence by the exercise of some reason tracking ability. In this way, one becomes competently, non-accidentally aware of the evidence. This, then, enables one to form the right, evidenced belief (by treating the evidence as evidence for the right thing). Becoming accidentally aware of the evidence, as in cases of veridical hallucination, equally enables one to do this (i.e. to treat the evidence as evidence for the right thing). But nevertheless, the status of the result (or of the entire structure) is quite different. Only if one is competently aware of the evidence can one achieve a justified belief. Merely accidental awareness of evidence is not good enough. And after all, we have to keep in mind that there is a third case: merely forming an evidenced belief without any possession of evidence. This is the case of entirely-luckily evidenced belief. The goal of evidenced belief, thus, can be reached in at least three different ways: entirely-luckily, luckily (by lucky possession of evidence) or competently (by competent possession of evidence). It is important to keep apart the goal and the different ways of how to reach it.

Coming back to the knowledge account of possession of evidence, we can conclude that it in effect eliminates one way in which one can possess evidence and arrive at an evidenced belief, the ‘intermediate’ way of lucky possession of evidence. For, plausibly, to know the evidence is to be competently in possession of evidence. Now, competent possession of evidence is basically the right idea for an account of justification, indeed.Footnote ²² But we should not build the competence into the possession of evidence. Rather, we should build it into our account of justification, as an explicit condition. Justification, then, is the competent achievement of evidenced belief. In contrast, merely rational evidenced belief can be had by accidental possession of evidence. Competently possessing evidence, by the exercise of reason tracking abilities, puts one into a position to justifiably believe.Footnote ²³

4. THE NEW EVIL DEMON CASE: JUSTIFICATION AND RATIONALITY

One might raise the following objection – call it the ‘new evil demon objection’: You have claimed that we have the intuition that in the tomato case, Daniel's belief that p (that there is a red tomato in front of him) is not justified. But how about the case of the brain in a vat? Many have claimed that we have the intuition that the brain in a vat has justified beliefs, just as we ordinary subjects do in our friendly environment.Footnote ²⁴ Suppose that a corresponding brain in a vat – call him ‘René’ – believes that p on the basis of having the same perceptual experience as Daniel (ceteris paribus). But there is no such tomato, and no object at all, so the belief is not based on evidence, since the perceptual experience is not correct (not satisfied). But many claim that in this case – the so-called ‘new evil demon scenario’ – René has a justified belief that p. If that is so, your claim that we have the intuition that Daniel's belief is not justified lacks any plausibility. So there it is, a clash of intuitions. Surely, it would be (almost) incoherent to have both the intuition that Daniel's belief is not justified and the intuition that René's belief is justified. After all, if anything seems rather clear then it is the fact that Daniel is epistemically at least as well off as René and, indeed, that he is somehow better off. But taken that many do share the intuition that René's belief is justified, we should be highly sceptical of your claim that we have the intuition that Daniel's belief is not justified, to say the least. In other words, if we have the intuition that René's belief is justified, it cannot be true that – as you claim – we also have the intuition that Daniel's belief is not justified.

In response to this objection we can point out that we have essentially three options: We could hold that both subjects have a justified belief; or that Daniel's belief is justified and René's is not; or that both lack justification. (The fourth theoretical option that Daniel's belief is not justified whereas René's belief is justified lacks any plausibility. I will ignore it from now on.) In the following, I shall argue that we should choose the latter, third option: both subjects lack justification.

Let me make a preliminary remark before entering into the arguments. To bring in the case of René at this point raises various methodological issues surrounding intuitions and their role(s) for philosophical theorizing. I cannot go into this here, simply for reasons of space. This field of philosophy is particularly controversial and, therefore, it would be quite desirable to steer clear of any commitments. I will try to do so. But surely, at some point this will leave room for further manoeuvres. An adequate treatment of the issues raised by dealing with several intuitions about a number of cases has to wait for another occasion. Here I will be content with providing a strategy of how to argue for the third option (‘both lack justification’), or at least a sketch of such a strategy.

To begin with, consider the first option, i.e. the view that both subjects have a justified belief. A major problem for this view is that it separates justification entirely from (possession of) evidence. If even René is supposed to have a justified belief, though he lacks any evidence for what he believes (since his perceptual experience is entirely incorrect),Footnote ²⁵ then surely Daniel has a justified belief as well. But this strongly suggests that the epistemic status of the beliefs that we are talking about is entirely independent of possession of evidence. Then, however, it becomes a real question what phenomenon we are talking about. ‘Justification’ can be used as a generic notion, covering quite different kinds of epistemic status, such as responsible or blameless belief, rational belief, reasons-related belief, entitled belief and possibly more. This has been noted already at the beginning. The intuition (presented above) that Daniel's belief is not justified was meant to concern one of these, namely, the reasons-related notion of justification. Concerning this concept of justification, the reasons-related concept, my claim was that intuitively, Daniel's belief is not justified. And it seems to me that concerning this concept of justification, René's belief is not justified either. The intuition that René's belief is ‘justified’ can only be correct if it expresses some other concept of justification. I suspect that it is driven by the rational-belief concept of justification and/or the concept of responsible, blameless belief (or even praiseworthy belief). And it has been granted from the very beginning that Daniel's belief is rational (and responsible and blameless, or even praiseworthy) – and so is René's belief. So Daniel's and René's belief share the epistemic status of being rational (and responsible and blameless, or even praiseworthy). This much is intuitively given and granted.Footnote ²⁶ But our main interest here is with the specific concept of reasons-related justification. And we would mis-describe our intuition that René's and Daniel's beliefs are rational by calling them ‘justified’. Here as elsewhere, we have to be careful to describe our intuitions in the right way and not to confuse similar concepts. (As Kripke has taught us: not an easy job.) Distinguishing between rationality and justification – or between a reasons-related concept of justification and a rational-belief concept of justification (and the other specific concepts of justification) – is crucial here. And once the distinction is made, the right way to describe our intuitions is to deny justification (in the reasons-related sense) to René's belief.Footnote ²⁷

Consider the second option. Can we hold that Daniel's belief is justified whereas René's is not? Prima facie this sounds like a plausible position. After all, Daniel is in possession of evidence – in contrast to René. But on reflection, it seems that we should be more demanding on what is required for justification, and deny that Daniel's belief is justified. The reason for this is simply that, even though there is a connection to real evidence, the connection is not of the right kind in order to achieve justification. Once we admit that a connection to real evidence is needed, we should be careful to distinguish between various different relations to evidence. And it is now a further question whether mere possession of evidence is the right link to evidence. And intuitively, the right connection is not just any kind of possession of evidence. Lucky, accidental possession of evidence is not the right link. One has to have the evidence in a competent, non-accidental way. Only then does one achieve genuine justification. Possession of evidence may be a good state, epistemically speaking. But it is not the same as justification. At least one should not be rash on deciding whether purely accidental possession of evidence is good enough for justification.

The second option, therefore, is an unstable intermediate position. Once one has moved from the first option to the second, one should move on to the third one, i.e. to the view that both beliefs lack justification. It provides perfectly sufficient conceptual resources for capturing all the epistemically significant differences. René lacks evidence, in contrast to Daniel (and a corresponding ordinary, good case of genuine perception). Daniel is better off than René since he is at least in possession of some reason for his belief. But he is still not in a position for having a justified belief since he is merely luckily in possession of evidence, and not reliably and competently. These are the right, appropriate descriptions of the cases. The third option, therefore, is the best one. The conditions on justification (in the reasons-related sense) exclude both René (quite obviously) and Daniel (not so obviously). The phenomenon we are after – belief which is appropriately linked to reasons (evidence) – is a bit more demanding than one might initially have expected. Considering cases like Daniel helps to make us aware of a component or condition that is usually or normally satisfied since we are often genuinely perceiving and, thus, in competent possession of evidence. It is unsurprising that we might tend to overlook this constitutive element of justification. But Daniel's case brings it to our attention. Properly described, and without confusing different concepts, we have the intuition that Daniel's belief lacks justification.

5. JUSTIFICATION AND UNIFIED VIRTUE EPISTEMOLOGY

Finally, let me end with some remarks on the broader view of justification which arises from the just-mentioned considerations. These considerations favour a virtue-theoretic account of justification. Justification is the competent achievement of evidenced belief. One is connected to objective evidence in an intimate, non-accidental way. One arrives at an epistemic goal – evidenced belief – in a competent way, that is, by exercising a corresponding competence. The competence relevant for justification is a reason tracking ability. For knowledge, it is a truth tracking ability. But the structure is the same, abstract structure of ‘success from ability’, as Greco puts it in his account of knowledge. In the case of knowledge, the explanation for why one has formed a true belief appeals to a truth tracking competence. The ‘AAA structure’ is present, as Sosa puts it: the belief is apt, that is, accurate (true) because adroit.Footnote ²⁸ In the case of justification, the explanation for why one has formed an evidenced belief appeals to a reason tracking competence. The belief has evidence in its favour because one has exercised a reason tracking ability. This common structure is what really makes the two cases described Gettier cases. Thus, the prospects for a unified virtue-epistemology of both knowledge and justification seem promising.Footnote ²⁹