1.1 Best explanation evidentialism

Conee and Feldman (Reference Conee, Feldman and Smith2008, 97) believe that ‘fundamental epistemic principles are principles of best explanation’. They suggest that what is justified for a person is propositions that are part of the best explanation of the person's experiences available to her and the conclusions of inferences that would enhance that explanation. Although they make this suggestion, Conee and Feldman (Reference Conee, Feldman and Smith2008, 98) admit ‘there are important details of this account that are yet to be developed.’ Here is a first step toward developing these details:

Explanation captures the general idea suggested by Conee and Feldman (Reference Conee, Feldman and Smith2008). Of course, it is not very helpful without an explanation of what it means for p to ‘explanatorily cohere’ with S's evidence.

Combining Explanation and Explanatory Coherence with Conee and Feldman's E yields:

BEE seems to be the sort of account that Conee and Feldman have in mind.Footnote ⁴ According to BEE, the basic epistemic principle is a principle of best explanation. Although BEE and similar accounts of epistemic support might seem attractive, they are problematic. Both Lehrer (Reference Lehrer1974) and Goldman (Reference Goldman and Dougherty2011) have exposed serious problems about this sort of account, which will be briefly considered here.

1.2 Lehrer and Goldman against BEE

In making his case against explanationist accounts of epistemic support like BEE, Lehrer uses an example that poses a significant problem for BEE. Lehrer's example involves using the Pythagorean Theorem to determine that a particular mouse is five feet from a particular owl:

Lehrer claims that this case is problematic for explanationist accounts of justification because he is ‘completely justified in his belief that the mouse is five feet from the owl’, but he ‘has no explanation of why the mouse is five feet from the owl’ (Reference Lehrer1974, 178). According to Lehrer, the justification of his belief does not depend on ‘explanatory relations’ at all. Rather, ‘it is enough that the man knows the Pythagorean Theorem, the distance to the pole, and the height of the pole, and deduces the conclusion’ (Reference Lehrer1974, 178). Assuming that relations of logical entailment are not themselves explanatory relations, it seems that Lehrer has exposed a flaw in accounts of epistemic support such as BEE.Footnote ⁵ Goldman presents a similar example involving a simple inference of arithmetic:

Goldman claims that it is clear that in his case he has justification, but explanationist accounts of epistemic support such as BEE cannot account for this justification. So long as entailment relations are not explanatory, Goldman's example presents a problem for BEE similar to that posed by Lehrer's example. In Goldman's example, <there are (at least) four animals> is a logical consequence of his evidence that there are two squirrels and two birds on his deck. However, it seems that there being four animals on the deck does not explain there being two squirrels and two birds on the deck. Thus, it again seems that there is a proposition which is justified for someone, but the proposition is not itself part of the best explanation of the person's evidence. BEE fails.

2.1 Explanationist Evidentialism

Lehrer's and Goldman's examples show that BEE cannot be the correct account of epistemic support because it is too restrictive. An adequate account must allow for more than simply an appeal to best explanations; it must also allow for logical consequence. Hence, rather than construing epistemic support as BEE, it should be construed as:

If, as seems quite plausible, relations of logical consequence are not explanatory, then despite the fact that EE is in the spirit of what Conee and Feldman suggest, it is in tension with the claim that all fundamental epistemic principles are principles of best explanation. However, given the difficulties noted above, it seems that construing all fundamental epistemic principles as principles of best explanation may be too restrictive to be accurate. Further, consideration of EE will show that it is a much more plausible account of epistemic support than BEE.

Before the merits of EE can be assessed, it is necessary to clarify three points. First, in order for p to be part of the ‘best’ explanation available to S at t, it must be the case that there is no rival (contrary) explanation with respect to p available to S at t that is as good of an explanation as, or better than, the best explanation of which p is a part. It is important to recognize that this does not require S to have a unique best explanation available to her. To see this, consider the following case: S has two explanations available to her, E₁ and E₂, both of which are equally good explanations of why S has the evidence that she does. In this case, S is justified in believing p just in case p is part of both E₁ and E₂. If p is not part of one of these explanations, then p is not part of the best explanation available to S, because there is an equally good explanation available to her that does not have p as a part. So S can be justified in believing p, even though there is not a unique best explanation available to her. What matters is that there are no explanations available to S that do not contain p and are as good as the best explanation she has available. Whether one explanation is better than another depends on which explanation is more explanatorily virtuous. The virtuousness of an explanation will depend on a variety of factors, including its explanatory power, simplicity and scope.Footnote ⁷

Now one might think that a further restriction is in order here, because there can be cases in which the best available explanation is a pretty poor explanation. If p is part of the best available explanation, but that explanation is not a good explanation, one might think that p is not justified. This is correct, if by ‘justified’ one means ‘strongly justified’ or ‘justified to the extent required for knowledge’. However, ‘justified’ in EE should be understood to simply mean that S has propositional justification that favors p over ~p. This justification can be very minimal. In fact, in cases in which p is part of the best explanation available to S at t for why S has e, and that best explanation is not a good explanation, the justification that S has for p is very minimal – it may be barely more than the justification she has for believing ~p. Recognition of the fact that the best explanation might not be a good explanation does not reveal that EE requires a further restriction, but it does bring to light a very important feature of EE. Since one can be more or less justified in believing a proposition, any plausible account of epistemic support must accommodate this fact. EE does. The quality of the best available explanation determines how much justification S has for p. P must be part of the best available explanation to be justified at all, but the better the explanation that p is a part of, the more justified p is for S.

Secondly, e should be understood to mean S's total evidence. This is important to keep in mind, because construing EE in terms of a proper subset of S's evidence would have the obvious flaw of ignoring the potential justificatory impact of defeaters. For simplicity, the examples below only focus on a portion of one's evidence. However, it is assumed in the examples that S's total evidence lacks defeaters that are relevant to the proposition being discussed.

Thirdly, it is important to say what it means for an explanation of why S has e to be ‘available’ to S at t and what it means for a proposition to be ‘available’ to S as a logical consequence of the best explanation available to S at t for why S has e. First, at t S has p available as part of an explanation for why S has e iff:

‘Seeming’ should be understood here in the sense that Tucker (forthcoming) terms the ‘Experience View’. According to this view, seemings are mental states with propositional content and a particular phenomenology – they ‘have the feel of truth, the feel of a state whose content reveals how things really are’.Footnote ⁸ Additionally, according to the Experience View, seemings are not themselves beliefs or inclinations to believe.Footnote ⁹ Rather, they are either sui generis propositional attitudes or experiences. Of course, this leaves open whether p is in fact part of the best answer to the question ‘why does S have e?’ It is plausible that S could have a mental state in which it seems to her that p is part of the best answer to the question ‘why does S have e?’, but in fact it is not. This is important, because S does not need to have infallible access to p's role in answering this question in order for an explanation to be available. Another point that is worth taking into account is that S does not have to have well-developed concepts of ‘evidence’ or ‘explanation’ in order to satisfy the requirements of having an explanation available. All that is required is for S to have the disposition to have a seeming that the truth of p is part of the best answer to the explanatory question ‘why does S have e?’; that is, ‘why does S have the evidence that she has?’ This explanatory question can be construed in a variety of ways. For example, in an ordinary case, in which S has a visual experience as of a red block, it is plausible that S has the disposition (on the basis of reflection alone) to have a seeming that the truth of <there is a red block> is part of the best answer to the question ‘why am I having this visual experience?’ This does not require S to have any concept of ‘evidence’ or ‘explanation’ beyond understanding when something is an answer to a why-question. All S really needs is the ability to understand the question ‘why do I have this?’ (where ‘this’ refers demonstratively to her evidence) and the ability to understand when something is the best answer to this question. Finally, it is important to note that in many cases, S is likely to only recognize a single answer to the relevant explanatory question. In these cases, the answer to this question is the best by default – it is the only available answer.

Secondly, p is available to S as a logical consequence of the best explanation available to S at t for why S has e iff:

It is worth mentioning that this availability condition does not require S to have well-developed concepts of ‘logical consequence’, ‘entailment’ or similar concepts. To illustrate this, consider a situation in which S has evidence that justifies the beliefs that p and that if p, then q. According to EE, S's evidence (assuming she has no defeaters) supports q so long as (1) S has the concepts required to understand q and (2) she is disposed to have a seeming that q is a logical consequence of the best available explanation of her evidence. It is plausible that S does not need the concept of ‘logical consequence’ or ‘entailment’ in order to have a seeming that q is a logical consequence of the best available explanation of her evidence; she simply has the seeming that q follows from the best available explanation of her evidence in the sense that someone who does not have the concept of modus ponens can have the seeming that if p and if p, then q are true, then q must be true.Footnote ¹⁰

Finally, it is worth noting that although EE is similar to explanationist accounts of justification put forward by others, such as Harman (Reference Harman1986) and Lycan (Reference Lycan1988), which claim that explanatory coherence among one's beliefs is what justifies those beliefs, it is different in at least one very important respect. Both Harman and Lycan restrict one's evidence to beliefs, but EE makes no restriction. Restricting evidence to beliefs is problematic because it opens the door for the ‘Input Objection’ that Laurence BonJour (Reference BonJour1985) described for coherentism. Essentially, the problem is that if evidence is restricted to beliefs, then it is possible for there to be cases in which S has justification for her beliefs (including beliefs about the external world) because of their explanatory coherence with one another, but in which she is completely cut off from experiencing the world around her. This is counterintuitive. Thus, restricting evidence to beliefs is a mistake; experiential states are evidence too.Footnote ¹¹

EE perhaps even more closely resembles the explanationist account of epistemic support put forward by Moser (Reference Moser1989). Moser's account also appeals to explanatory considerations to spell out epistemic support and avoids restricting evidence to beliefs. What is more, it avoids the problems facing BEE by allowing entailment to be part of the account of epistemic support. Despite these similarities with EE, the views are different in important ways. First, Moser's account is only in terms of occurrent mental states. That is, Moser provides an account of when one has occurrent justification for propositions that are being considered at the time. EE, however, can be applied more generally to both occurrent and dispositional mental states.Footnote ¹²

Secondly, Moser's account differs from EE when it comes to justification for entailments. According to Moser, S can have justification for believing an entailment, q, so long as (absent defeaters): (1) S is currently aware of p, (2) p is justified for S, (3) S understands q and (4) p entails q. Moser requires no awareness or apparent awareness of p's entailing q in order to have justification for believing q (propositional justification). Rather, it is simply required that p entails q. Moser only requires awareness of the entailment relation when it comes to justifiedly believing q (doxastic justification). In order to have doxastic justification for q, Moser claims that S must have de re awareness of the fact that p entails q. It seems that Moser's account makes propositional justification too easy and doxastic justification too hard. Consider an example of Feldman and Conee's (Reference Feldman and Conee2001, 4):

We can add to this example by filling in a few details: both the TA and the student are aware of the first sentence, they both have justification for accepting this sentence (they know it!) and they both understand the sentences that are logical consequences of the first sentence. Since the first sentence entails these further sentences, on Moser's account both have propositional justification for believing these further sentences. But this is implausible. After all, we are told, ‘the student is clueless’. The problem is that Moser's account does not require any sort of awareness/recognition of the entailment relations that hold between the sentences in order to have propositional justification.Footnote ¹³EE does not share this problem. According to EE, the student lacks justification for these further sentences because they are not available to her in the relevant way.

By requiring de re awareness of the entailment relation, Moser avoids the previously mentioned problem when it comes to doxastic justification, but he does so at the cost of making doxastic justification too difficult to come by. Given Moser's account, S cannot be justified in believing q unless it is actually entailed by p. This seems too strong. Further, when q is entailed by p it is not clear that S needs something as strong as de re awareness of the entailment relation that holds between p and q in order to be justified in believing q. Plausibly, S only needs a weaker sort of awareness, like the seeming that EE requires. So, despite many similarities, EE is importantly different from Moser's account of epistemic support.

2.2 Applying EE

In order to show that EE is an acceptable account of epistemic support, it will be instructive to begin by considering some cases in which it is intuitive that the agent has epistemic support for a proposition and see whether EE yields that result.

Let us begin with a simple case of perceptual experience. Intuitively, in normal circumstances (assuming S has no defeaters) S's visual experience as of a red block provides her with epistemic support for the proposition <there is a red block>. EE delivers this result. The proposition <there is a red block> is part of the best explanation of why S has an experience as of a red block that is available to S. Recall that in order for this proposition to be available to S, all that is required is for S to have the concepts required to understand <there is a red block> and for S to be disposed to have a seeming that <there is a red block> is part of the best answer to the explanatory why-question ‘why does S have e?’ on the basis of reflection alone. In this case, the relevant why-question can be expressed as: ‘why am I having this visual experience?’ or demonstratively as ‘why am I having this?’ It is plausible that even children and unreflective adults can have the disposition to have a seeming that <there is a red block> is part of the best answer to the question ‘why am I having this?’Footnote ¹⁴ Thus, it is plausible that an ordinary person will have <there is a red block> available as part of the best explanation of her current evidence in this case. Further, it is exceedingly plausible that in a normal case there will not be an equally good or better explanation (which does not include <there is a red block>) available to this person. Therefore, an ordinary person in this sort of case can have justification for <there is a red block> when she has a visual experience of a red block. Other cases of perception will come out similarly – EE will deliver the intuitively correct result.Footnote ¹⁵

EE also delivers the intuitively correct result in cases of a priori justification. If S understands the proposition <every blue book is a book>, it is plausible that S has a priori justification for believing that proposition. When S reflects on this proposition and understands its content, she has some sort of experience. One plausible account of the sort of experience that S has when reflecting on this proposition is that she has some sort of awareness of the proposition and of the relations among the conceptual components of the proposition.Footnote ¹⁶ Given this account, EE can explain the epistemic support that S has for the proposition <every blue book is a book> by noting that the truth of this proposition is part of the best explanation of her evidence. In this sort of case, S's evidence includes her awareness (or her apparent awareness) of the proposition <every blue book is a book> and of the apparent fact that the predicate of this proposition is contained in the subject. Even if the awareness that S has of this proposition and the relationship among its components is not factive, it is plausible that the truth of the proposition is part of the best available explanation for why S has this evidence.Footnote ¹⁷ So, according to EE, S's evidence supports her believing <every blue book is a book>.

A case of inductive inference is also worth looking at. When S is aware that all observed emeralds are green, intuitively, she is in a position to be justified in believing that the next observed emerald will be green.Footnote ¹⁸EE yields this result. It is plausible that in a case in which S has only observed green emeralds, part of the best available explanation of S's observations is that all emeralds are green. It is also plausible that in this sort of case S has the seeming (or is disposed on the basis of reflection alone to have the seeming) that the explanation containing <all emeralds are green> logically entails <the next observed emerald will be green>. Provided that S is so disposed, and that she understands the proposition <the next observed emerald will be green>, EE yields that she has justification for believing <the next observed emerald will be green>. Thus, EE yields the intuitively correct result in this sort of case.

Another kind of inductive case concerns the justification that one has when most, but not all, observed A's have been B. In cases where S has made a large and sufficiently diverse sampling of the A's – and a sufficiently large portion of the observed A's have been B – it is plausible that S is justified in believing <the next observed A will be B>. In this case, however, <all A's are B> will not be part of the best available explanation of S's evidence. Rather, something like <most A's are B> will be part of the best available explanation of S's evidence. This, of course, does not entail <the next observed A will be B>. However, one might think that S is justified in believing that the next A she observes will be B, because most of the A's she has observed have been B. Thus, one might worry that EE will not be able to properly account for the justification that we have in this sort of case because it cannot do so in the way that it does for cases in which all observed A's have been B.

Fortunately, there are at least two plausible ways for EE to handle this sort of case. First, as mentioned above, it is plausible that in this sort of case, <most observed A's are B> is part of the best available explanation of S's evidence. It is not unreasonable to think that the best available explanation of S's evidence in this case entails that the next A will probably (more likely than not) be B. So, EE yields the result that S is justified in believing that the next observed A will probably be B. One might argue that being justified in believing that the next observed A will probably be B just is to be justified in thinking that the next observed A will be B.Footnote ¹⁹ The reason for this is that one might agree with Feldman and Conee (Reference Feldman and Conee1985, 34 fn32) that ‘evidence which makes believing p justified is evidence on which it is epistemically probable that p is true’. As they note elsewhere, ‘what is epistemically justified is ‘likely’. This is what is commonly meant by ‘likely’ and ‘probable’' (Conee and Feldman Reference Conee, Feldman and Dougherty2011, 297). If one accepts a position like Feldman and Conee's, the fact that EE yields that one is justified in believing that the next A will probably be B means that EE yields that one is justified in believing that the next A will be B. Of course, one might disagree with this equating of epistemic justification and epistemic likelihood, and thus claim that EE does not yield that one is justified in believing that the next observed A will be B in this case. However, one might argue that this is exactly the right consequence in this case. That is, one might think that in this case, S should not believe that the next A will be B, but rather, she should only believe that it will probably be B.

There is a second way for EE to handle this sort of case. It is plausible that in this case there is some reason why observed A's are normally B. Since this is what normally happens, when an observed A is not B, something must interfere with the normal course of events. So, one may reasonably argue that in this sort of case, part of the best available explanation of S's evidence is <in all normal cases A's are B>. Additionally, it is not implausible to think that the default position is that any given situation is a normal situation. That is to say, it is not unreasonable to think that one should believe that a situation is normal unless there are reasons for thinking otherwise. Since in the case described it does not seem that S would have reasons for thinking that her situation is abnormal in some way, she should think that she is currently in a normal situation. So, it is plausible that part of the best explanation of S's evidence is <this is a normal case>.Footnote ²⁰ The propositions <in all normal cases A's are B> and <this is a normal case> entail <the next A observed in this case will be B>. Thus, EE can yield the result that S is justified in believing that the next observed A will be a B in a case where most, but not all, observed A's have been B.Footnote ²¹

A final sort of case concerns introspection. As Goldman (Reference Goldman and Dougherty2011) notes, it is plausible that one can have justification for believing that she has a tickle sensation when she does. However, one might doubt, as Goldman does, that explanationist accounts of epistemic support like EE can make sense of the justification one has for this tickle belief. According to Goldman, ‘the viability of this approach [explanationist accounts of epistemic support] depends, however, on there being introspective experiences of mental events or states of affairs distinct from the mental states that are allegedly introspected’ (Reference Goldman and Dougherty2011, 277). Goldman claims ‘I doubt there is any “intermediate” mental event that stands between the “objective” state of affairs in question (here, the tickle sensation) and the belief in this state of affairs’ (Reference Goldman and Dougherty2011, 277). Goldman maintains that in this sort of case there is nothing ‘that can function in the guise of a suitable explanandum’, so there is nothing for S's introspective belief that she is having a tickle experience to explain (Reference Goldman and Dougherty2011, 277). In Goldman's case S has justification for believing the proposition <S has a tickle sensation>, but he maintains that <S has a tickle sensation> is not part of the best explanation for why S has the evidence she does (her tickle sensation).

Although Goldman's example (and introspection in general) may at first seem to be a problem for EE, further reflection reveals that it is not. A supporter of EE should reject Goldman's claim that there is no intermediate mental state between the tickle sensation and the belief. After all, it seems that the mental state of S's attending to her tickle sensation is distinct from both the tickle sensation itself and the belief that S has a tickle sensation. To see this, consider a case in which S has a tickle sensation that ‘recedes’ from her consciousness. That is, at t ₁ S has a tickle sensation that she attends to, but at t ₂ S is distracted by a loud noise and fails to notice her tickle sensation. Later, at t ₃, S attends to the tickle sensation again. In this sort of case it is plausible that S has the tickle sensation (and the belief that she has the tickle sensation) throughout, but she only has the intermediate state of attending to the tickle sensation at t ₁ and t ₃.Footnote ²² Further support for thinking that there is such an intermediate state comes from anti-luminosity considerations. If one's mental states can fail to be luminous, in the sense that one may be unable to distinguish one kind of mental state from another, it is not unreasonable to think that there is some intermediate mental state between the state of affairs (the tickle sensation) and the belief about the tickle sensation.Footnote ²³ Given the plausibility of there being such an intermediate mental state, it is not unreasonable for the supporter of EE to claim that in the cases where S has justification for believing that she is having a tickle sensation, <S has a tickle sensation> will be part of the best available explanation of her having this intermediate mental state.

Now that EE has been described and how it works in a variety of cases explained, it is worth considering how it avoids the pitfalls of BEE exposed by Lehrer's and Goldman's original examples.

2.3 EE and the problems for BEE

Recall Lehrer's example from above, which uses the Pythagorean Theorem to determine that a particular mouse is five feet from a particular owl. Lehrer's example illustrates that one can be ‘completely justified in his belief that the mouse is five feet from the owl’ while having ‘no explanation of why the mouse is five feet from the owl’ (Reference Lehrer1974, 178). The problem for BEE is that it seems Lehrer's justification does not depend on explanatory relations. Rather, as Lehrer says, ‘it is enough that the man knows the Pythagorean Theorem, the distance to the pole, and the height of the pole, and deduces the conclusion’ (Reference Lehrer1974, 178). So, BEE seems to yield the result that Lehrer is not justified in believing that the mouse is five feet from the owl, but clearly he is justified.

Presumably, in Lehrer's case he has evidence that justifies his believing that the mouse is three feet from the flagpole, that the owl is on top of the four-foot tall flagpole and that the Pythagorean Theorem is true – as well as various other bits of relevant information (what three squared equals, what four squared equals, etc.). In terms of EE, this means that the following propositions would all be available to Lehrer as part of the best explanation of his current evidence: <the mouse is three feet from the flagpole>, <the owl is on top of the four-foot tall flagpole>, <the Pythagorean Theorem is true>, <the Pythagorean Theorem says that the hypotenuse of a triangle that has a three-foot long side and a four-foot long side is five feet long>, etc. Since these propositions are part of the best explanation available to Lehrer, <the mouse is five feet from the owl> is a logical consequence of this explanation. It is exceedingly plausible that <the mouse is five feet from the owl> is available to Lehrer as a logical consequence of the best explanation available to him at t for why he has the evidence that he does. After all, Lehrer says that he deduces this conclusion. So, EE yields the result that Lehrer is justified in believing that the mouse is five feet from the owl in this case.

Now consider Goldman's example, which involves an inference of simple arithmetic. In this case Goldman looks out at his deck and sees two squirrels and two birds. On the basis of his visual experience, he infers that there are (at least) four animals on his deck. The problem for BEE that Goldman exposes is that although this inference is justified, it does not seem that <there are four animals on the deck> is part of the best explanation of Goldman's evidence. As he says, ‘How does there being four animals explain there being two squirrels and two birds? It doesn't’ (Reference Goldman and Dougherty2011, 277). Again, this is a serious problem for BEE, but no problem at all for EE. Part of the best explanation available to Goldman for his total evidence (both his current visual experiences and his background evidence) in this case is <there are two squirrels on the deck>, <there are two birds on the deck> and <squirrels and birds are both animals>. Given this fact and the highly plausible assumption that Goldman has the concepts required to understand <there are four animals on the deck> and he is disposed (on the basis of reflection alone) to have a seeming that <there are four animals on the deck> is entailed by the best explanation available to him at t for why he has the evidence he does, EE again yields the correct result that Goldman has justification for believing <there are four animals on the deck>. Neither of the cases that made problems for BEE poses a challenge for EE.

With the superiority of EE to BEE established, it will now be useful to consider some further purportedly problematic cases as well as potential objections.

3.1 Purportedly problematic cases

Before considering specific objections that one might have to EE, it is worthwhile to briefly consider two further cases that are thought to be problematic for explanationist accounts of epistemic support. These cases, like the previous ones, come from Lehrer (Reference Lehrer1974) and Goldman (Reference Goldman and Dougherty2011). First, consider Lehrer's case:

It is unclear how exactly this case is supposed to pose a problem for any explanationist account of justification, even the flawed BEE. In this case Lehrer sees a dead man, and that – along with his evidence about how men come to exist and how those who died came to exist – gives him justification for believing that the man was sexually conceived. It seems reasonable to think that the proposition <the man was sexually conceived> is part of the best explanation of Lehrer's evidence (his perceptual experience of the dead man's body and his background evidence about how men are conceived), in which case this example poses no problem for explanationist accounts of justification in general or for EE in particular.

The final example from Goldman requires careful consideration in order to see why it fails to pose a problem for explanationist accounts of epistemic support. Goldman's case concerns preservative memory. He states, ‘If S was justified in believing P earlier, and S retains her belief in P now via preservative memory, then S is prima facie justified in believing P now’ (Reference Goldman and Dougherty2011, 277). Goldman goes on to claim, ‘like introspection, however, preservative memory lacks any type of mental experience or episode that invites explanation’ (Reference Goldman and Dougherty2011, 277). So, Goldman thinks he has illuminated a case that explanationist accounts of epistemic support, particularly those like BEE, are not equipped to handle. However, even accounts like BEE are not troubled by this example and, of course, superior accounts such as EE are not either.Footnote ²⁴

In their reply to Goldman, Conee and Feldman (Reference Conee, Feldman and Dougherty2011) offer a plausible response to this sort of example as a challenge to how S can have any evidence in support of such beliefs. They note that in cases of preservative memory by stipulation, S has not forgotten the preservatively remembered proposition. Since this is so, they claim that S ‘has the potential to bring Q to mind with the phenomenology of activating a memory, specifically, the memory that Q’ (Reference Goldman and Dougherty2011, 304). They go on to point out that S's ability to bring Q to mind in this way is not ‘merely because he has the stored belief that Q. Many of our stored beliefs are recalled as things we merely believe, not as things we know’ (Reference Goldman and Dougherty2011, 304). Conee and Feldman label this potential to bring a proposition to mind in this way in cases of preservative memory a ‘disposition to recollect’, which they explain ‘is a disposition to bring to mind the proposition as known’ (Reference Goldman and Dougherty2011, 304). They conclude that ‘there is something about his mental states at the time that provides justification for this stored belief, but the justifying is not done by any active conscious occurrence at that time. Evidentialism need not appeal only to conscious occurrences to justify stored beliefs’ (Reference Goldman and Dougherty2011, 304). So, it is plausible that S has dispositional evidence in Goldman's example. It is also plausible that the evidence that S has in this example supports her belief in the remembered proposition. Since in a case of preservative memory S has a disposition to recall the proposition remembered ‘with the phenomenology of activating a memory’, it is plausible that S also has a disposition to have a seeming that the truth of the proposition is part of the best explanation of why he recalls the proposition with that particular phenomenology (Reference Goldman and Dougherty2011, 304). So according to EE, these dispositions are sufficient for S to have justification for believing the proposition that is retained in preservative memory, so EE yields the intuitively correct result in cases of preservative memory. It is quite plausible that BEE can appeal to these sorts of dispositions to handle this sort of case as well.

3.2 Further objections

The first of the remaining objections is another from Lehrer (Reference Lehrer1974). He claims:

Of course, if Lehrer is correct in claiming that there is never a best explanation because there are always equally good competitors, then EE, and explanationist accounts in general, are in deep trouble.Footnote ²⁵ Fortunately, there is a good response to this objection. First of all, EE, and any halfway plausible explanationist account of justification, is restricted to the best available explanation. So long as the other explanations that Lehrer is appealing to are not available to S, they do not stop her evidence from providing epistemic support for propositions that are part of the best available explanation of her evidence.Footnote ²⁶ Although this is an important step toward responding to Lehrer's objection, the restriction to available explanations is not enough by itself. This is because even though S does not have any equally good rival explanations to her best explanation available to her, it might be the case that S has justification for thinking that there are such rivals for any of her best available explanations. If S has justification for this general claim about explanations, then it seems that she would have a defeater for any support that is gained by a proposition being the best available explanation of her evidence. However, it is unlikely that S has justification for believing this. For one thing, it is doubtful that most ordinary people have even considered this general claim about explanations, or that they have evidence that justifies accepting it. Further, it is plausible that there are not equally good competitors for every explanation.Footnote ²⁷ Without a good reason to think that there are always equally good competitors for every explanation, Lehrer's objection fails to pose a problem for EE.

The final objection is twofold. It concerns whether there are higher-level requirements for epistemic support. Given one's view on this, one might object to EE in one of two ways. On the one hand, some philosophers think that in order for S to have epistemic support for p, S must either have a higher-level belief to the effect that her evidence supports p or have some sort of direct awareness that her evidence supports p.Footnote ²⁸ These philosophers might object that EE makes epistemic support too easy because it does not require any such belief or direct awareness. On the other hand, other philosophers (with more externalist views of epistemic support) might think that there are no higher-level requirements at all.Footnote ²⁹ Thus, they might object that EE over-intellectualizes epistemic support because it requires S to have either (1) a disposition (on the basis of reflection alone) to have a seeming that p is an answer to an explanatory question or (2) a disposition (on the basis of reflection alone) to have a seeming that p is entailed by the best explanation available to S at t for why S has the evidence she does.

Although a full response to both of these views is beyond the scope of this article, a few things can be said in response to each objection.Footnote ³⁰ Against those who claim that EE makes epistemic support too easy by not including a higher-level belief requirement, it seems plausible that young children and unreflective adults can possess evidence that supports their believing various propositions. However, it is highly unlikely that children and unreflective adults form higher-level beliefs about what their evidence supports. It is also plausible that S's evidence can support believing a proposition for her even if she has not consciously considered the relationship between her evidence and that proposition. So, it is reasonable to think that S's evidence can support believing a proposition despite S not being directly aware of the relationship between her evidence and the proposition. Thus, the fact that EE does not require higher-level beliefs, or direct awareness of the relationship between p and one's evidence, is a strength of the account – EE makes epistemic support just easy enough.

Against those who claim that EE over-intellectualizes epistemic support by requiring S to be disposed to either (1) have a seeming that p is an answer to an explanatory question or (2) have a seeming that p is entailed by the best explanation available for why S has the evidence she does in order for p to be supported by her evidence, two points can be made. First, without this sort of requirement, accounts of epistemic justification would seem to imply that S has epistemic support for propositions that she cannot even understand. Without the higher-level requirement that S possess the concepts necessary to understand p in order for her evidence to support believing p, S could have epistemic support for a proposition that she lacks the requisite concepts to understand. It is implausible to think that one can have epistemic support for any proposition that she cannot understand. Secondly, there is reason to think that this sort of disposition is not too hard to come by. As noted above, this sort of disposition does not require one to have the well-developed concepts ‘evidence’, ‘explanation’, ‘logical consequence’ or ‘entailment’. All that is required is the ability to understand something as an answer to a why-question and the ability to have a seeming that, for example, the truth of p and if p, then q ensure the truth of q. Additionally, it is plausible that even the unreflective are good at using and recognizing explanations – even more so at recognizing propositions as answers to why-questions. In fact, there is evidence that children are not only good at evaluating explanations, but have ‘powerful and flexible theory-formation abilities’.Footnote ³¹ Young preschoolers ask explanatory why-questions and ‘almost as early as the first why-questions are the first attempts at explanations.’Footnote ³² Therefore it is not implausible to think that the dispositions required by EE do not put epistemic support out of reach of the young or unreflective.