Schaffer (2010) has recently put on the map a new kind of sceptical scenario: a scenario in which Descartes’ (1984 [1641]) all-powerful daemon ensures that one’s belief is held on a bad basis, or even no basis at all, even though it seems to one just like it is held on a good basis. Call this the debasing scenario. We can contrast it with the more familiar deception scenario, in which the daemon ensures that one’s belief is false even though it seems to one just like it is true. Schaffer argues that the possibility of the debasing scenario makes available a new and interesting kind of scepticism: Debasing Scepticism, which we can contrast with the more familiar Deception Scepticism made available by the possibility of the deception scenario.

A number of commentators have responded to Shaffer’s article by suggesting that he has not succeeded in putting on the map a viable new form of scepticism. Bondy and Carter (Forthcoming) argue that the debasing scenario is not so much as metaphysically possible in the first place. Brueckner (2011), Ballantyne and Evans (2013), and Conee (2015), by contrast, do not doubt the possibility of the debasing scenario. Rather, each construes in a different way how the argument will have to move from the mere possibility of the debasing scenario to the impossibility of knowledge, and raises a seemingly devastating objection to the thought that this move is viable, construed in the relevant way. The consensus amongst Schaffer’s commentators thus seems to be that Debasing Scepticism is an interesting idea, but doesn’t really add up to an interestingly plausible kind of Scepticism.

This paper aims to demonstrate that Debasing Scepticism is is an interestingly plausible kind of Scepticism: Schaffer’s critics are wrong to dismiss the argument as readily as they do. Section 1 introduces Debasing Scepticism in more detail. Section 2 purports to undermine Bondy and Carter’s argument for the impossibility of the debasing scenario. Section 3 aims to put on the map a new way of construing the move from the possibility of the debasing scenario to the sceptic’s conclusion. Finally, Sect. 4 argues that, given this new way of construing this latter move, Schaffer’s remaining critics do not succeed in refuting Debasing Scepticism.

1 Debasing Scepticism

We take ourselves to have a rich set of perceptual experiences—seeings, hearings, touchings—which gives rise to a correspondingly rich body of knowledge of the sensed environment. Together with the knowledge we take ourselves to have from other sources—testimony, introspection, memory, inference...—this constitutes an even richer body of knowledge about the external world.

The sceptic denies that it is possible for us to have knowledge, at least of a certain range of propositions, if not all. One way of arguing for scepticism is by appeal to the possibility of sceptical scenarios: scenarios in which our perceptual-cognitive situation is defective in some salient manner, but where we cannot distinguish our defective situation from one in which things have gone well.

Any sceptical argument which appeals to the possibility of a sceptical scenario contains two steps. First, there is the establishment of the metaphysical possibility of the relevant scenario. Second, there is the move from the possibility of that scenario to the impossibility (and hence non-actuality) of knowledge, either of a certain range of propositions, or of all.

The Deception Sceptic’s position is a case in point. Their first step is the establishment of the possibility of Descartes’ deception scenario.Footnote 1 In that scenario an all-powerful daemon arranges things to make it seem to one as if one is having a rich set perceptual experiences, so that one ends up with a correspondingly rich body of empirical beliefs, but those experiences are in fact systematically non-veridical, and the corresponding body of beliefs false. It is typically taken for granted that this scenario is metaphysically possible, thus granting the first step of the Deception Sceptic’s argument.

The second step of the Deception Sceptic’s argument is in my view best conceived as appealing to the following principles (or some suitably reformulated versions thereof):


  • Underdetermination\(_K\) Necessarily, if S knows that p then S possesses evidence {E\(_1\)–E\(_n\)} so that, for any not-p scenario, {E\(_1\)–E\(_n\)} decisively favours S believing that p over believing that they are in that scenario.


  • Phenomenal Evidence\(_P\) (i) Necessarily, for any proposition known by S via perception, p, S knows that p only because S believes that p on the basis of evidence supplied to them by their sensory experience; and (ii) Necessarily, for any piece of evidence supplied to S by their sensory experience, E\(_p\), E\(_p\) is a fact S is in a position to know even in deception scenarios.


  • Priority\(_P\) Necessarily, for any proposition known by S via perception, p, and for any deception scenario in which S is deceived concerning p, prior to S knowing that p via perception S is in a position to know that that scenario does not obtain.Footnote 2

The Deception Sceptic will then complete the second step of their argument in the following way. First, they will argue for the restriction of our sensory evidence to little more than truths about the way things appear (Phenomenal Evidence\(_P\)). Second, they will insist that we must be in a position to know that we are not in the deception scenario prior to knowing anything via sensory experience (Priority\(_P\)). And then finally they will argue that one can know that one is not in the deception scenario either by inference from one’s perceptual knowledge or via one’s phenomenal perceptual evidence directly. Priority\(_P\) precludes the former. The latter is precluded by Underdetermination\(_K\). If sound, this argument would render all knowledge via perception impossible.Footnote 3

To a first approximation, that is how the Deception Sceptic’s argument works. What about the Debasing Sceptic’s argument? Let’s start with the first step of the argument: the establishment of the possibility of the debasing scenario. The debasing scenario can be thought of in the following way. We are all used to cases of beliefs held in an ill-formed manner. Beliefs held out of prejudice, wishful thinking, off the back of fallacious reasoning, and off the back of a defective consideration of the available evidence, are all examples of routes to this epistemically undesirable status. The debasing scenario is one in which the daemon has arranged it so that the agent’s beliefs are held in an ill-formed manner, but they cannot distinguish this epistemically defective situation from a situation in which their beliefs are all well-formed, that is (what I shall assume here is an identical status): held in the light of adequate epistemic reasons. Perhaps the daemon does this by making our beliefs held out of wishful thinking, and then making it the case that we are self-deceived about this. Or perhaps the daemon turns the screw by making us fail to consider some salient piece of evidence, and then making it seem to us as if our consideration of the evidence has been maximally judicious. However the result is achieved, the agent’s beliefs are debased even though it seems to them just like they aren’t. This result can be achieved, it will be noted, without the daemon making any of our beliefs false, as in the deception scenario; an ill-formed belief which seems well-formed need not thereby count as a false belief. Moreover, the result can be achieved consistently with the agent continuing to possess decisive reason to hold the relevant belief; an ill-formed belief which seems well-formed need not thereby count as lacking (propositional) justification. Such a scenario seems possible. I will be defending its possibility in Sect. 2.

How should we construe the second step of the Debasing Sceptic’s argument? Schaffer’s commentators agree that we should construe the second step as relying on the claim that it is not possible for one to know that one is not in the debasing scenario. Beyond that, there is disagreement on how the second step might proceed. I will be offering my own interpretation of the second step in Sect. 3.

Debasing Scepticism is interesting because the danger the debasing scenario poses to our would-be knowledge seems to be wider in scope than that posed by the deception scenario. Whereas there seem to be some truths about which we cannot be deceived—I am currently thinking, say—it seems that every proposition could be believed in an ill-formed manner. Thus, the debasing daemon “threatens universal doubt, in the sense of threatening knowledge of any proposition p” (Schaffer 2010: 232).

2 The First Step Defended

Bondy and Carter (Forthcoming) aim to provide us with two reasons for thinking that the debasing scenario is not possible. The first is that the possibility of the scenario is ruled out by all extant accounts of the basing relation. The second is that the possibility of the scenario is ruled out by a plausible conception of the relation between basing one’s belief on reasons and the activity of justifying one’s belief, defended by Leite (2004). I tackle the former in Sect. 2.1 and the latter in Sect. 2.2.

2.1 The Debasing Scenario and the Nature of the Basing Relation

Bondy and Carter argue that the possibility of the debasing scenario requires there to be a possible world at which:

...[(i)] a demon causes some subject S to believe p on the basis of a bad reason R\(_1\) at time t. And (ii) at a later time t\(_2\) the demon makes S think that his reason for believing p was a different good reason R\(_2\). And (iii) at t\(_2\) it remains the case that S holds his belief only on the basis of R\(_1\), and not on the basis of R\(_2\). (Bondy and Carter, Forthcoming): 8–9)

They then argue that on either the causal theoryFootnote 4 or the doxastic theoryFootnote 5 of the basing relation, it is not possible for (i)–(iii) to hold together. Since any theory of the basing relation will be committed to some version of one of these theories (at least prima facie), we get the result that the debasing scenario is not possible (at least prima facie).

Let us start with the Causal Theory. According to a simple version of the Causal Theory, S’s belief that p is based on the reason that q if, and only if, S’s belief that p is causally sustained by S’s belief that q.Footnote 6 This theory immediately runs into the well-known problem of deviant causal chains. Following a plausible suggestion made by Turri (2011: 393), however, Bondy and Carter argue that deviant cases need to be handled by supplementing the simple Causal Theory in the following manner. What the proponent of it should say is that S’s belief that p is based on the reason that q if, and only if, S’s belief that p is causally sustained by S’s belief that q and the synchronic process of causation linking S’s belief that q with S’s belief that p is—the whole thing—the manifestation of a disposition the agent has which constitutes one of their cognitive traits.

Bondy and Carter argue, however, that once Turri’s modification is made, conditions (i)–(iii) cannot all be met together, so that the debasing scenario is not possible. Their thought here is that since the daemon is ensuring that S’s belief is causally sustained by their belief in (the bad reason) \(\hbox {R}_1\), their belief being sustained by \(\hbox {R}_1\) does not manifest one of the agent’s own cognitive traits, and thus, by the lights of the modified causal theory, S’s belief isn’t really based on \(\hbox {R}_1\) after all.Footnote 7

The problem with this argument is that the debasing scenario needn’t be construed as a scenario in which the daemon makes it the case that the agent holds a belief on the basis of any reason at all: it needn’t be construed as a scenario in which Bondy and Carter’s conditions (i) and (iii) are met. All that matters for the debasing scenario is that it seems to the agent as if their belief is well-formed, and hence held on the basis of a decisive reason, but that that is not really so. How the daemon ensures this is left open: it could be by making their belief that p based on some alternative reason which isn’t really an adequate one (a bad reason, in other words), and this might indeed be precluded by the modified Causal Theory. But the daemon could equally ensure the ill-formedness of the agent’s belief in myriad other ways, for example by making the belief sustained by some prejudice, or wishful thinking. In that case, the agent’s belief would not count as being based on any reason at all.

Indeed, the problem for Bondy and Carter is worse than this, for they have in fact supplied us with yet another way in which the daemon might operate, other than by causing the agent to believe on the basis of a bad reason: the daemon might operate by ensuring that S’s belief is causally sustained in a way that doesn’t manifest their cognitive traits, even though it seems to them as if it does. So the modified Causal Theory does not preclude the possibility of the debasing scenario; if anything, it strengthens the case for its possibility.

Next, what about the Doxastic Theory? According to the Doxastic Theory, S holds their belief that p in the light of the reason that q if, and only if, S also believes that q is a decisive reason to believe that p. Call the claim read from right-to-left Doxastic (Sufficient) and call the claim read from left-to-right Doxastic (Necessary). Here is how Bondy and Carter argue that Doxastic (Sufficient) would preclude the possibility of the debasing scenario:

...it’s explicitly included in the description of the debasing daemon scenario that at t\(_2\), S comes to think that he holds his belief that p on the basis of the good reason R\(_2\). So S thinks that R\(_2\) is a good reason for believing that p at t\(_2\). So according to Doxastic (Sufficient) S believes p on the basis of R\(_2\) at t\(_2\). (Bondy and Carter, Forthcoming: 9)

And yet, the agent is precisely not supposed to hold their belief that p on the basis of \(\hbox {R}_2\) at t\(_2\).

In reply, it seems to me that any plausible version of Doxastic (Sufficient) will have to include a ceteris paribus clause which isn’t met in the debasing scenario. To see why, consider cases of prejudiced belief, and belief sustained by mere wishful thinking. It’s plausible on the face of it that these cases are cases of ill-formed belief; that seems to be the most natural way to explain why agents in such cases lack justified beliefs, after all. But such epistemic maladies can be, and often are, accompanied by a disposition to defend the belief in question by appeal to some consideration the agent believes to be a decisive reason to hold the relevant belief. In particular, agents in such cases can be, and often are, in the grip of a kind of self-deception about why they hold the belief in question: they convince themselves that the belief in question is held in a healthy, rational manner, even though it isn’t. And as a result there is some consideration available to them which, through the relevant mechanism of self-deception, they believe to be a decisive reason to hold the belief in question. But plausibly, they do not base their belief on any such consideration nevertheless.

To handle the possibility of such cases, the proponent of Doxastic (Sufficient) should build a ceteris paribus clause into their principle, and say that that clause is not met in cases in which the agent suffers from the kind of self-deception at issue. But with this clause added, there is now no reason why the proponent of the possibility of the debasing scenario cannot simply say that the clause is not met in the debasing scenario, so that although the agent in that scenario treats \(\hbox {R}_2\) as a decisive reason to believe that p, and might even believe of it that it is, they do not hold their belief on the basis of \(\hbox {R}_2\) nevertheless.Footnote 8

Here finally is how Bondy and Carter argue that Doxastic (Necessary) would preclude the possibility of the debasing scenario:

...in the debasing scenario, S does not think at t\(_2\) that \(\hbox {R}_1\) is a good reason for p. So according to Doxastic (Necessary), at t\(_2\), S does not believe p on the basis of \(\hbox {R}_1\). (Bondy and Carter, Forthcoming: 9)

And yet, the agent is precisely supposed to believe that p on the basis of \(\hbox {R}_1\) at t\(_2\).

In reply to this, the defender of the possibility of the debasing scenario can point out once again that the scenario needn’t be construed as one in which the agent believes on the basis of some bad reason, for it needn’t be conceived as a scenario in which the agent believes on the basis of any reason at all.

We can conclude, then, that neither the Causal Theory nor the Doxastic Theory poses a problem for the possibility of the debasing scenario. Let us now turn to Bondy and Carter’s second reason for doubting the possibility of that scenario.

2.2 The Debasing Scenario and the Justificationaist Conception

We might be asked, by an interlocutor or by ourselves in a moment of reflection, Why do you/I believe that p? Responding successfully to this question requires engaging in an activity of justifying one’s belief. We can think of this as a certain kind of conscious mental activity: an activity which involves bringing to mind some consideration, q, to the truth of which one is committed,Footnote 9 which is supposed to settle the matter in favour of believing that p. Such activity, when carried out to completion, culminates in a conscious act of inference q, so p.Footnote 10 The activity may or may not be linguistically manifest.

What is the relationship between the act of justifying and the state of basing one’s belief on the relevant reason? Following Leite (2004), Bondy and Carter helpfully distinguish between two conceptions of the relationship: the Spectatorial Conception and the Justificationist Conception. According to the former, the act of justifying and the state of basing are constitutively independent of one another: it is possible to hold the agent’s state of believing that p in the light of q fixed, whilst varying how the agent is disposed to reason about whether p, and vice-versa. According to the Justificationist Conception, by contrast, the act of justifying and the state of basing are constitutively dependent on one another in at least one direction, if not in both. That is, the Justificationist subscribes to at least one of the following theses:

(JC\(_N\)):

Necessarily: if S believes that p in the light of reason q, then S is disposed (and hence is able) to engage in the activity of consciously reasoning from q to p.

(JC\(_S\)):

Necessarily: if S engages in the activity of consciously reasoning from q to p, then S believes that p in the light of the reason q.

(\(\hbox {JC}_N\)) says that basing one’s belief on a certain reason necessarily comes along with a disposition to reason from the reason in question, to the belief (and hence an ability to do so). (\(\hbox {JC}_S\)) says that the inferences one carries out with respect to p determines the basis on which one believes that p. If one infers p from q, this suffices for one to base one’s belief on q.

Bondy and Carter argue that each of (\(\hbox {JC}_N\)) and (\(\hbox {JC}_S\)) are plausible, and each rules out the possibility of the debasing scenario. I now want to argue that the Debasing Sceptic can perfectly well accept the truth of (\(\hbox {JC}_N\)) and (\(\hbox {JC}_S\)). Indeed, in Sect. 3 I will be supplying my own argument for the prima facie plausibility of the Justificationist Conception, and I will be suggesting that the Debasing Sceptic utilise that conception in the second step of their argument. The overall upshot will thus be that, far from the Justificationist Conception undermining Debasing Scepticism by undermining the first step of the argument, it positively enables it by enabling the second step.

Take (\(\hbox {JC}_N\)) first. Here is how Bondy and Carter argue that it rules out the possibility of the debasing scenario:

[(\(\hbox {JC}_N\))] closes off the possibility of the debasing demon scenario, because it entails that...S doesn’t believe on the basis of the bad reason \(\hbox {R}_1\). (Bondy and Carter, Forthcoming: 14)

The thought here is this: in the debasing scenario, the agent is not disposed to reason their way from the bad reason, \(\hbox {R}_1\), to p. (\(\hbox {JC}_N\)) therefore implies that they don’t believe on basis of \(\hbox {R}_1\) after all. And yet it is supposed to be built into the debasing scenario that the agent does believe on the basis of the bad reason \(\hbox {R}_1\).

The reader might well already have predicted what my response to this will be: as we have already seen, the debasing scenario need not be construed as a scenario in which the agent believes that p on the basis of any reason at all, let alone a bad reason. (\(\hbox {JC}_N\)) is thus perfectly compatible with the possibility of such a scenario.

Here now is how Bondy and Carter argue that (\(\hbox {JC}_S\)) rules out the possibility of the debasing scenario:

[(\(\hbox {JC}_S\))] also rules out the debasing demon, because it entails that at t\(_2\), S believes that p on the basis of the good reason \(\hbox {R}_2\). (Bondy and Carter, Forthcoming: 14)

The thought here is that the antecedent of (\(\hbox {JC}_S\)) is met at t\(_2\), so that the agent would count as believing that p on the basis of the good reason \(\hbox {R}_2\), at t\(_2\), after all.

An initial problem for this reasoning is that the debasing deamon might make it seem to the agent, at t\(_2\), that they believe that p on the basis of \(\hbox {R}_2\), but the agent needn’t have inferred p from \(\hbox {R}_2\) at t\(_2\) nevertheless. If that’s so, however, then at t\(_2\), the antecedent of (\(\hbox {JC}_S\)) needn’t be met. However, we can tweak Bondy and Carter’s reasoning here in the following way, so that their inference from (\(\hbox {JC}_S\)) to the impossibility of the debasing scenario is more defensible. In the debasing scenario, it seems to the agent as if they believe that p on the basis of some good reason, \(\hbox {R}_2\), and hence they are disposed to justify their belief via a course of reasoning which culminates in the drawing of the inference \(\hbox {R}_2\), so p. Let us suppose that that disposition is triggered, so that they make that inference. Then given (\(\hbox {JC}_S\)) that would suffice for them to believe that p on the basis of \(\hbox {R}_2\). But the proponent of the scenario intends it precisely not be the case that even were the disposition triggered, the agent would count as believing that p on the basis of \(\hbox {R}_2\).

My response to this takes as its starting point some reflections on the following well-discussed case from Peacocke (1998):

Someone can make a judgement, and for good reasons, but it not have the effects that judgements normally do – in particular, it may not result in a stored belief that has the proper influence on other judgements and on action. A combination of prejudice and self-deception, amongst many other possibilities, can produce this state of affairs. Someone may judge that undergraduate degrees from countries other than her own are of an equal standard to her own, and excellent reasons might be operative in her assertions to that effect. All the same, it might be quite clear, in decisions she makes on hiring, or in making recommendations, that she does not really have this belief at all. (Ibid.: 90)

Peacocke’s agent is disposed to reason her way from decisive evidence for thinking that foreign degrees are of the same quality as home degrees, treated as such by her, to the conclusion that foreign degrees have that status. And yet her behaviour manifests the belief that such degrees are not equal. Peacocke’s case constitutes a prima facie counterexample to (\(\hbox {JC}_S\)), for suppose Peacocke’s agent engages in their act of reasoning. Then by the lights of (\(\hbox {JC}_S\)), it would follow that she believes that foreign degrees are on-par with home degrees on the basis of the relevant evidence. But that is precisely not so, simply because it’s at least initially plausible that the agent in Peacocke’s case doesn’t so much as believe what they judge, as Peacocke himself suggests. To see why this has initial plausibility, suppose that Peacocke’s agent does believe that foreign degrees are of the same quality as home degrees. In that case, we’d have to say that the agent possesses a belief which is systematically precluded from having effects on their behaviour; it’s written into the case, after all, that the agent’s behaviour manifests a belief in the opposite. But if that’s so, the theoretically simpler thing to say is that the agent fails to have the belief in question.

If (\(\hbox {JC}_S\)) is to have any plausibility, then, the proponent of it will have to say something about Peacocke’s case. I now want to argue that the proponent of (\(\hbox {JC}_S\)) should accept a certain pluralist conception of apparent acts of inference in order to defend their view against the objection generated by Peacocke’s case. I will argue further, however, that once (\(\hbox {JC}_S\)) is conceived as requiring such a commitment, it no longer threatens the possibility of the debasing scenario.

The pluralist conception of apparent acts of inference I have in mind says the following. On the one hand, there are conscious mental acts of inference from q to p, the completion of which necessitates the agent believing that p for the reason that q. On the other hand, there are conscious mental occurrences during which it seems to the agent as if they are drawing an inference from q to p, but they are not because the occurrence in question has no impact on whether they believe that p, nor for what reason they believe that p (if they believe that p at all). The two conscious mental occurrences are different in kind, for the former has an essential property—it’s necessitating that the agent believes that p for the reason that q—which the latter doesn’t. But occurrences of the second kind are introspectively indistinguishable from occurrences of the first kind.

To be clear, the suggestion is that the occurrence of a merely apparent inference does not have an impact on whether the agent believes that p, nor for for what reason. Consistently with such an event occurring in the agent’s mental life, the agent might already have the belief that p anyway, in which case the apparent inference has no impact on the prior basis of that belief, if any. But equally consistently with such an event occurring in the agent’s mental life, the agent might not already have the belief that p. In that case the merely apparent inference has no power to generate the belief that p. The further thought is that we can contrast merely apparent inferences, conceived in the way just broached, with bone fide inferences, the engagement in which suffices for one to believe that p on the basis of the relevant consideration(s).

Assuming that (\(\hbox {JC}_S\)) is well motivated, it need not be ad hoc for the proponent of (\(\hbox {JC}_S\)) to subscribe to this pluralistic conception of apparent inferences. Pre-philosophically, we acknowledge that there is some important difference between the purported act of justifying carried out by Peacocke’s agent and normal acts of justifying. We would want to acknowledge that there is a sense in which the former, but not the latter, fail to manifest the agent’s genuine assessment of the situation; in purportedly justifying their belief there is a sense in which the agent is paying mere lip service to the evidence. The proponent of (\(\hbox {JC}_S\)) would be within their rights to say that the question of how to cash out this difference, which we should all acknowledge, is a matter to be decided by an independently motivated theory of the relationship between basing and justifying.

With this pluralism about acts of apparent inference in tow, the proponent of (\(\hbox {JC}_S\)) can readily handle Peacocke’s case: they can just say that Peacocke’s agent is engaged in a merely apparent inference, but that (\(\hbox {JC}_S\)) is a sufficiency thesis concerning only bona fide inferences. Moreover, with (\(\hbox {JC}_S\)) now conceived as committed to this pluralism about acts of apparent inference, Bondy and Carter’s argument from the truth of (\(\hbox {JC}_S\)) to the impossibility of the debasing scenario is also rendered unsound. For, it will be recalled, that argument appeals to the idea that the agent in the debasing scenario would reason from some adequate epistemic reason, to the conclusion that p even though the daemon is supposed to make it so that they don’t believe that p for that adequate epistemic reason. However, the proponent of the possibility of the scenario can now simply respond by saying that in the debasing scenario, the daemon makes it the case that with respect to their belief that p, the agent is in a situation analogous to Peacocke’s agent: they are made to have the disposition to engage in merely apparent inferences from sufficient reasons to the conclusion that p.

To be sure, the proponent of the possibility of the debasing scenario would acknowledge that there is a disanalogy between the agent in Peacocke’s scenario, and the agent in the debasing scenario: the former does not so much as believe that p, whereas the latter certainly does. But as pointed out above, this is perfectly consistent with the thought that the agents in each case only have available to them a merely apparent act of inference.

We can conclude, then, that Bondy and Carter do not succeed in establishing that the Justificationist Conception rules out the possibility of the Debasing Scenario. The first step of the Debasing Sceptic’s argument should be granted.

3 The Second Step Re-conceived

Let us now turn our attention to the second step of the Debasing Sceptic’s argument. I am going to suggest that there is a prima facie plausible framework for conceiving of the agent’s power to hold their belief in the light of reasons the correctness of which enables us to provide a fresh construal of the argument’s second step. Section 3.1 states the framework and provides it with prima facie justification. Section 3.2 provides the fresh construal of the argument’s second step. Section 3.3 responds to the objection that the argument, so conceived, cannot be extended to perceptual knowledge.

3.1 A Framework

The framework for thinking about the agent’s power to hold their beliefs for reasons is composed of three doctrines: (i) the Justificationist Conception described in Sect. 2.2; (ii) the claim that the inferences a capacity for which is said to be required for believing for a reason by the Justificationist Conception are basic mental actions; and (iii) a certain epistemic thesis concerning basic actions. Let us take each claim in turn.

I have already explained what the Justificationist Conception amounts to. Why think it plausible? Take (\(\hbox {JC}_N\)) first and consider an agent who believes that p in the light of the reason that q. It’s prima facie plausible that the ‘in the light of’ relation which is instantiated by the agent’s belief is a kind of relation which can be instantiated at all only if the agent can be held responsible for their belief in a certain sort of way. On the face of it, part of what it is to believe that p in the light of q is to open oneself up to praise and criticism, in so far as one holds the belief that p, at least partly on the basis of whether one’s reason is adequate.

It’s also prima facie plausible that an agent can be held responsible for their belief that p only if the agent is able to exercise control of what they think about whether p. So: believing that p in the light of the reason that q requires the agent to be able to exercise such control. But it’s prima facie plausible in turn that this activity just is the activity of consciously reasoning from a set of considerations which one treats as reasons, to a conclusion about whether p: the activity of justifying one’s belief. So we get the further result that believing in the light of a reason requires the ability to consciously reason from the set of reasons one has at one’s disposal, to a conclusion about whether p.

However, presumably one’s holding the belief that p in the light of q in particular will have an effect on the reasoning one is guaranteed to be able to carry out whenever one believes that p in the light of q: one will be disposed, during the course of that reasoning, to infer: q, so p. That gets us to (\(\hbox {JC}_N\)).

What now about (\(\hbox {JC}_S\))? Suppose one engages in an activity of justifying one’s belief that p by appeal to the reason that q. It’s prima facie plausible that this very activity constitutively involves the kind of responsibility-involving status already broached. It’s prima facie plausible, that is, that if one justifies one’s belief that p in a way that culminates in the inference q, so p, then one has opened oneself up to praise and criticism, in so far as one judges that p, at least partly on the basis of whether q really is an adequate reason to judge that p. This is part of what it is to engage in the act of justifying. Thus, there’s a kind of normative status which partly constitutes the act of justifying that mirrors the normative status that partly constitutes holding one’s belief in the light of a reason.

How are we to explain this correlation? The simplest way to explain it is by saying that when one infers p from q, part of what it is to do so is to be in the state of believing that p in the light of the reason that q: conscious inferences are, constitutively, the conscious manifestation of standing states of believing in the light of reasons. That way, it’s no wonder that inference constitutively involves the same type of normative condition as believing in the light of a reason. Any alternate explanation is bound to introduce a greater degree of complexity than this one. But notice that the explanation in question commits us to (\(\hbox {JC}_S\)).

So the Justificationist Conception has prima facie plausibility. The second doctrine which forms our framework is the claim that those inferences a capacity for which is said to be required for believing for a reason by (\(\hbox {JC}_N\)) are basic mental actions. There are really two ideas here: (a) that such inferences are actions and (b) that they are basic actions. A case in favour of (a) can be made straightforwardly. I’ve already said that believing that p in the light of the reason that q requires that one can be held responsible for one’s believing that p; that one can be responsible for one’s \(\phi \)-ing only if one is able to control for whether one \(\phi \)s; and that in the case of belief this control is exercised in inferences from one’s reasons to a conclusion about whether p. These claims were all part of the argument for (\(\hbox {JC}_N\)). But it’s prima facie plausible that the events which constitute our taking control of something such that we can be held responsible for it are actions: the kind of control required for responsibility is agential.Footnote 11

But it is also plausible that if drawing the inference q, so p is an action at all, then it is a basic action (under that description).Footnote 12 To a first approximation, something is a basic action just in case it is something which one can just do, without one’s doing it by doing something else which is a means of doing it.Footnote 13 Simple examples of basic bodily actions include arm raisings (but not attracting the teacher’s attention), moving one’s finger in a downward motion against the switch (though not boiling the kettle), and moving one’s cue arm (though not potting the blue). Some examples of basic mental actions include calculating the answer to a certain arithmetical question (though not solving the problem which the question constitutes), looking at the letters on the optician’s eye chart (though not completing the long-sightedness test), and imagining what would happen were one to try jumping the gorge (though not making oneself feel nervous). Right off the bat, it seems plausible that drawing the inference q, so p is yet another basic mental action. What else would one need to do as a way of performing an inference, after all?

This brings us to the final element of the framework: the epistemic claim that for any action basic under a certain description, it is at least metaphysically possible for one who is capable of acting in that way to know that they are acting in that way, or have acted in that way, under that description. This claim, insisting as it does only on a mere metaphysical possibility, is modally weak. It effectively says that in order for a certain occurrence in an agent’s life to count as an action they engage in, we must be able to find it intelligible—there must be some possibility—of their standing in a certain epistemic relation to it. The claim is prima facie plausible. Actions are occurrences in the agent’s life the initiation or continuance of which is an exercise of control on the agent’s part. It is difficult to make sense of what it is for an agent to exercise such control unless we credit the agent, in \(\phi \)-ing, with a capacity to know what they are doing (relative to a description of their \(\phi \)-ing as basic). But, I submit, S knows what they are doing in \(\phi \)-ing, relative to a description of their \(\phi \)-ing as basic, only if it is at the very least metaphysically possible for them to know that they are \(\phi \)-ing, or have just \(\phi \)ed, under the relevant description.

3.2 Completing the Second Step

With the framework just provided with prima facie justification in tow, we are now in a position to see how the Debasing Sceptic might utilise it in filling-in the details of the second step of their argument.

I want to begin this stage of the discussion with a reminder. Amongst the claims for which I argued in Sect. 2 is that the proponent of (\(\hbox {JC}_S\)) should accept a pluralist view of apparent inferences according to which apparent inferences are either bona fide inferences, which determine what the agent believes and for what reason, or merely apparent inferences, which do not. Since (\(\hbox {JC}_S\)) is part of the framework defended in Sect. 3.1, this pluralism should indeed be thought of as an element of the debasing sceptic’s position. I also argued that if (\(\hbox {JC}_S\)) is true, the proponent of the possibility of the debasing scenario should go ahead and claim that the agent in the debasing scenario only has an act of merely apparent inference available to them. Since (\(\hbox {JC}_S\)) is part of the framework defended in Sect. 3.2, this is indeed a further element of the Debasing Sceptic’s position. The Debasing Sceptic is thus committed to saying that I have just inferred p from q, where this is the thought had by an agent who takes themselves to be in the ordinary case and where it seems to them as if they have just inferred p from q, entails I am not in the debasing scenario.

From here the Debasing Sceptic can complete the second stage of their argument by first demonstrating that it is not possible for one to know that one has just inferred p from q, and second inferring from this that knowledge in general is not possible.

The first move relies on the following analogues of the Deception Sceptic’s Phenomenal Evidence\(_P\) and Priority\(_P\) principles:

Phenomenal Evidence\(_B\):

(i) Necessarily, for any proposition known by S of the form I have just inferred p from q, S knows that proposition only because S believes it on the basis of evidence supplied to them by their act of apparent inference from q to p; and (ii) Necessarily, for any piece of evidence supplied to S by their act of apparent inference, E\(_i\), E\(_i\) is a fact S is in a position to know even in debasing scenarios.

Priority\(_B\):

Necessarily, for any proposition known by S of the form I have just inferred p from q, prior to S knowing that proposition S is in a position to know that they are not in the debasing scenario.

Suppose an agent in the ordinary scenario engages in an inference from q to p. How would they know that they have just done so? Relying on Phenomenal Evidence\(_B\) the Debasing Sceptic will say that they know this at all only by appeal to evidence, and that evidence is restricted to what can be known even in the debasing scenario. Since (\(\hbox {JC}_S\)) implies that one cannot engage in the act of inference from q to p in the debasing scenario, the Debasing Sceptic will say that the relevant evidence is restricted merely to claims of the form it appears to me as if I have just inferred p from q, instead of the stronger I have just inferred p from q.

Next, the Debasing Sceptic posits Priority\(_B\), an analogue of the Deception Sceptic’s Priority\(_P\), and therefore presumably a thesis which has as much to be said for it as the latter. At this point, the Debasing Sceptic can argue that one can know that one is not in the debasing scenario only if either one is in a position to know it via inference from the claim that one has just inferred p from q, or by appeal to the phenomenal evidence provided by one’s act of inference directly. The former is straightforwardly precluded by Priority\(_B\). The latter is precluded by Underdetermination\(_K\), for the evidence it appears to me as if I have just inferred p from q underdetermines whether one is in the debasing scenario. Thus, they can conclude that one is not in a position to know that one is not in the debasing scenario. And since, by Priority\(_B\), one would need to be in a position to know that if one is to be in a position to know that one has just inferred p from q in the first place, it is not possible for one to know that one has just inferred p from q either.

This completes the first move of the second stage of the Deception Sceptic’s argument. It relies only on (\(\hbox {JC}_N\)), and the same style of reasoning relied on by the Deception Sceptic in the second stage of their argument. What now of the second move, which completes the argument for the sceptical conclusion?

It has already been argued that it must be possible for one to know that one has performed an action under a certain description of it as basic if the relevant occurrence is to count as an action in the first place, and that the episodes of inference a capacity for which is required for believing for reasons by (\(\hbox {JC}_N\)) are basic actions. The Debasing Sceptic has just argued that we cannot know that we have engaged in an inference from q to p. So they can conclude that such events in our mental lives do not constitute actions. The impossibility of our knowing our inferences robs us of the ability to engage in actions of inference.

However, since having a capacity to engage in actions of inference is laid down as necessary for basing one’s belief on reasons by (\(\hbox {JC}_N\)), it follows from our being unable to perform such actions that we cannot believe that p in the light of the reason that q. Since this argument can be run for any belief, no matter what the value of p, we get the result that it is not possible for any belief of ours to be held in the light of epistemic reasons. Our being incapacitated with respect to acts of inference in turn precludes us from holding beliefs in the light of reasons. Well-founded belief is impossible.

Finally, since well-founded beliefs are necessary for knowledge, we get the sceptical conclusion that it is not possible for us to know anything. Relying on the remaining elements of our framework, the Deception Sceptic reaches their extreme conclusion.

3.3 The Perceptual Case

It might be thought that the case of perceptual warrant generates problems for the Debasing Sceptic’s argument. That’s because it might be thought that in the perceptual case, one’s belief is held in the light of reasons supplied to one by one’s perceptual experience. This, it might be thought in turn, implies that the act of justifying one’s perceptual belief one is in a position to engage in cannot be conceived as an inference from one’s reason, to the proposition believed: there is no inferential transition from experience to belief.

It will be concluded that (\(\hbox {JC}_N\)) at best only covers those range of cases where one’s belief is sustained not by a perceptual experience, but by some other belief(s) one has. Given that, the argument presented in Sect. 3.2 can only hope to achieve a limited sceptical result concerning knowledge one has by dint of other articles of knowledge one possesses. It leaves perceptual knowledge, and perhaps non-inferential sources of knowledge in general, untouched. Since Debasing Scepticism is supposed to be significant partly because it has knowledge of every kind in its scope, if this line of thought is right then Debasing Scepticism would lose much of its significance.

There are a number of ways of responding to this worry. One is simply to point out that just because perceptually warranted beliefs are beliefs rationalised by experiences themselves, it does not follow that the act of justifying is not an inference. Perhaps my visual experience of the red ball itself stands in a rationalising relation to my belief that there is a red ball there, so that the latter counts as being held in the light of a reason supplied by the former. And perhaps the two states standing in that relation to one another requires that I have the ability to justify the latter by appeal to the reason supplied by the former. Still, it doesn’t follow that the act of justifying in question is non-inferential. It could be that the act of justifying constitutively involves the manifestation of (perhaps freshly acquired) beliefs I have in propositions that either are, or represent, my perceptual reasons, from which I then infer the truth of my perceptual belief. Thus, perceptual warrant might be thought to be no counter-example to (\(\hbox {JC}_N\)) after all.

But suppose that picture is not right. Suppose, that is, that we can find compelling grounds for thinking that the act of justifying one is able to engage in when one believes in the light of perceptual reasons must involve a transition from one’s perceptual reason qua reason one possesses because of one’s perceptual experience. In that case, the act of justifying cannot be conceived as a transition from one judgement (the content of which is the proposition which either is or represents one’s reason), to another (the content of which is the proposition believed). Rather it must be conceived as a transition from one’s experiential state itself to a judgement the content of which is the proposition believed.

Still, there would be a reply to the current worry available anyway. The thought would be that to the extent that we have good reason to think that the act of justifying in the perceptual case is non-inferential, that will just mean that we have to acknowledge the possibility of acts of justifying that are inextricably perceptual in nature. Call those acts perceptual-rational transitions. In order to acknowledge the possibility of such acts, we would have to reformulate (\(\hbox {JC}_N\)) in the following way:

(\(\hbox {JC}_N\)*):

Necessarily: if S believes that p in the light of reason q, then S is disposed (and hence is able) to engage in the activity of consciously justifying their belief that p by appeal to q.

It would then be left open whether the relevant act of justifying takes on the form of inference, or of a perceptual-rational transition, depending on whether it is a case of perceptual warrant.

Next, the framwork presented in Sect. 3.1 swould have to be reformulated in the following manner. We would have to say that when S engages in a perceptual-rational transition, that is a basic kind of mental action and is hence something that the agent can in principle know that they are doing, or have just done. And we would have to say that such actions can be performed only if the agent ends up counting as holding a belief that p in the light of a perceptual reason, so that in the debasing scenario it merely seems to the agent as if they engage in such an activity. The argument presented in Sect. 3.2 could then proceed in much the same way for perceptual belief as it does for belief held on the basis of other belief.

4 The Viability of Debasing Scepticism

I have defended the first step of the Debasing Sceptic’s argument against Bondy and Carter’s objections, and have put on the map a fresh way of construing the argument’s second step. In this final section, I want to defend the argument, so conceived, against objections from Schaffer’s remaining critics: Brueckner (2011), Ballantyne and Evans (2013), and Conee (2015). I will also consider a fresh objection suggested to me by one of my reviewers. I do not, of course, want to demonstrate that the argument, as it has been interpreted here, is sound. I am no sceptic. Indeed, for all I have said there might be a successful reply to Deception Scepticism currently on the market which can be made to work just as well against Debasing Scepticism. Rather, my aim is just to defend the claim that there is an interestingly plausible form of scepticism here by showing that the argument cannot be dismissed on the grounds offered by Schaffer’s remaining critics. Let me take each in turn.

Brueckner. Apparently following Schaffer (2010: 234) himself, Brueckner (2011: 296–297) reconstructs the second step of the Debasing Sceptic’s argument in the following way. Knowing that p requires that one’s belief that p is properly based. But knowing that p also requires being in a position to know that one knows that p. Thus, an application of (a version of) the Closure Principle gets us: if one knows that p, then one is in a position to know that one’s belief that p is properly based. But given that one cannot rule out that one is not in the debasing scenario, one cannot know that one’s belief that p is properly based, for any proposition p. So: one cannot know that p, for any proposition p.

Brueckner points out that the second step of the argument so conceived relies on a version of the KK principle, and plausibly rejects it on the grounds that the relevant principle is false. However, there is no need to interpret the second step of the argument in a way that involves an appeal to any version of the KK principle. The reconstruction of the second step presented in Sect. 3.2 does not do so.

Ballantyne and Evans. Ballantyne and Evans (2013: 3–5) argue that the second step of the Debasing Sceptic’s argument must proceed by an inference from the thought that the debasing scenario is possible, to the thought that one is not propositionally justified in believing that p, for any first-order proposition p about the way the world is. They then provide a further interesting argument for how any such inference is bound to fail.

There is no need to examine the further details of Ballantyne and Evans’s discussion, because it should already be clear how they have gone wrong: there is no need to construe the second step of the Debasing Sceptic’s argument as relying on an inference from the possibility of the debasing scenario, to the lack of propositional justification for belief in the target first-order proposition about the world—the reconstruction of the second step presented in Sect. 3.2 does not do so. Having said that, it does rely on an inference from Underdetermination\(_K\) and Phenomenal Evidence\(_B\) to the claim that one is not propositionally justified in believing that one is not in the debasing scenario, but Ballantyne and Evans do not apply their argument to that inference.

Conee. Conee plausibly claims that, however the second stage of the debasing sceptic’s argument is to be understood exactly, it will need to appeal at some stage to the claim that it is not possible for us to know that we are not in the debasing scenario. But, Conee suggests, we have a range of introspective evidence available to us which enable us to rule out that we are in the debasing scenario.

One source of introspective data Conee describes as follows:

Often, as we draw a conclusion from evidence that supports its truth, we have the marked impression that we are relying on the evidence to believe the conclusion...These impressions of reliance and deference, with their causal implications, are evidence that the reasons from which we infer our conclusions cause us to believe them. (Conee 2015: 6)

In reply to this, the Debasing Sceptic should ask: what are the ‘impressions of reliance and deference’ Conee is talking about other than its seeming to one as if one has just drawn an inference from one’s evidence, to the proposition believed? But the Debasing Sceptic will simply suggest in turn that one’s having undergone an apparent inference is good evidence for believing that one believes on the basis of the relevant evidence only if, prior to knowing that, one has ruled it out that one is not in the debasing scenario.

A second source of introspective data Conee describes as follows:

We have evidence from memory and perception for the existence of an extensive range of close correlations between our having justifying evidence for the beliefs that we apparently know and our forming and sustaining those beliefs. A few arbitrarily chosen examples: we believe that Jane Austen wrote Pride and Prejudice when we have recollections supporting that she wrote the book, we believe that we are sitting when we have sensory experiences giving evidence that we are seated...and so forth. (Conee 2015: 5)

Conee’s idea is that for a great many of our beliefs, and some pieces of evidence in our possession for those beliefs, we can know the following fact concerning them: when one consciously takes q to be a sufficient reason for believing that p at t, one also comes, or continues, to believe that p at t or immediately after. Call a proposition which states one of these correlations a reasons-belief correlation claim. Conee’s argument is that the knowledge we are in a position to have of a great many reasons-belief correlation claims licenses an abductive inference to the conclusion that our beliefs are by-and-large held in the light of reasons in our possession, and this in turn enables us to rule it out that we are in the debasing scenario.

In reply to this, the Debasing Sceptic should ask: how do we come to know the truth of a reasons-belief correlation claim? Presumably, the process would take the following form: first, the agent consciously takes q to be a decisive reason to believe that p, there is then a rational transition from such conscious taking to a judgement that p, and their having run through this process puts them in a position to know the truth of a reasons-belief correlation claim.

Now, leaving aside the complications that will have to be introduced in order to accommodate the perceptual case explored in Sect. 3.3, I take it that this process simply is the engagement in an apparent inference from q to p. But now we can ask: under what circumstances would its seeming to the agent as if they have just inferred p from q ground knowledge in a reasons-belief correlation claim? Presumably, it would do so only if it could ground knowledge that one has come to believe that p. However, given the framework within which I have suggested they operate, the Debasing Sceptic could say in response to this that in order for running through an apparent inference to ground knowledge that one has come to believe that p, one will already have to have warrant for believing that the apparent inference is a genuine inference. And to have warrant to believe that, one would already need warrant to believe that one is not in the debasing scenario. Thus, the Debasing Sceptic is within their right to argue that this second source of introspective data Conee appeals to is not available to us in the first place.

A Final Challenge. I want to consider a final way of challenging the Debasing Sceptic’s argument which has been suggested to me by one of my reviewers. The challenge has two components. First, there is the thought that Phenomenal Evidence\(_B\) is overly restrictive: included amongst one’s phenomenal evidence should not only be claims of the form it appears to me as if I have just inferred p from q but also memories of apparent inferences one has previously engaged in. Second, there is the following question, which constitutes the challenge to be addressed by the Debasing Sceptic. As long as one doesn’t have any positive reason to believe that one is in the debasing scenario and also no positive reason to think that one’s phenomenal evidence is unreliable, why shouldn’t one’s phenomenal evidence, broadly construed to include memories of apparent inferences, not be enough to decisively favour belief in the claim that one is not in the debasing scenario? Surely, if any fallible evidence is ever good enough for all-things-considered justification, such phenomenal evidence is.

My concern with this way of challenging the Debasing Sceptic’s argument can be brought out by asking the following question: what is the supposed epistemic relationship between one’s phenomenal evidence, understood in the broad manner at issue, and the claim that one is not in the debasing scenario? On the one hand, we might try to answer that question by saying this: the phenomenal evidence, broadly construed and absent the defeaters already described, decisively favours belief in the claim that one has just inferred p from q, over the claim that one hasn’t. But if that’s the case then it also decisively favours belief that we are not in the debasing scenario over belief that we are, because the claim that one has just inferred p from q entails that one is not in the debasing scenario. The problem with this way of cashing out the relation between one’s phenomenal evidence, broadly construed, and the claim that one is not in the debasing scenario, however, is that it falls foul of Priority\(_B\). That’s because on the cashing out at issue we’re to explain how one is warranted in believing that one is not in the debasing scenario partly by appeal to the thought that one’s phenomenal evidence warrants one in believing that one has just inferred p from q, and this presumably violates the requirement that one be in a position to know the former prior to one’s being in a position to know the latter.

It seems to me, however, that the only alternative would be to say that the epistemic relationship between one’s phenomenal evidence, broadly construed, and the claim that one is not in the debasing scenario is abductive: the best explanation of why it seems to one as if one has just inferred p from q, and why this also seems to one to be the case a number of times in the past, is that one is in the normal environment, carrying out bone fide inferences. But such abductive responses to Scepticism are plagued with problems which I won’t rehearse here.Footnote 14

We can conclude then that given our new way of conceiving the Debasing Sceptic’s argument, none of Schaffer’s remaining commentators succeed in demonstrating that the argument is unsound, and nor is a certain fresh objection convincing. This suffices to demonstrate that Debasing Scepticism stands as a new and interesting form of scepticism, not readily dismissible.

It might be suggested, in objection, that all that the current paper achieves is a reductio on the principles that constitute the framework presented in Sect. 3.1. That might well turn out to be so; whether it does depends on whether it’s decided that we might be better off rejecting one of the Sceptic’s core principles instead, namely: Underdetermination, Priority, or Phenomenal Evidence. But even if it’s true that the current paper constitutes a reductio on the framework presented in Sect. 3.1, that would still be an interesting result, given that I have supplied what I hope are plausible-looking arguments for the elements of that framework.