1. Introduction
Many moral error theorists endorse the following argument against moral realism:
(M1) Moral realism implies the existence of categorical normativity.Footnote 2
(M2) Categorical normativity does not exist.
(M3) Therefore, moral realism is false. [M1, M2]
Call this the Metaphysical Argument.Footnote 3 Moral realists, however, have a counterargument:
(P1) Epistemic normativity is categorical.
(P2) Epistemic normativity exists.
(P3) Therefore, categorical normativity exists. [P1, P2]
Call this the Parity Argument.Footnote 4 The Parity Argument attempts to undermine (M2) of the Metaphysical Argument by establishing a parity between moral and epistemic normativity. There are three possible responses for the moral error theorist. First, one could accept the parity between moral and epistemic normativity but reject moral normativity nonetheless (Cowie, Reference Cowie2014, Reference Cowie2019). Second, one could reject the parity between moral and epistemic normativity (Olson, Reference Olson, Reisner and Steglich-Petersen2011a, Reference Olson and Brady2011b, Reference Olson2014, Reference Olson, McHugh, Way and Whiting2018). Third, one could accept the parity between moral and epistemic normativity, and embrace scepticism for both (Streumer, Reference Streumer2013, Reference Streumer2017).
We aim to demonstrate problems with each of these responses. Streumer defends his position in part by arguing that we cannot believe in a global error theory, but we argue that Streumer's defence of this claim does not have the philosophical significance that he thinks it does. Olson's argument fails to remove the problematic relation of categorical normativity from epistemic normativity. Cowie's argument weakens the primary motivation for (M2), namely the alleged queerness of categorical normativity, without providing a sufficient alternative motivation for (M2). Highlighting these problems does not refute moral error theory, but it does support the soundness of the Parity Argument by demonstrating the philosophical costs incurred by rejecting one of its premises.
In section 2, we address Cowie's response. In section 3, we address Olson's response. In section 4, we address Streumer's response. We conclude in section 5 by drawing some lessons for any future challenges to the Parity Argument.
2. The Special Property Argument
Cowie (Reference Cowie2014) defends what we will call the Special Property Argument. This argument accepts that epistemic reasons for belief are categorically normative and that categorically normative reasons are metaphysically problematic, and thus we should be sceptical of epistemic reasons for belief. However, there is the further claim that epistemic error theory is self-defeating so epistemic reasons for belief have a special property that denying their existence would be self-defeating. The conclusion is that there are epistemic reasons for belief despite the fact that they are metaphysically problematic. Note that this means that Cowie cannot strictly accept the Metaphysical Argument against moral realism. Instead, he must first argue as follows:
(S1) Categorical normativity is metaphysically problematic to the extent that it justifies scepticism of it.
(S2) Moral normativity is categorically normative.
(S3) Therefore, scepticism of moral normativity is justified. [S1, S2]
While also arguing the following:
(S4) We should always reject theories that are self-defeating.
(S5) Epistemic error theory is self-defeating.Footnote 5
(S6) Therefore, we should reject epistemic error theory. [S4, S5]
So we should accept the existence of epistemic normativity in spite of its metaphysically problematic nature. Cowie suggests that this argument should be seen as analogous to the Quine-Putnam indispensability argument for the existence of mathematical objects.
Mathematical objects appear to be abstract […]. And this poses a problem for those naturalists who are inclined to deny the existence of abstracta […], [yet] [a]ccording to proponents of [the Quine-Putnam indispensability argument], we are warranted in believing some mathematical objects to exist because those objects possess a special property that warrants their admission into the ontology. They possess the special property of being indispensable to the overall theory of cause and effect that constitutes our best total theory of the world. (Cowie, Reference Cowie2014, p. 417)
Platonism about mathematical entities seems prima facie incompatible with philosophical Naturalism, so many Naturalists are nominalists about mathematical objects. The Quine-Putnam indispensability argument argues that, though there are legitimate objections to the existence of abstracta on metaphysical grounds, we have stronger reasons to accept the existence of abstracta if they are indispensable to our best scientific theories. Given that mathematical entities are indispensable to our best scientific theories, we should believe in their existence in spite of the fact that they are abstract. Importantly, the idea is supposed to be that this vindication of mathematical facts does not vindicate abstract objects in general. Thus, if this analogy holds, the vindication of epistemic reasons for belief does not vindicate categorically normative statements in general.
This line of thought is straightforward enough. Take some objects X, Y, and Z all with a metaphysically problematic property Q so we can be sceptical of X, Y, and Z. Yet, when we take into account that X has a special property S, then we should believe in X in spite of it having property Q. In the case of numbers, this property Q is the property of being abstract, but they also have a special property S of being indispensable to our best scientific theories. In the case of epistemic reasons for belief, property Q is categorical normativity, but they also have a special property S of being such that denying their existence is self-defeating. Moral reasons for action have the same property Q as epistemic reasons for belief – categorical normativity – but they lack a special property S.
Although straightforward, we think this reasoning is flawed and undermines the metaphysical motivations for moral error theory. If we are excluding X, Y, and Z from our ontology due to the queerness of property Q, then we must say that property Q renders them too queer to exist. In other words, nothing with property Q can exist. Yet once we accept X into our ontology due to a special property S, then it is no longer the case that property Q renders objects too queer to exist. Thus, we can no longer say Y and Z cannot exist because they have property Q. If we wish to exclude Y and Z from our ontology while including X due to the indispensability of X, then it is no longer justified merely to point to property Q; one would have to give a different, independent justification for such exclusion.Footnote 6
We'll use a simple example to clarify this point. Let's say that Alice does not believe in ghosts or ghouls. Bob asks her why and she responds ‘because ghosts and ghouls have the property of being spooky and spooky things do not exist’. But then suppose that in a few years our best scientific theories tell us that ghosts exist. Bob then asks Alice if she believes in ghosts, and Alice reluctantly says ‘yes’ since they are now indispensable to our best scientific theories. Bob asks her if she now believes in ghouls and she responds ‘no, because ghouls have the property of being spooky and spooky things do not exist’. This is not a reasonable response. Alice is incorrect on her own terms. She believes spooky things do exist. After accepting that ghosts exist, it is no longer reasonable for her to merely point to the spookiness of ghouls to justify her claim that ghouls do not exist.
Making the parallel with the Quine-Putnam indispensability argument, Alice could instead say that she does not believe in ghouls because they are unnecessary for our best scientific theories. Notice, however, this is a different metaphysical principle for ontological exclusion. One principle, call it the No Spookiness Principle, says we should not believe in anything with the property of being spooky. Another principle, call it the Scientific Dispensability Principle, says that we should not believe in anything which is dispensable to our best scientific theories. Before Alice admitted that ghosts existed, she could sensibly use either of these principles to ground her scepticism in ghouls. However, after Alice admitted that ghosts existed, it only makes sense to use the Scientific Dispensability Principle to ground her scepticism in ghouls.
We think that (M2) is grounded in a principle similar to Alice's No Spookiness Principle, a No Categoricity Principle. Does the Metaphysical Argument have a similar principle to Alice's Scientific Dispensability Principle? No. In order for the Scientific Dispensability Principle to ground her scepticism in ghouls, the Scientific Dispensability Principle cannot merely say that we ought to believe in entities that are indispensable to our best scientific theories, it must also say that we ought not to believe in entities that are dispensable to our best scientific theories. Otherwise, the principle could not be used to exclude entities from our ontology. An analogous principle for Cowie's argument would have to be a No Self-Defeat Principle, which does not merely say that having the property of being such that denying their existence is self-defeating gives us strong reason for believing in things that have such a property, but also that we have strong reason for disbelieving in things which do not have that property. While a Scientific Dispensability Principle is not a completely implausible principle, a No Self-Defeat Principle is a completely implausible principle. Virtually nothing has a property of being such that denying their existence is self-defeating, so one could use this principle to ground scepticism in virtually everything. Given that this principle is required to make the parallel between the Special Property Argument and the Quine-Putnam Indispensability argument work, we conclude that the parallel fails.
One possible response is to say that the application of the No Self-Defeat Principle should be seen as analogous to the function of a defeater.Footnote 7 A defeater, in Pollock's (Reference Pollock1974) terms, is evidence that an agent receives which undermines evidence that the agent already possessed. One could say that the fact that categorical normativity is evidence, call it EQ, that neither moral normativity nor epistemic normativity exists. The fact that epistemic normativity has a special property, however, is evidence, call it ES, that epistemic normativity exists. So ES defeats EQ but only for epistemic normativity; EQ is still evidence against the existence of moral normativity.
The problem with this response is that the purported parallel between defeasible reasoning via empirical evidence and philosophical reasoning via metaphysical principles is illegitimate. In an ordinary defeater case, an agent's evidence is defeasible empirical evidence (e.g. evidence provided from perception, testimony, or memory). The defeating evidence only defeats a particular instance of empirical evidence. Intuitively, abiding by metaphysical principles in philosophical reasoning is not analogous to receiving empirical evidence via perception. The Metaphysical Argument says that, given its essential nature, categorical normativity cannot exist. It does not say that a thing having categorical normativity is defeasible evidence that such a thing does not exist, or makes a thing's existence unlikely. Accepting a single thing with such a property refutes a metaphysical principle of this kind, so it does not make sense to compare it with defeasible evidence.
But perhaps one could modify the Metaphysical Argument to try to make it more analogous to defeasible reasoning. First, consider again our example with ghosts and ghouls. Suppose that instead of grounding her disbelief in ghosts and ghouls on a categorical No Spookiness Principle, she instead grounds it in a general, but not categorical, scepticism of magical beings. So Alice has a low, but non-zero, credence in ghosts, and ghouls, but also in vampires, werewolves, wizards, etc. Now suppose that she visits a graveyard and sees a demon-like creature eating corpses. It would seem to be perfectly reasonable for her to greatly increase her credence that ghouls exist, while not greatly increasing her confidence that other kinds of magical beings exist.Footnote 8 Connecting this to the Special Property Argument, the claim is that one could accept that categorical normativity exists, but maintain a low credence in any kind of normativity being categorical, unless one gets some defeating evidence. The defeating evidence for epistemic normativity is that denying its existence is self-defeating. This understanding of the Special Property Argument seems to do justice to Cowie's reasoning.
However, this modified parallel is still illegitimate. The reasoning for Alice is ultimately justified by the idea that magical beings exist but their existence is rare. But what could it possibly mean to say that categorical normativity is rare? Note that this is not the same as saying the existence of categorical normativity is unlikely, since Cowie's argument assumes that categorical normativity exists. To talk about the rarity of a metaphysical property like categorical normativity seems to make a category mistake. It is like saying that sets are rare or saying that grounding relations are rare. Categorical normativity is not something that exists in space and time (though one might think that it supervenes on things that do exist in space and time), and it does not seem possible to define a meaningful density measure relative to some comparison class (e.g. with a suitable measure it could make sense to say that prime numbers are rare relative to the natural numbers). Of course, this characterization of categorical normativity is why some are sceptical of its existence tout court, but, again, the presumption for this section is that we have already accepted the existence of categorical normativity.
The crucial difference is that defeasible reasoning via empirical evidence involves considering whether a certain thing that exists has a particular property, while philosophical reasoning via metaphysical principles involves considering whether the properties themselves exist. Consider weird properties like the property of being grue (Goodman, Reference Goodman1954) or being a fridgeon (Fodor, Reference Fodor and Pylyshyn1987). It would make sense for a person to exclude these gerrymandered properties from their metaphysical picture of the world. But if someone did allow for these kinds of properties in their metaphysical picture of the world, it would only make sense for them to make comparisons of the rarity of things with these gerrymandered properties (e.g. if nearly every emerald was observed to be purple, then it may make sense to say that grue emeralds are rare). It would not make sense for them to say that gerrymandered properties themselves are rare. Either they exist or they do not. The same, we think, is true of categorical normativity. It does not make sense to claim that categorical normativity itself is rare.
The main difficulty with Cowie's response to the Parity Argument is that accepting categorical normativity, in any form, strips the Metaphysical Argument of its power, since one can no longer rely on an unequivocal denial of categorical normativity. Cowie's attempt at making a parallel to the Quine-Putnam Indispensability Argument via the Special Property Argument is a clever attempt at vindicating moral error theory without denying the parity premise, but we have argued that it fails.Footnote 9 In trying to avoid epistemic error theory, the metaphysical motivation for moral error theory is too greatly diminished. Better, we think, to try to defend moral error theory by either rejecting the parity premise or by accepting an error theory about all normativity.
3. Problems for Reducing Epistemic Normativity
Olson has argued in favour of two ways of rejecting the parity between epistemic and moral normativity:
Hypothetical Reduction: Epistemic reasons are reducible to hypothetical reasons or reasons about norms associated with a pertinent role or activity while moral reasons are not. (Reference Olson, Reisner and Steglich-Petersen2011a, Reference Olson2014)
Functional Reduction: Epistemic normativity is reducible to the function of belief while moral normativity is not reducible to any function. (2018)
In his early work, Olson (Reference Olson, Reisner and Steglich-Petersen2011a, Reference Olson and Brady2011b, Reference Olson2014) claims that the moral error theory can be generalized to epistemology without incurring any problematic results as Cuneo (Reference Cuneo2007) claims, such as no reasons for beliefs or no arguments for anything. Olson suggests that even though there is no irreducible epistemic normativity, there are still epistemic reasons since epistemic reasons are either reducible to hypothetical reasons or reducible to reasons about correct norms associated with a pertinent role or activity.Footnote 10 Later, Olson (Reference Olson, McHugh, Way and Whiting2018) explicitly rejects the parity premise, suggesting that epistemic normativity can be reduced to the function of beliefs. Therefore, unlike queer moral normativity which is supposed to be irreducibly normative (Mackie, Reference Leslie1977; Olson, Reference Olson2014, Reference Olson, McHugh, Way and Whiting2018), epistemic normativity is not queer.
However, we think that these two options are problematic because they lead to either the failure of arguing for any genuine epistemic normativity or the failure of arguing for a genuine moral error theory.
Our argument against Hypothetical Reduction is as follows:
(N1) Genuine normativity in a specific domain does not allow for the relevant domain-specific reasons to evaporate.Footnote 11
(N2) Reducing epistemic normativity to a mere desire or a specified norm of a role or activity would make epistemic reasons evaporate.
(N3) Therefore, epistemic reasons are not reducible to a mere desire or a specified norm of a role or activity. [N1, N2]
In this argument, the essential move is to argue for (N2).Footnote 12 Let's consider Olson's attempt to reduce epistemic reasons to hypothetical reasons. The presumption seems to be that if one can understand epistemic reasons in terms of hypothetical reasons, then the parity between epistemic normativity and moral normativity is successfully broken. However, this is what we deny. We argue that even for hypothetical reasons, in order to preserve even the weakest sense of normativity, one must posit a favouring relation that is irreducible to the agent's desires or goals. This means that one cannot successfully reduce normativity to hypothetical normativity without appealing to an irreducibly normative relation. But this is precisely what is supposed to be problematic about moral normativity.
Here is a simple example. Suppose you have a contingent desire of passing a logic exam tomorrow which gives you a hypothetical reason to read the relevant logic book tonight. Suppose you also have a desire to watch TV tonight which gives you a hypothetical reason to watch TV tonight. There are four elements in the former hypothetical reason: the proposition ‘read the logic book tonight’, your desire to pass the logic exam, the context of evaluation, and the favouring relation between your desire to pass the exam and your reading the logic book. In order for your reason to read the logic book to be genuinely normative, it must contain the favouring relation that cannot be influenced by your desire to watch TV, or any other desire. If you do want to pass the exam, once you fix your desire of passing the exam, you must have a hypothetical reason to read the logic book tonight, no matter how strong your desire to watch TV is. Therefore, even given that hypothetical reasons are related to the contingent ends or desires of agents, these reasons are not determined by but merely indexed to the agents’ ends or desires. Ultimately, the normative force of hypothetical reasons is generated by the favouring relation, which is independent of agents’ ends or desires.
One may argue, in alignment with Quine's (Reference Van Orman Quine, Hahn and Schilpp1986) suggestion, that in this case the normative force stems from the goals or desires rather than a queer favouring relation. Once you settle the desire of passing the logic exam, it follows that you have a reason to read the logic book tonight. The normative force of this reason comes from your desire to pass the exam. There is no need to add on any extra normative relation, because the reason to read the logic book is based on your desire of passing the exam, along with a descriptive and natural fact that studying tonight will most likely lead to your desire being met. Like Quine's view, this reply does not deny that there is some sort of normativity in hypothetical reasons and there is an internal relation between reading the logic book and the desire of passing the logic exam; but it rejects the claim that the normativity is grounded in somewhere outside of agents’ desire or goals.
However, we suggest that this kind of response trivializes practical rationality and thereby evaporates hypothetical reasons,Footnote 13 because there will be no genuine failures in practical rationality.Footnote 14 Suppose the normative force of reading the logic book comes from the desire of passing the logic exam and suppose the only internal relation between reading the logic book and the desire of passing the logic exam is a descriptive and natural fact that doing so will most likely lead your desire of passing the logic exam to be met. If that is the case, then once you fail to read the logic book tonight, there are three possibilities:
(1) you simply fail to desire to pass the logic exam, or
(2) you simply fail to have the strongest desire to pass the logic exam, or
(3) you simply fail to appreciate the natural fact that doing so will most likely lead to your desire being met.
However, if these are only explanations for why you fail to read the logic book, then we could not criticize you for being practically irrational. Instead, we could only say that you simply fail to desire to pass the logic exam, or your desire to pass the logic exam is not strong enough, or you simply fail to appreciate the fact that reading the logic book will most likely lead your desire to pass the logic exam to be met. But firstly, this is an unintuitive implication. When we criticize someone for being practically irrational, we typically do not mean to criticize her motivational set for not having a desire X or only having a weak desire X, nor do we criticize her for failing to appreciate what means satisfy her desire. Furthermore, this leads to the fact that there are no genuine failures in practical rationality and the fact that there is no need for the correctness condition that separates the genuine failures from the success of practical rationality. As a result, there is no practical rationality and any sense of normativity for hypothetical reasons evaporates.
Next, we consider another related proposal from Olson: epistemic normativity is not queer because it can be reduced to correct epistemic norms associated with epistemic activities and/or natural relations that promote truth. Specifically, they are natural rules that promote truth or true beliefs.
However, this suggestion shares similar problems with the position in epistemology which suggests that evidence or the evidential support relation is merely a truth-conducive natural relation. Consider this example: at a party, a group of friends are playing with a Ouija board. They first try many ordinary questions, and the board gets a lot of them wrong. And then, they ask the board about lottery numbers and the Ouija board gives the correct answer. If a person were to buy a ticket and believe that they will win, this belief is clearly irrational. Would they thereby be judged to be rational if it was stipulated that this Ouija board had a magical connection to correct lottery numbers? Surely not. Truth and justification are connected, of course, but justification cannot merely be a kind of truth relation. Similar cases can be found in Bonjour (Reference Bonjour1980) and Cohen (Reference Cohen1983), where the former shows that truth-conduciveness is not sufficient for epistemic justification and the latter shows that truth-conduciveness is not necessary for epistemic justification.Footnote 15 These cases show that epistemic reasons are not reducible to a consideration that leads to the truth. Therefore, reducing epistemic justification to truth-conducive norms would make epistemic reasons evaporate.
Our point is this: reducing epistemic reasons to hypothetical reasons or reasons related to a truth-conducive norm does not necessarily undermine the parity premise since we think that there will still be an irreducibly normative element. On the one hand, rejecting the irreducibly normative favouring relation in hypothetical reasons will undermine practical rationality since it implies that there are no genuine failures of practical rationality, and if there are no genuine failures of practical rationality, then there are no genuine hypothetical reasons. On the other hand, trying to reduce the epistemic justification relation to mere truth-conduciveness makes the epistemic reasons evaporate. (N2) stands.
In his later work (Reference Olson, McHugh, Way and Whiting2018), Olson argues for functional reduction. He reduces epistemic normativity to something else other than a desire or a norm of an activity. He proposes a functional view of epistemic normativity; that is, epistemic reasons can be reduced to the function of belief of ‘helping subjects to navigate and explain the world by carrying correct, or largely correct, information’ (Reference Olson, McHugh, Way and Whiting2018, p. 115).
We think that this option is more promising in terms of preserving the genuine normativity of epistemic reasons. However, we do not think that this alternative approach succeeds as a response to the Parity Argument. First, it is not clear that reducing epistemic normativity to the function of beliefs would support believing error theory. Instead, it may support believing in a naturalistic form of moral realism. Second, there is no good specification of what counts as the correct epistemic end/goal that could serve in a teleological account of epistemic normativity. Our argument against functional reduction is as follows:
(N4) A genuine moral error theory is different from naturalistic realism on whether morality is ‘real’; that is, whether moral-normative properties exist.
(N5) Some naturalistic moral realists argue for a naturalized functional normativity of morality.
(N6) Then, a genuine error theorist cannot argue for a naturalized functional normativity of morality. [N4, N5]
Let us explain (N4) to (N6). Both moral error theorists and naturalistic realists deny that categorical moral facts exist. However, for naturalistic realism, our ordinary discourse has an ontological commitment to naturalized moral facts (or a reducible normative relation, using Olson's terminology) which fit in our natural and scientific world and are no less real than other entities. Thus, for naturalistic realists, there is no error in our ordinary discourse. According to moral error theorists, our ordinary moral discourse and concepts are full of errors because they refer to categorical moral facts (or an irreducible normative relation). Thus, according to error theorists, our ordinary discourse about morality has an ontological commitment to something that is not ‘real’. Now, if a naturalistic realist can plausibly argue for a naturalized functional normativity of morality, then an error theorist cannot argue for a naturalized functional normativity of morality without providing some good reason to hold the moral error theory over naturalistic realism. This is what Cuneo (Reference Cuneo, Schmechtig and Grajner2016) called the ‘destabilizing result’ of the moral error theory.
Olson makes two attempts to avoid this result. First, he claims that there is no plausible theory of naturalizing moral facts. However, we think that there are a lot of candidates. For example, David Copp (Reference Copp2009) sketches a plausible functional theory for all kinds of normativity, including moral normativity. Copp's basic idea is that, as humans, we face a lot of generic problems (i.e. the problem of sociality), and each of these problems generates a corresponding normative system that helps us better cope with the problem if we subscribe to it than would otherwise be the case (i.e. the normative system of morality). Copp, as a naturalist moral realist, can plausibly argue for an intelligent functional view of moral normativity, and this view seems no less plausible than Olson's functional view of epistemic normativity. If epistemic normativity can be reduced to functional roles, then it seems plausible that moral normativity could similarly be reduced to functional roles.
Olson's second attempt to avoid the destabilizing result of the moral error theory is to argue that we can accept the naturalized functional normativity of epistemology while rejecting the naturalized functional normativity of morality: a naturalized functional normativity of epistemology suggests that the function of belief is navigating and explaining the world to us via largely correct information. However, this proposal is vague. If this proposal means that epistemic reasons are merely truth-conducive reasons, then this simply leads back to the problem of evaporating epistemic reasons. It could mean that epistemic normativity is reducible to a truth-related goal, regardless of whether or not this goal is actually truth-conducive. However, this faces a new problem: how to specify this truth-related epistemic goal? Following Eder's (Reference Eder2021) discussion, if we understand the epistemic goal as a global ratio end such as having the best possible ratio between true and false beliefs, then we suffer from the trade-off problem: sometimes we are required to believe against the evidence in exchange for true beliefs.Footnote 16 But if we understand it as a local end such as pursuing truth or avoiding falsity, then either we add in some non-epistemic element to that goal in order to generate an epistemic reason to believe anything, or we fail to explain epistemic normativity at all. For example, if the epistemic goal is specified as avoiding falsity, then the best way to satisfy that goal is to believe nothing; but if it is specified as pursuing truth, then we have to recognize some reasons as epistemic reasons which are non-evidential but still aiming to the truth, such as recognizing a reason to breathe as an epistemic reason, since living will contribute to the acquisition of true beliefs. In both ways of understanding epistemic ends, the sense of epistemic normativity evaporates.Footnote 17 Because if there is no way to correctly specify what the epistemic end is, then there is no way to evaluate beliefs according to the epistemic end.
So we conclude that Olson does not have good reasons to reject the parity premise by reducing epistemic normativity, since neither Hypothetical Reduction nor Functional Reduction succeeds.
In this section, we first argued that the intimate relation between evidence and reasons for beliefs is not reducible to the agent's desires, goals, the correct norms associated with the activities that the agent participates in, or the function of belief. We conclude that Olson's attempt to break the parity premise fails. Next, we consider the remaining option that error theorists have: defending a global error theory.
4. Believable Errors
In this section, we address the third possible response to the Parity Argument: acceptance of both moral error theory and epistemic error theory. Streumer (Reference Streumer2017) accepts the parity between epistemic and moral normativity but argues that neither exists. Call this the Error Theory. Moral error theorists and moral realists alike have both rejected this view on pain of being self-defeating.Footnote 18 Though it is close to being self-defeating, we do not think it is, strictly speaking, since the view does not imply its own falsity. Instead, the theory implies that, if it is true, then there is no reason to believe that it is true. In Cuneo's (Reference Cuneo2007) terminology, the theory is toothless. Given how unpalatable a toothless theory is for a philosopher, it is tempting to disregard any theory which has this property. However, we will set aside the possibility of objecting to the Error Theory on these grounds.
If one does not object to the Error Theory on the grounds that it is toothless, it is open for one to object on Moorean grounds. By this, we mean that one can assert that one's confidence in certain judgments – call them Moorean premises – ought to be higher than any philosophical premises or theories which purport to undermine the truth of the Moorean premises. In this context, the Moorean premises would be propositions like ‘a person in pain has reason to believe that they are in pain’ or ‘a person looking at their hands has reason to believe that they have hands’. Streumer's theory says that a person in pain has no reason to believe they are in pain and a person looking at their hands has no reason to believe that they have hands. And this is not a contingent truth, but a metaphysical necessity. So not only has no person in history ever had a reason to believe that they are in pain or that they have hands, but it is metaphysically impossible for a person to have a reason to believe that they are in pain or have hands.
The Moorean-inspired response to Streumer's theory is to say that no matter how convincing the philosophical reasoning in support of it may seem, it is rational to reject a philosophical theory with such wild implications. This is not a dogmatic dismissal, but an expression of relative likelihood. The claim is that it is more reasonable to believe that there is something wrong with the complex, controversial reasoning in favour of the Error Theory than it is to believe that a person looking at their hands has no reason to believe that they have hands. This holds even if one cannot identify where specifically the argument went wrong. An uncontroversial example of this move in a different context are Zeno's paradoxes. Even if, upon hearing the paradoxes for the first time, a person cannot refute Zeno's arguments against the impossibility of motion, it is still rational for them to believe in the possibility of motion. They may rationally conclude that Zeno's arguments fail, even if they cannot pinpoint where Zeno's arguments fail. Similarly in the case of the Error Theory, one may conclude that Streumer's arguments fail, even if one cannot pinpoint where Streumer's arguments fail. Call this the Moorean Objection.
When one is faced with a Moorean premise, one can respond by attempting to debunk it. Streumer attempts to do this by arguing that we cannot believe the Error Theory. He thinks that this provides an explanation for why this view is so unacceptable to us; we literally cannot believe it. If the Error Theory is true, then it predicts that we could not believe it. Thus, Streumer reasons, the fact that we have a very high credence that the Error Theory is false cannot be used against his view. This does seem to be a relevant difference between other theories for which a Moorean Objection might be used (it may be difficult for us to believe that motion is impossible or that we do not have hands, but is it impossible for us to believe these things?). These other theories, while perhaps hard to believe, do not predict that it would be impossible for us to believe them.
Given that unbelievability is an essential feature of the Error Theory, Streumer thinks that objections to the Error Theory should not be predicated on the assumption that we can believe the Error Theory since that would beg the question. For instance, he directly addresses the Normative Objection, which claims that the Error Theory would undermine all of our normative judgments, including our deepest and most important moral convictions, like the claim that torturing children for fun is wrong.Footnote 19 To give up on this conviction in favour of another would require that we be more confident in other convictions, goes the objection, however, it is difficult to see what we could be committed to more strongly than the claim that torturing children for fun is wrong.
Streumer says this objection fails since ‘we cannot believe the error theory about all normative judgments that I defend, this theory cannot undermine any of our normative judgments at all, let alone our deepest and most important moral convictions’ (Reference Streumer2017, p. 176). In other words, Streumer thinks our inability to give up on our moral convictions goes hand in hand with our inability to believe the Error Theory. So, says Streumer, our inability to give up on our moral convictions is not evidence that the Error Theory is false; instead, it is evidence that it is true.
Now that we have explained why Streumer is arguing that we cannot believe the Error Theory, we can move on to Streumer's argument that we cannot believe the Error Theory. Streumer claims that a genuine belief must satisfy the following four criteria.
(B1) A person believes that P only if this person is very confident that P.
(B2) A person believes that P only if this person adequately understands P.
(B3) A person believes that P only if this person believes what he or she believes to be entailed by P.
(B4) A person believes that P only if this person does not believe that there is no reason to believe that P.
Streumer argues that if P = ‘the Error Theory’, then these criteria cannot be met. Thus, it cannot be the case that a person can believe the Error Theory. The reasoning is straightforward. If a person adequately understands the Error Theory, then they know that the Error Theory entails that there is no reason to believe the Error Theory. So in order for a person to believe the Error Theory, they must believe that there is no reason to believe the Error Theory, but that would contradict (B4). Thus, no one can believe the Error Theory.
One immediate worry with Streumer's claim is that belief talk seems to be equivocal and there are different theories of belief.Footnote 20 At first pass, it seems fair to restrict one's discussion of belief to mean something specific. Indeed, Streumer restricts his use to explicit, occurrent, full beliefs, and he takes all of (B1)–(B4) to be partly stipulative (Reference Streumer2017, p. 130, p. 146).Footnote 21 So Streumer concedes that there are senses of belief that do not abide by his criteria, yet maintains that there is a sense of belief in which we cannot believe the Error Theory.
But note that Streumer is not arguing that this is a mere quirk of his theory. He does not think that the unbelievability of the Error Theory in this stipulative sense is simply an interesting consequence of his view. Rather he claims that it has important philosophical significance. The significance, he claims, is that it rebuts objections to the Error Theory. So, even if one were to grant that there is a sense of belief such that we cannot believe the Error Theory, one can still ask whether the philosophical claims that Streumer makes actually follow from this. It is not enough for Streumer to simply demonstrate that there is a sense of belief in which it is true that we cannot believe the Error Theory. Rather, he must demonstrate that we cannot believe the Error Theory in the right sense of belief.
What is the right sense of belief? In order for the unbelievability of the Error Theory to have the philosophical upshot that Streumer takes it to have, it must be the sense of belief that is used in objections to the Error Theory. What is the sense of belief used in objections to the Error Theory? We think that it must be the ordinary sense of belief. That is, we think it cannot be a technical term or a philosopher's creation. Remember it is not just philosophers who think the Error Theory is implausible and are unlikely to believe such a theory. The essential points of the Moorean Objection and the Normative Objection could be articulated without philosophical definitions. For Streumer's claim to have any philosophical significance, it is not sufficient to merely show that there is a stipulated sense of belief in which we cannot believe the Error Theory. Rather, he must show that his stipulated sense of belief actually aligns with the ordinary sense of belief.
It seems doubtful that there is such an alignment. Indeed, there seem to be counterexamples to each of his conditions for belief. Regarding (B1), consider the following:
(1) Suzy believes that it will rain tomorrow, but she is not very confident that it will rain tomorrow.
(2) Smith believes that there are 354 jelly beans in the jar.
In the ordinary sense of belief, we think that both (1) and (2) make sense. However, both conflict with (B1). It is explicit in (1) that Suzy is not confident that it will rain, and it is implicit in (2) that Smith has low confidence that there are 354 beans in the jar (indeed, it is reasonable to assume that Smith's credence is below .5).Footnote 22 If this is correct, then (B1) is false for the ordinary concept of belief.
Regarding (B2), consider the following:
(3) Bailey believes that electrons can entangle with each other, though she does not really understand quantum entanglement.
This seems perfectly reasonable. Many non-physicists have read about quantum entanglement and are convinced that quantum entanglement is a real phenomenon. Do we require that these people fully understand quantum entanglement in order to believe it? We do not think so. Does one have to understand the mathematics of quantum mechanics to believe in entanglement? If so, then only physicists believe in quantum entanglement. Does one have to understand what quantum mechanics says about the world in order to believe in quantum entanglement? If so, then it is doubtful that we could correctly identify anybody who believes in entanglement, since physicists and philosophers are deeply divided on this issue. For the ordinary sense of belief, even if understanding is required, it cannot be as strong as the requirement given by (B2).
Preface cases seem to be a counterexample to (B3). An author can believe all of the claims that they make in a book, while also believing that they made a mistake, even though the former entails that they did not make a mistake. Streumer (Reference Streumer2017, p. 134) says that this is not a counterexample to (B3), since the preface case involves someone not accepting what is entailed by multiple beliefs, whereas (B3) is only a claim about what is entailed by a single belief. So Streumer takes the preface case to be the claim that someone could believe A, believe B, …, believe Y, believe Z while not believing the conjunction A & B & … & Y & Z. This seems to be a reasonable response, but only insofar as there are not any cases where a person believes a conjunction of propositions but fails to believe what that conjunction entails. Yet there do seem to be such cases. Suppose someone believes, quite reasonably, that a 100-year-old man is old and a two-year-old child is not old. Suppose they also believe, due to the sorites paradox, that this entails that a billionth of a second can make the difference between someone being old and not old. However, suppose this person maintains a belief that a billionth of a second cannot make the difference between being old or not old. This is an incoherent belief state, and they do not just believe it to be incoherent. They know it to be incoherent. But suppose that they maintain their beliefs nonetheless.Footnote 23 If Streumer were correct, then such a hypothetical person is conceptually impossible. But not only does this seem conceptually possible, it is plausible to think that this is actually true of people who understand the sorites paradox. Indeed, it is likely that many people believe different philosophical theses knowing full well that they conflict with one another. Ought we to say that such cases are, contra appearances, actually impossible? Again, it seems, we have a case where our ordinary understanding of the concept of belief has no problem accounting for a case that Streumer's stipulative definition of belief says is impossible.
The most crucial constraint for Streumer is, of course, (B4), yet here too there seem to be direct counterexamples to it. Suppose an epistemology professor convinces a student, via the Agrippan trilemma, that they have no reason to believe what they believe. It would be difficult to maintain that, upon being convinced, the student would give up on all of their beliefs. Presumably, there will still be many things that the student believes, but every such belief that the student has would be a counterexample to (B4), since he would be believing that P while also believing that there is no reason to believe that P. Of course, there is clearly a tension in the student's beliefs, but our claim is simply that it is possible for this kind of tension to exist.Footnote 24
It is available to Streumer to say that our hypothetical student does not really believe that the Agrippan trilemma shows that there is no reason to believe anything (or that Suzy does not really believe that it will rain tomorrow, or that Smith does not really believe that there are 354 jelly beans in the jar, or that Bailey does not really believe that electrons can entangle with each other). This kind of response is always available when one is using a stipulative definition; one can just further stipulate that such a case is not correctly employing the stipulative definition. And at times it seems that Streumer does exactly this:
I take my use of the term ‘belief’ to be partly stipulative: I take (B3) and (B4) to pick out a correct way to use the term ‘belief’, not the correct way. I therefore think that if a view about the nature of belief is incompatible with (B3) or (B4), it does not use the term ‘belief’ the way I do when I say that we cannot believe the Error Theory. (Reference Streumer2017, p. 146, emphases in original)
Given all of our considerations in this section, we hope that it is clear that this move essentially concedes defeat. If Streumer is not attempting to pick out the correct way to use ‘belief’, then he is not in a position to claim that the ‘unbelievability’ of the Error Theory has philosophical significance.Footnote 25
Consider the following analogy. Suppose Paul is arguing against dualism in favour of physicalism. His argument is the following:
(E1) Only physical things can be causally efficacious.
(E2) A thing exists if, and only if, it is causally efficacious.
(E3) Therefore, only physical things exist. [E1, E2]
Now suppose Paul is faced with the following counterargument:
(C1) If a certain counterfactual account of causation, call it CAC, is correct, then (E1) is false.
(C2) CAC is correct.
(C3) Therefore, (E1) is false. [C1, C2]
How could Paul respond? Available responses include challenging (C1) or challenging (C2): Paul could argue that CAC does not imply the falsity of (E1), or he could argue that CAC is false. An inadequate response, however, would be for Paul to say, ‘My definition of ‘cause’ is partly stipulative. I'm not claiming that my way of using ‘cause’ is the correct way to use the term, only a correct way. And on my way of using ‘cause’ (E1) and (E2) are true’. This is an inadequate response because it does not challenge the soundness of the counterargument. Of course, one could stipulate a sense of ‘cause’ in which only physical things could cause other things, but that wouldn't vindicate physicalism. One would have to ensure that only physical things can be causally efficacious simpliciter or, at least, that only physical things can be causally efficacious in the important or the relevant sense of ‘cause’. So even if Paul could show that there is a sense of ‘cause’ in which only physical things could be causally efficacious, this alone would be an insufficient response to the counterargument Paul faces.
The same point applies to Streumer. His claims regarding the believability of the Error Theory are only relevant to the objections to the Error Theory if his stipulative sense of ‘belief’ is the sense of belief in which such objections are couched. Yet we do not think this is the case because (a) we think that the important sense of belief is the ordinary sense of belief, and (b) it is doubtful that Streumer's conditions are true of the ordinary sense of belief. The upshot of our arguments is significant since it implies that Streumer does not have an adequate response to objections like the Moorean Objection or the Normativity Objection.Footnote 26
5. Conclusion
While we have focused on particular arguments from Cowie, Olson, and Streumer, they are representatives of the three general possible responses to the Parity Argument. We will conclude by drawing some lessons from our critiques of these authors – lessons that we think that any moral error theorist attempting to respond to the Parity Argument ought to keep in mind:
1. The motivations for the Metaphysical Argument should not be weakened.
2. The reduction of epistemic normativity, if possible, does not necessarily make moral error theory more plausible.
3. Epistemic normativity should not be overintellectualized.
Cowie's argument tries to give a principled way of maintaining that moral normativity ought to be rejected because of its queerness while epistemic normativity ought to be accepted in spite of its queerness. We pointed out that this weakened the motivation for the Metaphysical Argument since it accepted the existence of some categorical normativity. Cowie tries to make a parallel with the indispensability argument for the existence of mathematical objects espoused by Putnam and Quine; however, we showed that his metaphysical principles do not hold up under scrutiny.
Olson's argument tries to break the parity between epistemology and morality by reducing epistemic normativity. However, embracing reducible epistemic normativity might not be a good strategy for moral error theorists to respond to the Parity Argument. First, reducing epistemic normativity often leads to epistemic reasons evaporating. In our epistemic activity, the goal of truth-seeking usually plays a big role, but that does not mean that epistemic normativity can be reducible to the goal of truth-seeking. Second, Olson's updated moral error theory seems to collapse with some versions of naturalistic realism, so the pertinent debate between error theorists and realists on whether moral facts exist may become a verbal dispute. After all, a reducible moral fact is still a moral fact, according to some naturalistic realists.
One might respond with the question ‘Why should error theorists care about preserving genuine epistemic normativity? They are already advancing a radical philosophical thesis, so why would capturing something ‘close enough’ to genuine epistemic normativity be insufficient for their purposes?’ The issue here is for those error theorists who do not espouse Streumer's Error Theory. If they take themselves not to be rejecting normativity tout court, then they ought to explain how genuine epistemic normativity can be preserved.
We think that our attitude to moral error theorists who would accept revisionary epistemic normativity instead of embracing the Error Theory is similar to the one held by Derek Parfit toward certain forms of expressivism that reject that normative claims are in the business of getting things right or wrong. He suggested that if an expressivist maintained that normative claims merely express certain attitudes, then they should become moral error theorists (Parfit, Reference Parfit2017, p. 234). We think that if a moral error theorist's view about epistemic normativity cannot vindicate the idea that epistemic normativity is genuinely normative, then one should just join Streumer in defending the Error Theory. And if they insist that their revised, naturalistic account of epistemic normativity is genuinely normative, then they face the question of why a revised, naturalistic account of moral normativity must be rejected.
Finally, while Streumer aims to reject the existence of epistemic normativity, when he discusses the ‘unbelievability’ of his Error Theory he actually employs an overly intellectual notion of epistemic normativity. We emphasized that Streumer's notion of ‘belief’ was not in accord with the ordinary notion. This reveals itself when the locution ‘really’ is used by Streumer – if one does not satisfy (B1)–(B4) then one does not really believe.Footnote 27 To demand that a genuine belief must satisfy (B1)–(B4) is to overintellectualize the concept. By raising the stakes of what it takes to be a genuine belief, Streumer is overintellectualizing the concept of epistemic normativity that he is purporting to reject. But this is not epistemic normativity tout court. It's not just special beliefs that Streumer is identifying that are subject to epistemic norms, but ordinary beliefs as well. It is when we consider belief in the ordinary context – the primary realm of epistemic normativity – that his claims flounder.Footnote 28
Open access
