The problem of divine hiddenness

The problem of divine hiddenness is rooted in two theses widely and traditionally advocated by theists:

• Divine mastery (DM) Insofar as anything depends simply on God, it will always be as God wills it to be.

• Voluntary relationship (VR) God desires a voluntary and meaningful relationship with every person.

DM articulates the idea that God is supreme, or absolute, or the one and only creator. Theists typically claim that God is both omnipotent and omniscient. DM is entailed by these: the will of God is the intention that God endeavours to realize so, since there are no non-logical limits on divine power and knowledge, whatever it is God intends to realize should be realized. Notice, though, that DM is weaker than this: it could be that some state of the world depends not simply on God but also on something else. In such a case, all DM indicates is that God will never be the cause of the failure. This is consistent with the free-will defence to the problem of evil: not everything depends on God since God has given us the freedom to be able to shape reality to some extent ourselves. A theist is thus free to say that, while God desires everyone freely to choose to be good, God cannot will this because the state in question is not something that is solely up to God. Some critics of theism dispute this defence, but I will not. Most theists would accept DM although some finitists would reject it.

VR is also intended to be formulated in such a way that many theists would subscribe to it. There are, however, theists who would reject VR – some Calvinists, for example, would. Thus, insofar as the hiddenness argument depends on something like VR, it does not present a problem for all theists. It is worth noting that the relationship in VR could take many forms in different theistic practices: some might stress worship, some obedience, and some a personal relationship with God. Typically, the practices that constitute the voluntary relationship are ones that presuppose belief: relating to God is deliberate and intentional.

Together VR and DM strongly suggest (but do not entail) the following: insofar as it depends on God, everyone has the opportunity to develop a relationship with God. The reason they do not entail this is because DM is formulated using the term ‘will’ while VR is formulated using the word ‘desire’ and these terms have different extensions. If God wills everyone to face the choice of whether to have a relationship with God or not then God equips and situates everyone to make the right choice. Since belief is a precondition of deliberate choice, VR and DM support the following thesis:

• Universal opportunity (UO) Everyone has the opportunity to believe in God and anyone having this opportunity should come to believe.Footnote ³

Prima facie, UO is false. This is what generates the problem of divine hiddenness. Two problem cases should be distinguished: the ignorant and the critics. Everyone should face the choice, and yet it appears that a number of people are so ignorant of God as not to have the opportunity to believe. In the language of William James, for many, belief in God is not a live option. Further, if someone has the opportunity to believe and yet does not, there must be something wrong with her – she is, after all, acting in a way that frustrates God's plan and undermines her humanity. Yet there do appear to be inculpable critics. Given that theists hold that a relationship with God is the greatest spiritual good available in this life, why would God make a world in which rational and reasonably moral people either consciously reject the conditions for such a relationship or never have the opportunity to pursue it? Why would God make a world in which run-of-the-mill faults and mistakes preclude acquisition of the greatest spiritual good available to us in this world? Likewise, why would God make a world in which a person's accident of birth precludes access to the greatest spiritual good? These questions are deeply perplexing for any theist who maintains UO, and also maintains that there are many reasonable and morally upright unbelievers.

It is very important to distinguish the ignorant from the critics, because a plausible explanation as to why God would make a world in which prolonged ignorance is possible may not be a very satisfactory explanation for why God would make a world in which prolonged, considered rejection of belief is possible. Even this distinction is not enough, because there are many kinds of ignorance (and criticism): ignorance may be rooted in tragedy (childhood death), learning disability, mental illness, or social context. Swinburne, for example, suggests that the good made available through believers having the opportunity to proselytize is sufficient to explain why God would make a world in which some are socially isolated from theism.Footnote ⁴ Suppose that's fine as far as it goes, it doesn't at all explain why God would make a world in which some people are cognitively incapable of belief.

Poston and Dougherty's solution

Poston and Dougherty reply specifically to J. L. Schellenberg's formulation of the hiddenness argument.Footnote ⁵ They appear to take Schellenberg as holding that if theism is true then every inculpable person should always explicitly believe in God.Footnote ⁶ They reply that God's plan can actually be satisfied with something considerably less than this, and that a kind of divine distance consistent with this plan makes possible other goods. Their view can be summarized this way: while it is not the case that everyone always explicitly believes in God, each person really does have the opportunity at some time in her life to begin to develop a meaningful relationship with God. So far as this world is concerned, God's plan is satisfied as long as people have the opportunity to develop at least a low-grade or partial de re belief in God. Here, the phrase ‘low-grade or partial’ is intended to capture the level of commitment or certainty of the belief, not its vagueness.Footnote ⁷

As will be shown below, the de re/de dicto distinction allows one to ascribe a belief to a person which that person would not recognize and might even deny. Poston and Dougherty's thesis implies that many of the ignorant and critics, despite appearances and despite how they themselves would characterize their beliefs or respond to characterizations of their beliefs, really might at some times in their lives have a low-grade de re belief in God. In other words, some so-called unbelievers may not actually be unbelievers at all – to some extent they believe in God, they just do not recognize themselves as doing so and would not acknowledge doing so. Although they do not mention him, Poston and Dougherty's view seems to have some resemblance to Rahner's theory that many non-Christians are actually anonymous Christians, and that many atheists and agnostics are actually implicit Christians.Footnote ⁸ Rahner does not, however, use the notion of belief de re in his analysis so I will set him aside as this is the concept that does the lion's share of the work in Poston and Dougherty's analysis.

What Poston and Dougherty seem to mean, then, is that the inculpably ignorant and critical may have beliefs that are about God even though they don't de dicto believe that God exists, and may even de dicto believe that God does not exist. While these seeming unbelievers may lack positive beliefs about God that use ‘God’, they really could have positive beliefs about God in which God is sufficiently present for the beginnings of a meaningful relationship. In other words, while it is true that reasonable non-de dicto belief in God occurs, if a perfectly loving God exists then what would be excluded is reasonable non-de re belief, and this may never occur.Footnote ⁹ Poston and Dougherty's solution stands out from others precisely for this reason: they undermine the premise of the hiddenness argument that inculpable unbelievers exist. Their view suggests that every inculpable person has or develops at some time at least a low-grade de re belief in God.

I agree with Poston and Dougherty that Schellenberg's argument (at least, as they formulate it) depends on an implausibly strong thesis. In particular, I think they are right to call attention to the temporal nature of belief, and I agree that the existence of periods of doubt and ignorance do not, by themselves, pose a problem for theism. The problem of divine hiddenness is most acute in the case of those whose lives may be characterized as lives of non-belief. If such apparent unbelievers really were, in the requisite sense, believers then the problem of divine hiddenness would be dissolved. I read Poston and Dougherty's discussion of de re belief as aiming at precisely such a dissolution: de dicto unbelievers could turn out to be de re believers.

As developed, I do not think their solution works. I will begin with a very general analysis of the de dicto/de re distinction and then use the results of this to argue that, to the extent that they rely on this distinction, Poston and Dougherty's reply to the divine hiddenness argument is unsuccessful. At the very least, much more argumentation is needed to show that certain de dicto unbelievers have de re belief in God.

Before proceeding to further analysis, it is worthwhile to note that although Poston and Dougherty only make a possibility claim, this is insufficient. In order for de re belief to be a significant part of a reply to the hiddenness argument, it would have to be the case that it is not merely possible that some de dicto unbelievers are de re believers, but rather it should actually be the case that some de dicto unbelievers really are de re believers. This follows from the nature of the hiddenness argument. The argument is premised on the apparent real existence of seemingly inculpable unbelievers (this is why UO is prima facie false). There are only three types of reply that preserve the theology against which the hiddenness argument is directed: (a) the seemingly inculpable unbelievers are really culpable; (b) the seemingly inculpable unbelievers are really believers and only appear to be unbelievers; and (c) the existence of inculpable unbelievers serves some greater good.

Replies of type (a) and (b) show that UO is true while replies of type (c) show that even though UO is false, it doesn't matter. Poston and Dougherty's reply is of type (b). The de re/de dicto distinction allows them to give a sophisticated explanation of how someone could appear to be an unbeliever (namely, by actually being a de dicto unbeliever) while really being, in the required de re sense, a believer. If the set of people who are de re believers and also de dicto unbelievers is empty, then the mere fact that the description of the set is not self-contradictory is beside the point, since UO would still be, as an empirical matter, prima facie false. The evidence against UO is empirical; therefore establishing that something is a priori possible is insufficient for countering the claim that UO is false. Poston and Dougherty would thus be better served by saying something like: inculpable de dicto unbelievers are probably de re believers.

There is textual evidence that Poston and Dougherty implicitly recognize the need to make an existential claim rather than a possibility one. They characterize Schellenberg's argument in such a way that it depends on the following two theses: (a) ‘If a perfectly loving God exists, reasonable non-belief does not occur’; and (b) ‘Reasonable non-belief does occur.’Footnote ¹⁰ They then write,

I interpret them as holding that if (a) is read in such a way that it is true, then (b) is unlikely, and if (b) is read in such a way that it is plausible, then (a) makes an implausibly strong claim (i.e. so interpreted, (a) is probably false). These alternate readings are made possible by giving consideration to the degree of belief, the temporality of belief, and the de re/de dicto distinction. Consider the first part of this dilemma: if (a) is true, then (b) is unlikely. This means, if (a) is true, then (b) is likely to be false. This means, if (a) is true, then reasonable non-belief (properly understood) probably does not occur. Applying this to inculpable de dicto unbelievers implies that they are probably really de re believers. In what follows, I set issues of possibility aside and focus on what would have to be the case to assert that de dicto unbelievers really are de re believers. I grant that it is logically possible for someone to be a de dicto unbeliever and a de re believer. I find it extremely unlikely to suppose that the inculpably ignorant and critical are, generally speaking, in this set.

The de dicto/de re distinction

The way Poston and Dougherty formulate it, belief de dicto takes the form of belief that S, where S is a statement expressing some proposition. A belief de re is a belief about something that some predicate applies to it. This way of drawing the distinction parallels a view discussed by W. V. O. Quine in which he distinguished belief as a dyadic relation between a believer and a proposition (de dicto) from belief as a triadic one involving a believer, an object, and a predicate (de re).Footnote ¹²

Poston and Dougherty give an example to illustrate the distinction they have in mind. Suppose Tom believes that Mark Twain was a great author, and suppose also that Tom does not know that Samuel Clemens is identical to Mark Twain. We ask Tom, ‘Is Samuel Clemens a great author?’. He might reply by pleading ignorance or even by denying this. And yet, since Mark Twain is Samuel Clemens, and Tom believes Mark Twain to be a great author, it must be that Tom believes Samuel Clemens to be a great author. The distinction between de re and de dicto belief ascriptions resolves the seeming incoherence of this. We can say that Tom believes the statement ‘Mark Twain is a great author’ and also that Tom believes of Samuel Clemens that he is a great author. In the first case, we make a de dicto belief ascription whereas in the second we make a de re belief ascription.

As Burge formulates it, when we make a de dicto belief ascription we describe Tom as believing some closed statement; whereas when we make a de re belief ascription we describe Tom as believing an open statement of something.Footnote ¹³ The precise details of this distinction are difficult to work out and are the subject of scholarly discussion that involves not only matters of propositional attitudes but also issues regarding indexicals, demonstratives, and modal logic. For present purposes the following characterization of de re belief ascription seems sufficient: A de re ascribes belief s to B if and only if A's ascription of s to B can be reformulated as B believes p of o without loss of meaning. In a case like this, it does not matter to A whether or not B would accept A's formulation of B's belief or not. Typically though, de dicto belief ascriptions are quite different. When A ascribes s de dicto to B, A is attempting to describe B's belief s from the inside – from within the conceptual resources B has and uses. A successful de dicto belief ascription is one that is true to the sayings of the believer.

Whether this distinction serves generally for a discussion of the logic of propositional attitudes is beside the point here since it does serve Poston and Dougherty's purpose. Namely, it appears to allow one to say that quite apart from what any unbelievers might say they may still believe significant things of God. In other words, despite how we might ascribe beliefs to unbelievers in a de dicto fashion we are free to use their behaviour to ascribe to them de re beliefs that they may not themselves explicitly affirm. We can ignore what Tom has said or is inclined to say, while still accurately describing him as believing of Samuel Clemens that he is a good author. While Tom's sayings justify our ascribing this belief to him (because he says that Mark Twain is a great author), in doing so we are not attempting to be true to what Tom himself says or would say.

As I've described it, de dicto and de re are different ways that we ascribe beliefs to others.Footnote ¹⁴ It might be thought that de re and de dicto are two different kinds of believing. Poston and Dougherty sometimes write in a way that suggests one of the following: it could be that a person has de dicto and de re beliefs or it could be that a person believes something in either a de re or de dicto fashion. After all, they appear to hold that unbelievers can possess de re beliefs and this is suggestive of distinct ways of believing or distinct contents of belief. On the view I gave above, de dicto and de re are ways of characterizing the activity of the belief-ascriber. On these two interpretations, however, the distinction either describes the beliefs themselves (literally having a de re belief or a de dicto belief) or the relation of the believer to the object of belief (believing something in a de re or de dicto fashion). For my part, I cannot make much sense of either of these interpretations, and so I develop my line of thought by considering conditions that hold on belief ascriptions. It should be noted though that little of the analysis below hangs on this – even if belief de re is a particular type of belief or way of believing, ascribing such a belief to someone must still address the conditions below.

Conditions of ascribing de re and de dicto

In order to know whether belief in God can be de re ascribed to the ignorant and/or critics, we must examine more fully the conditions under which such ascriptions are legitimately made. For this purpose it is useful to introduce the notions of mode of reference and mode of presentation. Modes of presentation are essentially related to consciousness: in order to think about an object or have an attitude, desire, aversion, or such of an object, that object must be present to consciousness in some way. The very same object can be present to consciousness in many ways. So, given an object and a subject, there are multiple modes of presentation for that object to that subject. For example, the very same object may be touched, heard, etc. Because of our familiarity with and the long discussion of sense perception, we tend to think of modes of presentation in sensuous terms but the bare idea of a mode of presentation does not necessitate this. At a bare minimum, a mode of presentation is something within consciousness that can serve as a point of contact between consciousness and something external to it. This point of contact could, if other conditions are fulfilled, serve as a causal basis for thoughts or attitudes about the object and could also serve as a way of receiving information from the object or about the object and could serve as a basis for checking judgements about the object.

Modes of reference are essentially related to inter-subjective consciousness: a consciousness that communicates with others or records its thoughts for future use. Modes of reference are the ways by which a consciousness can direct another consciousness (including itself at a later time) towards some object. Signs and gestures are common modes of reference. The task of distinguishing all the various modes of presentation and reference and analysing them is a momentous one that is well beyond our scope here. Notice, though, that linguistic modes of reference apparently open up whole new realms of objects for thought – acquiring language gives one the ability to think about any number of things not previously present to consciousness in the right way (think of historical events, for example). Words can, it appears, serve as both modes of reference and modes of presentation.

When we make de re belief ascriptions, we must hold that the believing subject's consciousness has some mode of presentation of the object. For example, it makes no sense to de re ascribe beliefs about the Nile to a child who has never perceived it, an image of it, or heard of it in any way. There must be some contact, however tenuous or mediated, between a subject and an object if that subject is to entertain thoughts or attitudes about that object. However, it is not enough that something is present in some way to consciousness to de re ascribe beliefs because having something present to consciousness in some arbitrary manner is generally not sufficient for being able to have thoughts or attitudes about that object.

Suppose a particle physicist and someone wholly ignorant of science watch a collision of sub-atomic particles in a bubble chamber. In a very real sense, both of them have a mode of presentation of the particles involved because both of them have visual experiences caused by the particles. Because of what the physicist knows, those visual experiences can serve as a foundation for having beliefs about the particles involved. For the non-scientist, they can't. One could, in such a case, ascribe de re beliefs about the particles to the physicist but not to the non-scientist. Consequently, in order to de re ascribe beliefs to a subject, some form of contact between the consciousness of the subject and the object is necessary but not sufficient. Just what some of these other conditions are we will return to after briefly looking at de dicto ascriptions.

Just as there is a connection between de re ascriptions and modes of presentation there is a connection between de dicto ascriptions and modes of reference. The reason for this is clear from above: de dicto ascriptions take into account the fact that believers don't just believe, they speak about objects. Insofar as de dicto ascriptions aim at being true to the speech of the believer, the believer's mode of reference to the object determines, in part, the veracity of the ascriber's de dicto belief ascription. We cannot de dicto ascribe ‘Samuel Clemens was a great author’ to Tom since he does not have ‘Samuel Clemens’ available to him as an appropriate mode of reference to the object. A thorough analysis of this issue requires much more, but it is sufficiently established that there is a connection between de dicto belief ascriptions and the modes of reference available to the believer.

There is reason to think that we can only make full de dicto belief ascriptions to beings capable of using language that have mastered language games involving phrases like ‘I believe’, ‘I know’, ‘I hope’, etc. Unless we are to imagine thought to be some kind of internal speech in a private and mysterious language (but one which can conveniently be translated into, say, English), it is problematic to ascribe de dicto beliefs to someone incapable of formulating sentences like ‘I believe z’, or ‘No, I don't believe x; instead I really believe y.’ Full de dicto ascriptions are not possible unless the person the belief is ascribed to can actually play the role, linguistically, of a believer – without her having this ability, there really isn't such a thing as how she would describe her beliefs from her own point of view because without such linguistic mastery she is unable to reply to questions like ‘What do you believe?’ and ‘What do you think?’. Without her mastering such language games, there is simply no way to be true to what she says she believes because there isn't anything she says she believes or even would say she believes.

In other words, the truth conditions for de dicto ascriptions are not in place except for sufficiently competent language users and such conditions are robustly present in users that have mastery of words like ‘believe’, ‘think’, ‘know’, etc. Very likely it is through interacting with children and making quasi or pseudo de dicto ascriptions that we introduce them to the practice of using the word ‘believe’. Once they are introduced to this, they can self-report their beliefs and we can make full de dicto ascriptions that attempt to capture their beliefs from their conceptual point of view – having a cognitive perspective is an achievement, not a given. The better language users are at reporting their beliefs and responding to reports of their beliefs, the more meaningful it is to accurately ascribe beliefs to them de dicto.Footnote ¹⁵

Returning to de re ascriptions: recall that a person can only have a belief about something if that something is present to the believer's consciousness in some mode. However, that is not enough. The history of science is illuminating here: prior to the discovery of oxygen, it was postulated that burning was caused by the release of phlogiston. The activity of oxygen played a significant causal role in many scientists' beliefs about phlogiston; nevertheless those beliefs were not beliefs about oxygen. The same is true of the relationship between large-scale molecular motions and caloric. The history of science is replete with similar examples. Just because an entity plays a significant causal role in a person's cognition and even if that entity is the answer to the cognition in question, it doesn't follow that the cognition is really about that entity. Advocates of the phlogiston theory did not have de re beliefs about oxygen nor did they have implicit beliefs about oxygen. They believed in oxygen not at all. Likewise, it would be ridiculous to de re ascribe beliefs about oxygen to Aristotle on the basis of the fact that Aristotle explicitly discusses air and that his discussion of air is causally related to oxygen (since he, for example, breathes). While phlogiston theorists might be, in principle, able to commit to the existence of oxygen, Aristotle is not even able to do this – the ontological category is simply not available to him because of his historical context.

We see clearly then that, at least in the case of highly theoretical terms, the mere fact that the ascriber is in position to see that some outside thing is playing a causal role in a subject's cognition (and hence what is present in the subject's consciousness) does not suffice for ascribing to that subject a belief in that object. In the above examples, part of what blocks successful de re ascriptions are the sorts of ontological categories the subject uses or has available.Footnote ¹⁶ Phlogiston theorists explained phenomena in terms of ontological categories that are incommensurable with modern chemical theory. Aristotle was not in any position even to be aware of such categories. Ascribing an ontological commitment to oxygen either would be uncharitable.

The ontology of a subject makes a difference to what may properly be de re ascribed to that subject. This holds even if we are not considering highly theoretical entities. Suppose Frank points towards the tail of a black cat as it trails around a corner (the rest of the cat is not visible to Frank). Frank says, ‘That is black’. Of what shall we say Frank believes to be black: a hair, some fur, a finger-tip sized surface, the cat's tail, the cat, the whole corner area? In order to say that Frank believes of x that it is black, we must find a suitable ontological category for x. Typically we would assume a kind of cognitive congruency between Frank and ourselves – we assume that his ontological commitments are similar to our own and that the sort of thing we would point to if we were in a situation like Frank's is what Frank himself is pointing to. So, in assigning some specific x as the object of Frank's pointing, we must try to be true to Frank's ontology. To be sure, we need not use the same categorical terminology Frank would – but the category we assign x to should have a strong resemblance to the one Frank would. In particular, if the identity conditions of objects in the category we assign x to are radically different than the identity conditions of objects in the category Frank does or would assign x to then our belief ascription is, at the very least, deeply problematic if not a failure.

For example, if we say that Frank believes of the tail that it is black and Frank later says that he really believed of the cat that it was black, then there isn't too much of a problem (since what we ascribed is entailed by what Frank says he actually believed).Footnote ¹⁷ But, if we say that Frank believes of the cat that it is black and Frank later says that he really believed only of the tail that it was black, then our belief ascription failed – Frank did not believe what we said he believed (and our ascription would still fail even if Frank never told us this). So, it is not only in the case of highly theoretical terms that the ascriber must be true to the ontology of the believer; this is a general condition. In fact, all of the non-controversial examples Poston and Dougherty themselves give are examples in which the ascriber is true to the ontology of the believer.

Not only does this constraint follow from a consideration of general cases, it is reasonable if we consider the nature of de re ascription itself. In de re ascription one does not need to be true to what the believer says, but one must still be true to how that believer interprets reality. We can accurately depict a person's behaviour without necessarily being true to her interpretation of reality, but we cannot accurately depict a person's mental life without being true to her interpretation of reality.Footnote ¹⁸ Given the fundamental role played by a person's ontology in her interpretation of reality, this means that when depicting a person's belief, one must, to be accurate, aim at being true to that person's ontology. Being true to a person's ontology is thus an a priori constraint on depicting a person's beliefs and attitudes.

In addition to the mode of presentation and the ontology of a subject, the de re ascriber must also take into account the directedness or specificity of the subject's cognition, attitude, desire, or feeling. Suppose Frank did not see or ostend the cat (or any part thereof) but that he does step on the cat's tail and subsequently is briefly startled; the cat darts away. The cat, let's suppose, does not register visually or aurally with Frank (perhaps Frank even exclaims, ‘What was that?’). Assuming Frank to be competent language speaker, we can safely de dicto ascribe to Frank the belief that there was something he stepped on. Can we, without further ado, say that Frank believes of the cat that he stepped on it, or that Frank believes of the cat that it startled him? I don't think so. Any such claim ascribes more directedness to Frank's thought than is there. For all Frank knows, he stepped on a towel, or a fold in the carpet, or a sock, or a dirt clod, or a salamander, or any of a number of other things. From the standpoint of Frank, there is no way to assign some specific ontological category to the thing he stepped on – this in spite of the fact that the thing he stepped on does fit an ontological category he really has available to him and regularly uses.

From the mere observance of a connection between a subject's conscious behaviour and some object discerned by the observer (and therefore conceptualized as a specific something through the cognitive apparatus of the observer) the observer is not entitled to ascribe beliefs de re or attitudes about that object to the subject. To be sure, there is much the observer can say, for example, ‘Frank was startled by the cat’. But, as soon as the observer makes a belief ascription to the subject, the observer takes on the burden of being true to that subject's consciousness and cognition. So, whatever object the observer says the subject has a belief or attitude towards must be true to the subject's mode of presentation (her attention is thereby in contact with something), the subject's ontology (her attention is directed at something of category x), and the subject's specificity of directedness (her attention is directed at a specific X).Footnote ¹⁹

Ascribing de re and de dicto are intertwined processes when the subject of ascription is a competent language user. For example, our de re ascription to Tom of the belief that Samuel Clemens is a great author: the basis for this is his linguistic behaviour with respect to the claim ‘Mark Twain is a great author’. In other words, we de re ascribe the Samuel Clemens belief to Tom on the same basis that we could de dicto ascribe to Tom the belief that Mark Twain is a great author. In general it seems true that if the believing subject is a sufficiently competent and informed language user, for any de re ascription, a corresponding de dicto ascription can be made.Footnote ²⁰

Very often when we ascribe a belief to a subject, we can reword that ascription in the form of a question to which the subject can directly respond. The reason for this is plain enough. In a de re ascription the ascriber is not attempting to be true to the modes of reference of the believer, whereas in a de dicto ascription she is, and informed, competent language users can, generally, refer linguistically (though perhaps not very exactly) to anything that they can think about. Thus, to transform a de re ascription into a de dicto one, one must merely figure out how the believer herself would likely refer to the object.

There are two important roles regarding a believer's speech here On the one hand, when ascribing sufficiently complex belief (for example, belief that involves theoretical entities) the lack of certain modes of reference may indicate a lack of proper mode of presentation and/or an ontological absence. If Aristotle had a word for oxygen, then he would have had something in thought that could guide and direct his judgement. That he has no word for it suggests that he has no sufficient mode of presentation. Moreover, Aristotle's inability to refer to oxygen linguistically is evidence that oxygen is not part of his ontology, and hence is not something about which he has beliefs. On the other hand, a person's modes of reference may be sufficient to allow her to speak of the object, and in this case what the speaker says or would likely say may act as a defeater to de re ascription. For example, it counts against ascribing de re beliefs about oxygen to certain phlogiston theorists because they know what oxygen is alleged to be and they explicitly deny that it exists. While in some cases this defeater can perhaps be overcome, still a very strong reason needs to be given to discount what an informed and competent speaker says about his/her own beliefs.

Application of analysis to de re belief in God ascriptions

The above analysis has been of a very general nature. In this section, I use it to show that, generally speaking, one should not ascribe belief in God to critics and the ignorant. This follows fairly straightforwardly from the above discussion of conditions that hold on belief ascriptions.

Before applying the analysis, some discussion of the word ‘God’ is necessary. Unlike ‘phlogiston’, ‘oxygen’, and ‘electron’, ‘God’ is not a highly theoretical or technical term in that it is not a word the proper usage of which is fixed by the commitments or conventions of a group of specialists. Theologians, for example, could not by convention determine what God is – if theologians did to ‘God’ what astronomers recently did to ‘planet’ they would be rightly ridiculed. On the other hand, ‘God’ is not much like ‘George Bush’ either: it is not a proper name for a medium-sized, publicly-ostendable object. Nor is ‘God’ like ‘cat’, a term applying to a category whose instances can be readily and publicly ostended. Also, ‘God’ is very different from ‘thing’ or ‘stuff’ insofar as ‘God’ is a word loaded with descriptive significance (though there isn't a great deal of agreement as to what that descriptive significance is): someone who says, ‘that event was caused by God’ has said something much more robust than someone who says, ‘that event was caused by a thing’; likewise, ‘I felt the presence of God’ differs markedly from ‘I felt the presence of some stuff’.

One can de dicto ascribe belief in God only to those familiar with ‘God’-language who use that language to make claims about God that presuppose divine existence. For example, an atheist who says, ‘God is by definition omnipotent’ does not thereby have a de dicto belief in God (even though he does believe this claim de dicto and this claim is, in a sense, about God). Plainly the ignorance of the ignorant and the criticism of the critics imply that one ought not to ascribe belief in God de dicto to them. Such is fully in accord with Poston and Dougherty's position. Their suggestion is that we can, often enough to answer the hiddenness argument anyway, de re ascribe belief in God to critics and the ignorant.

From the above analysis, four conditions holding on de re ascriptions of belief in God can be distilled:

1. (1) The ascriber must hold that there is some mode of presentation between God and the subject.

2. (2) The ascriber must hold that God has a place within the ontology available to the subject.

3. (3) The ascriber must hold that the attitude, belief, feeling, etc. of the subject is sufficiently specific that it can be legitimately interpreted as aimed at God.

4. (4) The ascriber must be able to explain away de dicto defeaters – things the subject explicitly says or would say that tend to count against ascribing belief in God to the subject.

To see how general these conditions are, consider the case of Tom. We can de re ascribe belief that Samuel Clemens is a great author to Tom even though we cannot de dicto ascribe such a belief to him. Condition 1: Samuel Clemens is present to Tom via ‘Mark Twain’; condition 2: Samuel Clemens is a person and Tom makes frequent use of this ontological category; condition 3: since ‘Samuel Clemens’ and ‘Mark Twain’ designate identical people, the belief is specifically directed; condition 4: that Samuel Clemens pleads ignorance when queried as to whether or not Samuel Clemens is a great writer does not defeat our ascription because Tom does not know that Samuel Clemens is Mark Twain. Each of these four conditions presents a problem with de re ascribing belief in God to unbelievers.

The first condition is problematic for making ascriptions to both the ignorant and critics. In order to make such an ascription, the ascriber must hold that the unbeliever really does have a mode of presentation of God, and further that this presentation is actually involved in certain attitudes or beliefs that are directed at God – it is far from obvious that all inculpable unbelievers meet this condition.

The second condition is a problem when ascribing belief to the ignorant. If the ignorant do not have a God-idea, or an analogue to that idea, or the resources needed to readily construct such an idea, then one cannot ascribe a God-directed belief or attitude to them. In order to ascribe belief in God to an ignorant person, one must attribute to her an ontological category whose identity conditions sufficiently mirror those of God. This is deeply problematic.

In the case of the ignorant, the third condition is related to the second: God is, let us suppose, at least a thing – so God has a place in everyone's ontology. But, if God fits in a person's ontology in only a very general way, then one goes well beyond what is warranted when ascribing to the ignorant person beliefs or attitudes that are specifically directed towards God.

The fourth condition is particularly a problem for critics: the claims of agnostics and atheists count strongly against ascribing belief in God to critics – these are very difficult de dicto defeaters to overcome since atheists and agnostics typically know much, much more about theology than Tom does about Mark Twain. In order to ascribe belief to individuals successfully de re one must either show how these conditions are met or why they don't apply to a particular case.

Poston and Dougherty do not consider any specific conditions under which beliefs may be de re ascribed. Nevertheless, they do give an example of what they apparently hold to be a common circumstance in which we can safely ascribe belief in God to seeming unbelievers. They write,

Even before applying our four conditions of de re ascription, as written this appears to be false. Suppose I receive a large inheritance when my uncle dies. Then I have benefited from him and he is my benefactor. I am grateful. It doesn't at all follow that I'm grateful to God. Suppose I live in a land with freedom of religion. I benefit from this. I'm grateful – grateful to the people who had the idea, the people who fought for it, and the people who maintain it. It doesn't follow at all that I'm grateful to God. Suppose I have a very strong immune system that keeps me healthy. I am benefited by nature. I'm grateful. Again, it doesn't follow that I'm grateful to God.

The problem is this: while God may be my ultimate benefactor, He is not my immediate benefactor. So, if I'm grateful to the immediate source of my benefit, I'm grateful, but not to God – not even ‘so to speak’. As written, their example fails to show that any unbeliever meets condition 1: it doesn't at all follow that because someone can be described as grateful for benefits she received that she can also be described as having a mode of presentation of God. Neither the mode of presentation of the benefits nor the mode of presentation of the immediate cause of the benefits (in cases where there is something present as the cause) is, in any obvious way, a mode of presentation of God.

The above criticism turns on Poston and Dougherty having in mind people who are grateful to the sources of their benefits. Such people do not, for that reason anyway, have a sufficient mode of presentation of God. However, Poston and Dougherty's actual text uses singular language: they write ‘grateful for their source’, and ‘expresses gratitude to the benefactor’. These grateful folks, who have already thought of the idea of a single ultimate benefactor and have expressed gratitude to that benefactor, are, I think, merely words away from belief in God. It strains credibility to hold that, in the general case, inculpable unbelievers who are critics or ignorant are in an analogous situation. Thus, either Poston and Dougherty have successfully described people who (but for something like terminology) believe in God as actually believing in God or they have failed adequately to indicate how unbelievers meet condition 1. There are non-trivial cases of the first sort (some polytheists, some mystics, and some spiritualists could be described as ‘knowing God under a different description’ just as Tom knows Mark Twain under a different mode of reference) – but they don't appear to be enough to cover the vast majority of cases of the seemingly inculpably ignorant and critical.

Poston and Dougherty develop at length an example which they consider to parallel the case of the grateful unbeliever and in which we really can safely de re ascribe belief. Someone, let's say Alfie, is isolated in jail. Alfie hears a tapping, suspects it is another person, and taps back. While Alfie is uncertain that there really is another person there, he is grateful to that person if there. Suppose there really is someone there, Franny. It makes sense to say, de re, that Alfie has low-certainty gratefulness towards Franny. They mean for something similar to hold between God and the grateful unbeliever. With Alfie, as in the example with Tom, we can see how the first three conditions are met and how de dicto defeaters could be overcome.

Unfortunately, this example doesn't help us understand de re ascribing belief in God to unbelievers. Franny is, in this example, the immediate cause of something that does have a presence in Alfie's consciousness (certain noises). Unless it could be shown that God is similarly an immediate cause for at least some of our benefits, we are still left with no basis for holding that unbelievers really do have a mode of presentation of God. Moreover, Franny is a human being and hence fits in Alfie's ontology. Conditions 1 and 2 don't present prima facie problems in the Alfie example as they do in the case of God. Moreover, it is plain how Alfie's gratitude is directed specifically at Franny as the relevant factor in the causal chain leading up to the hearing of tapping noises. Our familiarity with Alfie's imagination makes this obvious: we know exactly what it is for Alfie to understand the taps imagistically as produced by another human being on the other side of the wall. We can de re ascribe belief to Alfie because we can imagine something very much like what she herself imagines. But with the unbeliever there need be no such imagining of an ultimate benefactor on the other side. There need not be any isomorphism between what the unbeliever actually imagines when considering her benefits and what the believer would imagine or thinks the unbeliever should imagine. The relationship between the consciousness of the ascriber and Alfie is not analogous to the relationship between the ascriber and the unbeliever. So, the example does not illuminate how unbelievers could have belief in God legitimately ascribed to them de re.

Here is another way to consider this example. We know Alfie hears the noise and that she thinks to herself that some person is causing the noise (notice again that we are in a position to de dicto ascribe a number of beliefs to Alfie – we can, for example, ascribe to Alfie belief that a person is on the other side tapping and be true to the sorts of things Alfie says or is likely to say). As outside observers, we trace backwards from the noises present in Alfie's consciousness to the person tapping. We know this is what she's grateful to because something like the category ‘human being’ or ‘person’ is at work in her thought – this means that we observers know where to come to a stop when tracing back the causal chain.

Suppose instead that Alfie feels driven to madness by the incessant noise and directs hatred towards the thing, whatever it is, causing the noise. We outside observers witness that the noise is caused by dripping water which is caused by a leaking tap which is caused by a gasket corroded by oxygen that was produced via photosynthesis in Amazonian plants made by God. Is there anything here that we outside observers can safely say his hatred is directed specifically at? We can reasonably speculate about what he would direct his hatred at if he were informed of the entire causal background of the pernicious noises, but we cannot, without arbitrariness, say what his hatred is specifically directed at. The situation we are in when we safely ascribe gratefulness towards Franny to Alfie is simply not analogous to the situation Poston and Dougherty consider of ascribing gratefulness towards God to unbelievers.

Conditions 1 through 4 pose significant problems for ascribing belief in God to unbelievers. In some cases, it may be possible to show that these conditions are met. For example, one might succeed in showing how belief in God can safely be ascribed to certain polytheists. However, there are so many unbelievers that it would be impossible to justifiably claim that in all (or even most) cases the four problems can be overcome. So long as any of these unbelievers is inculpable, the problem of divine hiddenness remains.

Here is a final consideration to show this. If we really were in a position to ascribe de re belief in God to unbelievers, the above examples show that we would also be in a position to ascribe beliefs de dicto to unbelievers that are very much like belief that God exists – in the example given by Poston and Dougherty we should be able to ascribe de dicto to many unbelievers something like belief that there is some ultimate benefactor. Simply by supplying a little information about language and the world we can move Tom from believing that Mark Twain is a great author to believing that Samuel Clemens is a great author (and, at some point, if he refuses to change his beliefs he becomes culpable). Likewise, if Poston and Dougherty's thesis about de re belief in God is correct, then it should be relatively easy to convince the ignorant and sceptical to believe in God since they would, in the words of Davidson, ‘be only words apart’.Footnote ²² While this may be true for some so-called unbelievers (who, for example, might call God by a different name), it is certainly not generally true of the seemingly inculpably ignorant and critical. This too shows that belief de re does not substantially contribute to solving the problem posed by divine hiddenness. UO remains prima facie false.