In Anti-Individualism and Knowledge Jessica Brown criticises views of content that combine Fregean Sense and anti-individualism. Brown assumes that all Fregean theories are motivated by a picture of the rational thinker as someone who will always have transparent access to the simple inferential consequences of his thoughts. This picture, Brown argues, is incompatible with anti-individualism about content. While traditional Fregean theories have indeed had such motivation, Brown’s mistake is in attributing this motivation to the modern Fregean anti-individualist. My goal in this paper is to bring to light a different, and seldom discussed, motivation for Fregean views of content which is immune to Browns objections.
In her book Anti-Individualism and Knowledge Jessica Brown presents an argument against views of content that combine Fregean Sense and anti-individualism. Her argument rests on two claims: First, that Fregean anti-individualism is incompatible with transparency of difference. Secondly, she claims that the sole motivation for Fregean views is a picture of rationality which, beyond being false, becomes unmotivated the moment transparency of difference is relinquished. I want to argue that Brown is wrong on two accounts: both regarding the incompatibility of Fregean Sense and transparency of difference, and regarding the central motivation for Fregean anti-individualism. At least this is the case when we are concerned with the brand of object-dependent Fregean anti-individualism, which has been defended by central figures in the field such as McDowell, Evans, Campbell, and Brewer. My essay is intended to serve two purposes. I wish to rebut Brown’s objection, but more importantly I want to shed light on, and defend, the actual motivation behind such Fregean anti-individualism.
The paper is divided into three sections. The first section presents the central notions and introduces Brown’s view of the debate. I discuss respectively: anti-individualism, Fregean views of content, and how they are combined in the specific target of the paper; what Brown calls Fregean singular anti-individualism. Lastly, I introduce Brown’s argument as to why this position must deny what she calls ‘Transparency of Difference’. The second section argues that, for reasons internal to their explanatory projects, major Fregean singular anti-individualists, like Campbell and Evans, can and ought to reject Brown’s claim that their position is incompatible with transparency of difference. However, I acknowledge the need for presenting a motivation for their Fregean view other than the picture of rationality suggested by Brown. The third section presents an alternative motivation for adopting the type of Fregean view that individuates content by means of object-dependent senses, a view shared by Campbell, Evans, and McDowell.
1 What is Fregean Singular Anti-Individualism?
Anti-Individualism about Content
Anti-individualism is the view that thought content is partly individuated by the external environment. The view comes in various forms depending on which types of thoughts one believes are partly externally individuated and depending on what features of the environment one takes to be relevant. Familiar versions claim that our natural-kind concepts are individuated by the metaphysical nature of the kinds in our environment (Putnam 1975), or that our specialist concepts are partly individuated in terms of the socio-linguistic community we are part of (Burge 1979). The type of anti-individualism I want to consider concerns singular thought and it focusses specifically on demonstrative thought. This is what Brown (2004, 1) calls “singular anti-individualism” and describes as the view that: “A subject’s thoughts are partly individuated by the particular objects in her environment”. The main defenders of this view that I want to consider are Campbell (2002), Evans (1982), and McDowell (1986). The general thrust of singular anti-individualism is that our singular thoughts, in particular our demonstrative thoughts, bear a special relation to the information we receive from our environment. The shared idea among the various defenders of this view is that through experience I am brought into contact with a particular object. The singular thoughts I base on that experience, or on some information received through the experience, inherently refer to that object. That specific type of singular thought would thus be unavailable to me in the absence of either the existence of that referent or my specific informational and/or experiential relation to it.
Fregean Views of Content
Fregean views of conceptual content claim that the individuation of the content of our concepts is more fine-grained than individuation in terms of mere differences in reference. Everybody agrees that two concepts that have different reference (when used by a specific subject on a particular occasion1) will possess different content. However, to the Fregean two concepts may share reference while they differ in content. In opposition, the Millian view claims that content is exhaustively individuated by reference. To the Fregean contents are fully individuated in terms of Senses, which are ways in which a referent is given to one. Thus for any Sense of a concept a certain reference is determined; yet the same object may be given to us in many different ways; thereby allowing multiple concepts with different Senses to share reference. Frege (1892) is rather inexplicit about his reasons for individuating content in terms of Sense rather than reference. However, his motivation clearly stems from dissatisfaction with how the Millian view handles the cognitive difference between informative and non-informative identity statements. Brown presents one motivation one may have for accepting a Fregean view.2 I want to present another. When Evans (1982) and McDowell (1986) introduced the idea of object-dependent Senses, one might, like Brown, simply view them as having reworked the notion of Sense within a philosophical project that shared the same basic motivations as those driving earlier individualist Fregean views. In contrast, I want to suggest that their views are driven by a completely different motivation. However, this alternative Fregean motivation is seldom made explicit and has received relatively little attention.
Fregean Singular Anti-Individualism
Fregean anti-individualism about singular thoughts is thus the view that the content of a singular thought is partly individuated by the objects in the thinker’s external environment and his relations to these, combined with the claim that such singular thoughts may differ in content, that is, in Sense, while sharing their reference. Thus the same external objects can be given to a subject in various different ways; each way corresponding to a different demonstrative concept, where all these concepts refer to the same object though they differ in Sense in accordance with the way the object is given. Note, while all singular anti-individualists will acknowledge that the availability of a singular thought depends on the existence of its referent, some singular anti-individualists also argue that the availability of Sense of the singular thought is dependent on the existence of its referent. The latter view acknowledges what have been called ‘object-dependent Senses’, in addition to its acknowledgement of object-dependent thoughts. It is solely this latter form of singular Fregean anti-individualism that I will defend, as only such views will be supported by the motivation I present in the third section of the paper.3
Brown takes her argument to target any form of Fregean anti-individualism. However, I want to focus on singular Fregean anti-individualist views that acknowledge object dependent Senses.4 This detracts little from the rebuttal of her claims as the major defenders of Fregean anti-individualism have defended this version of it (Brewer 2011, Campbell 2002, Evans 1982, McDowell 1987). It is important to notice that Brown doesn’t present any arguments to the effect that the Fregean position is wrong. She merely claims that it cannot be consistently motivated (Brown 2004, 228). Thus, all that is needed to rebut her specific argument is a coherent story that motivates the Fregean position. Of course, this story in itself needs some plausibility, but importantly one need not take every possible objection to the Fregean story into account, but merely display how the Fregean view is motivated given certain further reasonable commitments.
Qua Fregean, the Fregean anti-individualist insists that a rational subject can always grasp a priori simple instances of validity and contradiction, but qua anti-individualist she accepts that a rational subject cannot always grasp a priori simple instances of invalidity. But what could motivate the view that it is part of the concept of a rational subject that she can grasp a priori simple instances of some logical properties if one accepts that she cannot grasp a priori simple instances of other logical properties?
2 Transparency of Difference and the Challenge to Fregean Views
I want to start by considering Brown’s claim about the incompatibility of Fregean anti-individualism and transparency of difference. She motivates the claim by drawing attention to what have been called ‘switching cases’. In such cases, a subject is unbeknownst to her subjected to a change in the parts of her environment that are relevant to the individuation of her thoughts (Brown 2004, 213). Brown asks us to imagine a case in which a subject is viewing an apple and manages to visually track that apple over quite some time. Then a stealthy switch is performed, and unbeknownst to the subject a new apple is put in front of her. She manages to track the second apple for a while as well. In response to the first session of tracking she forms the judgement ‘that apple is green’. In response to the second session of tracking she forms the judgement ‘that apple is from New Zealand’. Notably, the subject has no a priori way of determining that the apples have been switched. Brown argues that, since the subject, who seems rational in every way, might on the basis of those premises form the conclusion ‘there is a green apple which is from New Zealand’, it is clear that transparency of difference fails. For as the singular concepts in the two judgements ‘that apple is green’ and ‘that apple is from New Zealand’ have different referents they surely have different contents. Yet by engaging in that form of inference the rational subject treats them as if they share content. Thus, what we have is a case where a subject S at a time t entertains two thought constituents that differ in content, while the subject has no a priori way of determining that this is the case. Hence, given that anti-individualism enables the type of switching case imagined by Brown, it follows that anti-individualism is incompatible with transparency of difference; or so Brown argues.
I think the Fregean singular anti-Individualist ought to deny the adequacy of Brown’s description of the switching scenario. Brown’s argument works by first motivating that the subject in question employs two well-functioning singular demonstrative concepts referring respectively to the pre-switch and the post-switch apple. Subsequently, she introduces the feature into the story that our rational subject is willing to immediately form an inference to a judgement that pools the information she has from the two apples in question. Crucially, Brown’s argument relies on the subject employing an inference that trades directly on the co-reference of the singular terms of the two demonstratives without the use of any identity statement as an intermediary premise. It is the use of this form of immediate inference5 which displays that our rational subject treats her two demonstratives as sharing content, and thus makes clear her transgression against transparency of difference. However, the Fregean singular anti-individualist should deny that these two elements of the story are compatible. If our subject is willing to pool her information about the two different apples by means of the form of inference suggested by Brown, then our subject hasn’t entertained two well-functioning demonstrative concepts in the first place. Conversely, if she is employing two well-functioning demonstrative concepts referring to the pre-switch and post-switch apples, then she cannot be disposed to accept the inference that moves immediately from her two singular judgements to the conclusion that there is a green apple that is from New Zealand. Rather, she must be relying on some intermediary identity premise of the form ‘that apple1 is identical to that apple2’; and her reliance on such an identity premise precisely displays that she acknowledges the difference in content between her two demonstrative concepts. The problem is that Brown overlooks a central explanatory role attributed to demonstrative concepts in the Fregean views she targets.6 Crucially, this explanatory role would be undermined if the Fregean acknowledged the adequacy of Browns description of the switching case.
The Justificatory Role of Demonstrative Concepts
Both Evans (1982, 195–196) and Campbell (1987) introduce a tracking requirement on our ability to form singular demonstrative thoughts. In order to demonstratively refer to an object, a subject must be able to perceptually track the referent of his singular concept across a certain period of time and across certain changes in appearance and properties. They argue that one simply cannot form demonstrative concepts in the absence of some minimal ability to track one’s referent. Rather, in such cases one can at best suffer the mere illusion of forming such a concept (Evans 1982, 133–135),7 leading to the illusion that one has issued a fully truth-evaluable judgement. It’s important to notice that someone who is suffering from illusions of thought doesn’t transgress against Transparency of Difference as formulated by Brown. Transparency of Difference requires that a subject must be able to appreciate a priori if two simultaneously entertained concepts differ in content. However, someone who suffers from illusions of thought may take himself to think something along the lines of the thoughts that that apple is green and that that apple is from New Zealand,8 but one or both mental events is in fact illusory thinking, due to the unavailability of the involved demonstrative concept. Hence, our subject hasn’t entertained two demonstrative concepts, but failed to notice their difference in content. Thus, there has been no transgression against Transparency of Difference as formulated by Brown.
Brown (2004, 91) notices this line of response, but she takes herself to counter it by applying what she calls a slow-switch scenario. In these cases, it is assumed that our subject has the ability to adequately track the one apple for an extended period of time prior to the unnoticed switch and, likewise, that she has the ability to adequately track the other apple for an extended period of time after the switch. Brown (ibid.) argues that as Evans and Campbell allow that tracking can continue across brief lapses of, say, visibility, they cannot respond by claiming that the present case is one where illusions of thought are present. Instead our subject must be awarded grasp of two proper demonstrative concepts that refer respectively to the tracked pre-switch apple and the tracked post-switch apple. This line of response is interesting, because it illuminates that Brown has misunderstood the very reason why Evans and Campbell place their tracking requirements on demonstrative thought in the first place. Campbell is particularly explicit about this.
Campbell’s primary concern is to determine how we can possess justification for the uses we make of the information we acquire through experience. His goal is to defend what he calls the classical view on which “Knowledge of what it is for a proposition to be true is what causes, and justifies, your use of particular ways of verifying, and finding the implication of, that proposition” (Campbell 2002, 24). This concern raises questions such as the following: How is it that we can possess two pieces of separate information, say, that something is green and that something is from New Zealand, and have justification for conjoining these pieces of information into one judgment, say, the judgement that something green is from New Zealand? One way of doing so is by employing an identity premise in our reasoning which states that the two pieces of information concern the same object. However, clearly not all such integration of information relies on the use of substantial identity premises. How can inferences that do not employ such premises yet still combine separate pieces of information be justified, given that it is a substantial empirical fact that two pieces of information derive from the same object? Campbell’s worry is thus ill stated as one that primarily concerns why certain identity statements are informative and others are not. Rather, his concern is with the following question: how can we be justified in forming inferences that integrate separate pieces of information without the use identity statements as premises (Campbell 2002, 128–130)?
Campbell (2002, 130) and Evans (1982, 180–181) answer by arguing that the very nature of what it is to possess certain concepts will automatically ensure that we are justified in some of the immediate inferential use we make of those concepts. It is a requirement on possessing an information-based concept, such as a demonstrative concept, that there is an extended sum of information acquired across time that one immediately takes to concern the referent of that concept (Evans, 146). However, that alone doesn’t yield justification for the inferences in which one unites that information. Hence, a further requirement on the possession of demonstrative concepts is introduced, namely that the information which one is in this way disposed to treat as coming from one object does in fact come from a single object (Evans, 147 footnote 10; Campbell 2002, 99). The reason why we are justified in integrating the information contained in a series of judgements is that those judgements involve the same concept, and that the possession conditions on such concepts themselves entail that my disposition to immediately integrate the information must be correct.9 Hence, Evans and Campbell can defend the following conditional: if I possess the demonstrative concept ‘that apple’, then my immediate patterns of integrating information as pertaining to that apple will be justified.
For two demonstrative concepts to share their Sense is precisely for them both to be tied to the same bundle of information; a bundle of information which our sub-personal information treatment system trivially treats as deriving from the same object, and which does in fact derive from the same object (Campbell, 99–100). The object from which that information stems is the referent of the demonstrative; it is this feature Evans draws attention to by calling demonstrative thoughts ‘information based’ (Evans 1982, 144). If parts of the information in the relevant bundle derive from different objects, or no object, then illusions of demonstrative thought occur. Whether it is merely one thought that is illusory, say, the thought ‘that apple is from New Zealand’, or all of the involved thoughts, will depend on the extent to which information from separate objects has been mixed in the informational-bundle that guides one’s use of the demonstratives in question. If only a little information from the post-switch apple is included in a large well-formed file regarding the pre-switch apple, then only the post-switch thoughts are illusory. Once the bundle of information starts to approach an equal mixture of information from the two separate objects, we get nearer to the situation where both demonstratives concepts are illusory.
In Brown’s presentation, the tracking requirement is introduced independently of its role in justifying certain immediate inferences. The consequence is that she can see no reason why the potential lapses in experiential contact with the referent cannot be exploited so as to introduce a shift in the referent of the temporally separated demonstrative judgements. To Brown the possibility that such a shift of the referent could occur while one remained in possession of the two demonstrative concepts doesn’t seem problematic at all. After all, the relevant tracking capacities are conserved both before and after the switch in question, only now these tracking capacities relate to two different referents resulting in the possession of two separate demonstrative concepts. The only difference is that one is now disposed to trivially treat information from two different objects as if it concerned a single one. However, this very possibility is detrimental to the justificatory role for which Campbell and Evans introduced the tracking requirement. Their point is that we can be justified in forming certain immediate inferences because the very nature of our demonstrative concepts ensures the validity of these inferences.
In light of the justificatory role they attribute to our conceptual capacities, Campbell and Evans have to argue that the moment one employs pieces of information coming from separate objects in reasoning that simply trades on co-reference, either one or both of the judgements involved in the reasoning must be cases of illusory thought. It is not simply a move they may choose to employ in response to Brown’s objections. It is inherent to their very reason for introducing the tracking requirement in the first place. Illusions of demonstrative thought will be involved in every case where one trivially trades on the identity between seeming demonstrative judgements containing information from different objects. Thus, for reasons internal to their explanatory projects, Evans and Campbell have to argue that in Brown’s switching case our subject fails to possess two demonstrative concepts of the general form that apple that have different content. Hence, it is not a case of someone who employs two concepts with different content without being able to appreciate this. It is a case of someone who suffers from an illusion of thought, which as part of that illusion is led to suffer illusions of reasoning as well. Thus, singular Fregean anti-individualism is compatible with transparency of difference.
3 The Motivation behind Fregean Views of Content
One could in principle end one’s objection to Brown’s argument at this place. One of the commitments she ascribes to the view she criticizes is one that its defenders can and should reject; hence the internal contradiction she presents dissipates. However, this would be an uncharitable reply, given the view of rationality that Brown takes to motivate every Fregean theory. If that picture of rationality was indeed the sole possible motivation for the Fregean view, then illusions of thought would themselves be deeply problematic for the Fregean. Brown has in mind a picture of rationality where the rational status of a subject’s judgements is by some means luminous to him, such that when due care and attention is given the subject will never make simple reasoning mistakes. Now, if this view were indeed the sole motivation for the Fregean position, then the defence above could at best grant temporary respite. For that notion of rationality seems as threatened by the view that we sometimes make invalid transitions between our mental states due to the presence of illusory judgements, as it is by the possibility that these invalid transitions are made due to our failure to appreciate differences in content between content bearing states. What is of relevance is that anti-individualism entails that a subject can be so positioned that he cannot by a priori means assure himself that a given judgement is based on valid reasoning.10
To properly motivate singular Fregean anti-individualism, we need to display how the thoughts about rationality and justification that do indeed motivate the view spring from concerns different than those suggested by Brown. To do so, we need to look closer at Frege’s original problem, which didn’t concern subterfuge and switching scenarios. Rather, it starts from the basic intuition that something to do with content has to explain the cognitive difference between those inferences where sameness of reference can be trivially traded on and those cases where informative identity premises are required for rational inference. At times we immediately and rationally move from the premises ‘A is F’ and ‘A is G’ to the conclusion ‘something is F and G’. What Frege (1892) draws attention to is that sometimes this form of inference is rationally unavailable, even though we accept two premises that attribute F’ness and G’ness to the same object. An ancient astronomer would be unjustified if he moved immediately from the premises ‘Hesperus is visible’ and ‘Phosphorus is famous’ to the conclusion ‘something famous is visible’; even though both names refer to Venus. In this case the conclusion is only rationally available to a subject whose reasoning relies on the identity premise ‘Hesperus is identical to Phosphorus’. Fregeans conclude that the premises employed cannot be understood as having the form ‘A is F’ and ‘A is G’. Rather, we have to accept that ‘Hesperus’ and ‘Phosphorus’ are different concepts in spite of their co-reference, hence reference cannot exhaust meaning. Thus, the premises in the Hesperus-Phosphorus case actually have the form ‘A is F’ and ‘B is G’, and that is why the inference to the conclusion needs to employ an identity premise of the form ‘A is identical to B’. This much of the history of Fregean views is uncontroversial. However, Frege is very inexplicit about what features of the astronomer’s story move him to change his view on the individuation of content.
One way of understanding Frege’s puzzle takes the central question to be why we cannot move immediately and rationally from ‘Hesperus is F’ and ‘Phosphorus is G’ to ‘something is F and G’. In this case the worry is that there is a truth-preserving inferential move which is valid on the Millian view of content, but which we cannot rationally employ. The Fregean then responds by concluding that the concepts employed are different despite sharing reference, which means that there is no longer a valid inference available which it would be irrational to employ. When the puzzle is phrased in this way, it seems obvious that the Fregean position must be motivated by some assumption which claims that all simple valid inferential moves must be transparent to the rational thinker. So, the Fregean introduces Sense to avoid the conclusion that there exist simple valid moves that are rationally unavailable. This way of setting up the issue leads us to Brown’s reasonable response, which is to question why rational thinkers must have such transparent access to simple valid reasoning when they have no similar insight into which instances of simple reasoning are invalid. As we have seen, the Fregean anti-individualist must accept that rational thinkers are sometimes in the dark as to whether they are employing invalid patterns of reasoning. This is either because they fail to notice differences in content, as Brown suggests, or because they fail to notice that they are undergoing illusions of thought, as Campbell and Evans suggest. If such a transparent insight into the validity of our reasoning was the only possible motivation for endorsing a Fregean view, then it would indeed be difficult to support a Fregean position while endorsing some form of anti-individualism.
There is, however, a very different way of understanding Frege’s puzzle. On this view, Frege’s central goal is to ensure that it is the very validity of our reasoning which makes us rational when engaging in inferences of the Hesperus-Hesperus type. The central question thus isn’t why there are certain seemingly valid patterns of reasoning which the rational subject does not engage in, but rather: How can it ever be the very validity of a subject’s reasoning which justifies him in performing the inferences that are in fact rationally available to him?
I want to suggest that the motivation behind Fregean anti-individualism is a drive to conserve this type of explanation. For as we will see it is unavailable on the Millian account, and this has grave consequences for the role which truth can play in our theory of human thinking all together.
Validity, of the type which concerns us, requires that it should be the same thing that guarantees that necessarily, if the premises are true, so too is the conclusion, as makes it obvious that the transition is correct.
If we accept a Millian theory of content, then it cannot be the validity of the inference above which accounts for our justification for employing it. For notice that on the Millian account the inference from ‘Hesperus is visible’ and ‘Phosphorus is famous’ to ‘something famous is visible’ is equally valid, and it shares the very same logical form as the merely ‘Hesperus’-involving inference; yet the former inference isn’t justified. Hence, something must separate the justificatory status of the two types of inference from each other. Yet, in both cases the subject sufficiently grasps the concepts involved and, according to the Millian, the inferences are both valid and logically identical. The obvious answer is that there is something about my grasp, use, or entertaining of the concepts which explains the difference in their justificatory position. That answer is available to the Millian, just like it was to the Fregean. What is unique to the Fregean position is that the Fregean takes this difference, which explains justificatory position, to be an element that suffices to determine reference. Sense, how objects are thought of or entertained, determines reference. In virtue of determining reference, Sense alone serves to settle truth-conditions, and therefore Sense deserves its status as the content of our thoughts. This means that when the Millian accounts for the justificatory element of our thinking, he has to present an account where what explains justification falls short of determining reference by itself. If he didn’t, his disagreement with the Fregean would be merely terminological. What the Millian would call content, the Fregean calls reference. What the Fregean call’s content, or Sense, would be the Millian’s favoured version of the justification determining element. However, their accounts would be structurally similar in every respect, as both would agree that the very element that accounts for the justified inferential use we can make of the concepts we grasp also determines their reference.
Various Millian accounts have been given that live up to the structural requirement which separates them from Fregean views. Fine (2007) has argued that certain of our thoughts are coordinated and others not and, on his view, we only have justification for immediately trading on co-reference between thoughts so coordinated. The point is then that, though all the ancient astronomer’s thoughts concerning Venus employ the same singular concept, those we would commonly describe as Hesperus-thoughts are all coordinated, as are his Phosphorus-thoughts, but there is no coordination between these groups. Burge has (1977) argued that the rational status of a subject is determined by how his reasoning relates to an attributive psychological representation that can be shared across differences in reference, or even in the absence of any reference at all. In a similar vein, Schellenberg (2011) 11 has argued that our concepts have beyond their reference a certain gappy-schema, where the gap is filled with a particular external object as referent. This means that filled gappy-schemas, that is, complete concepts, are given anti-individualist treatment. However, the gappy-schema itself is available in the absence of any specific referent, and indeed in the absence of any referent at all. On this account, sameness of gappy-schema is what accounts for our justification for immediately trading on co-reference. Many other variations are available. What is important is what separates all these views from the object-dependent account of Fregean Senses. They all uniformly agree that the element which accounts for inferential justification can remain constant independently of possible variations in reference. Hence, they deny the two key claims which together make up the object-dependent Fregean position: Differences in Sense accounts for differences in justified concept use and, importantly, Sense alone is sufficient to determine reference. The very same element which accounts for our justification for the use of our concepts is also sufficient to settle their truth-conditions. Hence, on the Millian view, two concepts can share their inference justifying element while differing in reference. As difference in reference entails difference in content, the Millian must acknowledge that the inference justifying element is independent of content.
It’s important to appreciate just how radically this type of answer differs from the Fregean account. That which justifies us in our concept use is individuated independently of differences in truth-conditions. That means, strictly speaking, that it cannot be our grasp of the truth-conditions of our concepts which justifies our use of them.12 In turn, this means that it cannot be our appreciation of the validity of our inferences which justifies us in employing them. Validity depends on truth-preservation; truth-preservation depends on truth-conditions; and truth-conditions depend on reference. In short, validity depends on content. If the justificatory position of my use of a concept depends on my sensitivity to something that doesn’t suffice to determine its reference, then the justificatory position of my use of that concept cannot depend on my sensitivity to something that determines the truth-conditions of the propositions I employ it in, nor the validity of the inferences in which it is involved. To the Millian one can thus remain rational, that is, justified in one’s concept use, while engaging in invalid inferences. It is precisely this general structure of Millian accounts which allows Brown to argue that our subject would remain rational in her inferences in spite of treating information from two different apples as if they concerned a single one. On her view the element, whichever it be, that accounts for inferential justification is separable from the determination of reference. For that reason, cases can occur where two thoughts differ in reference, but are so aligned in their justification determining elements that a subject is rational in treating them as co-referring. The subject’s inferences will of course be invalid, but they will remain rational.
This consequence, that we can be justified in performing invalid inferences, rather than merely blameless, is of course a significant consequence of the Millian theory, but for current purposes it is less important. While some of the Fregean anti-individualists involved in the debate require that justification be infallible (Notably, McDowell 1996), I do not take this to be the central motivation for their view. Moreover, the Millian can endorse infallibilism by imposing a further brute external requirement on justified inferences, namely that they be valid and that they be proper according to the reference-independent justification determining element. For example, in Fine’s version, one could simply specify that fully justified inferences which immediately trade on identity must not only employ coordinated thoughts, they must also employ co-referring thoughts. Such a stipulation may seem ad hoc, but for present purposes we can allow it, as it still leaves the Fregean worry untouched. What bothers the Fregean is that on no Millian account will it be a subject’s appreciation of the very validity of an inference which accounts for the justification she has for engaging in that inference. Even if sameness of reference is introduced as a further requirement for full justification, the fulfilment of this further requirement will still be external to that which motivates the subject in her reasoning, and thus accounts for her rationality. These accounts are thus completely contrary to what Campbell (1987, 297) is searching for according to the earlier quote. We are looking for an explanation of our justification for engaging in immediate inferences where we unify information from two separate demonstrative judgements. For the Fregean, an adequate answer requires that the very element which accounts for the validity of the inference also motivates the rational subject to perform the inference.
Why should we accept the Fregean constraints above as a requirement on an adequate theory of meaning? We come now to what I take to be the core motivation behind Fregean anti-individualist views that endorse object-dependent Senses. The worrying part about the Millian theories is that on such theories we seem to have lost the importance of truth in the overall account we provide of the use we make of concepts. If a subject’s motivation for her use of a demonstrative no longer depends on sensitivity to its reference, this means that her motivation for the use of the propositions containing the demonstrative cannot depend on her sensitivity to their truth-conditions. This follows because truth-conditions depend on reference. Given our current concern with inferences whose validity relies on the co-reference of the demonstrative concepts involved, this means that the rational motivation a subject has available to her as an agent for engaging in such inferences cannot depend on sensitivity to their validity. Truth has become obsolete in our account of the justified use we make of our concepts.13
The problem according to the Fregean is that once we have stripped reference and truth of their role in the justification of our inferential patterns, then it is difficult to see what role these notions should play in our theory of thought altogether.14 The input transactions between experience and concept use, as well as the output transactions between concept use and action, are as susceptible to Frege cases as the mere internal cognitive manipulations in inference. Hence, truth will equally be stripped from playing a role in explaining the rationality of perceptual judgements and the rationality of actions, speech acts included. At this point it becomes difficult to see why the notions ‘reference’ and ‘truth’ should play a part in our account of meaning at all.15 What can we make of the common-sense intuitions that speech and judgement aim at truth and that desire and action aim at making true, when truth no longer plays an ineliminable role in justifying our judgements and actions?
One option is to claim that truth functions as some form of transcendent goal or purpose in our conceptual behaviour. While we cannot be guided by a grasp of truth-conditions or reference itself, we can be guided by our appreciation of features that are truth-conducive, or reliable indicators of truth, in our given environment. It is in this sense that our misled apple watcher can remain justified in his invalid inferences. His acceptance of the invalid inferences is based on sensitivity to features which are commonly indicative of shared reference in our environment. After all, it is rare that skilled apple switchers are about, and most often it is the same object we encounter before and after short interruptions of visual contact with similar looking items at one location. Thus, while we cannot strictly be guided by truth itself, we can aim for truth by aiming for something else which is indicative of truth, and in that way a central role for truth and reference are conserved in our account of human concept use.
However, what sense can we really make of this suggestion that we aim at truth, while we are never actually directly guided by it? Note, we cannot require of concepts users that they have ever explicitly or implicitly entertained the goal of aiming for truth. Nor need they be able to acknowledge this as the goal they have upon being presented with that idea. Neither the concept of truth itself, nor our theory of what guides the rational propriety of our inferential practices and propositional attitudes, is something that one needs to grasp, or be prepared to acknowledge, in order to be a concept user. Rather, our theory about the role of truth is meant to make explicit the guiding aims which are already implicitly present (though not by being entertained as goals) in the concept use of ordinary thinkers. However, on the Millian account our rational concept use is solely guided by features which are insufficient to determine truth and reference. They may under certain conditions be reliable indicators of truth and co-reference, but why, given the picture of concept use presented, should we consider them to be employed as justificatory features because they are so indicative?
It is as if the Millian has started his theory of human thought from a presumption of the importance of truth and reference to an account of human psychology. With that in mind he has gone on to present a theory of our cognition which is so structured that one cannot by looking at that theory itself discern why such interest was taken in truth and reference in the first place. Given how the functional parts of the theory tie together, it seems that it was only because we came to the topic with a preoccupation with truth that we ever alighted on the importance of that notion. Given how the Millian explains and justifies our cognitive practices, we might as well have taken survival, success, social acceptance of behaviour, or some other non-alethic feature, to be the defining characteristic of proper use of concepts. In fact, it even seems more natural to do so, for truth only enters the picture as some transcendent regulatory goal for our practices postulated by a third-person theorist. Nothing in our justificatory practices displays our interest in it.16 And, of course, once truth and reference are left obsolete in our picture of human cognition, so is the notion of representation altogether. Our mental activity has indeed become the frictionless spinning in the void against which McDowell (1996) warns.
The motivation behind the Fregean approach to content is the conviction that we must present an account of human thought where we can make obvious, from within that theory itself, the central role that truth and reference play in the propriety of human cognition. If this is to be the case, then reference and truth cannot be divorced from the elements that rationalize human use of concepts. However, as Frege’s Puzzle makes clear, neither can we straightforwardly identify the rational cognitive role played by a concept or a thought with its reference or truth-value. Reference and truth-value are too coarse grained tools to capture the fineness of the justificatory structure of our thinking. Thus, the Fregean concludes, we must include an element in our thinking that serves the role of explaining the justificatory structure of our thought and suffices to determine reference. Hence, thoughts are individuated in terms of Sense, which determines reference, while they are individuated with finer grain than in terms of differences in reference alone. This is the motivation for the view that accepts object-dependent Fregean Senses.
In light of the motivation just stated, consider again Brown’s objection that the Fregean anti-individualist seems unduly occupied with our ability to grasp instances of validity, but seems nonchalant about our failure to appreciate simple inferences as either invalid or employing illusory thoughts. On the present account of the Fregean motivation we can explain this. The Fregean is occupied with explaining how grasp of truth, reference, and validity can ever play the role of justifying human cognitive behaviour. This means that, for the issue presently under discussion, there must be cases where our justification for forming an inference is due to our direct appreciation of sameness of reference. We need not always be able to grasp such sameness, nor need all our cognitive behaviour be directly justified in terms of such appreciation. As long as our theory can at least sometimes present truth and reference as the undisputed goals and guides of cognition, we can allow room for them as mere regulative goals imperfectly aimed for in other cases. Hence, though we can go wrong for many reasons, the Fregean is simply out to establish that we sometimes go right for the right reason, and that is why he is focussed on explaining our grasp of validity. A general presumption of transparency of access plays no role in motivating his arguments.
I have pointed out two flaws in Brown’s argument against Fregean singular anti-individualism. Her argument relies on the claim that Fregean singular anti-individualism is incompatible with accepting transparency of difference. I have shown why no such contradiction is entailed. Moreover, for reasons internal to their explanatory project, Brown’s central opponents, Evans and Campbell, have to present theories that conserve transparency of difference. The second pillar of Brown’s criticism relies on her assumption that all Fregeans motivate their position through the assumption that rational subjects must have luminous insight into the validity and invalidity of their simple reasoning. However, I have suggested that the goal, at least for Campbell and McDowell, is simply to preserve a theory of thought where our inferential justification is at times directly due to our appreciation of the validity of those inferences. Only such a view provides the notions of truth and reference with the central role they must play in any theory of cognition recognisable as one that characterizes human thought. Of course, this isn’t the end of the debate between Fregeans and Millians. The Millians have various responses that they can make considering the just presented criticism. However, recall that a refutation of Brown’s objections required only the presentation of some alternative motivation for Fregean views. I take that task to have been adequately accomplished. Hence, the particular argument against Fregean Singular Anti-Individualism presented by Brown (2004) has no bite. Moreover, a positive challenge has been raised for the Millian; she should account for the significance attributed to truth in her theory of thought.
Burge Tyler 2009. “Five Theses on De Re States and Attitudes.” In The Philosophy of David Kaplan. Edited by Almog Joseph and Leonardi Paolo . Oxford: Oxford University Press, 246–316.
Campbell John 2007. “If Truth is Dethroned what Role Is Left for It?” In: Library of Living Philosophers: The Philosophy of Michael Dummett . Edited by Auxier Randall E. and Hahn Lewis Edwin . Chicago: Open Court Press, 281–300.
Dummett Michael 1975. “Frege’s Distinction between Sense and Reference.” In: Truth and other Enigmas. Cambridge, Mass: Harvard University Press, 116–144.
Frege Gottlob 1892. “On Sense and Reference.” In: Philosophical Writings. Edited by Geach Peter and Black Max . Oxford: Blackwell.
McDowell John 1986. “Singular Thought and the Extent of Inner Space.” In: Meaning Knowledge and Reality. Cambridge, Mass: Harvard University Press, 228–259.
I will omit this qualification in the future.
This motivation, or one close to it, is indeed what has motivated a series of traditional Fregean views. However, Fregeans so motivated have also commonly rejected anti-individualism (see for example Dummett (1975, 1981) and Boghossian (1994)).
Thus adherents to Burge’s (2009) influential position cannot make use of the motivation that I propose for Fregean anti-individualism.
In the future I will for the sake of brevity omit the qualification with the object-dependent Senses whenever I talk of singular Fregean anti-individualism.
I will use ‘immediate inference’ to refer to inferences that trade on the identity of the referents of multiple singular thoughts without employing identity statements as intermediary premises.
Of course he cannot ascribe a real parallel of the illusory thought to himself, as that thought is precisely unavailable.
Both Burge’s and Schellenberg’s views aren’t properly Millian as they accept Sense as a feature of content. However, what is crucial is that they deny that Senses are object-dependent. This results in their inability to present an account where the same thing ensures both the validity and the rationality of an inference, which is precisely what I suggest thinkers such as Campbell, Evans, and McDowell search for. For present purposes I will somewhat unfairly refer to them as Millian views, due to their unwillingness to present a unified ground for validity and rationality.
Thus all these views differ from the earlier introduced Classical View, which motivates Campbell’s thoughts on the issue.
Beware that I will not in the following attempt to defend this line of thought against all objections. My sole goal is to present a line of consideration that provides an alternative motivation for adopting a Fregean view.
Of course, we employ the word ‘truth’, but we could simply give a minimalist, expressivist, or error account of such use.