Davidson on Pure Intending: A Non-Reductionist Judgement-Dependent Account

Abstract

I will argue that Davidson's account of pure intending can be construed as a first-person-based judgement-dependent account of intention. For Davidson, pure intending to do φ is to make an all-out judgement that φing is desirable. On this anti-reductionist account, intention is treated as an irreducible state of the subject. I will draw a comparison between this account and Wright's and I will show that Davidson's account can be viewed as a non-reductionist judgement-dependent account along the lines suggested by Wright. I then explain how this account can help deal with various perplexities in Davidson's later view of meaning and mental content.

Résumé

Je soutiendrai que la façon dont Davidson rend compte de l'intention pure peut être comprise comme une analyse de l'intention comme étant relative à un jugement dans une perspective en première personne. Selon Davidson, avoir la pure intention de faire A, c'est formuler un jugement tout bien considéré qu'il est désirable de faire A. Dans cette analyse anti-réductionniste, l'intention est traitée comme un état irréductible du sujet. J’établirai une comparaison entre cette analyse et celle de Wright et je montrerai comment la position de Davidson peut être considérée comme une analyse non-réductionniste rapportant l'intention au jugement, selon les directions suggérées par Wright. J'expliquerai ensuite comment cette analyse peut aider à éclaircir diverses difficultés qui se présentent dans la conception que Davidson se fait ultérieurement de la signification et du contenu mental.

1. Introduction

What is Donald Davidson's answer to the question ‘what it is to intend to do φ?’ This article seeks an answer to this question by focusing on Davidson's remarks on pure intending. More particularly, in “Intending” (Davidson, 2001a), Davidson seems to be offering such an account, according to which for a subject to have a pure, actionless intention to do φ is for her to judge that she has that intention. This account differs from his earlier account of intentional action, which aimed to explain intention in action in terms of the right sort of causal relation there can be between the agent's actions and her beliefs and desires. In his account of pure intending, intending to perform an action is not taken to be a qualitative state; nor is our knowledge of such a state introspectible, observational, or inferential (see Davidson, 2001a, p. 83). Given that, facts about my intending to do φ in the future cannot be viewed as directly constituted by facts about my performed action or those about a particular set of beliefs and desires of mine. The difference between these two accounts is significant, which I believe leads to a new sort of reading of Davidson's later view of meaning and intention — a reading that takes Davidson to be proposing a specific sort of non-reductionist, first-person-based judgement-dependent account of meaning and intention. Such a reading has thus far been absent in the extensive literature on him.

In this article, I will show that such a Davidsonian account shares the fundamental features of Crispin Wright's judgement-dependent (JD) account of intention. Wright's account is of major importance because it offers promising answers to certain metaphysical and epistemological problems about intention via advising a workable non-reductionist account of intention and a feasible account of self-knowledge. These outcomes have also been used by Wright to give a non-reductionist response to Saul Kripke's Wittgenstein's famous sceptical problem by offering powerful answers to the questions like ‘what makes it the case that a subject intends to φ?,’ ‘how it is that the subject can be credited with authoritative knowledge of the content of her intention?,’ and ‘how we can account for the objectivity of the subject's judgements about what she means and intends?’

All such issues are of great significance to Davidson too. Consequently, Davidson's account construed as a non-reductionist JD account can potentially shed new light on a variety of controversial aspects of Davidson's philosophy, some of which will be discussed in Section 5, Consequences. For instance, it can help to offer a brand new account of self-knowledge on behalf of Davidson, which lacks the defects of his official account of self-knowledge.¹ It helps to clarify his complex remarks on triangulation, by treating it as an analogy sustaining his non-reductionism, rather than an argument for meaning determination. Davidson's non-reductionism can also explain why his treatments of the Quinean indeterminacy-underdetermination distinction and Kripke's Wittgenstein's sceptical problem are often viewed as extremely puzzling or even wrong, that is, his view of these problems as epistemological (see, e.g., Hossein Khani, 2017, 2018a, 2018b, 2019; Kemp, 2012; Verheggen, 2017; Verheggen & Myers, 2016). These fruitful outcomes, among others, form the main motivation for my attempt to construe Davidson's view of intending as a non-reductionist JD account.

2. Davidson on Pure Intending

Davidson's early explanation of intentional action concentrates on describing the action performed by an agent as caused, in the right way, by the agent's certain attitudes — which, at the same time, rationalize that action. According to Davidson, “[w]henever someone does something for a reason, therefore, he can be characterized as (a) having some sort of pro attitude toward actions of a certain kind, and (b) believing … that his action is of that kind” (Davidson, 1963, p. 685). If an agent acts with an intention, she does so for a reason, and it is having that reason that rationalizes the action. Such reasons are pairs of the agent's desires and beliefs causing her actions.² The agent, however, might have the appropriate belief-desire pairs and they might cause her to perform that action, but because of some anomaly in the causal chain, the action is eventually done unintentionally. Thus, “an action is performed with a certain intention if it is caused in the right way by attitudes and beliefs that rationalize it” (Davidson, 2001a, p. 87). Nonetheless, no further clarification of the proviso “in the right way” is forthcoming since, as Davidson confesses, “I do not see how the right sort of causal process can be distinguished without, among other things, giving an account of how a decision is reached in the light of conflicting evidence and conflicting desires” (Davidson, 2001c, p. 232). The problem has deep roots: there is no strict psychophysical law directly connecting what happens in the physical aspect of the action with what happens in the agent's mental realm.³

Davidson's account of intentional action is reductionist, and the reduction, as Davidson puts, “is not definitional but ontological” (Davidson, 2001a, p. 88). In this account, the expression ‘the intention’ with which someone does something is “syncategorematic and cannot be taken to refer to an entity, state, disposition, or event. Its function in context is to generate new descriptions of actions in terms of their reasons” (Davidson, 1963, p. 690). Thus, no essential use of the concept of ‘intention’ is made in this account: the agent has done φ intentionally simply as the conclusion of her reasoning from those specific beliefs and desires of hers. To put it in terms of the Aristotelian form of practical syllogism, it “is a simple deduction concluding into an action. It models actions as logical consequences of beliefs and desires” (Glüer, 2011, p. 197).

There are at least two issues that Davidson's account fails to explain: the case of pure, actionless intentions, and the case of weakness of the will. Davidson's retraction of his early reductionist account was already in play in his “How Is Weakness of the Will Possible?” (Davidson, 2001b), where he attempts to explain how it can be the case that an agent eventually intends to do φ against her all-things-considered judgement that ψ is a better thing to do. But it is in his explanation of pure intending in “Intending” (Davidson, 2001a) that his retraction gets completed and results in an alternative view of intention. It now looks to him that the form of reasoning he should be after is to be of a probabilistic form rather than a deductive form, similar to the sort of reasoning we deal with in the process of decision making. What is Davidson's account of pure intending?

For Davidson, pure intending is “intending that may occur without practical reasoning, action, or consequence” (Davidson, 2001a, p. 83). Thus, our concern is how an agent arrives at an intention to do something rather than the action itself. Davidson now observes that “in any intentional action that takes much time, or involves preparatory steps, something like pure intending must be present” (Davidson, 2001a, p. 88). Such an ingredient, i.e., pure intending, is not something that could be reduced to any other component presented in his account of intentional action.⁴ Thus, the account takes the form of a non-reductionist account, as he now “identifies intentions as distinct attitudes that play an important role in mediating reasons and the actions they cause” (Ludwig, 2003, p. 15).

For Davidson, since “there is no action to judge simply good or desirable,” “[a]ll we can judge at the stage of pure intending is the desirability of actions of a sort” (Davidson, 2001a, p. 97). Davidson thus identifies intention with judgements of a sort. There are two general cases of pure intending that Davidson considers: the case of intending to perform an action in the immediate future, and the case of future-directed intentions. In “Intending” (Davidson, 2001a, pp. 99–100), Davidson deals with each case a bit differently by claiming that the case of performing an action in the immediate future concerns intending to perform a particular action, while my future-directed intentions concern my intentions to prefer a type of action. Later, however, Davidson gives up on such a distinction:

[W]e could pick out a particular past or present action, whereas there are not yet any future actions to be picked out. This now seems to me simply wrong. Future actions, like any other entities, have unique descriptions, and so can be “picked out” like other things. It is not even the case, as I apparently thought, that future actions can't be identified through the use of indexical devices. Of course they can. (Davidson, 1985, p. 198)

Following Davidson, I will not separate these two cases: in all such cases, pure intending as an irreducible state of the agent is present. Davidson's main point is that for one to intend to do φ is for one to make a judgement that doing φ is desirable. Such judgements are self-judgements: it is the subject herself who makes these judgements about the content of her own intentional states. However, Davidson's claim that to intend to do φ is to judge that doing it is desirable faces a problem.

2.1. Prima Facie vs. All-Out Judgements

The problem is that while, for instance, eating something sweet is desirable, it is undesirable to eat something poisonous. An object can be both poisonous and sweet. Accordingly, the same action of eating that thing is both that of eating something sweet and that of eating something poisonous. The same action is thus both desirable and undesirable. If the judgements under consideration can open the path to such contradictory outcomes, they cannot be considered as what constitutes the fact that the subject, S, possesses that particular intention, i.e., to eat something sweet.

In order to solve this problem, we must add some modifying conditions: an action is desirable only “in so far as an action has a certain characteristic” (Davidson, 2001a, p. 98). Davidson calls such conditional judgements “prima facie judgements” (see Davidson, 2001a, p. 98). Clearly, it would not be enough to say that S intends to perform an action merely because the action has a desirable feature: many actions share similar desirable features. Not only this, but “[t]here are endless circumstances under which I would not eat something sweet, and I cannot begin to foresee them all” (Davidson, 2001a, p. 99). When making a prima facie judgement, S judges that an action is desirable for some reasons; but there may be other reasons for not considering that action as desirable. The subject evaluates such reasons. Nonetheless, she may eventually intend to do φ against her prima facie reasons saying that ψ is a better thing to do. As Davidson emphasizes, “we may not know how the agent got from his desires and other attitudes — his prima facie reasons — to the conclusion that a certain action was desirable” (Davidson, 2001a, pp. 98–99). The simple judgement that I intend to perform φ is (and is to be) different from such prima facie judgements if S is to be able to form any intentions at all. The new type of judgement that Davidson introduces is “all-out or unconditional judgement” (see Davidson, 2001a, p. 98), which has the simple form of ‘doing φ is desirable’ rather than that of ‘doing φ is desirable if φ has a certain characteristic.’ Therefore, he introduces his new account as follows: “[i]n the case of pure intending, I now suggest that the intention simply is an all-out judgement” (Davidson, 2001a, p. 99).

Davidson identifies this unconditional judgement with intention. As Colin McGinn puts it in more detail, on Davidson's account, the agent engages in three states in her practical reasoning to reach an intention to do something: “first, he makes a number of prima facie judgments …; second, he judges on the basis of this that … he should do a, thus making a … conditional judgment; third, he makes the all-out judgment ‘a would be (is) best.’ When the agent reaches the third stage he is intending to do a” (McGinn, 1997, p. 135). But, as previously indicated, this should explain weakness of the will, i.e., the cases in which S's all-out judgement is that ‘doing b is desirable’ against her prima facie reason that doing a is the best. For Davidson, S may judge, after considering her desires and other reasons, that doing b is better than doing a. Such an (all-things-considered) judgement is a prima facie one. At the end, she may make the all-out judgement that doing a is desirable. The subject separates the prima facie reasons from her all-out judgement and thus may eventually intend to do a. Of course, “this is irrational all right, but it is perfectly possible” (McGinn, 1997, p. 135; for more on this, see also Bratman, 1985; Elster, 1999; Glüer, 2011, Chapter 4; Gustafson, 1986, Chapter 3; Lazar, 1999; Stout, 2021). This means that what constitutes truths about S's intention to do φ simply is her all-out judgement that she intends to do φ.⁵

As a final point, it is important to note that making such unconditional judgements cannot be free of background conditions. Such conditions often deal with the conceptual capability of the agent, such as her possession of the concepts required to form such judgements, her attentiveness to her intentions, her being free from self-deception, and so forth. Davidson adds a further condition: S's all-out judgement about the desirability of eating something sweet does not include eating a poisonous sweet thing because, as he puts it, “I do not believe I will eat a poisonous candy, and so that is not one of the actions of eating something sweet that my all-out judgement includes” (Davidson, 2001a, p. 99). Such beliefs put some constraints on my judgements, or better, on the range of actions I may judge desirable. For Davidson, “there is nothing absurd in my judging that any action of mine … that is the eating of something sweet would be desirable given the rest of what I believe about the immediate future” (Davidson, 2001a, p. 99). These conditions do not make my judgements prima facie; rather, they guarantee that the judgements remain all-out by excluding the conditions under which I do not judge that eating something sweet is desirable: “What a person intends must, I have urged, be consistent with what the person believes, but this merely reduces the number of alternatives to be considered, it does not help select one” (Davidson, 1985, p. 199). My belief that I do not eat a poisonous candy “is not part of what I intend, but an assumption without which I would not have the intention. The intention is not conditional in form; rather, the existence of the intention is conditioned by my beliefs” (Davidson, 2001a, p. 100). I will return to these conditions later (for more on Davidson's account, see, e.g., Bar-On, 2012; Bratman, 1985; Ferrero, 2013; Gjelsvik, 1996; Henden, 2004; Heuer, 2018; Hunter, 2015; McNaughton & Rawling, 2018; Tenenbaum, 1999).

The above discussion already points to the sort of similarity I think there is between Davidson's notion of background conditions and Wright's idea of optimal conditions in his JD account. But the similarity between these two accounts is much more. In what follows, I will formulate Davidson's account as a sort of JD account of intention, which to some important extent resembles the main characteristics of Wright's JD account. Let me briefly review Wright's account first.

3. Wright's Judgement-Dependent Account of Intention

Suppose that truths about my intending to do φ and truths about my judgements about having that intention co-vary. For Wright, this co-variance can be explained in two ways: either it is my best judgements about having that intention that determine whether I have it or my best judgements are merely good at tracking the independently constituted fact that I have the intention. One important question is: What are the conditions under which my judgements can be considered to be best judgements of mine? Such cognitively optimal conditions, or simply the ‘C-conditions’ (see Wright, 1992, pp. 108–109; 2001, pp. 192–194, 197–198), can be put as follows: S has the appropriate concepts required to form such judgements, such as that of intention, lacks any material self-deception, and is appropriately attentive to the question of what her intentions are (see, e.g., Wright, 2001, p. 201). Having assumed that S has formed her judgement under such conditions, Wright suggests that the following ‘provisional equation’ can capture the relationship between truths about S's intention and S's judgement about that intention:

WPE: C-conditions hold → (S intends to φ ↔ S judges that she intends to φ).⁶

We can call this account a ‘First-Person-Based Judgement-Dependent’ (FPJD) account of intention because, in WPE, it is the subject herself rather than anyone else who judges whether she has an intention. If S's best judgements about her intention determine, rather than merely track, the extension of the concept of intention, then intention can be viewed as judgement-dependent. Consequently, we can say that S's judgement about her intention constitutes truths about S's intention; otherwise, the judgements are at most merely good at tracking independently constituted facts about S's intention.

For Wright, whether intention can be viewed as judgement-dependent depends on whether WPE meets the following four conditions: (1) A Priority Condition: WPE is to be a priori knowable to be true; that is, a subject who possesses the concept of intention knows, by pure analysis of or reflection on the concept, that, under the C-conditions, she intends to φ if and only if she judges that she intends to φ (see Wright, 1992, pp. 116–117; 2001, p. 193). Otherwise, knowledge of the truth of WPE would at most be a posteriori, answerable to additional considerations, such as experience. (2) Substantiality Condition: The C-conditions are to be specified non-trivially (see Wright, 1992, p. 112; 2001, p. 194). The C-conditions must not include conditions that provide the subject with whatever-it-takes to form correct judgements about her intention all the time. Otherwise, there would be no guarantee that it is S's judgements alone that determine the extension of the intention concept. (3) Independence Condition: The C-conditions cannot presuppose truths about S's intention since truths about S's intention are supposed to be determined by S's judgements formed under such C-conditions (see Wright, 1992, pp. 122–123; 2001, p. 195). (4) Extremal Condition: There must be no better explanation of why WPE meets (1)–(3) than viewing intention as judgement-dependent (see Wright, 1992, pp. 123–124; 2001, p. 205). Otherwise, while (1)–(3) might be met by WPE, there might be some independently constituted fact, other than facts about S's best judgements, that contributes to the constitution of facts about S's intention. If WPE meets these conditions, we can claim that S's best judgements are playing an extension-determining role (see also Wright, 2001, p. 206; 1992, Appendix: The Euthyphro Contrast; 1988).⁷ This account is non-reductionist because, in WPE, the concept of intention appears on both sides of the biconditional. But does WPE meet all of the above conditions? Wright thinks it does but in a defeasible way.

For Wright, showing that WPE meets the Substantiality Condition is the most difficult task to do in his FPJD account since, at least, the condition that the subject must not be self-deceived seems to bring in the sort of whatever-it-takes provisos that the Substantiality Condition was supposed to exclude: “‘self-deception’ covers … any motivated condition which might lead to a subject's ignorance or error concerning his or her intentions” (Wright, 2001, p. 201). This ‘no-self-deception’ condition, as we may call it, seems to exclude all the conditions under which S might be wrong about her intention and thus violates the Substantiality Condition. Wright's solution is to treat conditions like this as positive presumptive, which in turn helps him to remove them from the C-conditions and consequently to preserve the substantiality of the C-conditions. According to this suggestion, “one is entitled to assume that a subject is not materially self-deceived, or unmotivatedly similarly afflicted, unless one possesses determinate evidence to the contrary” (Wright, 2001, p. 202).⁸ In the absence of evidence to the contrary, we are justified in assuming that the no-self-deception condition is met. What is such evidence?

According to Wright, “[w]hen possession of a certain intention is an aspect of a self-conception that coheres well enough both internally and with the subject's behaviour, there is nothing else that makes it true that the intention is indeed possessed” (Wright, 2001, p. 204). If S's self-ascriptions of intentional states cohere well with her self-ascriptions of other attitudes and with her outward behaviour, we have then no evidence to the effect that the subject is self-deceived. This “Interpretationist Constraint,” as I have called it (Hossein Khani, 2020b), points “to my ability to make sense of myself to others in my (speech-) community” (Wright, 2001, p. 88).⁹ The self-ascriptions of intention must be interpretable by others; otherwise, there can be no publicly available evidence on the basis of which one can assess whether S is self-deceived when S actually is. Thus, as Wright concludes, we are now allowed to view intention as defeasibly judgement-dependent “with defeat conditional on the emergence of evidence that one or more of the background, positive-presumptive, conditions are not in fact met” (Wright, 2001, p. 203). The similarities between Wright's account and Davidson's can now be seen more vividly. In what follows, I will focus on them.

4. A Davidsonian Judgement-Dependent Account of Intention

To begin with, Davidson's account of intending has the general form of a FPJD account since not only are the responses in question the subject's judgements, but the subject and the judge are the same. We can thus write the relevant provisional equation in the case of Davidson's account as follows:

DPE: Certain conditions hold → (S intends to do φ ↔ S judges that φing is desirable).¹⁰

There are two worries regarding this provisional equation, which I am going to concentrate on: (1) DPE, contrary to WPE, does not represent the non-reductionist nature of Davidson's account; (2) we need to specify what (additional) conditions Davidson may add to the C-conditions. I begin with the first worry: if Davidson's account results in viewing pure intending as an irreducible state of the subject, this feature must manifest itself in the DPE above, just as it did so in Wright's WPE. This is important for the purpose of this article because, as we will see, it is both such features of Davidson's account, i.e., its being non-reductionist and a JD account, that help to shed new light on his later view of meaning and intention.

DPE does not represent the non-reductionist feature of Davidson's account because the concept of intention does not appear on both sides of the biconditional in it. Thus, there seems to be a tension here: on the one hand, Davidson claimed that for S to intend to do φ is nothing but to judge that doing φ is desirable for her. For this reason, instead of the concept of intention, the concept of being desirable appears on the right-hand side of the above biconditional. Accordingly, DPE represents Davidson's account as a reductionist one. On the other hand, Davidson declared that his account does not aim to reduce intention to anything else; instead, it treats it as a primitive state of the subject: “[W]e cannot claim that we have made out a case for viewing intentions as something familiar, a kind of wanting, where we can distinguish the kind without having to use the concept of intention” (Davidson, 2001a, p. 102). The concept of intention is not reducible to any other concept, nor are facts about it reducible to any other facts about S's physical or mental states. This implies that the right sort of provisional equation must use the concept of intention on both sides of its biconditional.

The tension becomes clearer if we use the distinction that Richard Holton draws among three features of response-dependent concepts. For him, the concepts that are conceptually tied to certain human responses (i.e., response-dependent concepts) can be (1) “judgement-dependent,” (2) “echo,” and (3) “users’” concepts (Holton, 1992, pp. 180–181; see also Holton, 1991). A concept is judgement-dependent if the responses that are mentioned on the right-hand side of the biconditionals in the provisional equations, like WPE and DPE, are the subjects’ judgements. In this sense, Davidson's account is a judgement-dependent account. A concept is a users’ concept if those with whose judgements the account is concerned are the same as those who use that concept. Davidson's account treats intention as a users’ concept too since the same person who uses the concept judges about its extension. Holton defines an echo concept as follows: the judgements are to be “judgements about the applicability of the very concept which is being analysed. … [T]he concept that figures on the left-hand side is echoed on the right” (Holton, 1992, p. 181). Holton's example for such a concept is that of popularity:

HPE: Something is popular ↔ most people judge that they like it.¹¹

The concept of popularity is a judgement-dependent, users’ concept but not an echo concept because the equation “makes reference to people's judgements about whether they like the thing, not about whether it is popular” (Holton, 1992, p. 182). Davidson's account, if we take it at face value, does not seem to view intention as an echo concept. This makes the account look reductionist: S's judgements seem to be about whether φing is desirable for her rather than whether φing is intended by her. It seems as if the concept of intention is taken by Davidson to be reducible to that of desirability, while Davidson has made it clear that intention is irreducible to any other concept. Thus the tension. Nonetheless, I think there are ways to resolve this issue.

4.1. Davidson's Non-Reductionism

In what follows, I discuss the following three ways to tackle with the above tension, and I proceed by assuming that we all agree that Davidson does not aim to offer a reductionist account of pure intending: (1) we can make the reductionist appearance in Davidson's account innocent by viewing it as an appearance of a tolerable, ‘vindicative’ reduction, which allows for using ‘S intends to do φ’ in place of ‘doing φ is desirable for S.’ We can then rewrite DPE in the desired (non-reductionist) form of WPE: under C-conditions, S intends to do φ if and only if S judges that S intends to do φ. (2) Insofar as our concern is to explain whether intention (to perform an action) is judgement-dependent, the quality of the action (e.g., its being desirable, good, worth doing, etc.) does not contribute to the relevant provisional equation, unless there are good arguments for a reduction. Davidson has no such arguments and has claimed that our concern is S's all-out judgement about her intention to do φ, not her prima facie judgements about what qualities φ may have. (3) It seems that we have no other choice but to rewrite DPE in the non-reductionist form like WPE; otherwise, it would remain unclear what Davidson's analysis of the concept of desirability could be. He seems to have two options: (a) to treat it as an irreducible concept, in which case the question is why Davidson did not go for this analysis in the first place in the case of intending itself, and what he then really meant by his claim that intending is irreducible, or (b) to claim that desirability itself is reducible to some other concept, in which case the vicious regress of reducing one concept to another is waiting for him.

Starting with the first way of dealing with the tension, consider an example. Once it was discovered that water is nothing but H2O, the property of being water (denoted by ‘is water’) was treated as reducible to that of being H2O. The reduction, however, is vindicative: it allows us to keep using both the concepts, provided that we keep in mind that water is nothing but H2O. But consider the case of the polywater and water-with-impurities. Once it was discovered that the polywater was nothing but water-with-impurities, we are no longer allowed to use both ‘water’ and ‘polywater’ because there is no such thing as polywater. Our reduction here is intolerable and eliminative (for more on these two reductions, see Miller, 2013, pp. 185–186). This example can be misleading in that being water is not judgement-dependent since there is no a priori guarantee that what we judge to be water, i.e., what looks like or tastes like water, is water, i.e., H2O (see Wright, 2001, p. 198, for a similar point in the case of squareness). The case of intention is different in that it is judgement-dependent: for any subject — who has the concept, who is not self-deceived, and who is attentive to her intention — it is a priori knowable and non-trivially true that she has an intention if and only if she judges that she has it. In the case of Davidson's account, one may complain, although intention is judgement-dependent, it is dependent on S's judgement about the desirability of doing φ. Thus, intention to do φ is taken by Davidson to be reduced to that of φ's being desirable.

We can respond to the above objection by claiming that the appearance of reductionism here is of the vindicative sort, not eliminativist: we are allowed, by Davidson, to use both such expressions ‘intending to do φ’ and ‘φ's being all-out desirable’ interchangeably, provided that we keep in mind that these do not denote two different properties; nor are facts about them two different kinds of facts. Rather S's intention to do φ is (and is nothing but) judging that doing φ is all-out desirable for her. John McDowell makes a similar point when he says that the implication of Davidson's account is that “‘judging that doing such-and-such is all-out desirable’ and ‘intentionally doing such-and-such’ are interchangeable” (McDowell, 2010, p. 420). Elsewhere, Davidson himself emphasizes that “an all-out judgement that some action is desirable … is not distinct from the intention: it is identical with it” (Davidson, 1985, p. 197, emphasis added). If so, we have permission to rewrite DPE in the form of WPE. In other words, what DPE says is now equivalent to what DPE* says below:

DPE*: C-conditions hold → (S intends to φ ↔ S judges that S intends to φ).

For we can use ‘S's intending to do φ’ in place of ‘doing φ is desirable for S’ on the right-hand side of our original DPE; they are co-extensive and interchangeable. If there is an appearance of a reduction in DPE, it is an innocent or vindicative one.

The second option, however, helps to throw away such an appearance of reductionism. In Davidson's FPJD account, what quality the intended φ possesses is beside the point of such an account. For one thing, this comes from the fact that, for Davidson, the judgement that S makes with regard to her intention to do φ is not prima facie. An action can have a potentially infinite number of qualities, and Davidson already ruled out the claim that S judges that she does φ because S deliberates about such and such a quality φ may or may not have. Whatever action S intends to do, with whatever quality it has for S, such an intention of hers is determined by her all-out judgement about having that intention. Our aim in a FPJD account is to show what it is that constitutes truths about S's having an intention to do something (whatever it is). Davidson's answer was this: “I now suggest that the intention simply is an all-out judgement” (Davidson, 2001a, p. 99). Of course, the subject considers different actions with different qualities. She might have prima facie reasons to make the prima facie judgement that, all-things-considered, doing ψ is preferable to the alternatives. Eventually she makes the all-out judgement that her intention is to do φ. Davidson's point is that whatever she intends to do is determined by such an all-out judgement. It is important to note that if desirability of φ was taken by Davidson to be one of the qualities of φ — which S had to deliberately pick out in order to be able to form an intention to do it — it would make S unable to form such an intention. It would make S's judgement about her intention conditional and prima facie. For Davidson, we may not know how S got to the intention to do φ from her prima facie judgements; such judgements cannot constitute truths about S's all-out judgement that she intends to do φ — because, among other things, they fail to explain weakness of the will.

My point in this part becomes clearer if we note that, in Wright's account too, the action φ possesses, or is taken by the subject to possess, a certain sort of quality. But we do not need to say anything constructive about what quality φ has for S in writing the relevant provisional equation, unless we have good arguments in favour of a substantive reduction of the concept in question to another one. Davidson does not think we have any such argument and does not aim to commit himself to any such reduction. Therefore, we can rewrite DPE in the form of DPE* above: under C-conditions, S has an intention (to do φ) if and only if S judges that she has that intention (to do φ).

The above discussion is also linked to the third option. If Davidson really aimed to reduce the concept of intention to that of desirability, he would face a problem: either he must say what concept the concept of desirability is by reducing it to a further concept, which leads to a vicious regress of reductions, or he must go for a two-level analysis of pure intending, that is, an analysis of S's intending to do φ in terms of S's judgement about φ's being a desirable thing to do, and then an analysis of S's view of φ as a desirable thing to do in terms of S's judgement as to whether φ is a desirable thing to do. Not only would that make his claim that intention simply is an all-out judgement uncanny — by involving S in making higher-order judgements — but, again, to avoid the vicious regress, he must stop at that point. One might then wonder why Davidson, in his analysis of pure intending, did not say whatever he could say in his analysis of desirability. Obviously he does not have such claims in mind. His claim simply is that for S to intend to do φ is for her to judge that she has that intention, which is the same sort of claim Wright's non-reductionist FPJD account makes.¹²

I do not claim that there can be no other way to remove such a misleading appearance of reductionism in Davidson's account.¹³ All I say is that we have good reasons and enough justification to rewrite DPE in the form of Wright's WPE (i.e., the DPE* above), which can properly, as it should, display the non-reductionist feature of Davidson's FPJD account and capture his crucial remarks on pure intending. What about the second worry about the C-conditions?

4.2. The C-Conditions

In Wright's account, “[t]he C-conditions are roughly as follows: the subject S is not lying, is prey to no material self-deception, is not making a simple slip of the tongue, has an adequate grasp of the concepts requisite for the expression of the intention, and is adequately attentive to the question of the content of his intention” (Miller, 2018, p. 275). Davidson of course agrees that such conditions are to be met because, without these conditions in play, the subject cannot be treated as a suitable subject, i.e., as a rational agent who is conceptually capable of making the relevant judgements about her intention. However, he seems to add a further condition, according to which the judgement is also constrained by the beliefs and other pro-attitudes that one possesses regarding the course of future states of affairs so that if one believes that the action ψ is not a possible action to be performed, there can be no judgement that “such an action consistent with one's beliefs is desirable. There can be no such intention” (Davidson, 2001a, p. 101). More importantly, Davidson says that “[a] rational judgement is one that takes account of all an agent's relevant beliefs: these will include beliefs about other intentions an agent has” (Davidson, 1985, p. 200). If we add this sort of condition to the C-conditions, a similar problem to that which appeared in the case of Wright's account, i.e., the violation of the Substantiality Condition, re-emerges in the case of Davidson's account too. Once provisos like ‘S's judgement about φ is to be consistent with S's beliefs about the possibility of performing φ,’ ‘S must believe that S intends to do ψ after doing φ,’ and the like, appear in the C-conditions, we have included conditions that provide the subject with whatever-it-takes to make correct judgements about her intention to do φ.¹⁴ As with Wright's account, in Davidson's account, the Substantiality and Independence Conditions must be met. How can we deal with this problem?

I believe Wright's strategy to treat such conditions as positive presumptive also works well in Davidson's account. Unless there is evidence to the contrary, e.g., a conflict between S's self-ascriptions of intention and her outward performance or self-ascriptions of other intentional states, we are a priori justified in viewing S as, for instance, not believing that φ is impossible to be performed, just as we were a priori justified in viewing S as not being self-deceived in the absence of convincing evidence to the contrary. The subject's self-conception, as Wright points out, is to cohere both internally and with the subject's behaviour. Since pure intending is accompanied with no action, the evidence turns into S's self-ascriptions of intention and other attitudes. The evidence is to be publicly available to others in the form of avowals. For, otherwise, there would be no objective ground for assessing S's error about her intention.¹⁵ Others can check whether S's self-ascription of intention, such as ‘I intend to do φ,’ is compatible with her self-ascriptions of other attitudes, as well as her future behaviour. We ordinarily, by default, treat others as not being self-deceived about what they intend, as not intending to do something that they believe to be impossible, and so forth. With no evidence to the effect that any such conditions have failed to obtain, we can consider S's judgement about her intention to be best, i.e., to be formed under the C-conditions, and thus to be determining, in a defeasible way, truths about S's possession of that intention. We can see that the notions of interpreter and interpretability play a fundamental role in both Davidson's and Wright's accounts: the presence of others is what can provide us with an opportunity to check whether S is self-deceived about her own intentions or whether S has a proper self-conception of intention. In other words, in DPE*, like WPE, we have a restricted provisional equation: not only is the biconditional itself conditional on the obtaining of the C-conditions (and is hence provisioned), the aforementioned conditions, like the no-self-deception condition, are also treated as positive presumptive — DPE* is hence restricted to the absence of any evidence to the effect that S is self-deceived or in possession of some beliefs about the impossibility of performing φ. Davidson's previous claim that judgements about performing “an action [are to be] consistent with one's beliefs” can now be seen to be very similar to Wright's claim about the positive presumptiveness of conditions like the no-self-deception condition.

To sum up, there seems to be no serious obstacle in viewing Davidson's account of pure intending as a non-reductionist FPJD account along the lines suggested by Wright. In what follows, I review some of the consequences that such a reading of Davidson can motivate.

5. Consequences

Let me begin by showing how this reading can fit well with Davidson's later remarks on meaning. For Davidson, S can be said to mean something by an utterance if her utterance is interpretable in a particular way or, as Davidson sometimes prefers to put it, if the utterance is interpretable in the way S intended it to be understood (see, e.g., Davidson, 1986; 1987, pp. 449, 457; 1991, p. 1; 1992, pp. 260–261; 1993). As Alex Miller clarifies in his discussion of Wright's account, there are two options with respect to the concept of meaning: “If meaning such and such by an expression can be taken to consist in the possession of a certain sort of intention then the application of Wright's judgement-dependent account of intention to the case of meaning will be straightforward. But even if this is not the case, we can still construct something like the story we constructed about intention for the case of meaning” (Miller, 2018, p. 277). Wright chooses the second option by claiming that the concepts of meaning and intention exhibit very similar characteristics, i.e., authoritative first-personal avowals and what Wright calls “disposition-like theoreticity” (see Wright, 2001, p. 340).¹⁶ In this case, we can go for a straightforward FPJD account of meaning by showing that the following provisional equation is non-trivially true and a priori knowable:

DPE_M: C-conditions hold → (S means φ by ‘ψ’ ↔ S judges that she means φ by ‘ψ’).

The non-reductionist feature of this account implies that the state of meaning φ by ‘ψ’ is an irreducible state of the subject and that truths about what S means by ‘ψ’ are constituted by S's own judgement about what she means by ‘ψ.’ The background conditions, among others, would include S's beliefs about the linguistic abilities of the interpreter(s), her beliefs about the happenings in the environment, her desire that the sentence is understood in a particular way, and so forth.

Davidson sometimes treats meaning something by an expression as consisting in the possession of a certain sort of intention, such as S's intention to utter ‘that's green’ to be understood in a certain way, i.e., as meaning that's green. This is a case of pure intending to do φ. At some point, Davidson righty says that others, i.e., the interpreters, can say that S meant that's green by her utterance if that utterance can be interpreted in the way S intended: “[m]eaning … gets its life from those situations in which someone intends … that his words will be understood in a certain way, and they are” (Davidson, 1994, p. 11). The latter is a case of intentional action, as the intended action is performed and can now be judged by others as interpretable. Before uttering the sentence, there is yet no uttered sentence that others can judge to be interpretable in a particular way by engaging in the process of interpretation. In this case, what the speaker intends to do with that sentence is constrained by the speaker's other attitudes, such as her beliefs about the environment and the interpreter's abilities, as well as her desires, hopes, and expectations with regard to the way that sentence would be understood. As Davidson says, a speaker “speaks with the intention of being understood in a certain way, and this intention depends on … how he believes or assumes they will understand him” (Davidson, 1994, p. 13). The action of uttering a sentence on a specific occasion has a variety of features for the speaker and she has a variety of attitudes toward all these. But, as Davidson emphasized, we may never know how S got from such prima facie reasons to the all-out judgement that she intends to utter that specific sentence on this particular occasion at this time. The beliefs and desires the speaker has form the background conditions that already constrained the range of actions (utterances) S may intend to do.

Years after offering his account of pure intending, Davidson makes similar points in the case of meaning: “[i]n speaking or writing we intend to be understood. We cannot intend what we know to be impossible” (Davidson, 1991, p. 4). For Davison too, it is a positive presumption that the speaker does not believe that her utterance is impossible to be understood; thus she does not form an intention to utter such and such a sentence on such and such an occasion. The possession of these attitudes forms what we may call the Interpretability Constraint on Davidson's account, which is compatible with Wright's Interpretationist Constraint. Davidson, like Wright, believes that “speaking a language … merely requires that the speaker intentionally make himself interpretable to a hearer” (Davidson, 1992, p. 260). For “[t]he best the speaker can do is to be interpretable” (Davidson, 1984, p. 111). Wright puts his own version of this as the claim that “I need to be considered as an at least potential object of interpretation, with my claims about my own meanings essentially defeasible in the light of the shape assumed by my actual practice” (Wright, 2001, p. 87). These constraints, as previously discussed, help both accounts to avoid the violation of the Substantiality and Independence Conditions. Viewing Davidson's account of meaning as a FPJD account, therefore, helps to accommodate the sort of ‘Interpersonal’ view of meaning that Davidson offers in his later works, according to which “meaning something requires that by and large one follows a practice of one's own, a practice that can be understood by others” (Davidson, 1994, pp. 15–16).

This reading of Davidson differs significantly from the sort of ‘Interpretivism,’ ‘Interpretationism,’ or as we may call it the ‘Third-Person-Based Judgement-Dependent’ (TPJD) account which Alex Byrne (1998) and William Child (1994) have attributed to Davidson. According to this TPJD account, “as an a priori matter, the facts about mental content are precisely captured by the judgements of some Ideal Interpreter” (Byrne, 1998, p. 200; see also Child, 1994, p. 1). Both Byrne and Child heavily rely on Davidson's famous claim that “[w]hat a fully informed interpreter could learn about what a speaker means is all there is to learn; the same goes for what the speaker believes” (Davidson, 2001e, p. 148). I have my doubts about attributing such a TPJD account to Davidson (see, e.g., Hossein Khani, 2020b, 2021). This article rather attempts to show why Davidson's view is better matched with a FPJD account. I cannot engage in a detailed discussion of why attributing the TPJD account to Davidson is wrong; but it is worth noting that if the suggested reading of Davidson's account as a FPJD account works, then either attributing the TPJD account to Davidson is wrong or it is at best a rival reading. I believe, however, that the TPJD account is deeply problematic because, among other things, such an account is based on a misconstrual of the role of the interpreter in Davidson's view of meaning; it overlooks the Davidsonian distinction between actual interpretation and interpretability.¹⁷ A proper FPJD account, and in general any JD account, is not concerned with the process of concept-acquiring. Davidson attempts to deal with this process in his discussion of triangulation, in which actual interpretation plays a constitutive role. A JD account, however, begins by presupposing that the subject is conceptually adequate. The role of an interpreter is then to do his best to keep track of the independently constituted semantic facts, i.e., those already constituted by the speaker's, rather than the interpreter's, judgements, and to assess, by engaging in the process of interpretation, whether the speaker's self-ascriptions of intention are reliable and whether the speaker's linguistic behaviour is answerable to those intentions.

The role of the interpreter is still essential in Davidson's FPJD account because without the existence of others, whatever seems right to the subject is right so that if she is self-deceived, there would be no genuine opportunity for it to be disclosed. Interestingly, one of the main concerns of Davidson's later works on the topic has been to show that his interpersonal view can successfully accommodate the Wittgensteinian seems right/is right distinction.¹⁸ In the FPJD account, which this article attributes to Davidson, the interpreters and their judgements do not play a fact-constituting role, a role that is assigned to them by the TPJD account, which Byrne and Child attribute to Davidson. The role of the interpreter is to be captured in the Interpretability Constraint on Davidson's FPJD account of meaning and intention.

This reading of Davidson can also equip Davidson with an alternative account of self-knowledge. One of the main controversies about his remarks on self-knowledge has been to clarify what his ‘official’ account is. For, in his talk of this phenomenon, he employs various notions, such as interpretation, disquotation, evidence, self-ascriptions, and so forth. The attempts to put these remarks together have led to the attribution of different accounts to Davidson as his official account of self-knowledge.¹⁹ An account of self-knowledge must provide a satisfactory answer to two questions: the ‘Why’ question and the ‘How’ question. It must explain why there is, as Davidson strongly concedes, an asymmetry in knowledge of meaning and mental content between the speaker and the interpreter. Why do we credit the subject, but not the interpreter, with authoritative knowledge of what she means and believes? The account must also answer the question of how it is that the subject has such authoritative knowledge, while the interpreter's knowledge of what the subject means and believes is only indirect and inferential. Davidson's own answer, offered mainly in “First Person Authority” (Davidson, 1984; see also Davidson, 1987, 1989), seems to be that the existence of such an asymmetry is a necessary condition on the possibility of interpretation. But why is that so? There has been no consensus on what Davidson's answer to this question is. Despite such controversies, almost everyone agrees that Davidson's account fails to answer the ‘How’ question, even if we grant that it somehow deals with the ‘Why’ question (see, e.g., Child, 2007, 2013; Hossein Khani, 2021; Jacobsen, 2009; Wright, 2001, pp. 348–350).

The Davidsonian non-reductionist FPJD account of meaning and intention can offer a far more promising way to answer both such questions, and it can do so in a way that is compatible with Davidson's later remarks on meaning and intention. As an answer to the ‘Why’ question, there exists such an asymmetry in knowledge between the subject and the interpreter because what the subject means and believes constitutively depend on her own judgements about them, while the interpreter should go through the process of interpretation in order to capture such facts. As an answer to the ‘How’ question, the speaker, but not the interpreter, has authoritative knowledge of what she means and believes because it is enough that she makes a relevant all-out judgement under the right sort of conditions (for more on how such a FPJD account can deal with the problem of self-knowledge, see Wright, 2001, pp. 113, 125, 137–142).

Another important issue on which the suggested reading of Davidson can cast new light is Davidson's twisted remarks on the notion of triangulation. Many have attempted to construe them as establishing an ‘argument’ for how facts about meaning and mental content are constituted (see, e.g., Child, 2001; Glüer, 2006; Ludwig & Lepore, 2005; Verheggen, 2007). More recently there have been attempts to read those remarks as leading to some sort of non-reductionism about meaning by appealing to the circularity which seems to emerge in Davidson's discussion of triangulation: for Davidson, only those who can speak can have thoughts with determinate content and only those who possess mental states with determinate content can have linguistic responses with determinate meaning (for his discussion of triangulation, see, e.g., Davidson, 1999c, 1994, 1992, 1982, 2001d). For instance, Claudine Verheggen (2017) and Verheggen and Robert Myers (2016) have argued that such a circularity tells us that, in Davidson's view, no reductionist account of meaning and mental content is forthcoming. No plausible explanation, however, is given by them of what form such non-reductionism takes, what problems it faces, and how Davidson can deal with them. In general, it is not clear how Davidson's remarks on triangulation can establish non-reductionism at all or argue for it. Non-reductionism is simply an option, choosing which has its own consequences, that is, dealing with specific problems that are different from those a reductionist view faces. In other words, in order to be a non-reductionist, one need not show that all reductionist views fail, by leading, for example, to circularity. Even if some reductionist view, such as dispositionalism, works, it would not imply that non-reductionism is no longer an option. A non-reductionist has to deal with different problems, such as that of explaining self-knowledge and the issues with the generality of the content of such states.²⁰

Davidson's view as a non-reductionist FPJD account does not need to tell a story about how a subject may acquire concepts, or how meanings and mental contents may emerge. To emphasize, one should be careful about the distinction between the interpretability requirement (which plays a fundamental role in Davidson's explanation of how linguistic communication between two competent speakers succeeds) and the actual interpretation requirement (which Davidson relies on in his discussion of the process of learning a first-language and concept-acquiring). In the suggested Davidsonian non-reductionist FPJD account, triangulation does not emerge as an argument; instead, it appears at most as an analogy clarifying his general non-reductionist approach to meaning. Interestingly, Davidson himself has called his triangulation an “analogy” (see Davidson, 1982, p. 327; see also Davidson, 1997, p. 25).²¹

The offered reading of Davidson in this article can also show a promising way to make sense of Davidson's complex and puzzling reading of Willard Van Quine and Kripke's Wittgenstein. Davidson has always supported the Quinean indeterminacy thesis, as well as the Quinean indeterminacy-underdetermination distinction, but at the same time declares that indeterminacy, contrary to what Quine has argued for, does not have any ontological, sceptical consequences. As he says, the Quinean indeterminacy “does not entail that there are no facts of the matter” (Davidson, 1999b, p. 596). For Davidson, facts about meaning are “invariant” (see Davidson, 1999b, p. 596), though there can be different theories of interpretation, all of which equally well capture such facts. Thus, he claims that the Quinean indeterminacy at most leads to the (epistemological) problem that there are different ways of capturing the same facts about what the speaker means and believes. Unpacking his reading of Quine goes beyond the scope of this article (on Davidson's puzzling reading of Quine, see Hossein Khani, 2017, 2018a; Kemp, 2012). But one way (and I think the best way) to explain why Davidson reads Quine this way and what he thinks of the facts about meaning and mental content is to view him, as this article does, as a factualist, i.e., a non-reductionist about meaning — which goes against Quine's position. In this case, the problem that he faces would plausibly be epistemological, that is, that no amount of evidence can help to decide which theory should be chosen as the correct theory of interpretation, among those that are all compatible with the totality of such (fine-grained) facts about meaning and mental content.

Similarly, this non-reductionist reading of Davidson can explain why Davidson's objections to Kripke's reading of Wittgenstein, presented in Kripke's (1982) book, have been considered as puzzling and even plainly wrong (see, e.g., Hossein Khani, 2017, 2018b, 2019, 2020a; Verheggen, 2017; Verheggen & Myers, 2016). Roughly speaking, Davidson says that the problem that “troubled Kripke, seems to me to have a relatively simple answer” (Davidson, 1992, p. 258, original footnote omitted), which is, as he immediately puts it, the following: “The longer we interpret a speaker with apparent success as speaking a particular language the greater our legitimate confidence that the speaker is speaking that language …. Our strengthening expectations are as well founded as our evidence and ordinary induction make them” (Davidson, 1992, p. 258). Similar to his reaction to Quine's indeterminacy thesis, here too Davidson thinks that Kripke's Wittgenstein's problem leads to no scepticism about meaning and can be dealt with by appealing to the process of interpretation and induction.²² Within the common reading of Davidson — in which facts about what the speaker means and believes are constituted by the interpreter's judgement about them — this reaction to Kripke's Wittgenstein is simply wrong. For, Kripke's Wittgenstein's sceptic argues that there is nothing for the interpreter to interpret at all. But by viewing Davidson as holding a non-reductionist FPJD account of meaning, we can then take him to be offering a straight non-reductionist solution to the sceptical problem, according to which facts about meaning and mental content are sui generis. Accordingly, the most important problem to cope with is an epistemological one, i.e., a proper explanation of first-person epistemology of meaning, which, as previously indicated, can be offered by Davidson's non-reductionist FPJD account.

This reading of Davidson can also elucidate his view of the metaphysics of meaning and mental content, which Davidson has done his best to hide. He has refused to be labelled ‘anti-realist,’ ‘realist,’ ‘irrealist,’ and the like, which has led to a variety of conflicting interpretations of his position.²³ The suggested reading in this article construes Davidson as a factualist, in its non-reductionist form. There is much more to say about how this reading may help to shed new light on, for instance, Davidson's anti-conventionalist view of meaning and the problems with Davidson's treatment of the normativity of meaning. The limitations of space do not allow me to engage in detail in these topics, but if the interpretation of Davidson offered in this article has been successful in prompting new questions about Davidson's philosophy and inducing some alternative ways to approach them, then the article has been successful in achieving one of its chief goals.