1. Introduction

1What role does subjectivity play in determining intentional content? Some will say none: intentionality, on their view, can be characterized from a thoroughly objective, third-person perspective. Nevertheless, in what follows, I will distinguish two ways in which subjectivity might be thought to have a role – one first-personal, the other second-personal. I will then focus on the (arguably less explored) second-personal approach, considering further refinements of the role of the subject. The upshot will be that, on the best version of this approach, subjectivity is methodologically, but not metaphysically, involved in content determination.

2. Intentionality and Grammatical Point of View

2There are broadly three perspectives on intentionality in the philosophy of mind, namely (cf. Pautz, forthcoming) those of: the (internalist) phenomenal intentionality programme; interpretationism; and (reductive externalist) causal-information theories. Can they be subjected to an analysis in terms of differences in grammatical point of view – first, second, and third personal respectively?

3The phenomenal intentionality programme builds on Nagel’s (1974) work suggesting that experience has a subjective character – there is something it is like for the subject to undergo a conscious mental episode. It then develops this observation into a theoretical account that grounds intentionality, not just for sense perception, but also for moods and the emotions, and even (through an exploration of ‘cognitive phenomenology’) for thought (cf. Montague, 2010). In particular, the suggestion is that through phenomenological investigation – a paradigmatically first-personal method – the qualitative characters of various subjective experiences (i.e. what it is like for a subject to undergo them) can be discerned; that these will be found to involve, or supply, accuracy conditions – i.e. a certain sort of intentional content; and that this applies even in the case of ‘cognitive experiences’ (Pautz, forthcoming). While some theorists (e.g. Tye, 1995) embrace intentionalism about phenomenal consciousness, taking the (independently fixed) representational character of an experience to determine its qualitative character (due to its own transparency), advocates of phenomenal intentionality (e.g. Horgan & Tienson, 2002) see the connection here running in the opposite direction, with (subject) internally determined experiential qualities fixing a kind of narrow content, which can then be supplemented with appropriate causal connections to fix wide content. In this way, all intentional mental content is metaphysically grounded in (internally determined) phenomenal intentionality.

4According to interpretationism, by contrast, a subject’s intentional mental states are metaphysically determined as those attributed in a correct interpretation (Williams, 2020). Such an approach – dubbed ‘best systems theory’ by Pautz (forthcoming) – is characterized by its reliance on the global, or holistic constraints of rationality (Davidson, 1970): the correct attributions of attitudes are charitable, rationalizing the subject’s actions in light of the circumstances in which they find themselves (cf. Lewis, 1974). On at least some formulations of the view, however, (correct) attributions of attitudes are (those) made by idealized interpreters – characters who might be thought to introduce a second-person perspective into the picture (see below).

5Finally, causal-informational approaches attempt to reduce mental intentionality to certain sorts of (non-accidental, lawlike, or counterfactual supporting) correlations between (internal) states of the subject and those of her (external) environment. Such approaches may be thought to embody the ‘third-person world view’ (Kemp, 2012, p. 2) of natural science – Nagel’s (1989) utterly objective ‘view from nowhere’.

6What about the intentionality of language? Simchen (2017) distinguishes two approaches to metasemantics – that is, “the metaphysics of semantic endowment” (2017, p. 75): productivism, on which (referential and truth-conditional) content is fixed by “conditions surrounding the production” (2017, p. 175) of linguistic utterances; and interpretationism, on which what is relevant are “conditions surrounding the interpretive consumption or reception of such items (e.g. facilitation of good explanations of speakers’ verbal behavior)” (2017, p. 175). Arguably, the speaker’s, or producer’s perspective can be considered that of the first person, and the hearer’s, or consumer’s that of the second person. Finally, while Simchen does not consider such views, teleosemantic approaches to linguistic intentionality (such as that of Millikan, 1984) emphasize the roles of both producers and consumers in fixing content and may therefore suggest a third-person perspective, that of a (scientific) bystander observing and objectively characterizing interactions between two other parties (speaker and hearer).

3. Radical Interpretation and Epistemic Constraint

10Davidson (1973) introduced the problem of radical interpretation, asking what an observer of linguistic activity could know that would enable her to interpret a speaker’s utterances – in short, what evidence would suffice to enable her to interpret him. The interpretive task in question was assumed to be undertaken under the radical circumstances in which the interpreter could draw on no prior knowledge of the speaker’s language or attitudes. As Williams (2020) stresses, Davidson’s problem has its origins in Quine’s work (e.g. Quine, 1960): but Quine focussed on translation undertaken in the radical circumstances imagined above – the activity of producing a mapping from the expressions of one language to those of another (intuitively – though not on Quine’s account - those that are synonymous with them); Davidson, by contrast, considered interpretation, a process that targets a single language (though it may be undertaken in another), aiming to render its expressions and speakers intelligible.

11On Davidson’s articulation the problem has an epistemological bent – it is concerned with the evidence an interpreter can rely on in radical circumstances. Moreover, it is, on the face of it, concerned only with the interpretation of speakers of a language. (It turns out that those are the only agents with attitudes on Davidson’s view.) But Lewis (1974) reformulated the problem, generalizing it in such a way as to apply to non-linguistic agents, and making clear that it is ultimately metaphysical in character: how do the physical facts determine the intentional facts about meanings (if any) and attitudes? He also offered a different solution to the problem – though one that continued to emphasize the importance of principles requiring charity in interpretation, and an attempt to regard agents as rational. (Indeed, these commonalities between the two theorists might be regarded as definitive of interpretationist approaches to intentionality more generally – together they amount to the need to regard the fixation of intentional content as subject to the global constraints of rationality, broadly construed, as in Davidson 1970.)

4. A Pragmatic Role for the Interpreter?

Read as a piece of metaphysics, an appeal to an interpreter’s projects and beliefs would be bizarre. Truths about each individual’s beliefs would obtain only relative to another’s perspective, and those perspectives would be constituted in part by the others’ attitudes, generating a morass of circularities (2020, p. xix).

17Arguably, then, if we can interpret the words of others, it is because we share interests with them. Detecting tea, for instance, may have more to do with caring about (or being interested in) it than anything else (such as our epistemic capacities). And yet according to Wittgenstein (1953), “if a lion could speak, we could not understand him”. Why not? Primarily, I take it, because lions have different interests than we do. We cannot be (fully and appropriately) sensitive to what lions are up to (with their radically different form of life). Lions’ interests diverge from our own. And yet, if there is a divergence of interests between subject and interpreter, surely the correct interpretation is one that is sensitive to the (feline) subject’s interests, not the (human) interpreter’s! This suggests that relativism may indeed be a worry in explicating the metaphysics of intentionality in terms of a process of interpretation that takes into account the idiosyncratic interests of the interpreter.

18Interestingly, Hawthorne (2007) suggests a quite different role for the interpreter’s interests – one that does not appear to fall prey to this concern. He notes that agents in general have interests. Suppose, then, that (for some proposition p) a given (arbitrary) agent is interested in whether p. As Hawthorne notes, by minimal logical reasoning, it follows that they are interested in whether it is true that p. Thus, all agents have an interest in the truth. Hawthorne suggests that this fact can help to secure a role for the notion of truth, which is central to the (radical) interpretation that determines the contents of a given subject’s attitudes. Thus, it is not insofar as interpreters are epistemically constrained subjects, but rather because they are interested agents, that they have a key role to play in interpretation – and yet, this is not a reason in itself to expect any subject or interest relativity in correct interpretation.

19As I read him, Hawthorne is suggesting that a certain pragmatism lies at the heart of interpretationism. But it is (as a Kantian might say) the pure form of agency as such that determines a role for the (idealized, omniscient) interpreter on Hawthorne’s understanding of interpretationism. And this in turn suggests an anti-reductionist outlook: the (omniscient) interpreter needs to come at the task of interpretation in a way that is sensitive to high-level features like truth, and truth-conditional content, not just the low-level properties that are discerned by the natural sciences. In short, then, pragmatism need not yield relativism; but it does point away from reductionism – in effect, embracing the ‘circularity’ Williams regards as problematic.

5. Objections, Replies, Clarifications

20It will be worth considering objections to the foregoing reasoning, as the replies may help to clarify the position being described.

22It will not do to respond by reconstruing the original argument along the following, simpler lines: agents have an interest in the truth; so it is appropriate to interpret them using the concept of truth (rather than lower level physical notions). The problem is that, even if it is successful, this route to the conclusion that truth has a role to play in interpretation passes via the interests of the interpreted, not those of the interpreter – and it therefore fails to secure a role for the latter in radical interpretation.

23A consideration of Zawidski’s (2015) elaboration of Dennett’s interpretationism may help us to see a way forward. According to Zawidski, natural selection helps to secure the legitimacy of the interpretation of systems in intentional terms, on Dennett’s approach, in two steps. First, it favours systems that reproduce efficiently, and so can be regarded as pursuing goals by effective means – or, in other words, as intentional systems; and thus, it helps to secure the existence of many such systems. Second, it ensures that systems that engage in interpretation in this way are tracking real (rather than spurious) patterns – for otherwise it would not be advantageous for them to do so! (One might add that, in certain sorts of situations – e.g. those in which a fight or flight reaction seems appropriate – such ordinary, naturally occurring interpreters may be inclined to make their attributions of attitudes second-personal, seeking to answer the question, what are you doing/going to do?)

24Does this line of thought provide a reason to think interpreters will, after all, wield the concept of truth, given their interest in doing so (so that the transition from the second to the third step above is vindicated)? Better, I think, to reconstrue the (original) argument as ensuring the appropriateness of interpretation that attributes attitudes (and speech acts) with truth-conditional contents (where this involves both an element of accuracy and an element of utility – or, as Hawthorne says, significance). In other words, we move directly from the thought that interpreters have an interest in the truth, to the thought that it is therefore appropriate for them to employ truth-conditional contents in interpretation (whether or not they have the concept of truth). This then allows us to draw the further conclusion that there is a pragmatic role for the interpreter in radical interpretation.

25This raises a further question. Does the response given here conflict with what was said previously about relativism and/or the divergence of interests between us and lions? This would be ironic, as Zawidski develops the above line of thought as a way of furthering Dennett’s response to the relativism charge. Nevertheless, we can ask, to what extent does interpretation serve a kind of coordination function (between interpreter and interpreted)? Is it key that the contents attributed are in some sense shared between interpreter and interpreted, thereby reintroducing idiosyncrasy in the interpreter’s perspective? A consideration of some remarks of Tollefsen’s will, I believe, prove illuminating on this point – and at the same time bolster the case that the role of the subject in interpretation is genuinely second-personal. The short answer, though, is that it does not: all that must be shared between interpreter and interpreted is sensitivity to the truth.

26Tollefsen (2015) suggests that when, as interpreters, we adopt Dennett’s intentional stance, we endeavour to discern reasons the agent has which both explain and justify her actions. Accordingly, not just any reasons will do – they must be ones that make her action intelligible from her own point of view and not just ours. She writes: “an action (including an utterance) will be intelligible to us – in the relevant sense – only if it makes sense to us that it made sense to the agent that she did what she did. This reflective constraint guides us in our interpretive endeavours” (p. 100, emphasis original).

27This articulation of the reflective constraint certainly sounds as though two parties are essentially involved in interpretation on Tollefsen’s view: the reasons attributed to the agent must justify the action, both to the interpreter and for the interpreted. As Tollefsen herself puts it, “we must assume that the agent shares our norms of rationality” (p. 100); and “our practice of interpretation involves the positing of an alternative point of view” (p. 100, my emphasis). In other words, it takes two to tango… and to interpret. The interpreter, it appears, must adopt a second-personal point of view on this account.

28It is therefore somewhat surprising to see Tollefsen say:

I am introducing here what others have called a first-person point of view…. [W]e need not associate this point of view with consciousness…. A rational point of view… is a perspective and one that can be adopted by other agents. When we make sense of others, we often project ourselves into their rational point of view in order to be able to better provide reasons that are intelligible in the way that the reflective constraint requires (p. 100, emphasis original).

29The point here about the nature of a point of view (involving rationality, or sapience, rather than, say, sentience) is well-taken; what is a little surprising is to see it described as first-personal.

30It is worth remarking, however, that Tollefsen herself does not describe the perspective in this way in the passage cited: she says that others have so described it; this of course leaves open that she herself would not. But regardless of whether she would, it remains the case that the reflective constraint, as articulated, makes it clear that interpretation involves two distinct perspectives (not one or three). Not only that, there is perhaps a case to be made that in assuming shared norms of rationality, interpretation becomes inherently interactional – the interpreter opens themselves up to the possibility that they will need to revise their standards of rationality (e.g. their logic) in light of evidence that the interpreted’s standards are preferable. This is not something that occurs in straightforward, third-personal observation and theory building.

31Be all of this as it may, the key point here as regards the objection that idiosyncrasy, and with it relativism, has been reintroduced is that it is not contents per se that are assumed to be shared, but a kind of sensitivity to truth. This is not relative – if it were, it could not induce the interpreter to adjust their standards!

32If relativism is not a concern, what about circularity? Zawidski thinks that this (third) worry can be addressed roughly as follows. Intentional systems are those that exhibit a certain kind of behavioural pattern. Which kind? The one that is discerned by a certain naturally occurring sort of system recognized in developmental psychology. When the pattern in one system is discerned by another (of the kind developmental psychologists study), the second system is itself engaged in an intentional act. This appears to introduce the threat of circularity: we are saying which systems are intentional systems by appeal to the notion of an intentional system! But ultimately, the fact that interpretation (of the kind recognized by developmental psychologists) is appealed to when saying which sort of behavioural pattern is at issue simply means that the behaviour of the interpreting system itself also exhibits a pattern that can be discerned by systems that engage in interpretation (so understood). “In other words”, says Zawidski, “the intentional stance is recursively applicable” (2015, p. 603). This precludes discerning the patterns in question at a lower level of analysis – but “ontologically, there are just behavioral patterns and dispositions” (2015, p. 603). In this sense, interpretationism suggests a methodological anti-reductionism, but does not require any metaphysical additions to a naturalistic ontology.

33Finally, a fourth objection is simply that the argument given above against epistemic constraint in interpretation was too quick. In response, let me say the following by way of clarification. It may be the case that the explanandum of radical interpretation is subject to epistemic constraint: roughly speaking, what is to be explained is (high-level) observable behaviour (characterized non-intentionally). What is not subject to epistemic constraint, in my view, is the explanans: any facts whatsoever can be drawn on in constructing and confirming hypotheses about which attitudes and meanings serve to explain the observable behavior. (Of course, in order to receive an explanation in terms of attitudes and behavior, the behavior itself may need to be redescribed in intentional terms.) This gives us a clear sense in which interpreters are, by their very nature, epistemically limited, interested agents, without imposing any epistemic constraint on radical interpretation itself.

6. Conclusion: An Anti-Reductionist Methodology