Is Narrow Content's ‘narrow content’ narrow content?

Click to increase image sizeClick to decrease image size

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Y&H define thoughts as intentional states that are ‘aptly described by “that”-clauses’. (p. 1) The suggestion seems to be that thoughts are intentional states that are ‘aptly described’ by propositional attitude ascriptions of the form ‘A believes/desires/etc. that P’. It is not entirely clear how ‘aptly described’ should be understood here: must the state be ‘aptly described’ entirely or in part? Must it be described entirely accurately or merely sufficiently accurately for some purpose? Demanding interpretations of this phrase narrow the scope of Narrow Content in ways that don‘t seem entirely charitable. In particular, Y&H recognize that many internalists don't think narrow contents are (always or typically) the contents ascribed by ‘that’-clauses, so they presumably don't mean ‘described entirely and accurately’ (if they did, the views in question would be out of scope). Given the kinds of states Y&H focus on, it seems that they use ‘thought’ to mean any intentional state of a propositional form, i.e. any state that ‘says’ that something is the case.

2 The condition that contents must determine intensions is stated on p. 24. We couldn't find an explicit statement that this is sufficient for being a content, but this appears to be how Y&H use the term ‘content’ since when considering various content assignments they never worry that the intensions or intension-determining entities assigned by a relation might not in fact be contents.

3 Farkas (2008), who appears to be one of Y&H‘s main targets, argues at length that the right way to understand internalism and narrow content is in terms of subjective indistinguishability: the internal is the subjective point of view that is common to twins, where the subjective point of view is roughly a subject's consciousness. This seems to imply a commitment to narrow content being determined by QAPs, but there is a nearby, weaker view that is very much in the spirit of Farkas’ view and arguments. This nearby view takes narrow content to be determined not just by the phenomenal types of experiences, but by experiences (particulars instances of phenomenal types). As we suggest below, experiences are the main particulars that seem relevant to determining narrow content.

4 Using the term ‘psychological content’ for narrow content, Loar writes: ‘that-clauses on their oblique readings are sensitive, either directly or indirectly via translation, to how beliefs would linguistically be expressed, and that is […] only loosely related to psychological content’. (p. 164) Chalmers (2011b) readily concedes that the rules of propositional attitude ascriptions are complex and in no way guarantee that a correctly used ‘that’-clause simply denotes the ascribee's narrow content (it requires at most that the expressed proposition be ‘coordinate’ with the narrow content). Jackson (2003) suggests that ‘the respect in which the semantics of natural kind terms deliver broad conditions is precisely the respect in which they get contents intuitively wrong’, and that this is true of the semantics of names as well (see also Jackson 2004). Y&H ascribe the view that ur-contents are narrow to Farkas (2008, ch. 4) and Segal (2000). We could not find clear evidence of this view in Farkas' ch. 4. In correspondence, Farkas rejected the idea that she is interested in a notion of narrow content tied to the semantics of attitude ascriptions. Segal (2000) does subscribe to the view that, when used strictly correctly, the ‘that’-clauses of attitude ascriptions pick out narrow contents.

5 ‘Theories of narrow content get their interest not by simply claiming that there is a narrow content assignment but by claiming that there is a narrow content assignment that satisfies some further, non-trivial conditions that render them theoretically interesting. […] One flaw of the literature is that the nature of these further conditions is often left unclear. Nevertheless, conditions from one or both of two families [structural and theoretical constraints] are typically in play implicitly or explicitly’. (p. 39)

6 This is an approximate statement because it (indirectly, via the definition of a narrow content assignment) employs the notion of supervenience, and it is not mandatory (though it is common) to cash out the relevant notion of determination in terms of supervenience. More on this below.

7 Various ways of sharpening up this notion have been suggested. One way is the linguistic approach of Chisholm (1957, ch. 11), which equates the intentionality of mental states with certain linguistic features of propositional attitude ascriptions. This approach had its heyday, but it has now become clear that it leaves out too much of the original notion, especially if we are trying to capture narrow content (for reasons already discussed).

Another way of sharpening the notion of content is by focusing on representations' important metaphysical attributes. A peculiar fact about mental representation is that we seem to be able to represent situations that are not actual. Based on this, one of us (DB) has suggested a sharpening of the notion of an intentional state (and, by extension, content): intentional states are those mental states that consist in standing in a non-factive relation to one or more proposition-like entities (of course, this is not to say that any state (mental or not) of standing in a non-factive relation to a proposition is an intentional state). This definition of intentional states is far removed from the intuitive starting point, but it is helpful in capturing what is at stake in certain debates, such as those surrounding the intentionality of consciousness. See Bourget (2019).

A less committal definition of intentionality simply takes the most salient, uncontestable examples of intentional states and defines intentionality and content ostensibly. See Kriegel (2011) and Mendelovici (2018).

8 We disagree that the linguistic and rationality constraints clearly imply the representation constraint. Regarding the first, it seems perfectly possible, given the complex rules of propositional attitude ascriptions, that the contents that we literally (and perhaps correctly) ascribe in such ascriptions turn out not to be represented by the ascribees. Of course, this will seem hard to imagine if we define mental content as ur-content (as Y&H seem inclined to do). But if we think of mental content as that which is represented, on the Brentanian notion of representation described earlier, it is conceivable that mental content has little to do with the logical forms of linguistic descriptions of mental states. Regarding rationality, we just don't think internalists need to be committed to claims about apriority unless those claims are truisms that fall out of a definition of apriority along the lines of ‘having a necessary narrow content’, which of course means that narrow content needs to be defined independently of apriority (we can't define narrow content as that on which apriority supervenes and apriority as having a necessary narrow content). Short of this kind of definitional link that takes narrow content to be conceptually prior, it is unclear why narrow content and apriority should be tied together, so we don't think internalism should be understood in terms of apriority.

9 While we do think narrow contents play a rational role, we think the notion of narrow content is prior to that of apriority and related notions, and we think it would be a mistake to try to settle questions about their relationships without having a good grip on narrow content independently of any epistemic notions.

10 Y&H assume a coarse-grained conception of contents for most of the book, but at this point, they assume a fine-grained conception. We will go along with this for the sake of argument, since we agree that it would be a problem if quasi-internalism were pointless on a fine-grained conception of contents.

11 One common proposal (which Y&H discuss but do not ultimately endorse) is roughly that an intrinsic property is a property such that a thing‘s possession of it does not imply the existence of anything outside the thing. Properties making reference to world-bound entities are clearly extrinsic by this standard since they imply the existence of every actual thing. Properties making reference to world-bound entities remain extrinsic to subjects even if we say instead that the thing's possession of the property is modally independent of the existence of things outside the thing. Another proposal, which Y&H (ch. 5) consider superior but imperfect, is that the intrinsic properties are those that supervene on the natural properties of things and their parts. Tom Hanks’ property of being partly constituted by @nose is not intrinsic by this standard since Twin Tom Hanks could have a face whose parts exhibit all the same natural properties as their counterparts in Tom Hanks, but it would not have the property of being constituted by @nose. Another proposal considered by Y&H is that an intrinsic property of x is one that is not ‘about’ anything outside of x. Insofar as we understand this proposal, it seems to us that having-a-nose-developed-in-such-and-such-way-on-earth is a property that is ‘about’ things outside of Tom Hanks.

12 If all events are world-bound and hence excluded from APs, doesn't this remove too much from the internalist's supervenience base? We think not, since, presumably, there has to be a way of individuating events or event-like things other than by the totality of their intrinsic and extrinsic properties. Denying this would have bizarre consequences, like that on any accurate way of thinking about events, World War II would not have happened if you didn't read this footnote. Even if there is a technical sense of ‘event’ on which events couldn't have been other than they intrinsically and extrinsically are, there is also another kind of thing (perhaps we should call it a ‘schmevent’) that is not world-bound. We might characterize these things as instantiations of properties. These are the event-like entities that are supposed to form part of the internalist's supervenience base.

In any case, it may be worth noting that the arguments Y&H adduce in favor of the world-boundedness of events are not particularly well-developed or, arguably, compelling. They write, ‘One way to motivate this view is to begin with Jaegwon Kim's thought that the time of an event is essential to it, and combine this with the reasonably common impulse to treat the temporal and modal dimensions symmetrically’. (p. 156) Y&H cite works arguing that time and modality can be formalized in similar ways. We are not sure how to understand this quick argument. The idea seems to be that, if time and modality are symmetrical and events have their times essentially, they must have their possible worlds essentially as well. But the fact that time and modality can be formalized in similar ways is an extremely tenuous basis for such an inference. The other quick argument Y&H offer for the world-boundedness of events is that ‘short of world-boundness, there is nothing approaching a test for when an event in one world is the very same event as an event in another’. (p. 156) A footnote acknowledges that this is a weak argument, and we agree. It slips into thinking of possible worlds as places that we discover rather than a model for thinking about possibility and necessity (see Kripke 1980).

13 Y&H consider something close to a grounding characterization of internalism. They reject the grounding picture for roughly the following reason. Contents are abstract entities existing distinctly from a subject's internal state and to which subjects or thoughts are related. A relationship is grounded in facts about all its relata, not merely some of its relata. So, facts about an internal state cannot by themselves ground a relationship to a content: facts about the content itself must play a role. So, grounding internalism is false. We think this is an interesting argument, but it neglects two possible lines of response. First, not all internalists endorse the kind of relational view presupposed here. On some views, contents are parts of subject's internal states, either because they are adverbial modifications of subjects, components or aspects of intentional states, or intentional states themselves (Kriegel 2011; Pitt 2009; Mendelovici 2018). Second, an internalist who does embrace relationalism (as one of us does, see Bourget 2020) could simply tweak their grounding formulation of internalism. One option is to say that there are represented contents whose representation is not grounded in contingent facts outside the subject. This kind of grounding internalism is unaffected by Y&H's observation that relationships to abstracta are partly grounded in necessary facts about the abstracta.