<Forthcoming in Philosophical Studies. Draft of 4 Sep 2020. Please cite the published version.> Reply to Byrne Juhani Yli-Vakkuri Australian Catholic University John Hawthorne Australian Catholic University & University of Southern California In his comment on our book Narrow Content (NC), Alex Byrne (2020) questions the need to frame disagreements between internalists and externalists, as we did in the book, as concerning assignments of contents to thoughts. Byrne prefers to explore the merits of internalism within a ‘coarse narrowness’ framework, where we prescind from positing thoughts and instead discuss whether various content-involving properties of thinkers—including those that might be expressed by constructions of the form ‘believing that p’—depend only on inner goings-on Byrne is right that our baboon example was indecisive against his favored coarse framework, and his points in this connection are very well taken. We also agree that the coarse framework does not prejudge the debate between the sectarian and the ecumenical internalist. However, we would articulate ecumenism a bit differently than Byrne does within a coarse framework. Here is how he does it: … consider a view on which (i) believing that snow is white is extrinsic, and (ii) being in that state entails being in the intrinsic state of believing that the snowish-stuff is white, where the proposition that the snowish-stuff is white is the “internal” component of the broad content that snow is white (**). Byrne’s ‘basic’ ecumenical idea is that, for some contents, the property of believing that content is broad, but for other contents, the property of believing it is narrow. This is analogous to the view within our framework that a component of the ur-content assignment is narrow. In effect Byrne’s ecumenism is what you get when you restrict the sectarian thesis to an interesting subclass of contents. But that is not how we thought of ecumenism. Our ecumenism says that the ur-content assignment is broad but a different kind of content assignment is narrow. If one is looking for something analogous in the coarse framework, it is this: The class of belief properties is not narrow—i.e., at least some belief properties are ‘not in the head’. However there is relation belief* and a set of contents c that are the possible relata of believing* such that (i) the properties believing* any given c is a narrow property and (ii) believing* can do various interesting kinds of explanatory work. Note that it is compatible with this thesis that no belief property is narrow. So the issues are not merely terminological. We would like to point out a few other things in passing about Byrne’s way of framing the ecumenism debate. First, the idea that believing that the snowish-stuff is white is narrow obscures an important point. ‘The’ carries a presupposition of uniqueness. But where is that coming from? 1 There may be two snowish stuffs in the universe. Perhaps what Byrne has in mind is some expansion along the lines of ‘the snowish stuff near me is white’. But the property of believing that proposition seems to have little hope of being narrow: After all, if Juhani believes that the snowish stuff near Juhani is white, it hardly follows that a duplicate in a far-off land believes that as well. One worries that Byrne is setting up the debate in such as a way that the internalist will lose so quickly on those terms that they will not accept the terms in the first place. Second, one wonders how to cash out the ideology of ‘being the internal component’ in the coarse setting. What is it for one belief property to be the internal component of another? Is the idea that for a narrow belief property F to be the internal component of a non-narrow belief property G, G must entail F? If so, the cards are stacked quite heavily against the internalist at the outset, since it is hard to imagine any gloss on ‘snowish stuff” such that it is even halfplausible that every case of believing snow is white is a case of believing the corresponding snowish stuff proposition. Meanwhile. if the language of ‘internal component’ is to be given a different gloss, it would be good to know what that is. Did we have any good reason for operating within a thought-based framework? Yes: a coarse framework would have made it extremely hard to engage with the internalist literature. For example, Chalmers’ discussion of a priority makes use of compositional principles in a way that couldn’t even be expressed within a coarse framework. More generally, it is far from clear how to make good on the explanatory ambitions of many internalists in a coarse setting. Consider the internalist idea that narrow contents can best subserve the causal explanation of behavior. These discussions typically posit causal paths connecting thoughts to behavior, and the question then arises which properties of those thoughts are explanatorily important. The coarse reframing risks obscuring some prima facie important distinctions, such as Davidson’s famous distinction between a desire causing and merely rationalizing some behavior. Perhaps there is a way to make sense of all of this in the coarse setting (the hope would be that counterfactuals plus holistic features of the brain would be good enough for all the distinctions we need), but we did not want to engage with such issues in NC. Meanwhile, it is unclear how the coarse framework can offer any hope to the internalist who wishes to explain a priority. In the standard ecumenical framework, the a priority of a thought with a certain broad content is explained by appeal to another narrow content assignment that associates it with a different content. Without content assignments, how are we to explain, say, that somenone knows a priori that Hesperus is Hesperus? It is unclear what it would take for some narrow attitudinal property to be ‘associated’ with a belief that Hesperus is Phosphorus in a way that could explain its a priority. While we are—to put our cards on the table—quite sympathetic towards Byrne’s Stalnakerian hesitancy about representational posits for belief and desire states, we do think that representations will be important for explaining phenomena in the ballpark of a priority. We think that the felt obviousness of some claims we hear is due to features of representations we construct in response hearing them. Holistically ascribed attitudes will not, we suspect, be adequate to explaining the epistemological-cum-psychological phenomena that have preoccupied many internalists. Yet an internalist operating within the coarse framework has more resources that Byrne realizes. One theme of his discussion is that without thoughts available as indices, the internalist can no longer wriggle out of the Mirror Man challenge. However, it is worth noting that there are in principle some ways of trying to keep the internalist project going. If we are operating within the coarse framework, there is no place for a ‘one content per thought’ constraint, since we are no longer able to count thoughts. Given that, we might think of narrow contents in a more 2 holistic way: for example, we might say that any believer has a global belief state with a narrow content, where that narrow content is a function from indices to an entire belief state (i.e., the entire set of belief properties of that agent). And here some inventive indices can do the trick: Suppose for example we used worlds, entire intrinsic histories of an agent, and times as indices. Then Mirror Man would not pose a threat. If Mirror Man both believes that Fido is a dog and that Fido* is a dog then the narrow content will take as input the time, the world, and the intrinsic history of Mirror man, and then spit out a set of belief properties that includes both believing that Fido is a dog and believing that Fido* is a dog.1 There is no crash here. Once again, the main issue is that of explanatory worth: what do such narrow holistic contents explain? (Note, for example, that this holistic framework does not preclude there being a single narrow content shared by all thinkers, with the indices doing all the work.) Byrne thinks that there is a strong case for worlds being the only index and thus that I=W. He thinks that, if that is right, ‘a number of internalist positions in the book do not get off the ground’ (**). We note in passing that we do not think the case for I=W is quite as straightforward as Byrne thinks. For one thing, there are foundational issues in the philosophy of time here. On one respectable view, it can be true simpliciter than Juhani is sitting even though Juhani could have been standing and even though it will be that Juhani is standing. If tense and temporal operators are analogous to modal operators in this way, then it becomes natural to opt for a Kaplanian vision where worlds and times have importantly analogous relations to contents. How sunk is internalism if I=W? There are still ways to keep something in the internalist spirit going. Let’s imagine a version of Lewis’ (1979) vision according to which there is a relation of self-ascribing to properties that is narrow: If x self-ascribes a property so does x’s twin. On the internalist-inspired vision we have in mind, self-ascription is not belief, and the things that are self-ascribed are properties, not contents (for which I=W). But the following bold proposal is still live: Whenever a believes p, a self-ascribes is some property P, such that p = Pa. (Easy case: a believes a is hungry by self-ascribing hunger.) There is no clash with I=W here. And if the thesis was correct it would surely count as a kind of vindication of the internalist vision. That is not to say that we are optimistic. Suppose a believes that Aristotle was a philosopher. Is there really a self-ascribed property P such that its being self-ascribed is a narrow matter and P(you) = Aristotle was a philosopher? The prospects are not great. Here the internalist might try either one of two moves (there are precedents for each in the literature): (i) one does not strictly speaking believe Aristotle is a philosopher, since one does not typically grasp the semantic contents of singular claims. Rather, one believes some content like: the person one calls ‘Aristotle’ is a philosopher. (ii) It is true that one believes that Aristotle was a philosopher, but that claim needs to be understood using the resources of Kaplan’s ‘Quantifying in’: one believes that Aristotle was a philosopher iff there is a some content (yes, an I=W content) c such that one believes c and c represents to one that Aristotle was a philosopher. This last move takes us right back into the dialectic of Ch. 3. To motivate his skepticism about thoughts, Byrne does some ordinary language philosophy. He points out that ‘Over there/in his office Juhani believed that snow is white’ sounds infelicitous and concludes that it is dicey to work with an ontology of located thoughts. We are less sanguine about drawing metaphysical conclusions from such facts. Note for, one thing, that ‘Over there, Juhani is judging that snow is white’ and ‘Over there, Juhani is forming the belief that snow is white’ are fine, so this test can’t even tell prima facie against belief- 1 Thanks to Daniel Stoljar for discussion here. 3 formings and judgings as having location. Consider also that we readily say ‘Over there, the wall is red’ but that ‘Over there, the marble is red’ sounds odd (since it’s not as if the marble is located anywhere else). But we would not conclude from this that the redness of the wall has a location while the redness of the marble does not. Imagine we build a robot whose plans are stored in zone A of its executive system. We would happily say, pointing to zone A. ‘Its current plans are stored there’. But we might still find it a little odd, pointing to zone A, to say ‘Over there, the robot is planning to attack you’. On the other hand, we may after a while get used to that way of speaking. Why don’t similar remarks apply to human plans? Just as the initial oddity of ‘Over there, the robot is planning to attack you’ should not count as strong evidence against the thesis that the robot’s plans are stored in zone A, so the analogous oddity for us should not count as strong evidence against an analogous thesis about us. Byrne ends by raising a perception-based case for the thesis that narrow content can do important explanatory work. Here is the key argument: P1. How things perceptually seem determines a content: a perceptual content, the way the perceiver’s environment is presented as being. P2. How things perceptually seem is internally determined. C. Perceptual content is internally determined. Against challenges to P2 that appeal to singular contents of perception, Byrne remarks that one might ‘accept a layer of perceptual content that is existentially quantified (cf. Davies 1991)’ (**). We are not known as philosophers of perception and could not take on such issues in the book. Our main aim was to reframe the issue of internalism as a set of questions about whether any narrow content assignment could play various explanatory roles and satisfy various structural constraints. Inevitably, we had to restrict our focus to a limited number of combinations, and we welcome further exploration of others. But for what it’s worth we would like to register a few reservations about the prospects for narrow content here. First Davies’ proposal is not a way of saving narrow content. Suppose a visually represents Jones as red. True enough, Davies’ idea is to find a layer of content that is existential and which is not about Jones. But the contents he has in mind are not narrow. They are along the following lines: a visually represents there to be a red thing in front of a. While this content is not about Jones it is about a and thus will not be shared by a duplicate. Second, purging visual content of singular components is quite costly when it comes to the role of content in assessing veridicality. Suppose there is a red circle c to a’s right and a blue triangle t to a’s left, and that a, suffering from a trifecta of optical illusions on each side, dramatically misrepresents both: a visually represents c as a blue triangle on the left and t as a red circle on the right. It’s a perceptual disaster. Yet existential contents such as that expressed ‘there is a blue triangle to my right’ come out veridical. (Richer existential contents, such as that expressed by ‘There is a red thing to the right causing my phenomenally-red experience’, would have to be appealed to if existential contents are going to do the veridicality-testing job.) Third, we do not find it plausible that even if one focused on, say, the quality represented by phenomenally red experiences, that aspect of content will be narrow. Suppose Tom sees a white wedding dress illuminated by red light. Tom visually represents a property that the dress lacks. Twin Tom lives in a world where everything is normally bathed in red light. If we were to go to Twin Tom’s world, it is plausible that our visual system represents white wedding dresses to have a property—being red—that they do not have. If we knew how things worked in that 4 world we would not of course believe the dresses are red. We would not take experience at face value. But our visual system would (plausibly) represent the wedding dress as having a kind of surface it does not have. Not so for Twin Tom. It seems pretty obvious that Twin Tom’s visual experiences are veridical. That an experience is an experience as of something having the property of being red is not a narrow matter. References Byrne, A. (2020). ‘Comment on Yli-Vakkuri and Hawthorne, Narrow Content’, Philosophical Studies, this issue Davies, M. (1991). ‘Individualism and perceptual content’, Mind, 100: 461-84. Yli-Vakkuri, Juhani and John Hawthorne (2018). Narrow Content. Oxford: Oxford University Press. Cited as ‘NC’. 5