green numbers give full details.     |    back to list of philosophers     |     unexpand these ideas

Ideas of François Recanati, by Text

[French, b.1952, Institut Jean-Nicod, Paris. Director of the Centre National de la Recherche Scientifique.]

2012 Mental Files
05.1 p.57 Indexicals apply to singular thought, and mental files have essentially indexical features
     Full Idea: I defend the applicability of the indexical model to singular thought, and to mental files qua vehicles of singular thought. Mental files, I will argue, possess the essential features of indexicals.
     From: François Recanati (Mental Files [2012], 05.1)
     A reaction: I love mental files, but am now (thanks to Cappelen and Dever) deeply averse to giving great significance to indexicals. A revised account of files will be needed.
1.1 p.3 Descriptivism says we mentally relate to objects through their properties
     Full Idea: Descriptivism is the view that our mental relation to individual objects goes through properties of those objects. …This is so because our knowledge of objects is mediated by our knowledge of their properties.
     From: François Recanati (Mental Files [2012], 1.1)
     A reaction: The implication is that if you view an object as just a bundle of properties, then you are obliged to hold a descriptive theory of reference. Hence a 'singularist' theory of reference seems to need a primitive notion of an object's identity.
10.1 p.116 Files can be confused, if two files correctly have a single name, or one file has two names
     Full Idea: Paderewski cases are cases in which a subject associates two distinct files with a single name. Inverse Paderewski cases are cases in which there are two names but the subject associates them with a single file.
     From: François Recanati (Mental Files [2012], 10.1)
     A reaction: In the inverse there are two people with the same name, and someone thinks they are one person (with their combined virtues and vices). E.g. Einstein the famous physicist, and Einstein the famous musicologist. What a man!
10.2 p.120 Sense determines reference says same sense/same reference; new reference means new sense
     Full Idea: To say that sense determines reference is to say that the same sense cannot determine distinct referents - any distinction at the level of reference entails a corresponding distinction at the level of sense.
     From: François Recanati (Mental Files [2012], 10.2)
     A reaction: Does 'the sentry at the gate' change its sense when the guard is changed? Yes. 'The sentry at the gate will stop you'. 'The sentry at the gate is my cousin'. De re/de dicto reference. So changes of de re reference seem to change the sense?
11.3 p.140 Encylopedic files have further epistemic links, beyond the basic one
     Full Idea: The reference of a file is the object to which the subject stands in the relevant epistemic relation. In the case of encylopedic entries there is an arbitrary number of distinct relations. The file grows new links in an opportunistic manner.
     From: François Recanati (Mental Files [2012], 11.3)
     A reaction: I'm not convinced by Recanati's claim that encylopedic files are a distinct type. My files seem to grow these opportunistic links right from their inception. All files seem to have that feature. A file could have four links at its moment of launching.
12.2 p.150 There is a continuum from acquaintance to description in knowledge, depending on the link
     Full Idea: It is not too difficult to imagine a continuum of cases between straightforward instances of knowledge by acquaintance and straightforward instances of knowledge by description, with more or less tenuous informational links to the referent.
     From: François Recanati (Mental Files [2012], 12.2)
12.3 p.155 Singular thoughts need a mental file, and an acquaintance relation from file to object
     Full Idea: The mental file framework rests on two principles: that the subject cannot entertain a singular thought about an object without possessing and exercising a mental file about it, and that this requires an acquaintance relation with the object.
     From: François Recanati (Mental Files [2012], 12.3)
     A reaction: I'm puzzled by the case where I design and build a completely new object. I seem to assemble a file, and only bestow singularity on it towards the end. Or the singularity can just be a placeholder, referred to as 'something'. […see p.158]
13.1 p.164 Expected acquaintance can create a thought-vehicle file, but without singular content
     Full Idea: On my view, actual acquaintance is not necessary to open a mental file; expected acquaintance will suffice; yet opening a mental file itself is not sufficient to entertain a singular thought-content. It only enables a thought-vehicle.
     From: François Recanati (Mental Files [2012], 13.1)
     A reaction: I'm not clear why I can't create a file with no expectation at all of acquaintance, as in a fictional case. Depends what 'acquaintance' means. Recanati longs for precise distinctions where they may not be available.
14.1 p.183 An 'indexed' file marks a file which simulates the mental file of some other person
     Full Idea: Files function metarepresentationally if they serve to represent how other subjects think about objects in the world. ..An 'indexed' file has an index referring to the other subject whose files the indexed file stands for or simulates.
     From: François Recanati (Mental Files [2012], 14.1)
     A reaction: Presumably there is an implicit index on all files, which says in a conversation whether my interlocutor does or does not hold the same file-type as me. Recanati wants many 'types' of files, but I suspect there is just one file type.
16.2 p.214 Russellian propositions are better than Fregean thoughts, by being constant through communication
     Full Idea: The Russellian notion of a proposition is arguably a better candidate for the status of semantic content than the Fregean notion of a thought. For the proposition remains constant from one person to the next.
     From: François Recanati (Mental Files [2012], 16.2)
     A reaction: A good point, though I rebel against Russellian propositions because they are too much out in the world, and propositions strike me as features of minds. We need to keep propositions separate from facts.
16.2 p.215 The content of thought is what is required to understand it (which involves hearers)
     Full Idea: As Evans emphasises, what matters when we want to individuate semantic content is what would count as a proper understanding of an utterance; but 'understanding' defines the task of the hearer.
     From: François Recanati (Mental Files [2012], 16.2)
     A reaction: [cites Evans 1982: 92, 143n, 171] I like to place (following Aristotle) understanding at the centre of all of philosophy, so this seems to me an appealing idea. It makes misunderstandings interesting.
17.1 p.221 Fregean modes of presentation can be understood as mental files
     Full Idea: A mental file plays the role which Fregean theory assigns to modes of presentation.
     From: François Recanati (Mental Files [2012], 17.1)
     A reaction: I'm a fan of mental files, and this is a nice pointer to how the useful Fregean insights can be written in a way better grounded in brain operations. Rewriting Frege in neuroscience terms is a nice project for someone.
17.1 p.224 Definite descriptions reveal either a predicate (attributive use) or the file it belongs in (referential)
     Full Idea: A definite description may contribute either the singular predicate it encodes (attributive use) or the mental file to what that predicate belongs (referential use).
     From: François Recanati (Mental Files [2012], 17.1)
     A reaction: This nicely explains Donnellan's distinction in terms of mental files. 'Green' may refer in a shop, but isn't much use in a wood. What to make of 'He's a bit of a Bismark'?
17.3 p.231 Direct reference is strong Millian (just a tag) or weak Kaplanian (allowing descriptions as well)
     Full Idea: There are two notions of direct reference, the strong Millian notion (where the expression is like a 'tag' with no satisfaction mechanism), and the weaker Kaplanian notion (where reference is compatible with carrying a descriptive meaning).
     From: François Recanati (Mental Files [2012], 17.3)
     A reaction: I immediately favour the Millian view, which gives a minimal basis for reference, as just a 'peg' (Marcus) to hang things on. I don't take a Millian reference to be the object itself. The concept of a 'tag' or 'label' is key. Mental files have tags.
17.3 p.233 Reference by mental files is Millian, in emphasising acquaintance, rather than satisfaction
     Full Idea: The mental file account preserves the original, Millian inspiration of direct reference theories in giving pride of place to acquaintance relations and downplaying satisfaction factors.
     From: François Recanati (Mental Files [2012], 17.3)
     A reaction: I find this a very satisfying picture, in which reference links to the simple label of a file (which could be a number), and not to its contents. There are tricky cases of non-existents, fictional entities and purely possible entities to consider.
18.1 p.243 We need sense as well as reference, but in a non-descriptive form, and mental files do that
     Full Idea: My view inherits from Frege 'modes of presentation'. Reference is not enough, and sense is needed. …We must make room for non-descriptive modes of presentation, and these are mental files.
     From: François Recanati (Mental Files [2012], 18.1)
     A reaction: [compressed] Recanati aims to avoid the standard Kripkean criticisms of descriptivism, while being able to handle Frege's puzzles. I take Recanati's mental files theory to be the most promising approach.
18.1 p.246 If two people think 'I am tired', they think the same thing, and they think different things
     Full Idea: If you and I think 'I am tired', there is a sense in which we think the same thing, and another sense in which we think different things.
     From: François Recanati (Mental Files [2012], 18.1)
     A reaction: This is a very nice simple account of the semantic distinctiveness of indexicals, which obviously requires a 'two-tiered framework'. He cites Kaplan and Perry as background.
18.2 p.252 In super-direct reference, the referent serves as its own vehicle of reference
     Full Idea: In super-direct reference, the sort of thing Russell was after, there is no mode of presentation: the referent itself serves as its own vehicle, as it were.
     From: François Recanati (Mental Files [2012], 18.2)
     A reaction: To me this is a step too far, because reference is not some physical object like a chair; it is a mental or linguistic phenomenon. Chair's don't refer themselves; it is people who refer.
2.1 p.17 Two-D semantics is said to help descriptivism of reference deal with singular objects
     Full Idea: Descriptivism has trouble catching the singularity of objects, construing them as only directly about properties. …To get the truth-conditions right, it is claimed, the descriptivist only as to go two-dimensional.
     From: François Recanati (Mental Files [2012], 2.1)
     A reaction: I suspect that the descriptivist only has a problem here because context is being ignored. 'That man on the beach' can quickly be made uniquely singular after a brief chat.
2.1 p.17 In 2-D semantics, reference is determined, then singularity by the truth of a predication
     Full Idea: In the two-dimensional framework, what characterises the singular case is the fact that truth-evaluation (of possessing of the reference-fixing property) takes place at a later stage than reference determination.
     From: François Recanati (Mental Files [2012], 2.1)
     A reaction: This sounds psychologically plausible, which is a big (and unfashionable) plus for me. 1) what are we talking about? 2) what are we saying about it, 3) is it true?
2.2 p.18 A rigid definite description can be attributive, not referential: 'the actual F, whoever he is….'
     Full Idea: A rigid use of a definite description need not be referential: it may be attributive. Thus I may say: 'The actual F, whoever he is, is G'.
     From: François Recanati (Mental Files [2012], 2.2)
     A reaction: Recanati offers this as a criticism of the attempted 2-D solution to descriptivist accounts of singularity. The singularity is not strong enough, he says.
2.2 p.21 Singularity cannot be described, and it needs actual world relations
     Full Idea: As Peirce insisted, singularity as such cannot be described, it can only be given through actual world relations.
     From: François Recanati (Mental Files [2012], 2.2)
     A reaction: [Peirce - Exact Logic, Papers 3, 1967, §419] This is the key idea for Recanati's case for basing our grasp of singular things on their relation to a mental file.
2.2 p.21 Indexicality is closely related to singularity, exploiting our direct relations with things
     Full Idea: Singularity and indexicality are closely related: for indexicals systematically exploit the contextual relations in which we stand to what we talk about.
     From: François Recanati (Mental Files [2012], 2.2)
     A reaction: Recanati builds a nice case that we may only have an ontology of singular objects because we conceptualise and refer to things in a particular way. He denies the ontology, but that's the bit that interests me.
3.1 p.29 Problems with descriptivism are reference by perception, by communications and by indexicals
     Full Idea: Three problems with Frege's idea of descriptions in the head are: reference through perception, reference through communicative chains, and reference through indexicals.
     From: François Recanati (Mental Files [2012], 3.1)
     A reaction: In the end reference has to occur in the head, even if it is social or causal or whatever, so these are not problems that worry me.
3.2 p.29 There may be two types of reference in language and thought: descriptive and direct
     Full Idea: A widely held view, originating with Russell, says there are two types of reference (both in language and thought): descriptive reference, and direct reference.
     From: François Recanati (Mental Files [2012], 3.2)
     A reaction: I would rather say is there is just one sort of reference, and as many ways of achieving it as you care to come up with. With that view, most of the problems vanish, as far as I can see. People refer. Sentences are nothing but trouble.
3.3 p.35 Mental files are the counterparts of singular terms
     Full Idea: Mental files are the mental counterparts of singular terms.
     From: François Recanati (Mental Files [2012], 3.3)
     A reaction: A thoroughly satisfactory theory. We can build up a picture of filing merging, duplication, ambiguity, error etc. Eventually neuroscience will map the whole system, and we will have cracked it.
3.3 p.35 The reference of a file is fixed by what it relates to, not the information it contains
     Full Idea: What files refer to is not determined by properties which the subject takes the referent to have (information, or misinformation, in the file), but through the relations on which the files are based.
     From: François Recanati (Mental Files [2012], 3.3)
     A reaction: Maybe. 'Lot 22'. I can build up a hypothetical file by saying 'Imagine an animal which is F, G, H…', and build a reference that relates to nothing. Maybe Recanati overestimates the role of his 'epistemically rewarding' relations in file creation.
3.4 p.41 Sense is a mental file (not its contents); similar files for Cicero and Tully are two senses
     Full Idea: What plays the role of sense is not information in a file, but the file itself. If there are two distinct files, one for 'Cicero' and one for 'Tully', then there are two distinct (non-descriptive) senses, even if the information in both files is the same.
     From: François Recanati (Mental Files [2012], 3.4)
     A reaction: This may be the best idea in Recanati's book. A sense might be a 'way of coming at the information', rather than some set of descriptions.
4.1 p.42 Identity statements are informative if they link separate mental files
     Full Idea: An identity statement 'A=B' is informative to the extent that the terms 'A' and 'B' are associated with distinct mental files.
     From: François Recanati (Mental Files [2012], 4.1)
     A reaction: Hence the information in 'Scott is the author of 'Waverley'' is information about what is in your mind, not what is happening in Scotland. This is Recanati's solution to one of Frege's classic puzzles. 'Morning Star' and 'Evening Star' files. Nice.
4.1 p.42 A mental file treats all of its contents as concerning one object
     Full Idea: The role of a mental file is precisely to treat all the information as if it concerned one and the same object, from which it derives.
     From: François Recanati (Mental Files [2012], 4.1)
     A reaction: Recanati's book focuses entirely on singular objects, but we presumably have files for properties, generalisation, groups etc. Can they only be thought about if they are reified? Maybe.
5.1 p.57 Indexicals (like mental files) determine their reference relationally, not by satisfaction
     Full Idea: The class of indexicals have the same property as mental files, that their reference is determined relationally rather than satisfactionally.
     From: François Recanati (Mental Files [2012], 5.1)
     A reaction: Recanati is building an account of reference through mental files. This idea may be the clearest point I have yet encountered about indexicals, showing why they are of particular interest to philosophers.
5.1 p.57 Indexical don't refer; only their tokens do
     Full Idea: Indexicals do not refer; only tokens of an indexical refer
     From: François Recanati (Mental Files [2012], 5.1)
     A reaction: Thus 'Thurs 23rd March 2013' refers, but 'now' doesn't, unless someone produces an utterance of it. This is why indexicals are sometimes called 'token-reflexives'.
5.3 p.64 Mental files are individual concepts (thought constituents)
     Full Idea: I want mental files (properly speaking) to serve as individual concepts, i.e. thought constituents.
     From: François Recanati (Mental Files [2012], 5.3)
     A reaction: This is why the concept of mental files is so neat - it gives you a theory of reference and a theory of concepts. I love the files approach because it precisely fits my own introspective experiences. Hope I'm not odd in that way.
6.1-3 p.68 There are transient 'demonstrative' files, habitual 'recognitional' files, cumulative 'encyclopedic' files
     Full Idea: A 'demonstrative' file only exists during the demonstrative relation to something; …a 'recognitional' file is based on 'familiarity' (a disposition to recognise); …an 'encylopedic' file contains all the information on something, however it is gained.
     From: François Recanati (Mental Files [2012], 6.1-3)
     A reaction: [picked as samples of his taxonomy, pp.70-73] I'm OK with this as long as he doesn't think the categories are sharply separated. I'm inclined to think of files as a single type, drifting in and out of different of modes.
6.3 p.75 Files are hierarchical: proto-files, then first-order, then higher-order encyclopedic
     Full Idea: There is a hierarchy of files. Proto-files are the most basic; conceptual files are generated from them. First-order ones are more basic, as the higher-order encylopedic entries presuppose them.
     From: François Recanati (Mental Files [2012], 6.3)
     A reaction: This hierarchy might fit into a decent account of categories, if a plausible one could be found. A good prospect for exploring categories would be to start with mental file-types, and work outwards through their relations.
8.3 p.100 A file has a 'nucleus' through its relation to the object, and a 'periphery' of links to other files
     Full Idea: I take a file to have a dual structure, with a 'nucleus' of the file consisting of information derived through the relevant epistemically rewarding relation, while the 'periphery' consists of information derived through linking with other files.
     From: François Recanati (Mental Files [2012], 8.3)
     A reaction: This sounds strikingly like essentialism to me, though what constitutes the essence is different from the usual explanatory basics. The link, though, is in the causal connection. If we naturally 'essentialise', that will control file-formation.
2016 Mental Files in Flux
Pref p.-12 Mental files are concepts, which are either collections or (better) containers
     Full Idea: Mental files are entries in the mental encyclopedia, that is, concepts. Some, following Grice, say they are information collections, but I think of them as containers. Collections are determined by their elements, but containers have independent identity.
     From: François Recanati (Mental Files in Flux [2016], Pref)
     A reaction: [compressed] [Grice reference is 'Vacuous Names' (1969)] I agree with Recanati. The point is that you can invoke a file by a label, even when you don't know what the content is.
Pref p.-10 The Frege case of believing a thing is both F and not-F is explained by separate mental files
     Full Idea: Frege's Constraint says if a subject believes an object is both F and not-F (as in 'Frege cases'), then the subject thinks of that object under distinct modes of presentation. Having distinct mental files of the object is sufficient to generate this.
     From: François Recanati (Mental Files in Flux [2016], Pref)
     A reaction: [compressed] When you look at how many semantic puzzles (notably from Frege and Kripke) are solved by the existence of labelled mental files, the case for them is overwhelming.
5 p.71 A linguistic expression refers to what its associated mental file refers to
     Full Idea: Mental files determine the reference of linguistic expressions: an expression refers to what the mental file associated with it refers to (at the time of tokening).
     From: François Recanati (Mental Files in Flux [2016], 5)
     A reaction: Invites the question of how mental files manage to refer, prior to the arrival of a linguistic expression. A mental file is usually fully of descriptions, but it might be no more than a label.
5.2 p.77 A train of reasoning must be treated as all happening simultaneously
     Full Idea: For logic purposes, a train of reasoning has to be construed as synchronic.
     From: François Recanati (Mental Files in Flux [2016], 5.2)
     A reaction: If we are looking for a gulf between logic and the real world this is a factor to be considered, along with Nietzsche's observation about necessary simplification. [ref to Kaplan 'Afterthoughts' 1989, 584-5]
6.1 p.97 Indexicality is not just a feature of language; examples show it also occurs in thought
     Full Idea: People once took indexicality to be exclusively a property of language, ....but a series of examples seemed to establish that the thought expressed by uttering an indexical sentence is itself indexical (and is thus 'essential').
     From: François Recanati (Mental Files in Flux [2016], 6.1)
     A reaction: Perry's example of not realising it is him leaking the sugar in a supermarket is the best known example. Was this a key moment for realising that philosophy of thought is (pace Dummett) more important than philosophy of language?
7.1 p.111 How can we communicate indexical thoughts to people not in the right context?
     Full Idea: Indexical thoughts create an obvious problem with regard to communication. How can we manage to communicate such thoughts to those who are not in the right context?
     From: François Recanati (Mental Files in Flux [2016], 7.1)
     A reaction: One answer is that you often cannot communicate them. If I write on a wall 'I am here now', that doesn't tell the next passer-by very much. But 'it's raining here' said in a telephone call works fine - if you know the location of the caller.
7.1 p.111 The Naive view of communication is that hearers acquire exactly the thoughts of the speaker
     Full Idea: The Naive Conception of Communication rests on the idea that communication is the replication of thoughts: the thought the hearer entertains when he understands what the speaker is saying is the very thought which the speaker expressed.
     From: François Recanati (Mental Files in Flux [2016], 7.1)
     A reaction: It is hard to believe that any modern thinker would believe such a view, given holistic views of language etc.
7.2 p.119 There are speakers' thoughts and hearers' thoughts, but no further thought attached to the utterance
     Full Idea: There is the speaker's thought and the thought formed by the hearer. That is all there is. We don't need an additional entity, the thought expressed by the utterance.
     From: François Recanati (Mental Files in Flux [2016], 7.2)
     A reaction: This fits my view of propositions nicely. They are the two 'thoughts'. The notion of some further abstract 'proposition' with its own mode of independent existence strikes me as ontologically absurd.