Sharing Thoughts: Philosophical Perspectives on Intersubjectivity and Communication

Notre Dame Phil Reviews，Notre Dame Philosophical Reviews，圣母大学哲学评论，专注高质量学术哲学书籍评论。关注

This welcome collection addresses the question of how to individuate thoughts and their constituents, concepts, to explain intersubjective phenomena such as communication. The book contains an illuminating introduction by Verdejo and Valente, followed by thirteen original essays. For reasons of space and thematic unity, I will not discuss three excellent essays by Bermúdez, Longworth, and Peacocke.

To introduce the topic of this collection, pre-theoretical characterizations of several intersubjective phenomena appeal to thoughts shared between different subjects. According to a natural view of disagreement, subjects A and B disagree just when A accepts a certain thought that B rejects. And according to a natural view of communication, communication succeeds just when the hearer grasps the thoughtexpressed by the speaker. But what might thoughts be, such that they can be shared?

Of course, external truth-conditions can be shared across subjects. But an individuation of thoughts in terms of their external truth-conditions will not yield the right results. Subject A can accept the thought that George Eliot wrote Middlemarch while subject B rejects the thought that Mary Ann Evans wrote Middlemarch, and although George Eliot is Mary Ann Evans, A and B do not really disagree. Relatedly, when a speaker utters ‘he’, pointing to a man who just appeared on television, and her interlocutor misses the pointing gesture and takes ‘he’ to refer to the man they had previously been talking about, communication can fail even if the man on television turns out to be the man previously talked about (Loar, 1976).

It further proves difficult to individuate thoughts in a fine-grained way (more fine-grained than reference) while vindicating both thought-sharing views of intersubjective phenomena and other individuation constraints. There is a well-documented conflict between the thought-sharing view of communication and Frege’s constraint, according to which thoughts t1 and t2 are distinct if t1 is accepted and t2 rejected by the same subject without irrationality (Schiffer, 1978). Communication as thought-sharing is also in conflict with Campbell’s criterion, which tells us that mental tokens occurring in an inference are tokens of the same concept just when their coreference is assumed (‘traded on’) across premises (Campbell, 1987). Yet another conflict arises between the thought-sharing view of communication and a principle connecting thought and action, according to which two identical sets of propositional attitudes lead to the same actions under the same circumstances (Perry, 1977).

One possible reaction to these conflicts is to uphold the relevant intersubjective constraint and reject another constraint. In this volume, Prosser revises the thought-action principle to uphold communication as thought-sharing. Another option is to reject the relevant intersubjective constraint, as Pollock does with the thought-sharing view of (dis)agreement in her essay. Providing revisionary accounts of the phenomena that initially seem to demand shared thoughts can free us from making thoughts shareable.

Yet another option is to uphold all our constraints but regard them as constraints on different notions of thought. Some measure of unity can be retained by defining one notion in terms of the other, for example, by defining shared thoughts as equivalence classes of private thoughts (Cumming, 2013; Dickie & Rattan, 2010). Unfortunately, this compromise is unstable because the relation that underlies intersubjective phenomena seems to be intransitive, as explained by Goodman in this volume.

Interestingly, however, intransitivity allows one to regain theoretical unity in the following way. The relation that underlies intersubjective phenomena can be identified with the relation whose absence or presence is at stake in Frege’s constraint and Campbell’s criterion, respectively, and this unique relation R can be made intransitive. Since R is not an equivalence relation, it cannot hold in virtue of its relata bearing a further relation to a third entity or in virtue of a shared intrinsic property. This non-reducibility of R characterizes so-called relationist views of thought (Fine, 2007; Gray, 2017).

The non-reducibility to shared properties of thought-tokens does not mean that R cannot be positively characterized. All extant characterizations of R invoke the notion of coreference or conditional coreference, as in the condition that if both tokens refer, then they corefer. Some characterizations of R make R entail (conditional) coreference, as in representationally required coreference (Gray, this volume). Other characterizations do not yield this entailment, as in presupposed coreference (García-Carpintero, this volume).

Here is then a unificationist, relationist view of thought. The absence or presence of the same intransitive and irreducible (in the relevant sense) relation R between thought-tokens explains not only the rational believing of a material contradiction by a single subject and her trading on the coreference of tokens, but also communication and (dis)agreement between different subjects.

With this background in place, we can delve deeper into the contributions to this volume. Prosser and García-Carpintero both develop a unificationist, relationist individuation of thought, with an intransitive relation R that does not entail coreference. While Prosser defines R by the very ability to trade on coreference, García-Carpintero considers this trading as a manifestation of R, which he defines as the presupposed (conditional) coreference of tokens. The case of the Schroeters is slightly different. They also develop a unificationist account that employs an intransitive relation based on Campbell’s criterion. This relation is an immediate and epistemically basic appearance of coreference that does not entail coreference. However, there is an additional ingredient to the Schroeters' account: coreference itself. This additional ingredient yields an equivalence relation of ‘strict concept identity’.

Embracing a unificationist perspective in terms of an intransitive R, Gray focuses on the logic of this relation. Characterizing R as representationally required conditional coreference, Gray devises a formal language similar to predicate logic, except that constants may receive no interpretation, and (conditional) coreference between constant-tokens is encoded by an intransitive syntactic relation between constant-types, rather than by the identity of constant-type. The resulting logic is a three-valued one, and Gray claims that expected patterns of validity given an intransitive R are captured by a non-standard consequence relation called ‘Strawson entailment’.

In their respective essays, Goodman and Recanati both express skepticism about explaining intersubjective phenomena in terms of a relation based on Campbell’s criterion. Goodman compares a view of communication that invokes an intransitive relation R of encoded coreference based on Campbell’s criterion with a view that demands coreference reached through the right kind of process. Since it seems possible to explain communication without invoking encoded coreference, Goodman asks why one would invoke it. She argues that while encoded coreference is needed to stop a vicious regress in the intrapersonal case, there is no such motivation in the interpersonal case. This leaves the encoded coreference view of communication at a dialectical disadvantage. As for Recanati, he argues that Campbell’s criterion cannot be used to define notions of persisting or shared thoughts, because this criterion is about inferences, which Recanati claims are fundamentally synchronic and subjective. Recanati goes on to present his own account, fine-tuning variants of Frege’s constraint for each of the subjective, diachronic, and intersubjective cases.

Disunified accounts of thought in terms of one relation explaining intrapersonal phenomena and another relation explaining interpersonal phenomena are not represented in this volume. However, several contributors reject thought-sharing views of interpersonal phenomena, which is another way of renouncing the unificationist ambition. Although he operates in a specific framework that treats communication as a signaling gameand thought-content as information, Onofri’s main points are that successful communication does not require the sharing of anything more than referential content, and yet successful communication requires more than merely sharing referential content. It follows from this package of views that successful communication is not thought-sharing. According to Onofri, what successful communication requires beyond coreference is that coreference also obtains in a set of non-actual possibilities.

The account of successful communication offered in Dickie’s contribution may be paired with a rejection of the thought-sharing view. For communication with non-empty terms, Dickie’s account demands coreference reached through a process of the right kind. According to Dickie, a process of the right kind is one that makes coreference luck-free in relevant nearby situations, although it need not make coreference luck-free in the actual communicative situation.

As for Pollock, she revises the pre-theoretical account of (dis)agreement, in view of a holistic and internalist individuation of concepts, according to which a subject’s concept is defined by its relations to all her other concepts. This individuation does not allow thoughts to be shared, which makes the revision necessary. Pollock further targets advocates of a ‘sociolectal’externalistindividuation of a subject’s concepts in terms of her relation to her linguistic community. Pollock argues that the sociolectalist too must reject the pre-theoretical view of (dis)agreement, despite the easy shareability of sociolectally individuated thoughts. For example, the sociolectalist is forced to grant the following possibility: A accepts the thought that C’s party is in a fortnight, B rejects this very thought, but A and B intuitively do not disagree. This can happen, for example, if A believes that a fortnight is 14 days, B believes that a fortnight is 10 days, and both believe that C’s party is in 14 days.

Sawyer’s essay is a response to Pollock’s. Sawyer argues that the sociolectalist can still explain (dis)agreement in terms of accepting and rejecting shared thoughts. In our example above, Sawyer claims there is disagreement simpliciter between A and B, since one rejects and the other accepts the thought that C’s party is in a fortnight. But there is a further agreement ‘relevant to coordination in behavior’ between A and B, which is a matter of them both accepting another shared thought, i.e., that C’s party is in 14 days. This further agreement explains the intuitive absence of disagreement between A and B.

I came away from this book with several worries about unificationist accounts of thought using a relation based on Campbell’s criterion. My first worry does not concern their unificationist aspect. It is rather about the capacity of a relation based on Campbell’s criterion to explain the intersubjective phenomena of interest. Assumed coreference (or appearance of coreference, or trading on coreference) can be present when a speaker refers to one thing and a hearer refers to another, but in such cases communication fails. Worse still, the conjunction of assumed coreference with coreference itself does not guarantee communicative success. This is suggested by cases such as the one presented at the start of this review, in which speaker and hearer not only assume coreference but also corefer, and yet communication fails (Loar, 1976).

In view of this shortcoming, a characterization of R as knowledge of (conditional) coreference seems more promising. Since knowledge is more demanding than true belief, knowledge of coreference is more demanding than a true assumption of coreference. Knowledge of coreference might then be the mark of successful communication. However, as shown by Dickie in her essay, not all views of knowledge will support this thesis. Dickie argues that successful communication tolerates a certain kind of luck which knowledge does not tolerate. If this is right, knowledge of coreference is not necessary for successful communication (Tamburini, 2025).

It is further unclear whether a relation based on Campbell’s criterion can fulfill the unificationist ambition. Some characterizations of the relation do not obviously support this ambition. Take García-Carpintero’s presupposed coreference. García-Carpintero regards presuppositions as requirements on the common ground of a conversation. This notion of presupposition does not seamlessly extend to thoughts, whether of a single subject or of different subjects. Turning to Gray’s representationally required coreference, his formal language suggests a natural sense in which coreference may be representationally required:non-semantic relations between symbols are recruited as markers of coreference by the relevant representational system. While this view can make sense of relations between the thoughts of a single subject, it is unclear how it could extend to interpersonal phenomena.

Can we regard several thinkers as forming one representational system whose symbols are subject to the same rules?

The unificationist may reply that the marking of coreference by non-semantic relations between symbols is just one way in which coreference could be representationally required. In the same vein, unificationists such as García-Carpintero and the Schroeters contend in this volume that different cognitive mechanisms implement the same relation R in subjective and intersubjective cases. However, it seems to me that this appeal to implementation mechanisms does not help. To begin with, it is not obvious that the mechanisms supposed to implement R in the intersubjective case can really do so. For Prosser, who is the most explicit on this issue, the assumption by conversational participants that they are using shared words of a shared language allows them to trade on coreference.Prosser’s idea seems primarily designed for communication with words that carry reference-fixing meaning conventions. But there are words whose conventional meaning does not determine reference, such as demonstrative and personal pronouns. To trade on the coreference of pronoun-tokens, speaker and hearer must further converge on a way of identifying the reference of the tokens, as acknowledged by Prosser. Now, if non-reference-fixing conventions are the norm in natural languages, an account of intersubjective trading on coreference in terms of shared linguistic conventions has limited application.

Talk of different implementations of the same relation further raises the distinct possibility that we might be dealing with different relations. Assume that, in the case of a single subject, explanation bottoms out in non-semantic relations between mental symbols. And assume that, in the interpersonal case, explanation bottoms out in a type of interpretive process. An opponent to the unificationist project might not be impressed with an abstract characterization of a unique relation R that is variously implemented; she may insist that radically different ‘implementations’ are just a sign of distinct phenomena.

Although there is much more to say, I must stop here for reasons of space. I strongly recommend this collection. It is essential reading for anyone interested in how the individuation of thoughts relates to intersubjectivity.

REFERENCES

Bermúdez, J.L. (2016). Understanding I: Language and Thought. Oxford University Press.

Campbell, J. (1987). ‘Is sense transparent?’. Proceedings of the Aristotelian Society 88: 273-292.

Cumming, S. (2013). ‘From Coordination to Content’. Philosophers’ Imprint 13: 1–16.

Dickie, I. and Rattan, G. (2010). ‘Sense, Communication and Rational Engagement’. Dialectica 64: 131–51.

Evans, G. (1982). The Varieties of Reference. Oxford University Press.

Fine, K. (2007). Semantic Relationism. Malden, MA: Blackwell.

Gray, A. (2017). ‘Relational Approaches to Frege's puzzle’. Philosophy Compass 12 (10):e12429.

Heck, R.K. (1995). ‘The Sense of Communication’, Mind, 104: 79–106.

Heck, R.K. (2002). ‘Do Demonstratives have Senses?’. Philosophers' Imprint 2:1-33.

Heck, R.K. (2024). ‘Overview’, in Modes of Representations. Oxford University Press.

Loar, B. (1976). 'The Semantics of Singular Terms'. Philosophical Studies 30(6), 353–377.

Ninan, D. (2010). ‘De se Attitudes: Ascription and Communication’. Philosophy Compass 5(7):551-567.

Onofri, A. (2018). ‘The Publicity of Thought’. Philosophical Quarterly 68 (272).

Onofri, A. (2019). ‘Loar’s Puzzle, Similarity, and Knowledge of Reference’. Manuscrito 42 (2):1-45.

Perry, J. (1977). ‘Frege on Demonstratives’. Philosophical Review 86 (4):474-497.

Prosser, S. (2020). ‘The Metaphysics of Mental Files’. Philosophy and Phenomenological Research 100 (3):657-676.

Recanati, F. (2016). Mental Files in Flux. Oxford University Press.

Schiffer, S. (1978). ‘The Basis of Reference’. Erkenntnis 13: 171–206.

Tamburini, V. (2025). ‘Referential Understanding, Luck, and Knowledge of Reference’. Philosophy and Phenomenological Research 111 (2):590-606.

Valente, M. & Onofri, A. (2023). ‘A Puzzle about Communication’. Review of Philosophy and Psychology 14 (3):1035-1054.

Verdejo, V. (2018). ‘Thought Sharing, Communication, and Perspectives about the Self’. Dialectica 72 (4):487-507.

Although some authors in this volume and elsewhere prefer the terminology ‘content’, I will formulate everything in terms of thoughts and concepts.

Collective ascriptions of attitudes, for example ‘A and B believe that so and so’, are not discussed in the book. On the face of it, these ascriptions are true just when A and B entertain the same thought under the same mode (here belief).

It also seems that thoughts of a single subject should be allowed to persist across time. The dialectic then unfolds pretty much as it unfolds in the main text. See Recanati, this volume.

These examples suggest that sharing external truth-conditions is not sufficient. But sharing external truth-conditions might not be necessary either. In this volume, the possibility of ‘empty’ yet successful communication drives part of Dickie’s argument.

See (Heck 2002, Onofri 2018, Valente & Onofri 2023). I am here glossing over the difference between this principle and a modal principle that makes two thoughts distinct if some rational subject couldaccept one and reject the other at the same time (Evans 1982). For the consequences of such differences, see Recanati, this volume.

See Bermúdez 2017, Ninan 2010, Recanati 2016, Verdejo 2018.

If one does this, one can further define a unique yet disjunctive notion of sameness of thought (Cumming 2013).