Naturalizing epistemic norms



Institut Jean-Nicod, ENS, 29, rue d'Ulm 75005 Paris. Salle de Séminaire du Pavillon Jardin
Responsables : Joëlle Proust et Paul Egré

Epistemic norms (from now on: "ENs") refer to the dimensions on which mental contents can be evaluated with respect to their contribution to knowledge. Any learner needs to predict how exhaustively or accurately she can learn some material,  assess whether she understands what she reads, and  determine whether she should accept a proposition given a context of epistemic or instrumental deliberation. Little is known, however, about how ordinary people actually recognize ENs and use them in their epistemic decisions. There is no agreement as yet about whether all humans are sensitive to the same ENs, nor, even, about what they are. Epistemologists have mainly focused on truth, coherence, evidentiality and rationality, while anthropologists and psychologists have emphasized the import of additional norms such as relevance, consensuality and fluency. The seminar will be open to philosophers interested in addressing these questions, on the basis of all the methods available, among which formal and non formal epistemology, semantics, decision theory, and experimental psychology.



November, 26, 2-4 pm
Paul Egré (CNRS, IJN),

"Knowledge, Justification and Reason-Based Belief" (joint work with P. Marty and B. Renne)

Abstract: Can knowledge be defined as justified true belief? I will present elements from a recent joint paper with Bryan Renne and Paul Marty in which we argue that the answer can legitimately be positive or negative depending on how the concept of justification is articulated. Gettier's counterexamples to the JTB analysis typically challenge internal justifications. Our main argument is that we can give an alternative version of the JTB analysis in which the concept of justification is understood is more externalist terms. To achieve this goal, we argue for the need to represent reasons in epistemic logic, and for the need to distinguish between adequate reasons and veridical reasons. The distinction also allows us to make progress on the typology of Gettier cases.

December, 10, 2-4 pm
Joelle Proust (IJN),

"Time and action: impulsivity, habit, strategy".

Granting that various mental events might form the antecedents of an action, what is the mental event that is the proximate cause of action? The present article reconsiders the methodology for addressing this classical question: Intention and its varieties cannot be properly analyzed if one ignores the evolutionary constraints that have shaped action itself, such as the trade-off between efficient timing and resources available, for a given stake. On the present proposal, three types of action, impulsive, routine and strategic, are designed to satisfy the trade-off above when achieving goals of each type. This analysis applies equally to actions aiming at world or at cognitive outcomes. While actions of the first two types depend on non-conceptual appraisals of a given intensity and valence, strategic intentions have a propositional format and guide action within longer-term executive frameworks involving prospective memory. Interestingly, different epistemic norms are underlying the regulation of mental actions of each variety.

January, 7, 11:30- 13:00. Séance conjointe avec le séminaire Linguae
Paul Egré (CNRS, IJN),

"Vague Judgment: a Probabilistic Account"

This paper investigates the idea that vague predicates like "tall'', "loud'' or "expensive'' are applied based on a process of analog magnitude representation (see Fults 2009, van Rooij 2012, Solt 2012), whereby magnitudes are represented with noise. I present a probabilistic account of vague judgment, inspired from early remarks by Emile Borel (1907) on vagueness, and use it to model judgments about borderline cases. The model involves two main components:  probabilistic magnitude representation on the one hand, and a notion of subject-relative criterion. The framework is used to compare judgments of the form "x is clearly tall'' vs. "x is tall'', using the idea of a shift of a criterion shift. The model can be viewed as giving a naturalistic counterpart to the strict-tolerant semantics of vagueness (Cobreros et al. 2012). I then extend it to fit data concerning borderline contradictions of the form "x is tall and not tall'' (Egré, Gardelle and Ripley 2013).

February, 4, 2-4 pm
Benjamin Spector (CNRS, IJN),

"Decomposing knowledge",

It seems uncontroversial that knowledge can not be in general identified to some 'internal' state of the agent to whom knowledge is attributed. This is so because knowledge is factive. To know whether X knows p, we have to know whether p is true, and the truth-value of p is most often completely independent of X's internal epistemic state. In this sense, "externalism" for knowledge appears to be the default assumption. However, a natural idea to consider is that `X knows that p' might be equivalent to  `p and K', where K is a proposition whose truth-value only depends on X's epistemic state. For instance, K could be `X believes that p and has internal justifications for this belief'. Gettier's problem suggests that this specific proposal is not tenable, and Williamson argues that no such K can be found.

In this talk, I will argue that even though we might well be incapable of providing a perfect paraphrase of `X knows that p' as described above, the best theory of natural language must nevertheless assume that the mental lexical entry for the verb 'know', and for factive verbs in general, can be decomposed into a presuppositional part and an assertive part, with a specific consequence in the case of `know'. In the case of `know', the presuppositional part encodes the fact that `know' is factive, and the assertive part would be a non-factive and purely internal attitude. The fact that we cannot paraphrase this non-factive attitude is in itself no more surprising than the fact that many other words cannot be perfectly paraphrased, and is any case irrelevant to the question whether there is an `internal' attitude corresponding to knowledge (i.e. knowledge minus factivity). The argument will be based on a section of a recent paper that I co-authored with Paul Egré on the interpretation of interrogative clauses when they are arguments of a presuppositional (typically factive) attitude predicate (as in  `John knows who came'). We need to assume that the rule whereby such constructions are interpreted must make reference to the `non-presuppositional' part of the relevant attitude verbs. If our account is correct, it follows that the mental lexical entry for `know' includes two components, one of which can be thought of as referring only to the agent's internal mental state.


March, 4, 2-4 pm
Emmanuel Chemla, Alexandre Cremers
& Lyn Tieu (LSCP)

"Children's exhaustive readings of questions"

Embedded questions have been argued to give rise to multiple readings, which are related in terms of strength. Cremers and Chemla (2014) provide experimental evidence that questions embedded under 'know' are ambiguous between 'weakly exhaustive' (WE), 'intermediate exhaustive' (IE), and 'strongly exhaustive' (SE) readings. The SE reading entails both the IE and WE readings, and the IE reading entails the WE reading. Certain proposals in the semantics literature derive the stronger readings from weaker ones through the same process of pragmatic enrichment that underlies scalar implicatures, i.e. exhaustification (Klinedinst & Rothschild 2011). Given previous developmental studies of scalar implicatures that suggest children typically perform this pragmatic enrichment less often than adults do (Noveck 2001, Chierchia et al. 2001, Papafragou 2003, among many others), such proposals might lead us to expect that children may initially prefer weak readings. The present study investigated French-speaking children's comprehension of such embedded questions, and found that 5-year-olds were sensitive to the multiple readings of questions embedded under 'savoir' ('know'). Compared to adults, however, children were more tolerant of weaker readings. We discuss connections between our results and existing literature on children's performance on scalar implicatures: in both cases children appear to be aware of the ambiguity between weaker and stronger forms, but are more tolerant of weak meanings than adults.

April, 1
, 2-4 pm
Adrian Cussins (National University of Colombia),

"Hot, Wild and Thoughtful"

How can perception deliver content which is sufficient for empirical knowledge of an objective environment?  There are three reasons why this appears impossible.  One is the problem of how perceptual content could be both, at once, receptive and spontaneous, as McDowell discusses towards the beginning of "Mind and World".  Another is that we lack an account of how content could be either receptive or spontaneous.  We do not understand how perceptual experience can justify observational belief in a distinctive way, different from how one belief may justify another.  Standard accounts do not allow perceptual justification to be punctate rather than holistic, nor to provide that cognitive resistance of the world which is necessary for the receptivity of perception.  Nor do we understand the possibility of cognitive content which is capable of sustaining a distinction between fiction and reality.  A third reason shows how it follows from the nature of animal being that animal perception is not objective.  Perception in animals is in the service of the guidance of niche-adapted activity.  Since human perception is a form of animal perception, human perception should not be able to deliver world knowledge.

I show how empirical knowledge entails nonconceptual content, meaning construction from the structure within atomic concepts, and punctate justification from hot and wild encounters with the environment.  We must re-think the nature of perceptual content as mediational content which is wild (ie it is possible to have a perceptual experience whose content is genuinely a representational content, but is such that, initially, it makes no sense at all to the subject); which is motivationally hot (not neutral in relation to judgment or activity); which does not respect the attitude-content distinction characteristic of the theory of propositional attitudes; which is affective and involves affordances, solicitations and a characteristic subjective valence; and which has a normative structure appropriate to activity guidance.  We should rethink semantics as dynamic meaning constructions through which mediational and referential contents are transformed into truth-evaluable contents.  Perceptual justification occurs internally to meaning construction, rather than as an inferential process subsequent to the availability of complex conceptual contents.  Meaning construction in thought is governed by the epistemic virtue of thoughtfulness, and only indirectly by the norm of truth. 

The problem of empirical knowledge is not resolved by epistemology, but by a theory of meaning which allows us to reconceive cognition in terms of an interplay between the mundane normativities (of activity guidance and thoughtfulness), and the elite normativity of truth.

This talk will focus on perceptual justification sufficient for the co-application of spontaneity and receptivity, and its dependence on content which is hot, wild and thoughtful.


 June, 10, 2-4 pm

Christoph Michel (University of Stuttgart),
"Context Sensitive Rationality and the Construction of Attitudes"

Abstract: Two concepts have been at the heart of our behavioral sciences and their understanding of our mind in reasoning, deciding and acting: rationality and attitudes. These two fundamental concepts of analysis have always been seen as being closely linked. However, developments of our understanding of human rationality and reasoning during the last decades have yet failed to make a deep impact upon how to model our building bricks for mental explanations: attitudes such as beliefs and preferences. In this talk I shall consider implications of pluralism in the theory of rationality for attitude theory and for the idea that attitudes are subject to rationality constraints. If, as I am going to argue, context-sensitivity holds as a meta-norm for rationality, this entails a demand for context-sensitivity also on the level of attitudes. This demand creates tensions with the naïve internalist realism in the representationalist, functionalist and dispositionalist orthodoxies that have been dominating in the philosophy of mind. It appears that context-sensitivity creates a pressure to take constructionism more seriously also in the metaphysics of propositional attitudes.



Interdisciplinary seminar (logic, mathematics, philosophy, psychology) co-organized by Joëlle Proust and Paul Egré.

On the seminar

Epistemic norms (from now on: "ENs") refer to the dimensions on which mental contents can be evaluated with respect to their contribution to knowledge. Any learner needs to predict how exhaustively or accurately she can learn some material,  assess whether she understands what she reads, and  determine whether she should accept a proposition given a context of epistemic or instrumental deliberation. Little is known, however, about how ordinary people actually recognize ENs and use them in their epistemic decisions. There is no agreement as yet about whether all humans are sensitive to the same ENs, nor, even, about what they are. Epistemologists have mainly focused on truth, coherence, evidentiality and rationality, while anthropologists and psychologists have emphasized the import of additional norms such as relevance, consensuality and fluency. The seminar will be open to philosophers interested in addressing these questions, on the basis of all the methods available, among which formal and non formal epistemology, semantics, decision theory, and experimental psychology.



Next meeting:


Thursday March 13, 2014

13h30 to 15h30


Diana Raffman (University of Toronto)

"Vagueness, Normativity, and Belief"




Some theorists of vagueness think that the predicate 'belief' is vague (e.g., Hilpinen 1980, Hajek 2014).  I think that 'belief' probably is vague, though for reasons different from the ones usually cited.  In particular, on the conception of vagueness that I favor, 'belief' is vague primarily because its application is arbitrary over a certain range of cases, not because it is "sorites-susceptible" or admits of borderline cases.  I discuss some implications of the latter view for the nature of belief and the practice of belief ascription.


Preceding meetings



Wednesday December 4, 2013, from 2:30 to 4:30 pm,


Edward Garrett (SOAS, University of London),  

"A Kantian Perspective on Predicates of Personal Taste"


The literature on philosophical aesthetics arising from and reflecting on Kant’s Critique of the Power of Judgment has emphasized certain themes in the investigation of taste that have played a mostly minor role in semantic work. This paper is structured around three broad contributions that a roughly Kantian approach to taste can make to linguistics, concerning (1) the nature of aesthetic judgment and its linguistic encoding, (2) the universality of judgments of taste, and (3) the ‘scope’ of taste phenomena.

There are, in principle, a number of ways to explore these three contributions. In this paper, I draw substantially for linguistic support on data from Standard Tibetan. This decision is motivated by a grammatical feature that makes Tibetan an especially useful language to examine in this context: Tibetan routinely marks expressions of personal taste through the use of evidentials, simplifying the diagnosis of PPTs in this language.

An examination of the behavior of PPTs in Tibetan sheds light on each of these three contributions. First, certain otherwise unaccounted for evidential distributions, relating to the availability of specific evidential shifts and inference patterns, begin to make sense when sentences with Tibetan PPTs are analysed as expressing aesthetic judgments in the Kantian sense. Second, the ubiquity of the direct evidential tag construction, used with PPTs to converge upon a common characterization of shared space, evidences the normative or coercive element of judgments of taste in Tibetan. Third and finally, small-scale corpus research reveals that a range of adjectival predicates can be categorised as PPTs, extending well beyond the usual suspects.

Most of the Tibetan data in this paper comes from audio or video recordings made by the Tibetan & Himalayan Library (THL) in the early 2000s. THL’s entire audio-visual corpus is available online. Many of its recordings have been transcribed into Tibetan, and some also have been translated into English or Chinese. Many of these recordings have been timecoded and can be viewed online using an interactive media player.


This paper therefore aims to establish points of connection between three distinct literatures: first, work on the semantics of PPTs; second, work in philosophical aesthetics arising from and reflecting on Kant’s Critique of the Power of Judgment; and third, work on the semantics and pragmatics of Tibetan evidentials. These elements will be blended together in an attempt to advance our understanding of PPTs.


Thursday October 3, 2013, Emmanuelle Danblon,  Rhetorics, Université Libre de Bruxelles:

Rhétorique naturelle, rhétorique technique: Le cas de la palabre 


Selon Jean-Godefroy Bidima (1997), la palabre est un modèle de discussion qui permet la rencontre active avec l’autre, sans éviter le face à face, alors que l’Occident contemporain propose un modèle de tolérance passive symbolisée par la formule : « Énoncez votre vérité, j’énonce la mienne et qu’on ne se gêne pas » (Bidima 1997 : 40). Je voudrais faire le point sur certains traits de la rhétorique dans ses productions actuelles vues depuis l’angle africain. La rhétorique occidentale est d’abord obsédée par le pacifisme. À ce titre, elle propose des modèles rhétoriques essentiellement centrés sur la résolution du conflit par un consensus généralisé. Elle est ensuite obsédée par une certaine conception de la tolérance, qui se traduit, selon les termes de Bidima, par une accumulation de « soliloques parallèles où l’on garde intacts les préjugés de chacun » (ibid. p. 40). Consensualisme obsessionnel ou relativisme culturel comme comble de la tolérance, la rhétorique occidentale aurait-elle quelque chose à apprendre de la palabre traditionnelle ?


Thursday June 27,  Isabelle Drouet (University of Paris 4):

Causal reasoning, causal probabilities, and conceptions of causation.



The present paper deals with the tools that can be used to represent causation and to reason about it and, specifically, with their diversity. It focuses on so-called “causal probabilities”—that is, probabilities of effects given one of their causes—and critically surveys a recent paper in which Joyce (2010) argues that the values of these probabilities do not depend on one’s conception of causation. I first establish a stronger independence claim: I show that the very definition of causal probabilities is independent of one’s conception of causation. Second, I investigate whether causal probabilities indeed take the same values under their different possible definitions.


Wednesday May 29

Kevin T. Kelly and Hanti Lin (Carnegie Mellon University):

Propositional Reasoning that Tracks Probabilistic Reasoning 


According to the familiar Bayesian story, one has degrees of belief and one revises them in light of new propositional information by conditioning. According to qualitative theories of belief revision, one has propositional beliefs that one revises in light of new propositional information.  Of course, propositional reasoning will always be coarser than full Bayesian reasoning, but one would hope at least that the former could somehow correspond to or aptly represent the latter. Traditionally, that correspondence has been understood in terms of acceptance: a rule that associates propositional belief states with probabilistic belief states.  For example, the Lockean acceptance rule accepts all propositions whose probabilities exceed a given threshold.  That rule occasions the familiar lottery paradox (Kyburg 1961).  We will show that it also falls prey to new and more stubborn paradoxes when conditional probabilities are taken into account.  We propose an alternative approach to acceptance that avoids all of the paradoxes and that, furthermore, is guaranteed to track Bayesian conditioning, in the sense that acceptance followed by propositional belief revision always yields exactly the same result as Bayesian conditioning followed by acceptance. The proposal combines an odds-based acceptance rule proposed originally by Levi (1996) with a non-AGM belief revision method proposed originally by Shoham (1987).  We also show that the familiar AGM approach to belief revision (Harper 1975 and Alchourrón, Gärdenfors, and Makinson 1985) cannot be realized in a sensible way by any uncertain acceptance rule that tracks Bayesian conditioning in the sense just described.  


Wednesday May 15, 2013


Seth Yalcin (UCB):

On Expressivism


I develop a conception of expressivism according to which it is chiefly a pragmatic thesis about some fragment of discourse, one imposing certain constraints on semantics. The first half of the paper uses credal expressivism about the language of probability as a stalking-horse for this purpose. The second half turns to the question of how one might frame an analogous form of expressivism about the language of deontic modality. Here I offer a preliminary comparison of two expressivist lines. The first, expectation expressivism, looks again to Bayesian modelling for inspiration: it glosses deontically modal language as characteristically serving to express decision-theoretic expectation (expected utility). The second, plan expressivism, develops the idea (due to Gibbard 2003) that this language serves to express ‘plan-laden’ states of belief. In the process of comparing the views, I show how to incorporate Gibbard’s modelling ideas into a compositional semantics for attitudes and modals, filling a lacuna in the account. I close with the question whether and how plan expressivism might be developed with expectation-like structure.


Wednesday March 20, from 2:30 to 4:30 pm,

Jean Baratgin (Université Paris 8, Saint-Denis & Institut Jean Nicod (ENS, Paris)

Uncertainty and de Finetti’s three-valued logic


The new paradigm in the psychology of reasoning adopts a Bayesian, or probabilistic, model to study human reasoning. This approach is supported by two findings:

- most people judge the probability of the indicative conditional, P(if A then C), to be the conditional probability, P(C|A), as implied by the Ramsey test.

- the existence of a so-called defective truth table in which people judge that if A then C is (i) true when A holds and C holds, (ii) false when A holds and C does not hold, and (iii) neither true nor false when A does not hold.

Our presentation focuses on this second point. Contrary to the traditional binary approach based on truth functional logic, with truth-values of truth and falsity, a third value that represents uncertainty is introduced in the new paradigm. Various three-valued truth table systems are available in the formal literature, including one proposed by de Finetti. We examine their descriptive adequacy, considering the usual connectives, and in particular the conditional. Within this framework the so-called defective truth table becomes an explainable and coherent response. Our main result is that the three-valued de Finetti’s logic (and only this one) shows a very good descriptive adequacy when uncertainty is introduced as a third truth-value.


February 20, 2013 

Paul Egré (Institut Jean-Nicod) will present his joint work with Vincent de Gardelle (LPP) andDavid Ripley (Melbourne, UConn.) :

"Vagueness and order effects: evidence for enhanced contrast in a task
of color categorization".

Abstract :

This paper proposes an experimental investigation of the use of vague predicates in dynamic sorites. We present the results of two studies in which subjects had to categorize colored squares at the borderline between two color categories (Green vs. Blue, Yellow vs. Orange). Our main aim was to probe for hysteresis in the ordered transitions between the respective colors, namely for the longer persistence of the initial category. Our main finding is a robust phenomenon of negative hysteresis or enhanced contrast, present in two diff erent tasks, a comparative task involving two color names, and a yes/no task involving a single color name. We propose an explanation of this e ffect in terms of the strict-tolerant framework of P. Cobreros, P.Egré, D. Ripley, and R. van Rooij (2012), "Tolerant, classical, strict", The Journal of Philosophical Logic, pp. 1-39, in which borderline cases are characterized in a dual manner in terms of overlap between tolerant extensions, and underlap between strict extensions.The framework, more generally, relies on two notions of assertion, whose relevance will be questioned from a broader epistemological perspective.


January 23, 2013

Helen de Cruz  (Radcliffe Humanities Oxford and Leuven University)

 "The epistemic significance of common consent"


Philosophers have long focused on individual reasoning and experience
as the main and sometimes even exclusive proper grounds for belief. In
this picture of human reasoning, relying on common consent as a
guideline to forming true beliefs seems like an awful strategy. Even
in experimental psychology, the dominant paradigm (Asch-type
experiments) pits consensus against truth. However, recent work in
social epistemology prompts us to reassess common consent. According
to social epistemology, the opinions of others constitute a source of
knowledge, for instance, taking into account the opinion of a
dissenting epistemic peer can be a valuable way of "coping with our
own infirmities" (Christensen). This paper aims to assess the
epistemic significance of common consent by looking at in the context
of recent social epistemology, reviewing recent arguments from common
consent by Zagzebski and Kelly. These arguments outline conditions
under which deferring to majority opinion can be rational. I then
survey how empirical findings in social psychology can be interpreted
in the light of these normative frameworks. Finally, I examine
theoretical models of cultural transmission in the light of the
evidential value of common consent.


Wednesday October 31, 2012


Igor Douven (University of Groningen):

Indicative conditionals revisited



It has long been held that high conditional degree of belief in an indicative conditional's consequent given its antecedent is necessary and sufficient for the acceptability of that conditional. Recently, it has been argued, both on theoretical and on empirical grounds, that the notion of evidential support plays a crucial role in the acceptability of indicative conditionals, next to that of high conditional degree of belief. It is still an open question under which principles acceptability is closed if we adopt this proposal. The present paper makes a beginning with answering that question.

June 13, 2012, Denis Bonnay (Paris Ouest Nanterre) and Mikael Cozic (Paris-Est Créteil) :

"Trust and Bayes",


How should an agent take into account the probabilistic opinion of other agents in a group? The traditional Bayesian answer would be to use Bayes rule and higher-order probabilities. In the 80's, Lehrer and Wagner proposed a different and much simplier model based on the attribution of epistemic weights, which were meant to express degrees of trust. The interest of the model is to allow for a detailed study of the conditions under which agents may reach consensus (by repeated updating on each other’s beliefs).However, the absence of a principled justification for their update mechanisms casted on some doubts on the significance of the results. In this talk, we will discuss whether such a justification can be given and prove a representation theorem for Lehrer and Wagner’s updates with respect to Bayesian updates.

Opening meeting:  May 16, 2012 

Joëlle Proust

"Acceptance and their norms: a two-tiered view".


An area in the theory of action that has received little attention is how mental agency and world-directed agency interact. The purpose of the present contribution is to clarify the rational conditions of such interaction, through an analysis of the central case of acceptance.There are several problems with the literature about acceptance.First, it remains unclear how a context of acceptance is to beconstrued. Second, the possibility of conjoining, in acceptance, an epistemic component, which is essentially mind-to world, and a utilitycomponent, which requires a world-to-mind direction of fit, is merelyposited rather than derived from the rational structure of acceptance.Finally, the norm of acceptance is generally seen as related to truth,whichturns out to be inapplicable in a number of cases. We will argue, first, that the specific context-dependence of acceptances is derived from their being mental actions, each embedded in a complex hierarchy of acceptances composing, together, a planning sequence. Second, that acceptances come in several varieties, corresponding to the specific epistemic norm(s) that constitute them. The selection of a particular norm for accepting answers to considerations of utility –to the association of an epistemic goal with an encompassing world-directed action. Once a type of acceptance is selected, however, the epistemic norm constitutive for that acceptance strictly applies. Third, we argue that context dependence superimposes a decision criterion on the output of the initial epistemic acceptance. Strategic acceptance is regulated by instrumental norms of expected utility, which may rationally lead an agent to screen off her initial epistemic acceptance.