Archive Electronique - Institut Jean Nicod
Not a member yet
1997 research outputs found
Sort by
Evoked and Transmitted Culture models: Using bayesian methods to infer the evolution of cultural traits in history
No full textInternational audienceA central question in behavioral and social sciences is understanding to what extent cultural traits are inherited from previous generations, transmitted from adjacent populations or produced in response to changes in socioeconomic and ecological conditions. As quantitative diachronic databases recording the evolution of cultural artifacts over many generations are becoming more common, there is a need for appropriate data-driven methods to approach this question. Here we present a new Bayesian method to infer the dynamics of cultural traits in a diachronic dataset. Our method called Evoked-Transmitted Cultural model (ETC) relies on fitting a latent-state model where a cultural trait is a latent variable which guides the production of the cultural artifacts observed in the database. The dynamics of this cultural trait may depend on the value of the cultural traits present in previous generations and in adjacent populations (transmitted culture) and/or on ecological factors (evoked culture). We show how ETC models can be fitted to quantitative diachronic or synchronic datasets, using the Expectation-Maximization algorithm, enabling estimating the relative contribution of vertical transmission, horizontal transmission and evoked component in shaping cultural traits. The method also allows to reconstruct the dynamics of cultural traits in different regions. We tested the performance of the method on synthetic data for two variants of the method (for binary or continuous traits). We found that both variants allow reliable estimates of parameters guiding cultural evolution. Overall, our method opens new possibilities to reconstruct how culture is shaped from quantitative data, with possible application in cultural history, cultural anthropology, archaeology, historical linguistics and behavioral ecology
Speech acts and Communicative Intentions for Urgency Detection
No full textInternational audienceRecognizing speech acts (SA) is crucial for capturing meaning beyond what is said, making communicative intentions particularly relevant to identify urgent messages. This paper attempts to measure for the first time the impact of SA on urgency detection during crises,006in tweets. We propose a new dataset annotated for both urgency and SA, and develop several deep learning architectures to inject SA into urgency detection while ensuring models generalisability. Our results show that taking speech acts into account in tweet analysis improves information type detection in an out-of-type configuration where models are evaluated in unseen event types during training. These results are encouraging and constitute a first step towards SA-aware disaster management in social media
Dominant jerks: People infer dominance from the utterance of challenging and offensive statements
No full textInternational audienceCould there be upsides to rudely challenging people’s positions? If no one calls out the speaker of a challenging or offensive statement, it might be because the audience is afraid to challenge the speaker, thereby suggesting the speaker holds a dominant position. In two experiments (N = 635), participants read vignettes in which a speaker uttered a statement that was challenging (it directly clashed with the audience’s prior views) or unchallenging (it agreed with the audience’s prior views). We also manipulated whether the audience accepted or rejected the statement after it was uttered. In Experiment 1 the statements were about mundane topics, while in Experiment 2 the statements were offensive. In both experiments, speakers uttering challenging statements that the audience nonetheless accepted were deemed more dominant and more likely to be the boss of the audience members. This shows that people use audience reactions to challenging statements to infer dominance, and suggests that people might use the utterance of challenging statements to demonstrate their dominance
Being green or being nice? People are more likely to share nicer but potentially less impactful green messages
No full textInternational audienceCitizens can play an important role in disseminating scientific information about climate change, if motivated to do so. However, expressing green positions has the potential to negatively affect people's reputation, by making them look judgmental for instance. In three experiments among US and UK participants (N = 1197) we investigate the reputational costs of sharing statements about climate change that vary in accuracy and in potential impact. In Experiment 1, we show that participants judge more negatively someone sharing a bleak (but arguably more accurate) statement about climate change (e.g., calling it "climate breakdown"), compared to a control statement. Experiment 2 replicates this finding with control statements (e.g. "The richest 1% in the world is responsible for most of the greenhouse gas emissions") compared to accusatorial statements (adding "because most citizens in countries like the United States consume too much energy"). Experiment 3 shows that participants are less willing to share more accusatorial statements-even though they are thought to exert a greater effect on their audience. Our results further show that the fear of appearing judgmental and unfriendly might make people less likely to share bleaker or more accusatorial-even if more accurate or potentially effective-statements about climate change
VAGO: un outil en ligne de mesure du vague et de la subjectivité
No full textInternational audienceVAGO is an online tool relying on an annotated lexical database and expert rules to provide a measure of vagueness and subjectivity in textual documents. The development of VAGO is the result of the cooperation between the INSTITUT JEAN-NICOD (UMR 8129 of CNRS) and the MONDECA company. VAGO is based on a four-fold typology of vague expressions, distinguishing generality, approximation, one-dimensional vagueness, and multidimensional vagueness. %Currently, the database consists mostly of adjectives and a limited lexicon for English and French. In this demonstration, (i) we introduce the user to the motivations behind the VAGO typology, (ii) we make explicit the technological chain used for the implementation of VAGO, and (iii) we show how VAGO can help in the detection of false or unreliable information. Online demo: \url{https://youtu.be/L6cc05SlA5E}VAGO est un outil en ligne de mesure du vague et de la subjectivité dans le discours, fondé sur une base de données lexicale annotée ainsi que sur des règles expertes. VAGO est développé dans le cadre d'une coopération entre l'INSTITUT JEAN-NICOD (UMR 8129 du CNRS) et la société MONDECA. VAGO repose sur une quadruple typologie des expressions vagues, distinguant la généralité, l'approximation, le vague unidimensionnel et le vague multidimensionnel. Nous utilisons la typologie pour étiqueter les expressions comme marqueurs de subjectivité ou d'objectivité. Dans cette démonstration, nous présentons (i) les motivations de la typologie de VAGO, (ii) la chaîne technologique mise en place dans la réalisation de VAGO, et (iii) l'utilisation de VAGO pour l'aide à la détection d'informations fausses ou peu fiables. Vidéo de démonstration : https://youtu.be/L6cc05SlA5
Understanding force cancellation
No full textInternational audienceTo address the Frege-Geach objection, proponents of the 'forceful' version of the act-theoretic approach to propositions appeal to the idea of force cancellation. How is that idea to be understood? In this paper, three models of force cancellation are discussed (and their shortcomings pointed out): the mereological model, the Brentanian model, and theintermediatetransmutation model. Extant versions of these models are meant to account for force cancellation in speech, but they do not easily extend to force cancellation in thought. To overcome that limitation, a psychologistic version of the transmutation model is put forward, based on 'simulation theory'
Early concepts of intimacy: Young humans use saliva sharing to infer close relationships
No full textInternational audienceAcross human societies, people form “thick” relationships characterized by strong attachments, obligations, and mutual responsiveness. People in thick relationships share food utensils, kiss, or engage in other distinctive interactions that involve sharing saliva. We found that children, toddlers, and infants infer that dyads who share saliva (as opposed to other positive social interactions) have a distinct relationship. Children expect saliva sharing to happen in nuclear families. Toddlers and infants expect that people who share saliva will respond to one another in distress. Parents confirm that saliva sharing is a valid cue of relationship thickness in their children’s social environments. The ability to use distinctive interactions to infer categories of relationships thus emerges early in life, without explicit teaching; this enables young humans to rapidly identify close relationships, both within and beyond families
It's not about 'about' - comparatives, negation and intervals
No full textInternational audienceSolt (2014, 2018) discovered an intriguing pattern regarding the distribution of the approximator 'about'. While 'about n' is typically infelicitous under negation, this pattern is reversed with 'more than about n', which is fine under negation, but not in a simple, unembedded context. Solt proposed an ingenious account based on certain assumptions about the meaning of 'about' and principles of language use, and, specifically, the fact that 'about' is an approximator that manipulates a granularity parameter. I argue that the pattern uncovered by Solt is not specifically tied to approximators, as it can be reproduced with disjunctions of numerals and interval-denoting expressions ('between n and m'), and is therefore part of a broader generalization. I offer an account based on (a) the universal density of measurement scales (Fox and Hackl 2006), (b) a semantic analysis of degree constructions that involves in a crucial way the notion of 'maximal informativity' (Buccola and Spector 2016, with roots in Rullmann 1995; Fox and Hackl 2006; Schlenker 2012; von Fintel et al. 2014) and (c) a pragmatic ban on redundant numerical expressions. I then discuss some limitations of the proposal, in comparison with Solt’s
