Sophie Bavard, Maël Lebreton, Mehdi Khamassi, Giorgio Coricelli & Stefano Palminteri (2018). Reference-point centering and range-adaptation enhance human reinforcement learning at the cost of irrational preferences. Nature Communications 9, 4503

Abstract:
In economics and perceptual decision-making contextual effects are well documented, where decision weights are adjusted as a function of the distribution of stimuli. Yet, in reinforcement learning literature whether and how contextual information pertaining to decision states is integrated in learning algorithms has received comparably little attention. Here, we investigate reinforcement learning behavior and its computational substrates in a task where we orthogonally manipulate outcome valence and magnitude, resulting in systematic variations in state-values. Model comparison indicates that subjects’ behavior is best accounted for by an algorithm which includes both reference point-dependence and range-adaptation— two crucial features of state-dependent valuation. In addition, we find that state-dependent outcome valuation progressively emerges, is favored by increasing outcome information and correlated with explicit understanding of the task structure. Finally, our data clearly show that, while being locally adaptive (for instance in negative valence and small magnitude contexts), state-dependent valuation comes at the cost of seemingly irrational choices, when options are extrapolated out from their original contexts.

Communiqué de presse

Cavanagh, P., Casati, R., Elder, J., Tight shadows shrink depth. Journal of Vision September 2018, Vol.18, 493. doi:10.1167/18.10.493.

Abstract:
When an object casts a shadow on the surface behind it, we get a strong impression of depth separating the object and the rear surface. Kersten and Mamassian (1996) demonstrated this with a cast shadow that moved in and out below a static square, making the square appear to approach and recede in depth. There was no visible light source and the shadow motion could have been caused by a moving light source. Instead, viewers assumed a fixed light and a moving object where the depth of the object scaled with the offset of its shadow. Here we report a particularly strong light source bias in the case of "tight shadows", shadows that have very small offsets from the object casting them, as is often seen with flash photography. The small offset between object and shadow is due to the light source being near the point of view of the observer or camera. However, rather than attributing the tight shadow to the light source location, observers instead attribute it to a very small separation between the object and the rear surface. This bias – "tight shadow equals small separation" – leads to paradoxical impressions of objects impossibly close to their backgrounds. However, it is consistent with the expected object to background separation averaged over all possible light source directions, e.g., lights to the left for right shadows. In this framework, tight shadows are most likely to be caused by shallow relief.

Guez, A., Peyre, H., Le Cam, M., Gauvrit, N., & Ramus, F. (2018). Are high-IQ students more at risk of school failure?. Intelligence, 71, 32-40

Abstract:
While it is well-established that intelligence tests positively predict academic achievement, there remain widespread beliefs that gifted students experience difficulties at school and are particularly at risk of school failure. Many studies have provided evidence to the contrary, however few were based on representative population samples. This paper intended to assess whether prior results on the academic success of gifted children could be generalized to a large sample from the general French population. We analyzed a database of French middle school students (N = 30,489), including scores in a fluid intelligence test in grade 6 and a variety of school performance measures in grade 9 (results at a national exam, teachers' grades, academic orientation in high school). In addition, self-efficacy and motivation were assessed. Our results replicate and extend previous findings: high-IQ students scored much better on all academic performance measures, which was corroborated by higher levels of motivation and self-efficacy. Consistently with the previous literature, there was a robust positive relationship between fluid intelligence in grade 6 and academic performance in grade 9 in the whole sample, which was also observed within high-IQ students. Exploratory analyses revealed that IQ moderated the association between social background and children's achievement, such that the positive link between parental education and achievement levelled off for high-IQ children. The positive association between high-IQ and achievement was similar for boys and girls.

Recanati F. (2018). "From Meaning to Content : Issues in Meta-Semantics". In D. Ball et B. Rabern (dir.) The Science of Meaning : Essays on the Metatheory of Natural Language Semantics. Oxford : Oxford University Press, pp. 113-137.

Abstract: According to a widespread picture due to Kaplan, there are two levels of semantic value: character and content. Character is determined by the grammar, and it determines content with respect to context. In this chapter Recanati criticizes that picture on several grounds. He shows that we need more than two levels, and rejects the determination thesis: that linguistic meaning as determined by grammar determines content. Grammatical meaning does not determine assertoric content, he argues, but merely constrains it—speaker’s meaning necessarily comes into play. On the alternative picture he offers, there are four basic levels, only one of which is determined by the grammar. Pragmatics is what enables the transition from each level to the next.

Recanati F. (2018). "Contextualism and Singular Reference". In J. Collins et T. Dobler (dir.) The Philosophy of Charles Travis: Language, Thought, and Perception. Oxford: Oxford University Press, pp. 181-196.

Abstract:
This chapter discusses the relations between three approaches to the referential/attributive distinction: the Gricean approach advocated by Kripke and others, the two-dimensional approach pioneered by Kaplan and Stalnaker, and the Millian approach favoured by Donnellan. In contrast to the two-dimensional approach, the Millian approach honours the intuitions which led to the rejection of descriptivism, but it is subject to Gricean criticism based on the speaker’s reference/semantic reference distinction. The chapter shows that, suitably elaborated and revised, the Millian approach can be made immune to that criticism. The resulting view, it is argued, applies beyond the case of definite descriptions. It also supports Austin’s and Strawson’s speech act theoretic approach to reference and truth—an approach which Grice initially dismissed and which Travis insightfully attempted to defend and resurrect.

AGENDA

S'abonner

Se désabonner