Scientific Publications of the University of Toulouse II Le Mirail
Not a member yet
92205 research outputs found
Sort by
Reconnaître sans se souvenir : la persistance de la reconnaissance implicite dans les stades sévères des maladies neuro-évolutives
No full textInternational audience•Contexte : Cette recherche a pour objectif de montrer qu’il existerait une persistance de la reconnaissance implicite chez les personnes atteintes d’une maladie neuro-évolutive (MNE) à un stade avancé. Il s’agit de mettre en avant la fonction de la mémoire implicite comme support de la capacité de reconnaissance et de rappel d’une personne ou d’une situation relationnelle.•Méthodes : Cette étude repose sur un protocole de recherche structuré en 11 séances avec un expérimentateur inconnu des participants. Les rencontres ont eu lieu quotidiennement sur une période de 2 semaines. La 11ème séance, programmée 6 jours après la 10ème séance, visait à mesurer la capacité de reconnaissance de l’expérimentateur par les participants. Nous les avons rencontrés autour d’un dispositif relationnel : jeu de construction avec des pièces en bois (couleurs et formées variées). Afin de tester l’hypothèse d’une reconnaissance implicite chez les personnes atteintes d’une MNE à un stade sévère, deux situations expérimentales, une neutre (N) et une positive (S+), ont été choisies dans le but de vérifier l’impact de la valence émotionnelles sur la mémorisation des sujets. Nous avons évalué cette reconnaissance en combinant des éléments verbaux et non verbaux afin de mesurer et d’interpréter les indices de reconnaissance.•Résultats / Discussion* : À ce jour, 40 participants ont été inclus (20 rencontrés dans une valence expérimentale positive et 20 dans une valence neutre). Les premiers résultats de cette recherche montrent que, quel que soit la modalité de valence attribuée, les personnes âgées atteintes d’une MNE parviennent à reconnaître l’expérimentateur. Ces résultats semblent montrer qu’une reconnaissance implicite est possible même à un stade avancé de la maladie. *Étude en cour
Les adeptes d’activités physiques, sportives et récréatives de pleine nature et la transition écologique : quels rapports à la nature et quels engagements en faveur de l’environnement ?
No full textInternational audienceAlors que la transition écologique semble inéluctable dans le domaine des activités physiques et sportives (APS) comme ailleurs et que par nature les pratiquant-e-s des activités physiques, sportives et récréatives de pleine nature (APPN) sont considéré-e-s comme (plus ?) sensibles aux enjeux environnementaux que les autres pratiquant-e-s, peu de chercheur-e-s examinent leur point de vue sur les risques que le changement climatique font peser sur leur pratique, les adaptations ou non au changement climatique, leur engagement ou non sur le front de la protection de l’environnement et les moyens pour y parvenir. A partir des discours recueillis dans le cadre d’un programme de recherche soutenu par l’institut National de la jeunesse et de l’éducation populaire auprès d’une vingtaines de pratiquant-e-s impliqué-e-s dans deux activités de pleine nature (surf sur eau, escalade en extérieur) contrastées au plan social, du genre, de l’âge et à l’appui de la sociologie de la socialisation (Dubar, 1991, Lahire, 2002) et sociologie de l’engagement et du militantisme (Fillieule et Pudal, 2010), nous démontrerons l’ambivalence qui s’impose chez les pratiquant-e-s au sujet de la place de la transition écologique, comprise comme « un ensemble d’ajustements techniques, logistiques et symboliques destinés à réduire l’empreinte environnementale des pratiques sportives tout en réaffirmant leur ancrage local » (Vivier, 2021). L’analyse des échanges montrent que les adeptes interviewé-e-s n’ont pas toutes et tous une conscience aigüe des enjeux climatiques, leurs ajustements sont mineurs et rares sont celles et ceux à s’engager en faveur de la protection de l’environnement. Les autrices de cette communication proposent aussi d’ouvrir le débat/la discussion sur le concept « écoféminisme » (Larrère, 2012) dans les sciences sociales du sport. Dans cette perspective, la question est de savoir si les femmes − pratiquantes d’APPN − sont plus concernées par les enjeux climatiques, non pas essentiellement/nécessairement parce que ce sont des femmes, mais parce que certains processus (socialisations entre autres) suscitent des inquiétudes, des postures, des manières genrées de prendre soin de la nature et de soi-même
Expériences érémitiques, inspiration orientale et traditions locales dans le Nord-Ouest de la péninsule Ibérique (VIIe-Xe siècle)
No full textInternational audienc
Taming the Triangle: On the Interplays Between Fairness, Interpretability, and Privacy in Machine Learning
No full textInternational audienceMachine learning techniques are increasingly used for high‐stakes decision‐making, such as college admissions, loan attribution, or recidivism prediction. Thus, it is crucial to ensure that the models learnt can be audited or understood by human users, do not create or reproduce discrimination or bias and do not leak sensitive information regarding their training data. Indeed, interpretability, fairness, and privacy are key requirements for the development of responsible machine learning, and all three have been studied extensively during the last decade. However, they were mainly considered in isolation, while in practice they interplay with each other, either positively or negatively. In this survey paper, we review the literature on the interactions between these three desiderata. More precisely, for each pairwise interaction, we summarize the identified synergies and tensions. These findings highlight several fundamental theoretical and empirical conflicts, while also demonstrating that jointly considering these different requirements is challenging when one aims at preserving a high level of utility. To solve this issue, we also discuss possible conciliation mechanisms, showing that a careful design can enable to successfully handle these different concerns in practice
Immigration Lawyers in Troubled Waters
No full texthttps://e-legal.ulb.be/volume-9/droit-s-aux-frontieres/immigration-lawyers-in-troubled-watersIn France, as elsewhere in Europe, defending the rights of migrants entails exposure to multiple forms of risk. Practicing as a lawyer does not automatically provide protection compared to other non-specialist supporters; rather, it involves confronting a specific set of challenges. These include difficult working conditions and inadequate remuneration, rooted in a longstanding devaluation of this area of law and its association with so-called “poor people’s litigation.” They also derive from the constant stream of legislative and procedural reforms—both national and European—that progressively restrict migrants’ rights, their effective exercise, and the possibilities of meeting the conditions required to contest decisions that often render those rights illusory.The present period, however, has also witnessed another challenge for immigration lawyers (and, at times, judges adjudicating related cases): orchestrated smear campaigns. These public attacks, disseminated through social media and certain magazines, go so far as to publish the names of the supposed “culprits of the migrant invasion,” accused of creating a “pull factor,” squandering public funds, and allegedly profiting from a “lucrative immigration business.”This paper therefore examines the social function performed by many migration lawyers in such dark times. Building on Liora Israël’s and Laurent Willemez work on the political mobilizations of law by lawyers and judges in contemporary France, I focus on the relationship between immigration law in practice and social justice, by foregrounding the increasingly constrained role of lawyers as defenders of democratic principles under threat. More specifically, I seek to show how this social function is undermined in distinct but interconnected ways: first, through the constant reshaping of the legal framework; and second, through the material constraints under which the legal defense of migrants must be conducted. In conclusion, I argue that these two dynamics, which operate as obstacles to the defense of migrants and, more broadly, to the safeguarding of fundamental democratic principles now at risk, are further aggravated by a sociopolitical climate marked by heightened hostility towards certain groups, stigmatized as intellectuals, the “woke,” or “human rights advocates.”The research conducted focused on the professional routines and forms of commitment among immigration lawyers. What unites these professionals is their commitment to defending migrants' access to legal rights, even if their areas of legal expertise may differ (public law, criminal law, etc.). Defending the rights of non-citizens is inherently difficult. First and foremost, this is because those rights are steadily diminishing, and the governments of an increasing number of European countries appear firmly determined to narrow them even further. This is not the only common constraint: the conditions under which this profession must be practiced defending what remains of the rights of non-UE foreigners are themselves challenging due to a variety of economic, symbolic, and procedural constraints. 25 interviews were carried out between 2018 and 2024, mainly in Lille, Bordeaux, Toulouse, Nice, and Boulogne-sur-Mer, with lawyers defending the rights of non-UE persons, as well as their associative and activist supporters. This inquiry was supplemented by courtroom observations in Boulogne-sur-Mer and Toulouse, and by observations during volunteer legal training sessions in the Roya Valley. The interviews sought to understand the meaning attributed by actors to moral and legal arguments in the defense of their clients’ rights. Discussions with lawyers revolved around their professional daily life, their personal networks, their relationships with colleagues and judges, then with clients and the associative networks in which they are involved, and more generally around their trust in justice and in the “weapon” of law. Access to the case files of the interviewed lawyers proved invaluable, as it enabled insight into the construction of defense strategies and served as a basis for follow-up questioning.The article is structured as follows. The first section situates the research question within the broader landscape of existing social science scholarship, at the intersection of studies on legal intermediaries and those on anti-intellectualism and anti-woke movements, before outlining the empirical design of the study and the materials on which it draws. The second section examines the first factor that undermines the work of these legal experts: the continual reconfiguration of the French legal framework. The third section, based on research into the professional practices and forms of mobilization among immigration lawyers in five French cities, illustrates the diverse material and symbolic challenges involved in defending the rights of foreign nationals. The final section addresses the reactionary strategies of intimidation and delegitimization aimed at lawyers and judges accused of defending unpopular causes and obstructing the increasingly unrestrained xenophobic agendas of right-wing movements and concludes by underlining the dangers for democratic societies of underestimating the threats to the social role of migration lawyers
Un outil de gravure en bois de cervidé dans la grotte de Cussac (Dordogne, France) ?: Analyses et discussions autour d'un objet particulier découvert à près de 620 mètres de l'entrée
No full textInternational audienceThe bone industry discovered in Cussac Cave (Dordogne, France) is represented by three fragmentary, associated elements, whose structural, morphometric, and technological characteristics permit attribution to a single long, cylindrical, and pointed antler artifact. Isolated and found at a considerable distance (ca. 618 m) from the entrance of the cave, it is of significant interest both in terms of its potential chrono-cultural attribution and its presumed function. After being lost or abandoned, this object underwent various taphonomic alterations. Discovered broken, encrusted, and adhering to the floor of the cave, it raised several questions with regard to its raw material, chrono-cultural attribution, status as a processing tool as opposed to a piece of hunting equipment, its “operating mode” (according to the translation given in Guéret et al., 2014 of ‘mode de fonctionnement’) and its “technical function” (sensu Claud et al., 2019), and reasons for its presence in the cave. The different steps in its analysis and associated results are presented here, including in situ observations, sampling, restoration, taphonomic and techno-functional analyses, ZooMS identification, and 14C dating. They permit the reader to follow the inductive approach that guided our investigations and led us from empirical elements placed in perspective (on the object itself and its archaeological context) to explore a possible use of the Cussac object as an engraving tool in the production of parietal art. Notably, the cave contains rock art panels (e.g., the Réticulé panel in the downstream branch) with markings consistent with the use of a soft-tipped organic tool (as concluded from a 2022 published experimental study on the morphometric characterization of marks produced by different objects).To test the plausibility of this hypothesis, the use-wear analysis of the object was complemented with an exploratory experimental component conducted in a neighboring cavity with a comparable geological environment, providing a preliminary reference for wear traces generated by engraving with antler tools (performed on weathered limestone and, to a lesser extent, on clay).Based on this study, the following results can be reported: the object likely measured approximately 200 mm in length and 10 mm in width and thickness in its original state; it was shaped from a small-diameter reindeer antler segment (possibly a beam). Moreover, its post-depositional history is complex; its morphotype links it to other discoveries, currently recognized only in Gravettian contexts. However, it stands out due to the care of its manufacture and the status of its place of discovery in a rock art burial cave. While direct dating of the object was unsuccessful due to insufficient collagen preservation, its chrono-cultural attribution corresponds with stylistic evidence from the parietal art as well as other elements, such as radiocarbon dating of charcoal and human remains and the chronological attributions of the few lithic artifacts recovered in the cave.The analysis identified macro- and microscopic functional marks, the nature and organization of which permit, on one hand, the identification of an active part, round-tipped and bearing short transverse incisions (and a potential second one); and, on the other hand, the inference for its operating mode. This involved a deliberate rubbing action, performed with an acute rake angle, involving brief but repeated and continuous contact of the working surface with a relatively supple but not soft material, and intermittent contact of the apex with a harder material. However, the preserved and observable wear areas do not fully represent the full extent of its original appearance due to taphonomic effects and the porous nature of bone surfaces, and must therefore be considered as a minimal approximation. It can nevertheless be noted that the object’s operating mode and structural properties are incompatible with use involving high mechanical stress. At the same time, the associated nature and pattern of its use-wear traces and the non-cutting, non-piercing, non-perforating character of the object excludes its use as a hunting weapon, as an intermediary perforating tool or pressure flaker, as well as use as a soil- or termite-excavating tool unlikely, based on reference collections used in this study. These observations allowed us to narrow down plausible functional hypotheses. At the same time, use-wear analysis correspondences were established with the experimental tools used. We thus propose, with all necessary caution, to consider it as a possible engraving tool, while remaining aware of the interpretive limits of reasoning from a single object, particularly one with a complex preservation history. This limitation is primarily circumstantial, as no other bone tools potentially associated with engraving activities have, to date, been studied from a functional perspective.The contributions and limitations of our experimental approach are also discussed, notably the number of experimental tests conducted and the raw material used for the experimental tools. While further experiments could enrich the reference framework and improve the protocol, the results presented in this article permit— with necessary precaution — the hypothesis of a parietal engraving tool to be considered “operational,” if not fully demonstrable, and at minimum, they do not invalidate it. The state of preservation of the object and its unique status that precludes comparison with other such specimens remain a limitation that cannot be overcome at present.Our review of reported or potential cases of engraving tools in Palaeolithic cave contexts shows that the Cussac piece constitutes the first bone tool – and the only manufactured one – for which this hypothesis is discussed based on a detailed study of technical and functional attributes, combined with an experimental component. In this sense, the study provides an unprecedented didactic basis from which to address these questions. More specifically, it contributes to our understanding of and questions concerning the relationship between underground environments and material culture at Cussac Cave.La grotte ornée et sépulcrale de Cussac a livré pour l’heure un unique objet en matière osseuse (appointé mousse), retrouvé isolé, cassé et à une grande distance de l’entrée. Les étapes et résultats de sa chaîne d’analyse ̶ ici présentés ̶ permettent de suivre la démarche inductive qui nous a conduit à nous interroger sur son possible usage en outil de gravure pariétale. Cette hypothèse prend appui sur les résultats de l’analyse de l’objet croisés aux autres données archéologiques disponibles. Pour tester cette hypothèse, l’analyse tracéologique de l’objet a été adossée à un volet expérimental conduit dans une cavité offrant un environnement géologique comparable à celui de Cussac afin de disposer d’un premier référentiel de traces d’usure générées sur outils en bois de cervidés par l’acte de graver. Les résultats de notre étude sont les suivants : l’objet (env. 200 mm de long pour 10 mm de large et d’épaisseur) est façonné sur un tronçon de bois de renne de petit module ; son histoire post-dépositionnelle est complexe ; son morphotype le rapproche d’autres découvertes, pour l’heure, uniquement reconnues en contextes gravettiens. Il se démarque toutefois par le soin apporté à sa fabrication et le statut fonctionnel du lieu de sa découverte. Son attribution chronoculturelle, en l’absence de datation radiométrique (essai infructueux), fait toutefois écho aux données de l’art pariétal, de l’industrie lithique et aux datations 14C sur charbons de bois et un reste humain. Son analyse fonctionnelle a permis d’identifier une extrémité active (et une autre potentielle) et d’en inférer un mode de fonctionnement. Ce dernier, ainsi que les stigmates fonctionnels et les propriétés morpho-structurelles de l’objet sont notamment incompatibles avec une utilisation en armature de projectile, compresseur, outil intermédiaire ou outil perforant. Parallèlement, des correspondances tracéologiques ont été établies avec les outils expérimentaux que nous avons utilisés. L’ensemble de ces données, avec les précautions nécessaires, autorise à considérer l’hypothèse d’outil de gravure pariétale comme « opératoire », à défaut d’être pleinement démonstrative et a minima, elles ne l’invalident pas. Cette étude constitue une base de travail didactique inédite. La pièce de Cussac est en effet le premier outil en matière osseuse (et le seul manufacturé) pour lequel cette hypothèse est examinée sur la base d’une étude détaillée des stigmates techniques et fonctionnels et d’un volet expérimental exploratoire
Neuroinflammation and disrupted functional connectivity in coma: a graph-theoretical study
No full textInternational audienceComa leads to persistent cognitive and behavioral deficits, yet the mechanisms underlying recovery remain poorly understood. Addressing this challenge requires multimodal approaches that integrate imaging, modeling, and clinical data. TSPO-PET imaging reveals widespread neuroinflammation in key resting-state network regions of coma patients, while fMRI graph-based analyses show marked functional network disruption. However, how neuroinflammation relates to functional connectivity disruption remains unclear.To address this gap, we developed new graph-based modeling strategies to jointly analyze neuroinflammatory and functional brain networks. Using TSPO-derived graphs, we show that neuroinflammatory network organization parallels functional connectivity reorganization in coma. Multiplex graph analyses further reveal a shared multimodal network structure within groups, with clear differences between healthy controls and coma patients. Together, these results highlight cross-system network reorganization in coma and support a unified framework for simultaneously studying functional connectivity and neuroinflammation
A predictive coding framework for safe and versatile control of supernumerary robotic limbs
No full textRobotic supernumerary limbs (SLs) can enable humans to perform alone tasks requiring more than two hands. By increasing the number of the body's effective degrees of freedom, SLs could have a major impact on industrial and medical applications, and profoundly change the way humans interact with their environment. However, current SL usage is limited, with only specific applications and a lack of a comprehensive framework for versatile ubiquitous use. This limitation arises primarily from the intertwined challenges of ensuring safety and coordination between the natural limbs and SLs. This paper presents a novel framework for SL sensorimotor augmentation based on predictive coding, a process rooted in hierarchical neural structures that allow the brain to seamlessly manage coordinated complex actions. Our framework comprises three levels where each generates predictions for and receives prediction errors from the level below. The α-layer detects the user's movement intent and ensures safety through generated constraints. The β-layer controls actions and related sensory feedback. In this layer, we propose a novel general voluntary control (GVC) mechanism that offers a continuum between fully autonomous SLs behaviors and direct control, thereby allowing a user to control SLs with minimal cognitive effort. The γ-layer executes the SLs' control and handles their interaction with the body and the environment. For each framework component, we review relevant literature and illustrate implementation strategies that support safe and versatile SLs control. We further demonstrate, through systematic simulations, the usability and robustness of the proposed GVC mechanism, paving the way toward versatile human-SLs interactions
Few Shades of Supervision for Discourse Segmentation
No full textInternational audienceElementary Discourse Units (EDUs) constitutes the interface between language grammar and lan- guage use. On the one hand, they result from compositional semantic processes that combines individual word meanings into proposition-level representations. On the other hand, EDUs form the building blocks of most text, discourse, and dialogue frameworks. In written genres, where punctuation is available and reliable, segmenting EDUs is sometimes seen as a nearly solved problem, as least for high-resource languages. However, this is not the case for spontaneous speech transcripts. In this paper, we use a significant (8-hour) French corpus, manually segmented into EDUs, to evaluate several large language model (LLM)-based approaches for this task. We compare various fine-tuning strategies, including those relying on weakly supervised labels, in relation to the amount of ”gold” manual annotations that can be available. We also experiment with in-context learning, where example instances are provided to condition a generative model (few-shots learning) or in a purely generative approach (zero-shot). Our findings indicate that classical fine-tuning is still the most effective approach, requiring only a reasonable amount of gold-annotated data to achieve the best performance in our experiments. Beyond traditional quantitative evaluation, we conducted a systematic qualitative analysis, identifying directions for further improvement. These include integrating prosodic considerations while handling pauses when they co-occur with disfluencies or complex discourse markers uses. Finally, we argue for the significance of this task and the resulting units, compared to acoustic and syntactic proxies, especially for quantitative linguistics focusing on spontaneous speech