Au-delà du cerveau: la cognition incarnée, située et distribuée

Cognitio 2006

Colloque jeunes chercheurs en sciences cognitives

Montréal, 19, 20 et 21 août 2006.

A PDP model of complex sentence production
Hartmut Fitz, Universeit van Amsterdam, Netherlands
Recursive productivity is considered a core property of naturallanguage and the human language faculty (Hauser, Chomsky & Fitch, 2002).It has been argued that the capacity to produce an unbounded varietyof utterances requires symbolic capabilities. Lacking structuredrepresentations, connectionist models of language processing arefrequently criticized for their failure to generalize symbolically(Hadley, 1994; Marcus, 1998).Addressing these issues, we present a neural-symbolic learning modelof sentence production, called the recursive dual-path model, whichcan cope with complex sentence structure in the form of embeddedsubordination of multiple levels.The model has separate pathways, one for mapping messages to words andone for sequence learning. The message is represented through bindingof thematic roles to concepts by weight and is inspired by spatialprocessing of visual input. In selecting syntactic frames, thesequencing system is guided by an event-semantics layer which providesinformation about clause attachment, tense, aspect, and the relativeprominence of message components.The model is tested on a structurally complex language built fromsimple clause constructions which are basic to human experience(Goldberg, 1995). We investigate the model's learning behaviorconcerning complex multi-clausal utterances and show that itsperformance matches differential trends in humans. Furthermore, weexplore its ability to produce novel embedded structures and to map`familiar' constituents to novel roles at novel sentence positions.The recursive dual-path model is joint work with Franklin Chang,postdoc researcher at NTT Communication Science Laboratory Kyoto,Japan, and based on:Chang, F. (2002) Symbolically speaking: A connectionist model ofsentence production. Cognitive Science, 26(5), 609-651.
Apprentissage lexical et activations cérébrales : une étude chez des sujets sains et des patients aphasiques
Gaelle Raboyeau, Université de Montréal, Canada
L’aphasie est un trouble de la compréhension et/ou de la production du langage dû à une lésion cérébrale focale, suite à un accident vasculaire cérébral. La récupération linguistique des patients aphasiques implique des mécanismes cérébraux qui ne sont actuellement pas bien compris. Certaines études montrent que cette récupération est sous-tendue par des aires homologues à la lésion dans l’hémisphère droit tandis que d’autres études présentent l’implication des zones gauches autour de la lésion. De plus, on ne connaît pas actuellement les processus cérébraux de l’apprentissage chez les sujets sains qui permettrait de mieux dissocier les phénomènes de plasticité cérébrale impliqués dans l’apprentissage lexical en général de ceux spécifiques aux patients.Mon étude avaient donc un double objectif : déterminer, grâce à la neuro-imagerie fonctionnelle, les aires impliquées dans les phénomènes de réapprentissage du lexique chez des sujets sains adultes puis confronter ces résultats à ceux obtenus chez des patients aphasiques. Le protocole était constitué d’une période d’entraînement lexical intensif d’une durée d’un mois, pendant lequel les patients réapprenaient de mots français et les sujets sains réapprenaient de mots d’une langue seconde acquise scolairement mais partiellement oubliée (Espagnol ou Anglais). Deux examens de neuroimagerie (un avant et un après l’entraînement) durant lesquels les sujets réalisaient une tâche de dénomination d’images, servaient à identifier les modifications cérébrales liées à l’apprentissage lexical.Chez les sujets sains, le réapprentissage lexical semble être sous-tendu par des activations frontales droites, ceci indépendamment de la langue. Chez les patients, le réapprentissage lexical fait intervenir des activations frontales droites similaires à celles des sujets sains. Les résultats de cette étude montre que l'activation de l'hémisphère droit chez les patients ne correspondrait donc pas à une activité compensatoire " lésionnelle " (provenant de la lésion) mais plutôt "fonctionnelle" (provenant de la fonction d’apprendre) puisqu'elle est présente chez les sujets sains.
Awareness and Agency in the Visual Field: Sensorimotor Subjects
Drina Eva Bosnjak, York University, Canada
My paper examines the effects enlisted epistemology has on our ontologies. Sensorimotor vision describes vision, and agents of sight whose particular way of seeing is contingent upon their embodiment. My paper proposes that cognitive agency corresponds to varying degrees of awareness. By arguing that a necessary requirement for cognitive agency is awareness, and that awareness varies in degree, I consider varying degrees of awareness, and cognition, as indicative of different degrees of cognitive agents. I begin my argument with the theoretical framework that considers awareness, and cognition, embodied. All cognitive awareness arises from the interaction of cellular automata. Different automata and sensorimotor patterns arouse different perspicacity . Different perspicacity interacts with sensorimotor patterns to arouse different degrees of cognitive awareness. Awareness is a necessary condition for cognitive agency. Different degrees of awareness reflect different degrees of cognitive agency. Our inquiry into minds should allow for different degrees of cognitive awareness. It is unlikely that the complexities of individual minds can be accounted for by using a theory of fixed cognitive identity. The perceptual aspect I will be focusing on for the duration of this paper is vision as it supports cognitive awareness and agency. This perceptual aspect is largely responsible for the precise mapping of spatial location and recognition. It will vary from respectively evolved cognitive agents. Our exploration of epistemic terrain, especially when dealing respectively evolved agents ought to account for the malleability and variation implicit in exploring other terrains. Suggesting that cognitive agency should be reexamined in light of agency demonstrable through sensorimotor vision, I seek to disrupt categories that fail to account for a cognitive agency of variation.[1] Perspicacity is awareness of sensory perceptual orientation.
Comparing Online and Lab Methods in a Problem Solving Experiment
Frederic Dandurand, Universite McGill, Canada
Thomas R. Shultz, Universite McGill, Canada
Online experiments have recently become very popular. Reips (2000,2002) reports several advantages for web-based (or internet-based)experiments over traditional lab experiments, such as reduced demandcharacteristics, automation and generalizability of results to widerpopulations.The aim of this study was to replicate Dandurand, Bowen, and Shultz'(2004) problem solving lab-based experiment as an internet experiment.Consistently with previous results, we found that participants whowatched demos of successful problem solving sessions and participantswho read instructions outperformed participants who were simply toldif they solved problems correctly or not.Furthermore, we found no significant effect involving location(internet vs. lab), meaning that online and internet results areconsistent. Several other researchers (e.g., Krantz & Dalal, 2000;Reips, 2002) also found internet-based and lab-based experiments toyield consistent results.However, because the internet affords short attention spans, moredistractions and lower participant commitment, online research appearbest suited for short and simple experiments and studies. For long andcognitively demanding tasks, we recommend you make sure you haveaccess to committed and motivated participants (e.g., subject pool) orthat you provide external incentives (e.g., monetary).
Concepts: Where Fodor Went Wrong
Marc Champagne, UQAM, Canada
In his book Concepts: Where Cognitive Science Went Wrong (1998), philosopher Jerry Fodor presents an ambitious bid to reorient theorizing away from holistic models of the mind towards an atomist construal of mental representation. Underlying Fodor's unforgiving critiques of many an accepted research program is a desire to construct an account of conceptual compositionality which forbids inference to play any role in the individuation of conceptual content. Given the breadth of its targets and the scope of its revisionism, we propose to submit Fodor's trenchant dialectic to a likewise fate, examining two key points where we believe his arguments have run afoul. On the syntaxic front, Fodor's argument that the world could not possibly be a source of the mind's systematicity will be rejected as resting on a modal conflation between the mind's ability to form the converse of a relation and the necessity of relations holding in the world. On the semantic front, we will argue that Fodor's insistence that informational content can be accessed by a multiplicity of means entails a transitive mind-to-mind-to-world division of labour, and that Fodor's desire to avoid the repercussions of this reveals a latent construal of causal-cum-nomic contact as relaying some kind of iconic replica -- an empiricist position he claims to repudiate.
Conscience, évolution artificielle et cognition sociale
Mouhamadou El Hady, Université Paris 8 Vincennes à Saint-Denis, France
Les robots produits grâce à l’évolution artificielle peuvent-ils devenir conscients ? Floreano (2001) soutient que c’est déjà le cas. Définissant la conscience comme le fait pour un organisme d’être attentif à ses propres états internes en lien avec l’environnement, il affirme que tout organisme ou machine doué d’un corps, autonome et capable d’adapter son comportement à son environnement remplit les pré requis pour être dit conscient. Nous nous interrogeons d’abord sur la définition minimale de la conscience qu’adopte Floreano. Malgré l’intérêt de cette définition liant conscience et comportement finalisé et donnant la primeur à l’action plutôt qu’à la pensée et à la subjectivité, nous montrons qu’elle est trop libérale et nous mènerait à attribuer la conscience à tout organisme exhibant des routines comportementales complexes (l’écureuil qui stocke des noix pour l’hiver par exemple). Ensuite nous tirons les conséquences de cette critique et complétons les conditions d’attribution de la conscience proposées par Floreano. Ses critères sont quasiment béhavioristes car la conscience est selon lui révélée par l’adéquation de l’action du robot à un environnement fluctuant. Dans un tel cadre, la subjectivité est presque épiphénoménale. Nous argumentons quant à nous pour la thèse selon laquelle il faudrait au minimum deux autres choses qu’il n’intègre pas pour qu’un artefact puisse être dit non seulement autonome, mais également conscient: un mécanisme d’indexation des états internes et un mécanisme de communication de ces états. Chez l’homme, ce sont respectivement la mémoire et le langage qui remplissent ces rôles. Quoique conditions sine qua non, ces mécanismes ne sont cependant pas des conditions suffisantes pour l’avènement de la conscience. Nous terminons en montrant que leur nécessité découle du fait que la conscience n’est pas seulement une faculté individuelle mais est principalement une construction sociale. Aussi, s’il n’est pas exclu la conscience artificielle voie le jour, cet avènement ne saurait se faire qu’après la constitution d’une société de robots.
Consciousness Beyond the Now: a Phenomenology of the Future
Craig Mackie, University of Ottawa, Canada
“Future is a contradictory study for phenomenology, it entails going outside the giveness of phenomena.” - James Mensch In Edmund Husserl’s lectures on time-consciousness, On the Phenomenology of the Consciousness of Internal Time (1893-1917), there arises a new construal of phases in the flow of consciousness: retention-now-protention. Unified consciousness, say of an experience of a melody, is constituted by a type of synthesis of intentionality. This synthesis is founded by the process of internal time-consciousness. The structure of Husserl’s time-consciousness is such that consciousness is self-constituting and pre-reflectively self-aware. Objects are retained in what Husserl calls primary memory, they are present and known within the continuum of given nows, held though an act of retention (this is different from remembering a represented object in secondary memory) in order to maintain continuity with the now. They are also anticipated in an act of protention. Thus we have Husserl’s borrowed metaphor of a double tailed comment, the locus in the middle represents the now, the preceding and receding tails represent protention and retention. This is the structure of the intentionality of internal time-consciousness. What happens when consciousness of an object, say a melody, is stopped? Displaced? We become acutely aware of the directionality of consciousness oriented to the future. This is an apperceptive act: it is based on an experience that has not happened yet. Thus, the pull of consciousness towards the future, through the structure of the fulfillment of protention, is another way in which consciousness is self-founding, self-constituting. My question: how can consciousness, at its very base, be directionally motivated by an act that presumes the existence of an experience that is non-existent, i.e. yet to come? I will demonstrate that, within Husserl’s structural analysis of retention-primary givenness-protention, consciousness is structurally self-supporting such that the possibility of apperception is self-propelled. I will draw on papers from the current phenomenologist Dan Zahavi, Tonie Kortoons and James Mensch.
Diversité culturelle et réutilisation d'objets d'apprentissage
Isabelle Savard, TELUQ-UQAM, Canada
La problématique que je souhaite étudier se rattache aux grandes questions touchant le « E-Learning », le partage des savoirs et des Objets d’Apprentissage (OA)1 interopérables. L’interopérabilité et la réutilisabilité des ressources pédagogiques offrent des occasions de partage des plus enrichissantes aux concepteurs pédagogiques des quatre coins du monde et permettent d’espérer un meilleur accès à du matériel pédagogique de qualité et significatif pour les apprenants des pays moins favorisés.Toutefois, il reste à déterminer quelle place aura la diversité culturelle dans ces environnements qu’on semble vouloir les plus uniformes possible. Il importe de trouver des façons de répertorier ces OA afin d’assurer leur réutilisation non seulement par un plus grand nombre d’utilisateurs possible mais aussi en tenant compte de leurs origines diversifiées.D’une part, plusieurs questions se posent quant aux possibilités d’intégrer la dimension culturelle. D’autre part, on se demande dans quelle mesure ces considérations viendraient changer la tâche du concepteur pédagogique qui devrait tenir compte de la diversité culturelle. Ces questions suscitent des réflexions aux sujets de l’importance du contexte social et culturel de l’apprenant, des stratégies d’enseignement et d’apprentissage au sein des futurs environnements pédagogiques et du nouveau rôle attendu des concepteurs pédagogiques, qui méritent d’être approfondies. Notre projet de recherche vise à contribuer à ces réflexions.Un stage de recherche sera effectué au Virtual Centre for Innovative Learning Technologies (VCILT) [http://vcampus.uom.ac.mu] de l’Île Maurice et la présentation fera état des hypothèses vérifiées lors de ce stage et des résultats obtenus. L’objectif général du stage est de recueillir des éléments nécessaires à la modélisation des connaissances sur la diversité culturelle Il devrait permettre notamment : de recueillir des données au sujet des pratiques des concepteurs pédagogiques des différents continents notamment en matière de réutilisation d’Objets d’Apprentissage (OA) et de démarche de scénarisation et de recenser des facteurs culturels qui peuvent influencer l’apprentissage dans la pratique des concepteurs pédagogiques et les caractériser en vue de les modéliser.1. Selon la définition adoptée par IMS, Global Learning Consortium on peut considérer un objet d’apprentissage comme étant toute ressource ou tout actif qui peut servir de soutien à l’apprentissage. En général, une ressource est considérée comme un objet d’apprentissage quand des métadonnées d’objet d’apprentissage y sont associées et qu’elle peut être découverte dans une banque d’objets numériques et affichée à l’aide d’une application de téléapprentissage. (traduction récupérée du site d’Edusource http://edusource.licef.teluq.uquebec.ca/ese/fr/glossary.htm)
Entre acteur situé et gestion de projet distribué : du soi à l’intelligence collective
Pierre-Luc Lalonde, École Polytechnique de Montréal., Canada
La gestion de projet est une des formes d’organisation le plus en vue des dernières décennies (Hazebroucq & Badot, 1996). En fait, le projet est de plus en plus considéré comme le principal outil d’innovation et de changement au sein des organisations (Cleland & Ireland, 2001). En ce qui a trait au débat entourant l’intelligence collective et l’innovation sociale (Midler, 1995), la gestion de projet apparaît en tant que discipline incontournable des sciences de gestion. En effet, comment les acteurs organisationnels, autrefois cloisonnés dans leur corps de métier propre, construisent-ils les connaissances nécessaires à la réalisation des projets? Un vecteur de cette nécessité propre à la construction du savoir est sans aucun doute lié aux dimensions de la communication et de la distribution de l’information. Une importante étude de cas international concernant un projet distribué entre des firmes américaines, canadiennes et italiennes, a permis de se pencher sur les pratiques de communication mises en place lors du projet. Notre étude permet d’une part d’illustrer le réseau d’information qu’ont construit les acteurs et d’autre part de souligner les biais occasionnés par une telle configuration du réseau d’acteurs. L’étude de cas suggère que la centralisation des informations, jumelée aux modes d’interactions du gestionnaire avec les principaux acteurs du projet, aurait occasionné l’évitement d’une forme ou l’autre de délibération collective. À l’instar des maints chercheurs dans le domaine de la gestion, nous croyons qu’il faille favoriser des « lieux » de délibération - des « espaces de discussion » (Detchessahar, 2001) - afin d’enrichir les savoirs nécessaires à la conduite des projets organisationnels. Toutefois, nous pensons aussi que les résultats apportés par de tels lieux susceptibles de favoriser la réflexion collective, donc de lutter contre la myopie collective, seront limités tant et aussi longtemps que des changements profonds au sein des sciences de gestion ne prendront pas racine (Lalonde, 2005). Nous montrerons comment le développement des sciences de la communication, comment le passage d’un modèle de la communication basé sur le modèle de l’émetteur-récepteur – fondé par les pionniers de la cybernétique et notablement inscrit au sein des sciences de gestion – au modèle de la pragmatique communicationnelle principalement attribué aux efforts de Watzlawick (1973) mais développé par l’équipe multidisciplinaire de l’École de Palo Alto, nous permet de suggérer la position hic et nunc, éminemment située pour ne pas dire existentielle, des acteurs du projet. Cette insertion du modèle watzlawickien de la communication au sein des sciences de la gestion nous positionnera face à une obligation en vertu de laquelle un détour devient incontournable. La présentation se clora sur les possibilités contenues à même ce détour, celui-là constitué par l’anthropologie philosophique, pour penser l’innovation sociale et l’intelligence collective.
Étude des stratégies cognitives mises en place par les sujets lors de la recherche d'informations sur Internet
Emma Holder, Université René Descartes – Paris 5, France
Ces dernières années, le public concerné par l’accès à Internet s’est considérablement étendu et diversifié et les recherches les plus récentes sont nombreuses à insister sur la nécessité de s’appuyer sur une approche plus différentielle et qualitative des comportements de navigation. Notre travail est ainsi centré sur l’étude des stratégies cognitives individuelles mises en place par les sujets au cours de quatre tâches de recherche (deux portant sur des sites de commerce électronique présélectionnés, deux autres portant sur l’ensemble du Web). Nous avons donc, en premier lieu, établi un modèle procédural par sujet et par tâche en fonction du cheminement suivi jusqu’à l’information cible pour ensuite regrouper ces modèles en sous-ensembles architecturalement homogènes et constituer enfin six catégories générales de recherche sur Internet. Il semble que le choix entre ces différents cheminements dépende du type de tâche (Web vs. Sites) et qu’il existe une forte variabilité inter et intra individuelle. Nos résultats suggèrent notamment que l’expérience antérieure d’Internet permet aux sujets les plus expérimentés, non pas de se forger une unique méthode de résolution, mais bien d’acquérir une « métaconnaissance » du fonctionnement du Web. Nous avons également étudié les représentations que les sujets ont des sites et du Web au cours même de leur recherche, et non plus a posteriori. Nous avons ainsi relevé le nombre d’anticipations correctes émises par les sujets, à chaque clic ou appui sur le clavier concernant la page à venir. Les résultats montrent qu’il existe une liaison entre qualité des anticipations et vitesse d’exécution d’une part, et expérience antérieure d’Internet d’autre part. Enfin, d’une manière générale, les stratégies les plus fréquemment mises en œuvre sont celles qui donnent lieu à la plus forte proportion d’anticipations exactes.
Gestural Communication in Nonhuman Primates: An investigation of intentions
Helen Karpouzos, York University, Canada
This conference focuses on nonhuman primate gestural communication. More specifically, examining intentional gestural communication and intentional understanding in orangutans and gorillas at the Toronto zoo. The notion of understanding intentions and mental states in others is argued to be a uniquely human capacity (Suddendorf & Whiten, 2001). However, intriguing evidence from nonhuman primate studies has suggested otherwise. If behaviors accepted as intentional states in humans are also exhibited by nonhuman primates, where the cognitive divide falls with respect to understanding intentional states comes into question. The first question asks when humans themselves come to appreciate intentional states. To determine whether prelinguisitc infants understand intentions in others, pursuit of implicit investigations has been ongoing. The second question asks what gestures nonhuman primates use in communicative contexts and how they differ from those of human infants. Most reasonably, they do not. Therefore, human infancy researchers must either deny that infants who gesture comparably to nonhuman primates do not understand intentions or that nonhuman primates do. By comparing intentional acts of human infants and what is accepted by human infancy researchers as indicative of intentional communication to those of nonhuman primates, claims that nonhuman primates do not behave intentionally will be refuted.
Lessons from decompiling an embodied cognitive model
Audrey Girouard, Tufts University, USA
Noah W. Smith, Tufts University, USA
Frank E. Ritter, The Pennsylvania State University, USA
Cognitive models and intelligent agents are becoming more complex and pervasive. It is time again to consider high-level behavior representation languages and development environments that make it easier to create, share, and reuse cognitive models. One of these languages is Herbal, a high-level behavior representation language. Users represent knowledge in Protégé, an ontology editor. Herbal compiles this knowledge into cognitive models in Soar, a rule-based cognitive architecture. Herbal includes the ability to automatically link the resulting models to dTank, a simple, distributed tank game co-developed with Herbal. To understand the theoretical implications of the process of compiling cognitive models from high-level descriptions more clearly, we generated by hand the Herbal high-level description of a well-written, medium-sized (50 rule) Soar model that plays dTank. This process is a type of decompilation process, of going from low- to high-level language, that yields lessons for both the compiler and the process of modeling. Many of the constructs in the model were supported by Herbal, particularly elaborations and simple and regular actions. In some cases, Herbal’s representation prevents the user from generating incomplete, incorrect or atheoretical code—we saw hand written code that cannot be generated by Herbal because it is incorrect or overgeneral. This process also highlights problems with Herbal. Certain types of theoretically sound hand-written rules do not yet possess an exact translation to the high-level language (mostly knowledge about which action to chose, and links across known representations). We have several suggestions for constructs to be added. This process of decompilation illustrates how users are creating models and could do so more easily and less error prone with more appropriate languages, in addition to helping develop Herbal, and, if automated, this decompilation process done by hand could lead to a decompilation feature in Herbal to help explain raw Soar code.
Maladie d'Alzheimer et Environnements Virtuels Évocateurs: Évaluation de la Mémoire Autobiographique et des apports thérapeutiques d'une nouvelle forme de thérapie de type revue de vie.
Stéphane Protat, Université de Sherbrooke, Canada
Aujourd’hui, nous ne pouvons ignorer que la Maladie d'Alzheimer (MA) présente aussi des répercussions sur le fonctionnement psychologique et social de la personne, en altérant sa sensation d'identité et ses capacités de communication. Entité cognitive préférentiellement préservée aux premiers stades de la MA (Piolino et al., 2003), la Mémoire Autobiographique du passé lointain (MAuto) représente à nos yeux le meilleur modèle mnésique nous permettant de nous risquer à l’interdisciplinarité. En effet, la MAuto ne symbolise pas un simple contenant d’informations, mais constitue un module dynamique, nous obligeant à nous tourner autant vers des notions neurologiques, psychodynamiques que cognitives et sociales (Welzer et Markowitsch, 2005). L’objectif va être ici d'évaluer l'efficacité de l'emploi de nouvelles technologies – que nous appellerons Environnements Virtuels Stimulants (EVS) – comme une forme moderne de Thérapie de Revue de Vie (Finnema et al., 2000) pouvant maximiser la stimulation cognitive de la MAuto grâce à l'immersion sensorielle. Étant donné que les environnements virtuels sont capables de délivrer un large panel de stimuli écologiques, alors lorsque la mémoire fait défaut, ces environnements enrichis viennent la suppléer en lui fournissant des indices nécessaires à l'adaptation du sujet (Vincelli et al., 2002). Les quelques études portant sur les thérapies de réalités virtuelles et la MA rapportent que cette population peut bénéficier d'améliorations cognitives, psychoaffectives et relationnelles (Hoffmann et al., 2003). Nous comprenons que ce projet de mise au point d’un nouvel outil clinique de thérapie nous fournira des éléments intéressants comme entre autre, l’évaluation écologique de la MAuto, la mise en évidence des effets positifs attendus sur le bien-être et la sensation d’identité du malade, mais fait aussi appel à des considérations épistémologiques alternatives à l’instar de la vision neurophénoménologique de Varéla (Varéla et al., 1993) nous permettant d’enrichir ces résultats d’une réflexion « méta». Bibliographie : Finnema E, et al. The effects of emotion-oriented approaches in the care for personssuffering from dementia: a review of the literature. International Journal of GeriatricPsychiatry. 2000; 15(4):141-161. Hofmann M, et al. Interactive Computer_Training as a Therapeutic Tool in Alzheimer'sDisease. Comprehensive Psychiatry. 2003; 44(3):213-219. Piolino P, et al. Autobiographical memory and autonoetic consciousness: tripledissociation in neurodegenerative diseases. Brain. 2003; 126:2203-2219.Varéla F., Thomson E., et Rosch E. L’inscription corporelle de l’esprit. Seuil. 1993. 377p.Vincelli F, Molinari E, Riva G. The communication between therapist and patient in virtual reality: the role of mediation played by computer technology. Stud Health Technol Inform. 2002;85:549-51.Welzer H and Markowitsch HJ. 2005. Towards a bio-psycho-social model of autobiographical memory. Memory, 13(1): 63-78.
Situated Cognition, The Extended Mind Thesis and The subject of Cognition
Yual Chiek, University of Alberta, Canada
This paper has two major aims. First, I argue that the situatedness of cognition is a defensible position.Second, I argue that if cognition is understood to be situated, it must also be taken to be extended; thatis, the only way to properly understand situated cognition is by thinking of some cognitive acts as beingconstituted by a proper coupling between the brain, and features of the environment – cognition isextended. To this end, I criticize Richard Rupert's position that embedded cognition is preferable to theextended mind thesis. To support the first claim, I discuss cases of morphological computation fromneuroethology and argue that morphological computation provides a picture in which the organism isan aggregate of interactive structures; structures that interact with the environment through increasinglyweaker causal and spatial couplings, with intra-cranial structures having weaker couplings than extra-cranial structures. This “structures all the way down” view has interesting implications for theconscious subject of cognition. One implication is that the conscious subject is itself a product of thestructure-environment interactive coupling. If this is correct, it is possible to extend the conscioussubject provided the proper coupling between the in-head structures responsible for consciousness andsome feature of the environment is made.
Synaesthesia: Metaphor and Meaning
Trevor E. Keogh, University College Dublin, Ireland
Susan Robinson, St. Patrick's College, Drumcondra, Dublin 9, Ireland
Synaesthesia is a neurological condition where stimuli in one sensory modality elicit anomalous experiences in another modality, or within a different dimension of the same modality. Theories of synaesthesia are couched primarily in psychological, neuroanatomical, or neurophysiological terms. Psychological theories of synaesthesia operate on the principle of either associative learning or failure of differentiation. The current study provides and overview of these theories and deals centrally with theories of associative learning. In its traditional form, the associative learning theory has been dismissed as implausible, but new research suggests that many synaesthetic correspondences are learned in childhood. A new interpretation of associative learning is offered by arguing that synaesthetic correspondences are formed as a result of meaning mappings between conceptual domains. Metaphor is used to structure the conceptual system and meaning is the primary determinant of what associations are formed. The new interpretation addresses ten key questions hitherto insurmountable to previous theories. The distinction between intra- and inter-modal synaesthesia is also considered. This paper presents a case for understanding synaesthesia as a condition that is greatly influenced by cognitive processes and less so a condition that results from purely neurophysiological factors. It is further argued that synaesthesia is a useful cognitive construct for those that possess it and endures because of it so deeply embedded in the conceptual system. Future research suggested includes the investigation of inducer/concurrent pairing and seeking support for the proposed theory by identifying correspondences (particularly in colour->music synaesthesia) as meaningful and systematic associations and not arbitrary anomalous pairings. Further research could also investigate whether the similarities between synaesthesia and eidetic memory are due to common psychological mechanisms.
The effect of visual distortions on postural stability assessed in a fully immersive virtual reality environment
Jean-Marie Hanssens, Université de Montréal, Canada
Background. In most cases, presbyopic vision is corrected using progressive lenses. In spite of the geometric advanced of such products, many progressive lens wearers have difficulty adapting to them because of the distortions perceived in the bottom of the lenses. The purpose of the present study was to evaluate the influence of the visual distortions induced by progressive lenses on the postural stability of the wearers. Methods. In order to simulate the visual distortions induced by progressive lens wear, twelve young observers (22 to 35 years of age) were placed in a fully immersive virtual reality environment where the virtual floor, defined by a 10 x 10 m black and white checkerboard pattern, oscillated sinusoidally. The amplitude (1 to 4°) and temporal frequency (0.03 to 2.0 Hz) of the oscillation were systematically manipulated (for a total of 21 conditions). Postural reactivity, defined by the amplitude of lateral (side-to-side) sway and the instability index, was measured as a function of virtual checkerboard distortion using magnetic sensors placed at the head and close to the centre of mass of the subjects and transformed using Fast Fourier Transformations. Results were analysed using Split-plot ANOVAs.Results. Results showed that postural instability increased with the amplitude of stimulation, suggesting a direct relationship between amount of postural movement and visual distortions. Furthermore, subjects gained stability as the temporal frequency of the distortions increased. However, even in the high temporal frequency conditions, postural reactivity depended significantly on the distortions. Our results suggest that the adaptation difficulties to the progressive lenses could be related to the visual distortions induced in the lower visual field. Moreover, participants reported feelings of “seasickness” during task execution comparable to what some presbyopes feel during progressives lens wear.
The impact of social interaction on visual recognition memory in infants and adults
Andrew Brown, University of Sheffield, UK
The Visual Paired Comparison Task (VPC) is a widely-used measure of recognitionmemory. In this task, a static image is typically presented for several secondsduring a habituation phase. During the test, the same image is simultaneouslypresented with a new image. A preference for looking at the novel stimulus atthe test is widely reported as the traditional demonstration of recognitionmemory. However, a small body of literature suggests that a preference forlooking at the familiar stimulus can also constitute evidence of recognitionmemory. It has been proposed that stimulus type and presentation style may beimportant factors in determining the type of preference observed at test. Forexample, the typical novelty preference is not observed in 7-month-olds if thehabituation stimulus is an interactive puppet that moves and responds to theinfant (Nachman et al., 1986)In two experiments we examined the effect of interactive stimuli on recognitionmemory in the VPC task. In Experiment 1, twelve 12-month-olds and twelve18-month-olds were habituated to moving puppet presented on video. Only12-month-olds exhibited a novelty preference. The lack of a novelty preferenceat 18-months of age is consistent with changes in social abilities occurring atthis age, rather than an indication of a failure of recognition memory. InExperiment 2, 40 adults were tested in 3 conditions where the amount ofinteraction with the habituation stimulus was varied. In the no-interactioncondition, the traditional novelty preference was observed. However, even thebriefest interaction with the stimulus abolished the novelty preference.
The importance of motor command in agency attribution: active vs. passive movements.
Catherine Preston, University of Nottingham, UK
In order to identify an action as belonging to the self the actor can use information from both the original motor command of the movement and proprioceptive feedback received. Previous literature has shown that reduced proprioceptive feedback can inhibit accuracy in self other judgements (Farrer et al 2003) but, strategies focusing on the motor command or a forward model system, would predict planning of the action to be of greater significance (Blakemore et al, 1999). The aim of this study was to assess the importance of the motor command in agency attribution, using both active and passive movements. Although recent studies have examined this aspect, the movements used were neither goal directed nor physically coincident with the feedback received (e.g. Tsakiris et al. 2004). In the current study, using a robot arm motion tracking device with physically coincident visual feedback, participants were asked to determine whether visual feedback received was synchronous (self) or asynchronous (other) with the movement of their hand. In half of the trials the participant was required to move their hand (holding the robot arm) to one of four targets, in the other half of the trials, the participant’s hand was passively moved by the robot. Judgements for the active movements were found to be significantly more accurate than for the passive movements in 12 neurologically normal participants. This suggests that the motor command of a movement is an important aspect in accurately determining agency.
The Morphological Turn in Cognitive Science: a perspective on New A.I. and embodied, situated cognition
Fernando Almeida e Costa, University of Sussex, UK
During the last 20 years, research in New A.I. has been providing existence proofsfor a major theoretical shift in cognitive science: one which denies the deeply engrainedhuman belief that cognition is something done by the brain.Evolutionary robotics posits the cognitive system as constituted by dynamicallycoupled sub-systems which include body morphodynamics, environmental dynamics andinternal neural dynamics. For example, leg movement coordination in insect-robots(Beer, 1995) doesn’t rely on a coordinating centre, and critically depend on the localinteraction between leg and environment. It is very hard, if not impossible, to attribute aparticular functional property to one of the sub-systems.I propose that this shift is a component of a broader morphological turn that isclearly detectable from the second half of the 20thcentury onwards, in different areas ofresearch. Apparently unrelated examples of its manifestations include early investigationsin A.I., e.g. the studies on morphogenesis and its relation with cognitive structures byAlan Turing, the semiophysics firstly proposed by René Thom, or the morphological, nonlogico-combinatorial structuralism of Claude Lévi-Strauss in cultural anthropology.I will present the general framework of this morphological turn, by developing thetheoretical implications of the three examples given above. Then I will elaborate on theconsequences that can be extracted to cognitive science, from morphodynamicallyinspired robotics, where the cognitive behaviour is a result of the opportunisticexploitation of all intrinsic and extrinsic physical properties available. I will concentrateon the research on passive dynamic walkers, evolvable hardware and morpho-functionalmachines.The broad results for cognitive science entail the denial that the “properly cognitive”level is to be found at an algorithmic level that should constrain and dominate thephysical properties. Cognitive activity is crucially based on the exploitation of allphysical properties available to the agent in its interaction with the environment andactually implies the minimization of control at the algorithmic level. The functionalistprinciple of the irreducibility of the physical medium to the cognitive level is put aside.Thus, the possibility is open for a non reductionist physics of meaning.Finally, this perspective is totally consistent with high-level cognitive abilitiesforming a continuum with low-level ones. The hypothesis of an evolutionary path linkingthe emergence of the particular feet morphologies and, as a consequence, the ability ofwalking, to language, put forward by palaeontologist André Leroi-Gourhan, will bebriefly discussed in this context
The Role of M-capacity and Executive Processes in Mathematical Cognition
Alba Agostino, York University, Canada
Mental attentional capacity (M-capacity) is a general purpose, content free resource for effortfully activating task-relevant schemes/information-units. The power of M-capacity increases with age in childhood and is a main developmental factor limiting the complexity of children’s cognitive processing. Executive function (EF) refers to a broad range of activities that control and regulate cognition, including efficient application of one’s M-capacity. Examples are ability to shift efficiently between mental sets, to monitor and update information in working memory, and to inhibit irrelevant information. Recent studies suggest that insufficient M-capacity and poor EF skills may in part account for the difficulties of children with lower mathematical ability. However, to date, research has not investigated satisfactorily the combined roles of M-capacity and EF in mathematical cognition. Investigators have tended to examine EF with ‘impure’ measures and use math tasks that can be solved without the use of EF (e.g., rote learning). The present study measured M-capacity and three distinct EFs (i.e., updating, shifting, and inhibition) using well-validated measures. The goal was to clarify the relevance of these processes to both developmental and individual differences in mathematical reasoning. One hundred and fifty-five children in grades 3 through 6 completed a battery of EF measures, a series of M-capacity tasks, and a multiplication word-problem test. Preliminary results show that score on the multiplication test is correlated with all the predictors, and most strongly associated with measures of M-capacity and the updating EF. Structural equation modeling will be used to clarify the relationship between M-capacity, EF, and mathematical cognition. We hypothesize that the best model will be one in which EF mediates the relationship between M-capacity and mathematical cognition. Theoretically, EF abilities play a vital role in the development of mathematical reasoning.
Un agent tutoriel cognitif inspiré de l’esprit humain et de sa conscience
Daniel Dubois, UQAM, Canada
Seuls les tuteurs humains parviennent actuellement à interagir avec un apprenant entenant compte de toute sa complexité ainsi que du contexte d’enseignement. Nous émettons l’hy­pothèse qu’en utilisant l’esprit humain comme modèle, nous pouvons réaliser des agents artifi­ciels aux capacités, comportements et attitudes plus proches de ceux des tuteurs humains.Nous utilisons comme base de travail la théorie de Bernard Baars (1988) sur l'esprit et laconscience, ainsi que son implémentation informatique dans l’agent IDA du professeur Franklin(2005). La théorie du Global Workspace décrit l’esprit comme une vaste collection de processussimples et automatiques qui collaborent pour effectuer la cognition. L’émergence des niveauxsupérieurs de l’esprit humain est rendue possible par le phénomène qu’on appelle la conscienced’accès, qui permet, entre autres choses, de faire connaître une information à tous les sous-sys­tèmes inconscients ainsi que de fermer la boucle délibérative.Notre agent cognitif vient enrichir les capacités tutorielles de CanadarmTutor, qui sait queréagir aux actions posées par l’astronaute dans un environnement virtuel reproduisant la Stationspatiale internationale et le bras télémanipulateur canadien Canadarm2. Notre système tutorielcognitif se compose d’un ensemble de modules relativement autonomes qui, d’une part, modéli­sent le domaine d’enseignement et l’apprenant, tant dans son savoir que dans ses caractéristi­ques psychologiques et affectives, et d’autre part, reproduisent les fonctions majeures de l’esprit(perception, mémoires, sélection des actes, etc.). Les microprocessus, qui réalisent informatique­ment les processus simples de Baars, servent entre autres à relier tous ces modules à lamémoire centrale de travail où convergent les informations requérant un traitement explicite"conscient". Les divers niveaux et types de "conscience" contribuent, à divers moments, à lasélection des informations les plus importantes. L’agent devient ainsi capable de tenir compte demultiples sources d’information, mais de n’affecter ses ressources volontaires limitées qu’auxplus importantes.
Understanding our understanding of strategic scenarios: A new model of chess chunks.
Paulo Brum, Getulio Vargas Foundation, Brasil
There is a crucial debate concerning the nature of chess chunks: classical chunking theory and its followers state that chunks are built by encoding particular combinations of pieces-on-squares (POS); that "close" pieces (in an "Euclidean" sense) would form a chunk— with recent amendments postulating templates (based on either particular pieces or particular squares). An alternative theory is that chunks are encoded by abstract, semantic information. In this paper we extend recent experiments, and show that chess players are able to perceive strong similarity between two positions if the pieces retain the same abstract roles in both of them. This places doubt that piece-on-square information should be the key information encoded in chess chunks, and we propose that the key encoding is composed by abstract roles pieces and sets of pieces play—a theoretical standpoint in line with the research program in semantics that places analogy at the core of cognition.
Get Firefox! [Design du site internet : Nicolas Payette]