Conceptuccino


Navigation und Suche der Universität Osnabrück


Hauptinhalt

Topinformationen

Preliminary Program

Preliminary Program of the CARLA Summer School

The preliminary program consists of three types of sessions:

  1. Lectures (red):

    • Robert L. Goldstone: “The Learning and Use of Concepts by Humans”
      Issues related to concepts and categorization are nearly ubiquitous in psychology because of people’s natural tendency to perceive a thing as something. We have a powerful impulse to interpret our world, to “see something as X” rather than simply seeing it. I will discuss research on the nature of human concepts, the function of concepts, how they are represented, and how they are connected to the world and each other.
    • Peter Gärdenfors: “Conceptual Spaces as a Foundation for the Semantics of Word Classes

      The theory of conceptual spaces presents a way of representing the meaning of words in terms of geometry and vector spaces. A key notion is that of a domain, for example the colour domain, consisting of a number of dimensions (hue, brightness and intensity in the colour domain) that have a certain geometric structure. In brief, the meaning of a word can be seen as a convex region in some conceptual domains. In linguistics, word classes are in general defined in syntactic term. I will show that word classes can be given a semantic characterization. In my book The Geometry of Meaning (2014), I have analysed adjectives, nouns, verbs and prepositions in terms of structures in conceptual spaces. The meaning of an adjective is represented by a convex region of a domain. The meaning of a noun is constructed from regions in a number of more or less central domains. Verbs can be divided into manner and result verbs. Manner verbs are analysed in terms of force vectors (or force patterns). Result verbs are vectors representing changes in some domain. Locational prepositions are represented semantically as regions of space and directional prepositions are paths in space, but there are also prepositions that are represented by force vectors. The analysis will be extended to demonstratives, articles and quantificational expressions. The semantic model will be applied to some linguistic phenomena.

    • Nicholas Asher & Julie Hunter: “Concepts and Conceptualization: From the Lexicon to Situated Conversation”
      Our course will examine two areas in which concepts and conceptualization are crucial. The first is lexical semantics. In Lexical Meaning in Context, I argued that lexical meanings have both an external dimension, what has been conceived of as referential or truth conditional semantics modelled by intensions in formal semantics, and an internal dimension, which I argued can be conceived of in terms of a system of semantic types. Such semantic types have a rich structure, a content that can be described in terms of proof-theoretic ideas, and a series of complex interactions with the external semantics. With colleagues I have recently explored modeling such semantic types in distributional semantics. Our course will review some of the highlights and then move to a second area in which concepts and conceptualization are crucial. In recent work “A Formal Semantics for Situated Conversation,” forthcoming in Semantics and Pragmatics, Julie Hunter and I argue that nonlinguistic events can take the place of linguistically expressed bits of content in a discourse structure. The integration of nonlinguistic events into a discourse structure along with linguistically expressed discourse units provides the nonlinguistic events with a particular conceptualization. We will explore this process in our course using some of the machinery developed for lexical semantics.
    • Christiane C. Fellbaum: “A Linguistic Perspective on Concepts”
      “Concept” is a squishy notion and, indeed, some philosophers deny the usefulness of this term. Yet it is common currency among linguists, psychologists and knowledge engineers. Rather than attempting an exhaustive definition, my talks will focus on the question as to how language – in particular the lexicon – reflects the way speakers select and label entities in the world, their properties and events. Adopting a crosslinguistic perspective, the discussion will include children’s acquisition and the representation of lexicalized concepts for applications like computational semantics.
    • Michael Spranger: “Co-acquisition of Concepts and Language: Computational Models and Robot Experiments”
      The course will focus on interactive development of concepts and language. Natural languages show tremendous amount of cross-cultural variation even for cognitively central concepts such as spatial language. This has important consequences for models of language learning. The course introduces the field of computational models of language learning in populations of artificial agents. We explain how artificial agents (and robots) can learn to use gesture for referential tasks and subsequently co-acquire the semantic (concepts) and syntactic structure (grammar) of a language - that is the focus is on multi-modal, interactive concept learning. The atelier focuses in particular on the integration of multiple modalities such as gaze, gesture, vocalizations and speech in models of language emergence. Participants will be 1) introduced to the problem of symbol grounding, 2) how gestures emerge as a communicative tool in artificial learner agents 3) how to build models for lexicon learning 4) how grammar and complex semantics can be co-acquired.
    • Evening Lecture by Max Garagnani: "Action-perception circuits for word learning, conceptual grounding, and beyond: a unifying neurocomputational account"
      Embodied semantic theories maintain that word meaning is grounded in the perception and action systems of the human brain. These theories are supported by a growing number of experimental results, indicating that processing of words which belong to specific semantic categories leads to selective activation of modality-preferential areas. Neuroimaging data,
      however, also show that a number of multi-modal brain areas are consistently activated by word and sentence comprehension processes, regardless of the stimuli’s semantic category. This suggests the existence of general semantic centres (or “hubs”) involved in processing the  meaning of all word types. The presence of both category-specific and category-general semantic areas begs for a unifying mechanistic explanation, able to reconcile the emergence of both in the cortex. In this talk I will present a neurocomputational model which provides such an explanation at cortical-circuit level, in terms of neural mechanisms underlying conceptual grounding in action and perception. The model’s main distinguishing features are that: (i) it replicates connectivity and anatomical structure of the relevant brain areas, and (ii) it uses only mechanisms that reflect known cellular- and synaptic-level properties of the cortex. In addition to integrating and elucidating the emergence of both category-specific and category-general semantic areas in the cortex, I will argue that the same neurocomputational architecture goes a long way in explaining a range of experimental data in the language as well as other domains, making a number of novel predictions which receive support from recent neuroimaging results. On these premises, I will highlight elements of a unifying theory of cognition based on action-perception circuits whose emergence, dynamics and interactions are grounded in known neuroanatomy and neurobiology.

  2. Interactive Sessions (green):

    • Welcome
      In addition to providing general organizational information, this session allows participants to introduce themselves, their affiliations, and their research areas. They will also have the chance to raise interesting research questions they would like to address in the course of the summer school.
    • Discussion Groups
      Three discussion sessions will focus on one of the three focus topics each. In each session, the participants will form small groups of five persons, discuss a given set of questions for 60 minutes, and then present the results to the plenary for discussion (30 minutes). This format fosters the active involvement of the participants and an interactive learning atmosphere. Immediate feedback from peers and lecturers will benefit both the learning and the research of the participants.
    • Research Forum
      In the first 30 minutes of this session, each participant will sketch the research question on which he or she is currently working. Afterward, the participants will split up into small groups of five persons and will use the remaining time to discuss these questions. In this session, the participants will receive useful feedback on their own research from peers with different backgrounds. Moreover, they will be able to relate their own research to the research that their peers conduct as well as to the overall topics discussed during the summer school.
    • Panel Discussions
      Two moderated discussions will take place with all invited lecturers on a panel and with questions collected from participants in advance with regard to (but not restricted to) the three focus topics. The goal is to help the participants to better understand the scope and limits of the different approaches to concepts and to form or improve their own research questions and methods.
    • Networking Session
      An informal plenary discussion on existing academic events, research institutions, and networks (e.g., mailing lists) concerning research on concepts will take place. Moreover, we will discuss opportunities for future collaborations among the participants and potential follow-up events, for instance, a DFG-funded research network. This will allow the participants to discover research collaboration possibilities within and outside of their respective fields.
    • Visionary Outlook
      In this session, lecturers will outline in a short presentation (of roughly five minutes each) current challenges and possible future directions of concept research from their perspectives. The goal is to equip the participants with further ideas for their future research.

  3. Social Program (blue):

    • Reception
      The reception at the Felix-Nussbaum-Haus provides participants with the opportunity to get in informal contact with their peers and to look at some great artwork.
    • Evening City Tour
      A guided tour through the old city center of Osnabrück.
    • Excursion to Tecklenburg
      The excursion to Tecklenburg offers opportunities to relax, to deepen contacts with other participants, and to enjoy a beautiful landscape.