DCS 672 - Meaning Machines

Fall 2004


Main Page Contents
Schedule
Requirements
News
About this course
Syllabus
Materials

Schedule

Class
Wednesday 1:10-4:10, Hill 120.

Office Hours
Tuesday 3-5, Psych A103.


Requirements

Weekly readings, serving as a principal discussant of suitable papers during the semester, and a final project involving a paper and/or programming. Interdisciplinary project collaborations are especially welcomed.


News

  • Thu, Dec 2
    Slides posted.
    Final project deadline extended, to Dec 15.

  • Mon, Nov 15
    Slides posted.

  • Thu, Nov 4
    Slides posted.

  • Mon, Oct 25
    Slides posted.
    Alternative readings for week of Nov 12 (clarification).

  • Fri, Oct 22
    Guidelines for breakout session posted.
    Guidelines for project proposal posted.

  • Thu, Oct 14
    Slides posted.

  • Thu, Oct 7
    Handout posted; slight adjustment of schedule.

  • Thu, Sep 30
    Slides posted.

  • Tue, Sep 21
    Readings for Stalnaker and Lewis posted.

  • Mon, Sep 20
    Slides posted.
    New PDF for readings files with better-quality scanning
    (no paper feed glitches).

  • Thu, Sep 9
    Slides posted.

  • Fri, Sep 3
    Class date moved.
    Rough copies of Ballard and Brennan readings.

  • Thu, Aug 18
    This page created.

    Class meeting time is still somewhat flexible: Tuesday and Wednesday afternoons and Wednesday mornings are also candidates. If you have a conflict with the set meeting time let me know, and indicate your availability for possible alternatives.

    Most readings are available here; in some cases, I'm still tracking down clean copies.
    Where possible, I link to the publisher's electronic copy.
    Otherwise, I link to the author's.
    As a last resort, I link to a locally scanned version. This will be accessible within the rutgers.edu domain only. Scanned material will also be made available through electronic reserve.

    Not every page of every reading is crucial. As readings get close, I'll highlight the key issues and key passages that I think it makes most sense to focus on.


About this Course

This course looks at semantics specifically from the perspective of cognitive science. In cognitive science, the project of semantics is to explain how intelligent agents can come to share mental content using natural language. This project folds together investigations of mental content normally pursued in the philosophy of mind, investigations of the structure of knowledge of language normally pursued in linguistics, and investigations of embodied perception and action normally pursued in psychology.

Two complementary ideas motivate this integrated investigation. The first is the suggestion that mental content is indexical. In other words, an agent's mental representations derive their meanings from the agent's interactions with its environment, not from the agent's general abilities to perform inference or make decisions. The second is the suggestion that language use is collaborative. In other words, when we exchange meaningful utterances, what we are doing, fundamentally, is working together to achieve a shared understanding of what we say in context.

If these suggestions are right, agents' knowledge of language cannot be separated from their experience in the world or from their abilities to interact cooperatively with other agents. Put most contentiously, we will only have an adequate scientific account of linguistic meaning when we can build machines that work with us to capture the meanings of our utterances in representations that are suitably grounded in the machines' own perception and action. Hence the title of this class: Meaning Machines a phrase due to Deb Roy, Assistant Professor at the Media Lab, MIT.

Computation provides the setting for this investigation in part, of course, because these meaning machines hold practical interest. More importantly, computation provides a rigorous methodology for formulating theories in cognitive science and an attractive framework for cognitive modeling. Every cognitive scientist should aspire to a concrete understanding of programs and processes in explanations of mental intentionality and intelligent behavior.


Syllabus