Group Photo
Symbols, Embodiment, and Meaning
A workshop and debate
December 16-18, 2005
University of La Laguna
Tenerife, Spain
Sponsors:
|
Organizing Committee:
Dr. José M. Díaz (jmdiaz@ull.es)
Dr. Hipólito Marrero (hmarrero@ull.es)
Mabel Urrutia (murrutia@ull.es)
Yurena Morera (ymorera@ull.es)
Vicente Moreno (vmoreno@ull.es)
|
Read: Framing_the_debate.pdf
Chairpersons and Affiliations
Meeting overview
Program Schedule
-Day 1: Embodiment and grounding of language
-Day 2: Embodiment and the brain
-Day 3: Embodiment and computers
Participants
Post-docs, Students and other attendees
Instructions for Papers
Sessions, Speakers, and Posters
Travel and Lodging information
-Arrival at Tenerife
-Site of the workshop
Funding for Post-docs/Graduate Students
Chairpersons and Affiliations
Manuel de Vega
Professor of Psychology
Department of Cognitive Psychology
University of La Laguna
Campus de Guajara
38205 La Laguna, Tenerife, Spain
Phone: +34 922 317511
Fax: +34 922 317461
Email: mdevega@ull.es
Arthur Glenberg
Professor of Psychology
University of Wisconsin-Madison
1202 W. Johnson St.
Madison, WI 53706, US
Phone: +1 608 262 8992
Email: glenberg@wisc.edu
Arthur Graesser
Professor of Psychology
Department of Psychology
202 Psychology Building
University of Memphis
Memphis, TN, 38152-3230, US
Phone: +1 901 6782742
Email: a-graesser@memphis.edu
Meeting overview
One of the most striking problems in cognitive sciences is to understand how meaning arises in our minds. This workshop is an interdisciplinary effort to explore the cognitive, neurological, and computational nature of meaning. The issues are relevant both on theoretical and technological grounds. As an illustration of the latter, if we want to design a computer program that uses natural language in a creative and flexible way (such as in a conversation), then we need to provide the system with a sophisticated capability to process meaning and world knowledge.
As often happens with some fundamental concepts in science, we all have an intuitive notion of meaning (or several polysemically related notions of meaning), but an adequate scientific specification is more evasive than would be expected. For the layman, meaning is an obvious and almost trivial property of language. Words and sentences "have" meaning themselves. For instance, we think of dictionaries as huge collections of word meanings. However, the impression that meaning is just "out there" is possibly misleading. Many researchers would argue that meaning involves human or biological activity, rather than simply pointing to external physical symbols, so unveiling the underlying structures and processes of meaning is an extremely challenging task. There are two traditional approaches that markedly differ: (1) Meaning is symbolic and (2) meaning is embodied.
An extreme version of the symbolic conception of meaning has deep roots in formal logic and early models in computational sciences. It has been applied to explain a variety of phenomena, including visual processing, imagery, motor skills, semantic memory, problem solving, reasoning, and discourse comprehension. The radical symbolists claim that meaning is represented in a universal format, involving amodal, discrete symbols, which are governed by syntactic rules or production rules. By contrast, scholars who adopt an extreme version of the embodiment approach postulate that meaning is entirely grounded in perception and action and that brain areas partially overlap with such processes. That is, meaning is always modality-specific (e.g., visual, auditory, motor or emotional) rather than amodal; it is analogically related to its referent rather than arbitrary; and it is productive by means of mental simulations situated in a world rather than formal syntax. As one might expect, there are various intermediate positions between these two extremes. There are symbolic theories that are grounded in perception and action, and that are governed by production rules that construct mental simulations of experience. There are embodied theories that accommodate representations that have minimal specifications of perception and action.
The two research communities not only assume different theoretical accounts of meaning, but they also differ in their methodological preferences, their research programs, technical vocabulary, and even their preferred criteria of theoretical validation. Both communities run psychological experiments with human participants to produce empirical demonstrations of their claims. However, it is more frequent for symbolists to provide formal evidence of their theories, including analytical derivations, cognitive modelling, computational tests, and computer simulation of complex systems (as in the fields of artificial intelligence and artificial life). The complexity of the embodied approach has made such quantitative treatments too high of a bar, although some recent models have made serious headway. It is somewhat more frequent for researchers in the embodied communities to scrutinize brain mechanisms and neuroscience methods, although there are those in the symbolic camp that also are exploring biologically inspired computing. One problem with both the symbolic and embodiment approaches is that their postulates too often remain largely implicit, so the mutual criticisms rely on straw men, rather than on a rigorous analysis of the approaches. The proposed workshop will stimulate a formal debate of ideas among symbolists and the embodiment defenders by creating a forum in which representative specialists from both sides converge. In addition, the workshop will present opportunities for the design of joint research projects to test the contrasting proposals and to advance theoretical development.
Program Schedule
The workshop will last 3 full days. Each of the three days will be devoted to a morning and an afternoon debate, plus time for two sessions that report especially illuminating empirical research. Poster sessions will also be organized daily. Participants in a debate will provide, at least four weeks in advance of the workshop, a position paper that can be read by all participants in the workshop. The paper will develop the speaker's position and provide definitions, citations, data, and technical analyses that would be difficult to convey in a time-limited debate format. The position paper will be analogous to a legal document introduced into a court: evidence available to both sides of the debate.
In addition to formal debates, the workshop will include time for presentation of some ground-breaking empirical research, and informal discussions among the debaters. The poster sessions will allow the graduate and post-doctoral students to present their own research, which will be also discussed by all the participants. Finally, the framework of the workshop will allow for discussions of future collaborative research.
Three main topics will be debated, one on each of the three days of the workshop: (1) Embodiment and grounding of language, (2) embodiment and the brain, and (3) embodiment and computers. Below a more detailed descriptions of these topics, as well as an outline of the program. The individuals assigned to the topics have all agreed to participate. However, we are in the process of securing commitments from other researchers as well.
Day 1: Embodiment and grounding of language
This debate on Embodiment and grounding of language is related to Searle's Chinese room argument. The argument is meant to demonstrate that symbol systems, including language, need to be grounded to be proper, interpretable representations of meaning. The debate topic is the extent to which language must be grounded and the extent to which embodiment theory provides a basis for such grounding. Thus, each side in the debate will choose and defend one of the following positions:
1. Searle is wrong: symbol systems need not be grounded to represent or produce meaning.
2. Language needs to be grounded, but the mechanism of grounding plays little role in most language processing.
3. Searle's argument is no longer relevant to contemporary symbolic models that are complex, multilayered at coarse and fine grain levels, and grounded in perception and action. Searle's argument is either incorrect or indeterminate.
4. Grounding language in embodied representations is important for understanding some language (e.g., about concrete situations), but not others (e.g., descriptions of abstract situations).
5. All levels of language understanding are grounded in action and perception.
Day 2: Embodiment and the brain
Neurosciences are contributing to understanding of the brain's architecture for language, including linguistic meaning, with a level of detail previously unknown. The brain itself is an organ of the body, and thus, in a trivial sense, human cognition is necessarily embodied. However, beyond this truism, could the neuroscience data illuminate the current debate?
1. Word meaning activates brain regions that partially overlap those responsible of perception and action. Therefore, the embodiment of meaning has been empirically demonstrated.
2. Neurological data do not solve the question because, even if the processing of word meaning overlaps sensory-motor areas in the brain, this fact does not preclude that the processes themselves are symbolic.
3. In addition to the above claim, the classical perisylvian areas of language in the left brain hemisphere perform computations that are more symbolic-like (morpho-syntactical) than embodied.
Day 3: Embodiment and computers
Some approaches to embodiment propose that only biological systems can be embodied, whereas other approaches admit the possibility of embodied computer agents/robots. A related question is the degree to which an embodied system can be simulated on a computer. Each side in the debate will choose and defend one of the following positions:
1. Because computers are not biological systems, they cannot be embodied or even simulate embodied cognition. Nonetheless, computers might be able to model embodied cognition in the same sense that a computer can model a thunderstorm by solving complex equations.
2. A computer program can simulate embodied cognition in humans, but cannot be literally embodied in the same way that human meaning is embodied.
3. A computer program can be embodied, but the nature of the embodiment is different from humans because the world experienced by the computer is different.
4. For a computational cognitive architecture for modeling human behavior to be satisfactorily accurate and complete, not only must perceptual and motor constraints be incorporated, but also mental activities that appear to be "purely cognitive" must be modeled in terms of perceptual and motor representations and processing.
5. A computer/robot can be fully embodied in the same way that meaning is embodied in humans.
Participants
(Note: There will be no public announcement. All participants are invited individually.)
Louise Antony
Department of Philosophy
Ohio State University
Lawrence W. Barsalou
Department of Psychology
Emory University
Manuel de Vega
Department of Cognitive Psychology
University of La Laguna
Arthur Glenberg
Department of Psychology
U Wisconsin-Madison
Robert Goldstone
Department of Psychology
Indiana University-Bloomington
Antoni Gomila
Department of Psychology
University of the Balearic Islands
Arthur Graesser
Department of Psychology
University of Memphis
Marcel Just
Department of Psychology
Carnegie Mellon University
David E. Kieras
Dept of Electrical Engineering & Dept of Computer Science
University of Michigan
Walter Kintsch
Institute of Cognitive Science
University of Colorado-Boulder
Andreas Knoblauch
Department of Neural Information Processing
University of Ulm
Max Louwerse
Dept of Psychology/Institute for Intelligent Systems
University of Memphis
Mitchell Nathan
School of Education
University of Wisconsin-Madison
Friedemann Pulvermüller
MRC-Cognition and Brain Sciences Unit
Cambridge University
Francis Quek
Department of Computer Science
Virginia Tech
Mike Rinck
Department of Clinical Psychology
Radboud University Nijmegen
Deb Roy
MIT Media Laboratory
Anthony J. Sanford
Department of Psychology
University of Glasgow
Larry Shapiro
Department of Philosophy
University of Wisconsin-Madison
Luc Steels
Department of Computer Science
University of Brussels
Rolf Zwaan
Department of Psychology
Florida State University
Post-docs, Students and other attendees
Alberto Avilés (U Murcia)
Manuel G. Calvo, Ph.D. (U La Laguna)
Manuel Carreiras, Ph.D. (U La Laguna)
Alberto Dominguez, Ph.D. (U La Laguna)
Elena Gámez, Ph.D. (U La Laguna)
Peter Gorniak (MIT)
David Havas (U Wisconsin-Madison)
Tanner Jackson (U Memphis)
Patrick Jeuniaux (U Memphis)
Mike Jones (U Colorado-Boulder)
Johanna Kaakinen, Ph.D. (Florida State U)
Immaculada Leon, Ph.D. (U La Laguna)
Marc Ouellet (U Granada)
Rupal Patel (MIT Media Laboratory)
Shirley-Ann Rueschemeyer, Ph.D. (Max Planck Institute, Leipzig)
Julio Santiago, Ph.D. (U Granada)
Yury Shtyrov, Ph.D. (MRC-CBU)
Ji Son (Indiana U-Bloomington)
Mitchell Valdés, Ph.D.(Director of the Centro de Neurociencias de Cuba)
Instructions for papers
1. At the request of several participants, DeVega, Glenberg, & Graesser will write a brief general paper having to do mainly with terminology and issues. The point will be to have a common reference point for people to agree with, modify, or disagree with. This paper will be sent to you by October 15.
2. For all other papers:
a) address one or more of the issues we have laid out in proposal.
b) there should be explicit agreement with terminology paper by De Vega, Glenberg, & Graesser, or explication of your stance
c) suggestions for resolving one or more of the issues
d) optional: future challenges and prospects for embodiment theory and symbol theory
e) a draft of your paper is to be ready to distribute to others by November 15. The point of the paper is two-fold. The first is to be able to present data/argument that may not be amenable to oral presentation, and then to be able to refer to the paper during your talk. The second is to have a start on what is sure to become a best seller for Erlbaum. As you prepare your paper, please shoot for 3000-5000 words.
Sessions and speakers
1. There will be six sessions over three days, each led by a group of speakers. An Open Discussion session will conclude the Day 3 schedule. The sessions are scheduled as follows:
Schedule updated Dec. 7
Day 1
i. 9-11:30 Cognition 1: Glenberg, Graesser, Shapiro
11:30-12:30 Discussion (& coffee)
ii. 2-4:00 Neuroscience: Just, Pulvermüller, Knoblauch
4:00 - 5:30 Discussion (& coffee)
Day 2
iii. 9-11:30 Language 1: Kintsch, Zwaan, Sanford
11:30-12:30 Discussion (& coffee)
iv. 2-4:30 Robotics/AI/Cognitive Modeling: Quek, Roy, Steels
4:30 - 5:30 Discussion (& coffee)
Day 3
v. 9-11:30 Language 2: Barsalou, de Vega, Louwerse
11:30-12:30 Discussion (& coffee)
vi. 2-4:30 Cognition 2: Goldstone, Gomila, Nathan
3:30 - 5:30 Discussion (& coffee)
2. We have two poster sessions scheduled for 5:30 - 7:30 each evening. However, there is no need to be constrained by the confines of typical poster sessions. Instead, think of these times as an opportunity to present new research, new ideas, demonstrations, and so on, in whatever format would be most useful.
Day 1 Poster Session:
Shtyrov & Pulvermüller: A neurobiologically-grounded model of syntactic processes
Jones & Kintsch: Holographic representation of meaning.
Julio Santiago, Marc Ouellet, Antonio Román: Time (also) flies from left to right
Julio Santiago, Ana Torralbo & Juan Lupiáñez: Flexible conceptual projection of time onto spatial frames of reference
David C. Landy: The Role of Perceptual Grouping in Building Abstract Objects
Day 2 Poster Session:
Patrick Jeuniaux, Max Louwerse, Jie Wu, and Ehsan Hoque: Embodied Conversational Agents: Multimodal Communication and Embodiment
Mabel Urrutia, Vicente Moreno, Manuel de Vega, & Yurena Morera: A double task study of embodiment in counterfactuals
Elena Gámez, Hipólito Marrero, Yurena Morera, & José M. Díaz: Interpersonal verbs and Approach vs. Avoidance behavioural tendencies: a
question of embodiment?
Alberto Dominguez, Pierre A. Hallé, Fernando Cuetos, and Juan Segui: Is visual word recognition a motor-guided ability? Phonotactic repair during reading
Ji Y. Son: What kind of teacher is experience? Looking at signal detectors learning Signal Detection Theory
Travel and Lodging information
Arrival at Tenerife
Tenerife is a main tourist destination for many Europeans. This explains why a small island has two airports: Tenerife Norte (North) and Tenerife Sur (South). We recommend that you to arrive at Tenerife North, because by highway it is much closer to Garachico. In addition, it might be easier for us to pick up all of you if you arrive at the same airport (if for some reason any of you come to Tenerife Sur, we will pick you up anyway!).
I checked in internet some typical airway prices from different origins:
London - Tenerife Norte: 200.03 euros
Amsterdam - Tenerife Norte: 281.25 euros
San Francisco - London - Madrid - Tenerife Norte: 659.97 euros
New York - London - Madrid - Tenerife Norte: 469.79 euros
Chicago - London - Madrid - Tenerife Norte: 574.97 euros
These prices are just for your guidance, and they correspond to flights operated by British Airways and Iberia. Europeans and North Americans should follow different procedures for obtaining tickets. In either case, please obtain your tickets soon because the workshop dates are close to the Christmas Holidays when many people travel to the islands. Also, if tickets can be purchased more cheaply than anticipated, we will be able to fund more post-doctoral researchers for the workshop.
Euopeans: Purchase your tickets on your own, and you will be reimbursed at the workshop.
North Americans: The procedure is a bit more complicated because we must satisfy both NSF and University of Wisconsin auditors. The easiest procedure involves two steps. First, use your favorite website (e.g., Orbitz.com) to find flights that are convenient and, if at all possible, under $1000. Second, contact Burkhalter Travel (http://www.burkhaltertravel.com/; info@burkhaltertravel.com; 608.833.5200). Tell them that you are a member of the Glenberg Workshop group, and that you would like to purchase the ticket you identified in step 1 (unless, of course, they can find you something cheaper that is at least equally convenient). Burkhalter will automatically charge the NSF grant, rather than you, thus saving the hassle of reimbursement. If your trip to Tenerife will be added on to other travel, please work with Burkhalter in determining what costs should be charged to the Workshop grant and what costs should be charged to your personal credit card.
If you prefer to purchase the ticket on your own, you will save yourself (and Glenberg) enormous headaches if when you purchase the tickets you include a cost comparison of at least five other flights/airlines at the same or similar times/dates to indicate that you have chosen the least expensive arrangements (Burkhalter does this automatically).
We will cover your tickets and 5 nights in the hotel. I suggest that you arrive in Tenerife on December 14, which will allow you -especially if you come from the US- to have a whole day to adapt to local time.
As soon as you know your schedule, please let me (mdevega@ull.es) know it so that we can arrange for pick-up at the airport and make the hotel arrangements. I also need to know whether you will come alone or accompanied.
Site of the workshop
Click image tod download larger map
Garachico is a historical village in the North of the island of Tenerife (Canary Islands). Because it is far away from the main tourist places, Garachico still preserves the architecture and atmosphere of the old Spanish colonial settlements.
The workshop sessions will be held in a former XVI century monastery currently used as a cultural center ("Casa de la Cultura").
Lodging
Just in front of the aforementioned monastery, in the same square, is the Hotel la Quinta Roja, a XVI century palace where you will be lodged. You will find more information about the hotel on their web page:
http://www.quintaroja.com/indexin.htm
Funding for post-docs/graduate students
We had hoped to have money for each invitee to bring along a post-doc or graduate student. Because of funding cuts, we will not be able to
manage that. Instead, we will have a bit of a competition. If you would like to bring a post-doc please send the person's vita and a one-page justification. The justification should describe this person's preparation so the he/she can contribute to the debate and learn from it. Also, you may include a description of any research this person might present during a poster session.
Should you bother? We should be able to fund 5-6 people from the U.S. and perhaps an equal number from Europe.
Please submit the vita and justification by September 15 to Art Glenberg.
|
|
|
