1/24 Cogsci Colloquium, Lawrence W. Barsalou
         Department of Psychology, University of Chicago

Title: Perceptual Symbol Systems
2-3:30p.m., 280 Park Hall

1/25 Psychology Colloquium, Dr. Letitia Naigles,

 "Syntactic Bootstrapping the Multiple-frames Way:
Naturalistic and Experimental Evidence in Verb  Acquisition"
2:00-3:30, 280 Park Hall

1/31 CogSci Colloquium, Michael Webster,
         Dept. of Biological Sciences, UB,

 Title: "Male mating strategies in a colonial breeding Neotropical bird"
2:00-3:30, 280 Park Hall

2/7 CogSci Business Meeting

 Fri 2/9 Slips of the Tongue Research Group, 12:00-1:00, 631 Baldy

Hall (Phonetics Lab)

2/9 Philosophy "Teas", Alan Clune, "Direct and Indirect Theories of Perception"
3:15 pm, Philosophy Dept. Library

2/9 Linguistics Colloquium, David Fertig, UB, Dept. of Modern Languages and Literatures,

"Some Constraints on the Directionality of Analogical Leveling,
2-4 pm, 684 Baldy Hall

2/14 CogSci Colloquium, Robin Cohen,  Dept. of Computer Science, University of Waterloo,

 Title: "User Feedback in plan recognition"
2:00-3:30, 280 Park Hall

2/16 Philosophy Department, Barry Smith, "Bosnia: the Cognitive Geometry of War"
3:00 pm, 280 Park Hall

2/21 CogSci Colloquium, TBA

2/21 Buffalo Logic Colloquium, Sixth Mtg., Jacek Pasniczek, Dept. of Philosophy and Sociology,
Maria Curie-Sklodowsky University, Lublin, Poland and
UB Dept.of Philosophy,
"Nonstandard Possible Worlds and Generalized Quantifiers",
4-5:30 pm, 684 Baldy Hall

2/23 Computer Science Colloquium, Sreejit Chakravarty, UB, Dept. of Computer Science,
"VLSI CAD Researdch at UB: An Overview",
3-4:15 pm, 14 Knox Hall,
A reception follows the talk in 224 Bell Hall

2/28 CogSci Colloquium, Mike Alvard, UB, Dept. of Anthropology,
Title: "Testing the ecologically noble savage hypothesis"

  2/28 Buffalo Logic Colloquium, Seventh Mtg.,
 Ky Herreid, Barry Smith and John Corcoran,
UB Dept. of Philosophy, Buffalo Logic
Dictionary Project, 4-5:30 pm, 684 Baldy Hall

 Fri 3/1 Philosophy Colloquium, Lydia Goehr, Columbia University
"The Double Life of Music and Musicians in Exile",
3 pm, 684 Baldy Hall

 3/6 CogSci Colloquium, Brown Bag Presentations by: 1) Jeff Higginbotham, CDS "Using a communication frame approach to develop a  communication aid for individuals with severe expressive disabilities."
2) joint presentation by Barry Smith, Philosophy and David Mark, Geography

 3/6 Buffalo Logic Colloquium, Eighth Mtg., John Corcoran,  UB Dept. of Philosophy,
"Semantic Omega Properties",
4-5:30 pm, 684 Baldy Hall

 3/8 Computer Science Colloquium, Bharat Jayaraman,  "Algorithm =  Logic + Control Revisited",
3-4:15 pm, 14 Knox Hall,
reception follows the talk in 224 Bell Hall

3/13 CogSci Colloquium, DOUG MEDIN, Dept. of Psychology, Northwestern University,
Title: Trees: Taxonomic and Otherwise

 Fri 3/15 Slips of the Tongue Research Group,  1:00 pm, 631 Baldy Hall (Phonetics Lab)

 3/15 Computer Science Colloquium, David Milun,  "A Taste of Java"
3-4:15 pm, 14 Knox Hall,
reception follows talk in 224 Bell

 3/20 Spring Break

 3/27 CogSci Colloquium, JEAN-PIERRE KOENIG, UB, Department of Linguistics,
Title: WORDS, PATTERNS, AND CATEGORIES

  3/27 Colloquium co-sponsored by CogSci and Philosophy, Jose Bermudez,
4-5:30 in 684 Baldy Hall

 Fri 3/29 Philosophy Colloquium, Carolyn Korsmeyer,  SUNY Buffalo,
Work in Progress: "Taste, Food and Philosophy"
3 pm, 280 Park Hall

4/3 CogSci Business Meeting.

 4/10 CogSci Symposium on Animal Consciousness, DAVID SMITH,
Dept. of Psychology, UB

Thur 4/11 CogSci and Linguistics Co-sponsored Colloquium,  RONALD LANGACKER,
Dept. of Linguistics,
University of California at San Diego
Title: Reference-Point Subject Constructions 2:00-4:00 p.m, 280 Park Hall

Thur 4/11 CogSci and Linguistics Co-sponsored Colloquium, GEORGE LAKOFF
Department of Linguistics
University of California, Berkeley
Title:Moral Politics: What Conservatives Know That Liberals Don't
5:00-6:50, 280 Park Hall

Thur 4/18 CogSci Distinguished Speaker Series,  JEROME BRUNER,
5-6:50 p.m., NatSci 225,
Title: "Concepts of Self"

4/24 CogSci Colloquium, JANELLEN HUTTENLOCHER, University of  Chicago

 Thur 4/25 Neurology Seminar, Peter Huttenlocher, 8:30 am

 5/1 TBA

Abstracts

Wednesday, January 24, 1996,
280 Park Hall
2:00 - 3:30 p.m.

Lawrence W. Barsalou

Department of Psychology
University of Chicago

Perceptual Symbol Systems

Traditional theories of knowledge assume that arbitrary amodal symbols are transduced from perceptual states to represent concepts. An alternative assumption underlies perceptual symbol systems, namely, concepts are built from fragments of perceptual states, extracted via selective attention. The first half of the talk presents a theory of perceptual symbols. On this view, perceptual symbols are not conscious subjective images but are grounded in the neuronal states of perceptual systems. Rather than being holistic 'pictures in the head,' perceptual symbols are analytic, schematic, multimodal, and dynamic. Organized systems of perceptual symbols underlie the competent simulation of entities and events in their absence, and they enable important symbolic functions, including productivity, propositional construal, and the representation of abstract concepts (e.g., truth, negation, disjunction).

The second half of the talk presents empirical support for this theory. Two sources of evidence indicate that perceptual representations play central roles in conceptual tasks: instructional equivalence and perceptual work. Instructional equivalence occurs when neutral subjects, who receive no imagery instructions, perform similarly to subjects asked explicitly to use imagery. In experiments on feature listing, when neutral subjects list features for concepts, they produce essentially the same information as subjects asked to construct and then describe images. In experiments on property verification, when neutral subjects verify the properties of concepts, their patterns of reaction times are very similar to those for imagery subjects. Effects of perceptual work occur when perceptual variables affect performance in conceptual tasks, for both neutral and imagery subjects. In feature listing, 'visible' features are more likely to be produced than 'occluded' features, suggesting that subjects are scanning images to produce features. In property verification, perceptual factors central to image processing, such as size and visual similarity, predict the time to verify properties. The observation of instructional equivalence and perceptual work in two standard conceptual tasks suggests that perceptual representations play central--not peripheral--roles in conceptual representation and processing.

Back to Top

Back to Calendar

Wednesday, January 31, 1996
280 Park Hall
2:00-3:30 pm

Michael S. Webster

Department of Biological Sciences
University at Buffalo

"Male mating strategies in a colonial breeding Neotropical bird"

A widely accepted model for the evolution of animal mating systems predicts that male mating strategies will be determined by the temporal and spatial distribution of females. I studied a marked population of the Montezuma oropendola, a neotropical member of the American blackbird subfamily, to determine how male mating strategies were influenced by female dispersion. Females of this species nest in dense aggregations, with up to 100 females nesting in a single colony tree. High-ranking male oropendolas defended groups of females at nesting colonies and prevented other males from copulating there.
Observed male mating success was strongly biased toward high-ranking males; the top-ranking (alpha) male obtained 90-100% of all observed copulations at a focal colony in 3 of 4 breeding seasons. However, low-ranking males unable to copulate at the colony adopted two alternative behaviors to obtain mates. First, low-ranking males followed females and courted them away from the protection of the alpha. I used multilocus DNA fingerprinting to determine the paternity of 21 sampled nestlings from 4 study colonies. Seven of these nestlings matched with the alpha male at their colony, 4 matched with the beta male, and the remaining 10 did not match up with any sampled male. These results indicate that alpha males sire more progeny than other males, but that lower-ranking males are able to achieve some reproductive success by copulating with females away from the colony. Second, low-ranking males disrupted the copulation attempts of higher-ranking males at the colony. Although low-ranking males were unable to copulate with females at the colony, such disruptions may increase the chances that such males can copulate with females away from the colony. Regardless of why copulation disruptions occur, their effect was to significantly reduce the alpha male's ability to copulate at large colonies. I develop an extension of earlier conceptual models that incorporates copulation disruptions and predicts that males will adopt different mating strategies at colonies of different sizes.

Back to Top

Back to Calendar

Wednesday, February 14, 1996

ROBIN COHEN

Department of Computer Science
University of Waterloo

"USER FEEDBACK IN PLAN RECOGNITION"

 In this talk, we first describe the problem of plan recognition which has been addressed in the field of artificial intelligence. Traditionally, plan recognition systems take as input some observations about a user and a library of possible plans in some domain and return as output an interpretation of the underlying plan of that user. We then discuss five different research projects which are currently underway: (i) Allowing updates to plan libraries during plan recognition (with Spencer). Algorithms are developed which employ the results of previous plan recognition steps to produce an updated interpretation of the user's plan. We also discuss the development of tools to assist users in carrying out these updates, and the lessons learned about our possible user community. (ii) In advice-giving settings, using plan recognition in the production of cooperative responses (with van Beek). We determine the circumstances when it is not necessary to resolve plan ambiguities, in order to produce the advice-giving response. We also propose to enter into a clarification dialogue with a user, when ambiguities matter, and describe algorithms for generating these clarification dialogues (with van Beek and Schmidt). (iii) Considering a system which employs user modeling during plan recognition (with Ardissono). We demonstrate how such a system can extend the possibilties for clarification dialogues with users. (iv) Developing tools to assist designers of plan libraries (with Paris). (v) Extending the possibilities for dialogue structures in advice-giving settings which employ plan recognition by recasting the clarification task in terms of time-bounded persistent goals (with Donaldson). Our overall conclusion is that it is useful to allow user feedback during the plan recognition process, and to view plan recognition as something other than unassisted automated reasoning.

Back to Top

Back to Calendar

Wednesday, February 28, 1996
280 Park Hall
2:00-3:30 p.m.
North Campus

MICHAEL ALVARD

Department of Anthropology
University at Buffalo

"TESTING THE ECOLOGICALLY NOBLE SAVAGE HYPOTHESIS"

Two alternative hypotheses regarding subsistence hunters and their prey are tested. The first hypothesis states that native peoples are conservationists and make hunting decisions accordingly. The second hypothesis, derived from optimal foraging theory (OFT), states that short-term resource maximization guides hunters' behavior. Data were collected from the Piro, subsistence hunters residing in the rain forests of southeastern Peru. The alternative hypotheses generated predictions concerning hunters' inter- and intra-specific prey choice, and patch choice. Evidence for local prey depletion was sought. The harvests of the Piro and a nearby Machiguenga village were compared to examine for effects of technology, ecology, consumer demand and market participation. Finally, the Piro harvest was compared to maximum sustainable yield estimates. Direct observation of 79 hunts, and 122 interviews regarding unobserved hunts noted prey encounters and pursuits. Species, sex, age, and prey weight were recorded. Household interviews characterized the entire harvest for a village subsample. Marked and measured trails determined the location of prey encounters. Dental eruption and wear patterns from collected mandibles were used to age the kills. Prey vulnerable to local extinction - spider monkey, howler monkey and tapir were pursued by hunters at most opportunities. All species in the optimal diet were pursued, as predicted by OFT. Hunters did not choose age and sex categories that minimize impact on prey populations. Males and females were killed in proportion to their abundance in the wild. Hunters did not focus on immature and older individuals, but rather chose prime adults, and occasionally immatures of large-bodied species. Spider and capuchin monkeys, and game birds were encountered less often near the village, indicating depletion. Piro shotgun hunters did not harvest more meat per consumer-day than did the bow-using Machiguenga, suggesting diminishing returns for meat. The absolute harvest was greater at Diamante because of its larger population. Estimates indicate the large primate and tapir harvests are not currently sustainable at Diamante, while the harvests of less vulnerable species (collared peccary, deer and capybara) are sustainable. The conclusion is that the Piro do not live in 'harmony' with their environment, but rather hunt according to the predictions of OFT.

Back to Top

Back to Calendar

Wednesday, March 13, 1996
280 Park Hall
2:00-3:30 p.m.
North Campus

DOUG MEDIN

Northwestern University
Department of Psychology

"TREES: TAXONOMIC AND OTHERWISE"

 The talk is concerned with the role of expertise in the conceptual organization of biological kinds ( in this case, trees). One theoretical position is that the structure of things in the world acts as a strong constraints on categorization: categories are RECOGNIZED, not CONSTRUCTED. An alternative view is that goals, theories and belief systems play and significant role on category organization and related reasoning. These issues are evaluated by both studies of different types of experts and by a comparison of experts with novices. Among other things the results suggest that widely accepted generalizations concerning conceptual behavior may only hold for relative novices.

Back to Top

Back to Calendar

Wednesday, March 27
2:00-3:30 p.m., 280 Park Hall

JEAN-PIERRE KOENIG

Department of Linguistics
University at Buffalo

"WORDS, PATTERNS, AND CATEGORIES"

Under a widespread view of grammar, the knowledge underlying our linguistic abilitities consists of two unrelated components: (i) a set of universal rules/principles, (ii) a lexicon containing idiosyncratic information about words. Such a view conflicts with many recent results: lexical knowledge exhibits mid-size generalizations (language-specific, but not idiosyncratic generalizations); the lexicon does not consist in an underordered set of entries, but a highly structured hierarchically organized network; productive lexical generalizations still exhibit exceptions. In this talk, I propose a model of lexical knowledge which accounts for these properties. The basic idea is to assume that (i) stored lexical items can contain partially specified schematic patterns which are abstractions over individual entries; (ii) these items constistute a multidimensional taxonomic hierarchy. In such a model, words and patterns are not qualitatively distinct: they are all categories, although of a different degree of abstraction. Although this organization of lexical knowledge derives from purely linguistic considerations, it finds interesting echoes in other Cognitive Sciences (Artificial Intelligence/Knowledge Representation and Cognitive Psychology).                     

Back to Top

Back to Calendar

Wednesday, March 27
4:00-5:30 p.m., 280 Park Hall

JOSE BERMUDEZ

Department of Philosophy
University of Stirling

"LEVELS OF EXPLANATION IN PSYCHOLOGY -
PERSONAL AND SUBPERSONAL"

This paper considers a view of the nature of psychological explanation which is very widely held among philosophers and forms part of an important paradigm in cognitive science. This is the view that there is a distinctive way of explaining and predicting the behaviour of rational agents P variously termed the 'knowledge level', the 'semantic level' or the 'personal level'. On this view, although the central concepts of personal level explanation are realised subpersonally in computational sub-systems and ultimately at the neural level, explanation at the personal level is autonomous, dealing with the activities of persons (rather than computational sub-systems, or neural events) and subject to distinctive principles of rationality. Although this view is widespread and appealing, it seems to me that there are serious problems that it has to overcome if it is to be defensible. This paper considers some of these problems P including: a) the difficulties of explaining the behaviour of neuropsychological patients without bringing in subpersonal facts, b) the difficulties in explaining in personal level terms why skills manifest themselves in the precise way that they do, c) the difficulties in maintaining a contrast between syntactic subpersonal facts and semantic personal facts, d) the difficulties posed by experimental study of human reasoning habits.

Back to Top

Back to Calendar

Back to Top