Date: Fri, 30 Nov 2001 11:53:17 -0800
Reply-To: Colette Faucher <colette.faucher@WANADOO.FR>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Colette Faucher <colette.faucher@WANADOO.FR>
Subject: CFP and participation : "Causality and Categorization",
FLAIRS 2002 workshop
FLAIRS 2002 Workshop : "Causality and Categorization"
May 15, 2002, Institute for Human and Machine Cognition, Pensacola, Florida,
Website : http://perso.wanadoo.fr/colette.faucher/causality.html
Topic of the workshop :
This workshop is intended to further the study of the role of causal
knowledge in the categorization process (how to acquire categories or
concepts, use them, represent them, and so on).
More precisely, at least the following questions seem worth discussing :
- how causal information intervenes in the concept learning process,
- what is the importance of causal information to completing the usual
accounts of the representation of concepts (subject-predicate, prototypical
and by means of exemplars),
- how causal knowledge can be inferred from instances, to be represented
within concepts, in the framework of unsupervised learning systems.
This workshop is related to the special track : "Categorization and Concept
Representation : Models and Implications".
Planned format :
Four invited talks are planned (duration of each talk : 45 minutes).
Interested researchers are invited to submit papers relevant to the topic.
Each presentation must not exceed 20 minutes.
Papers must be written using MS Word, RTF or PDF formats according to AAAI's
standard format for authors. They must be sent to both Clark Glymour and
Only a few papers (about 6) will be presented to have time to discuss both
the talks and the presented papers. Papers presented at the workshop WILL be
included as workshop briefs in the published proceedings of FLAIRS. They
also will be possibly published in an international journal.
The meeting will be split into a morning and an afternoon session.
Important dates :
Paper Submission Deadline : February 1, 2002
Notification of Acceptance-Rejection : February 15, 2002
Camera Ready Copy Due : March 4, 2002
Conference Date : May 15, 2002
Conditions of attendance :
Except for the invited speakers and those people presenting submitted
papers, attendance will be by application only so that there can be fruitful
exchange between participants in a less formal and more intimate way than
during the corresponding track (see above).
Please, send your application (short CV and selected publications) to Clark
Glymour and Colette Faucher.
Presentation of the invited talks :
Patricia Cheng, UCLA Department of Psychology, Los Angeles, CA, USA
Title : "Functional Basic-Level Categories and Causal Theories"
One might think that causal discovery depends on the definition of the
entities among which causal relations are to be discovered. Categorization
would therefore precede the discovery of causal relations. This paper argues
for a dependence in the opposite direction: Causal discovery is the driving
force underlying our mental representation of the world, not only in the
sense that it is important to know how things influence each other, but also
in the sense that causal relations define what should be considered things
in our mental universe. The paper provides evidence that categorization does
not precede causal discovery; instead, the two operate together as a single
process, with optimal causal discovery being the driving force. In
particular, the paper presents experiments showing that "basic-level"
functional kinds are defined by causal theories of the functions in
question, rather than by similarity relations. The latter are merely
byproducts of the causal theories. Using identical stimuli, the experiments
manipulated the level of inclusiveness of the target function to be learned.
This manipulation shifted the basic level in a hierarchy, as indicated by
such properties as similarity structure, verification performance, and
choice of description.
Bob Rehder, Department of Psychology, New York Univ.,New York, NY, USA
Title : "A Causal-Model Theory of Conceptual Representation and
This talk introduces a theory of categorization that accounts for the
effects of causal knowledge that interrelates or links the features of
categories. According to causal-model theory, people explicitly represent
the probabilistic causal mechanisms that link category features, and
classify objects by evaluating whether they were likely to have been
generated by those mechanisms. Participants were taught causal knowledge
that linked features of a novel category into a causal chain. In three
experiments, causal-model theory provided a good quantitative account of the
effect of this causal knowledge on the importance of both individual
features and inter-feature correlations to classification, and did so
without postulating differences in subjective feature weights or
higher-order properties. By enabling precise model fits and interpretable
parameter estimates causal-model theory places the "theory-based" approach
to conceptual representation on equal footing with the well-known
Steven A. Sloman, Cognitive & Linguistic Sciences, Brown Univ., Providence,
Title : "The Psychology of Causal Reasoning"
I report empirical tests of a probabilistic framework for causal modeling
that captures strong intuitions about human thought and reasoning, including
intuitions about the nature of counterfactual reasoning and the distinction
between observation and action. The fundamental claim of the framework is
that people represent the world by decomposing it into autonomous
mechanisms. The experiments reported examine a key assumption of the
framework "its representation of actual and counterfactual intervention" in
order to evaluate its viability as a source of cognitive models of
categorical reasoning. The experiments focus on counterfactual inference.
Implications for conceptual structure are discussed with emphasis on how
people think about the functions of human artifacts.
Michael R. Waldmann, Department of Psychology, Univ. of Göttingen,
Title : " Categories and Causal Models : A Tale of Chickens and Eggs"
The standard view guiding research on causality presupposes the existence of
networks of causes and effects in the world that cognitive systems try to
mirror. This position also underlies current research on the relationship
between categories and causality. According to the view that categorization
is theory-based, traditional similarity-based accounts of categorization are
deficient because they ignore the fact that many categories are grounded in
knowledge about causal structures. And indeed in a number of experiments,
which will be summarized in the first part of the talk, we have shown that
prior assumptions about the causal status of learning events governs the
process of learning new categories. These studies show that learners use
prior knowledge to create representations about causal models rather than
associating cues with outcomes, as standard associative theories or
regression models assume. However, in the second part of the talk I will
present evidence that shows that the opposite direction, which thus far has
largely been neglected, also holds. Categories that have been acquired in
previous learning contexts may influence subsequent causal learning. It can
be shown that different conceptual schemes may lead to dramatically
different causal models with identical learning data. Thus, there is a
bi-directional interaction between categories and causal models, similar to
the relation between paradigms and scientific discovery.
Workshop co-chairs :
John Pace Scholar and Senior Research Scientist, IHMC, University of West
Alumni University Professor of Philosophy
Department of Philosophy
Carnegie Mellon University 135 Baker Hall
Pittsburgh, PA 15213, USA
E-mail : firstname.lastname@example.org
Phone : (412) 268-2933
Fax : (412) 268-1440
Associate Professor of Computer Science
Faculté des Sciences de Saint-Jérôme, Université d'Aix-Marseille III,
Avenue Escadrille Normandie-Niemen, 13397, Marseille, Cedex 20,
E-mail : email@example.com and firstname.lastname@example.org
Phone : (+33) 4 91 05 60 58
Fax : (+33) 4 91 05 60 33