17th ANNUAL WORKSHOP, 26-29 August 2000, Oxford, England
ANNOUNCEMENT (provisional list of abstracts is further down this page...)
The aim of the annual workshops is the multidisciplinary approach
of cognition in all its aspects, including modelling, experiments
and applied research:
- psychology (cognitive, developmental), perception,
- artificial intelligence (general aspects),
- associative memory and neural networks, neuroscience,
- linguistics (also computational), disorders of language,
- education and instruction,
- cognitive ergonomics,
- philosophy, history of concepts.
Papers accepted for the workshop may be published in the journal
of the ESSCS, 'Cognitive Systems'.
Accomodation in single and double rooms is available in Wadham
College, right in the centre of the old city. A block booking has
been made in the College UNTIL 20 May 2000. It includes breakfasts
and two dinners in the evening in the dining hall. Three nights and
breakfasts, two dinners, cost 160 pounds sterling. Lunches are
available in a nearby pub.
Length of abstracts: between 20 and 40 lines (1-2 kByte), to
be sent by ordinary mail, and possibly also by E-mail. They
will be judged as to relevance to a multidisciplinary approach.
(Full papers are also welcome, especially with a view towards
publication in 'Cognitive Systems'.)
Please send abstracts as soon as possible, but not later than
the 20th of May, in order that a provisional programme may be
put onto the webpage. Abstracts that arrive later, may be
considered, room permitting in the programme.
The fee is 18 Pounds Sterling for members of the ESSCS,
25 for others, for students the fee is 10 Pounds Sterling.
Membership is Euro 12 per annum.
ESSCS (Dr. G.J. Dalenoort)
Dept. of Psychology (E&A), University of Groningen
Grote Kruisstraat 2/1, 9712 TS Groningen, The Netherlands)
Telephone Netherlands +31-50-3636448 (G.J.Dalenoort) / 3636454
(or 3636472), Fax +31-50-3636304, E-mail esscs@ppsw.rug.nl
adress info is also available via the contact page
EUROPEAN SOCIETY FOR THE STUDY OF COGNITIVE SYSTEMS
17th Annual Workshop, Oxford (UK) 26-29 August 2000Date: 26 June 2000
Provisional list of authors and titles of papers.
(Names of authors who will not be present are placed between ( ) ).
M. Abraham, ENST-Bretagne, BREST, France,
Cognitive approach for an animated pictographic dictionary aimed at building sentences.
R. Allott, Seaford, England
Time and consciousness.S. Barattelli, (U. Schade), & K. Hadelich, University of Bielefeld, Germany
Production of Relative Clauses: A Model and some Experiments.G.J. Dalenoort, University of Groningen, Netherlands,
The correspondence between models for the neural level and for the cognitive level.J. Fritz and (K. Kotrschal), Konrad-Lorenz Forschungsstelle and Univ. of Vienna,
An adaptive-switch model on the formation of behavioural traditions.
P. Gerjets, K. Scheiter, W.H. Tack, University of the Saarland, Germany
Resource-adapted selection of strategies in learning from worked-out examples.
P.A.M. Gelepithis, Kingston University, England
Complex Intelligent Systems: Juxtaposition of Foundational notions and a Research Agenda
N. Hayes, Huddersfield, United Kingdom
Cognition in its social context: social identification and social representation.
K. Kessler, University of Bielefeld, Germany
Models of the interaction of spatial and linguistic representations in the brain.
S. Kobori, D. Fujii and (T. Nakamura), Ryukoku University, Shiga, Japan.
Problem Solving Processes in a Card Game.
K.B. Koster, University of Groningen, The Netherlands
Piaget rediscovered: A new metatheory for constructive learning ?
E. Laurent, (H. Ripoll & J. Baratgin), University of the Mediterranean (Faculty of Sport Sciences), Marseille, France
Visuo-spatial cognition in sport: an approach based on the study of expertiseI. Lemmens & P. Braspenning, University of Maastricht, NL
A Formal Analysis of Smithsonian Computational Reflection.
M. Pichat et G. Ricco Universite Paris 8, Saint-Denis (93), France
A multi-contextual study of advanced mathematical thinking: The case of the derivatives of functions
K. Plunkett, Oxford University, England
- Acquisition and Atrophy in Connectionist Models of Inflection
P. Saariluoma, University of Helsinki, Finland
Current address: International institute for applied systems analysis Laxenburg, Austria
From perception to apperception.
K. Scheiter, P. Gerjets, University of the Saarland, Germany,
E. Heise, Georg-August Univ. Goettingen, Germany
Hypertext navigation and conflicting goal intentions: Using log files to study distraction and volitional protection in learning and problem-solving.
P.H. de Vries (Univ. of Groningen, NL)
Performance of cognitive tasks in conceptual networks
Monday afternoon there will be a discussion on:
'The role of models for cognitive science'
Titles and abstracts (alphabetical order on surname of first author)
M. Abraham, ENST-Bretagne, BREST, France,
Cognitive approach for an animated pictographic dictionary aimed at building sentences.
Keywords: - linguistics (also computational), disorders of language, - education and instruction, - cognitive ergonomics,
We propose to help children with cerebral palsy compensate for
certain neuro-physical deficiencies in vocalization by building a
speaking machine, based on a three-step cognitive model of the
language. For paralyzed and aphasic children, communication is
often reduced to poor physical signals, unlike real language.
Alternative augmentative communication uses communication boards
with icons, so that "speaking" becomes "pointing". This substitution
raises crucial questions about visual language and about the links
between language and images, if we wish to computerize the
communication board.
The initial problem to be solved does not concern the child's
intelligence, but verbalization, and how to access information on the
screen when unable to use a mouse. In this case, we have to provide
automatic browsing, browsing as Æ naturally Ø as the user would have
done if he had been able to do so. We also have to provide icons as
intuitive as possible in order to quickly find the suitable word via
the icon. The system consists of: ? a pictographic dictionary, where
each pictogram represents a word of the lexicon. In order to be easily
identified, the verb icons are animated. ? the grammatical functions
necessary to build a sentence. Several problems arise in the knowledge
organization at both static and dynamic levels: i) in the representation
of words, especially those which represent actions and activities;
ii) in the dynamic presentation of the icons; iii) in the browsing
system to reach a choosen icon; iv) in the rebuilding of the
sentences (which grammatical functions are to be proposed, and how
should we represent them? ). The first version of the machine developed
speaks French; faced with translation into English, we show the
limitations of using icons as an intermediate language.
Maryvonne Abraham
ENST-Bretagne , B.P.252, F29285 BREST CEDEX
tel : 33 - 2 - 98 00 12 52 / fax 33 - 2 - 98 00 10 30
E-mail: maryvonne.abraham@enst-bretagne.fr
R. Allott, Seaford, England
Time and consciousness.
Time and consciousness: Is consciousness in time or time in
consciousness? Of course, the question is what is consciousness but
first what is time? Time is an ancient puzzle. Parmenides said nothing
changes; Heraclitus said there is only change - you cannot step in the
same river twice. St. Augustine was puzzled: everyone knows what time
is until you start to think about it and then you do not know. Bergson
attempted to explain la durÈe; Karl Popper wrestled with the subject in
his recent Parmenides. It is startling suddenly to recognise that
everything is momentary: matter, the cosmos, the world, ourselves, our
brains. William James' specious present is exactly that - only an
appearance. In reality our brains have only the present moment; our
brains are instantaneous patterns changing from moment to moment,
formed from the collection of structures, ultimately molecules, chemical
elements, protons, electrons, quarks, which constitute the neural
material. Our past is part of our present brain, the future we expect is
part of our present brain. Kant famously proposed that time and space
are necessary forms under which sensation and perception are
accommodated in our minds, in our brains; we could have no knowledge of
the dingen an sich. Recently Anthony Quinton has suggested that Kant
left totally unexplained how, if we see only through time and space as
arbitrary forms of intuition, we in fact cope with reality. Kant,
unfortunately for him, was writing before Darwin; evolutionary
epistemology (as formulated by Lorenz, Popper, Campbell, Wuketits and
others) proposes that the natural selection of brain processes has
provided us with practical concepts of time and space to allow us to
manage reality. The momentariness of everything and thus of ourselves
means that we are inconstant changing patterns, changing aggregations
of material; we are not the same from instant to instant. The cells in
our bodies are continually dying; the constituents of the cells are
being replaced; the aggregations of electrons and protons which form us
(perhaps the patterns which electrons and protons themselves are) are
travelling at incredible speeds, as the world turns, as the solar
system rotates, as the galaxy and supergalaxies turn. What are the
implications of this view of time for our understanding of
consciousness ? Physicists and cosmologists have their accounts of time,
space and space-time; a theoretical physicist, Julian Barbour, has
recently argued that even for physicists time does not exist. But time
and space as forms of the understanding, with due acknowledgement to
Kant, come before any use physical scientists may make of them. Time
and space come before mathematics and physics. It may be that the idea
of the momentariness of matter, of the cosmos, can have important
consequences even for the most refined and elaborate cosmological and
physical theories (quantum gravity, spacefoam, superstring theory, M-
brane theory).
Robin Allott <RMAllott@percepp.demon.co.uk>
http://www.percepp.demon.co.uk
tel/fax: +44 1323 492300
S. Barattelli, (U. Schade), & K. Hadelich,
University of Bielefeld, Germany
Production of Relative Clauses: A Model and some Experiments.In modelling the cognitive processes underlying language production,
there is a fruitful competition among modular and interactive
models. Modular models stand in the tradition of Fromkin (1971),
Garrett (1975), and Levelt (1989), whereas interactive models
are inspired by the local connectionist approach proposed by
Dell (1986). The interactive models, however, were primarily used for
modelling the production of single words, which can be done
without considering syntactic processes.
We would like to fill this gap. Our work is based on a local
connectionist model for the production of noun phrases as "die
lange rote Schraube" ("the long red bolt") within a situated object
naming task (Schade & Eikmeyer, 1998; Eikmeyer, Schade, Kupietz
& Laubenstein, 1999). It aims at modelling the production of
German sentences with noun phrases modified by relative clauses.
Such structures are syntactically complex: On the one hand, German
relative pronouns agree with their antecedents with respect to
case, number, and gender. On the other hand, relative clauses
can be embedded recursively. Obviously, a critical aspect in the
production of relative clauses is the capacity of short-term memory,
since lots of information pertaining to case, number, and gender
have to be stored, especially if there are recursively embedded
clauses.
In order to obtain empirical data, which can be used for model
construction and evaluation, we conducted a series of experiments.
The subjects had to complete sentence fragments as "Der bekannte
Rennfahrer, der die erstklassigen Mechaniker, die den grosszuegigen
Sponsor..." ("The famous race driver, [who]Subj [the top
mechanics]Obj, [who]Subj [the generous sponsor]Obj..."). The
independent variables belonged to two classes. On the one hand,
the subjects differed with regard to age and language disorders:
Three experiments were run with young subjects without language
deficits. One experiment had a sample of elder subjects with
normal speech and in one experiment we used patients with
different forms of aphasia.
On the other hand, the materials varied with respect to case and
number of the head NPs of the clauses, the number of recursively
embedded relative clauses, and the syntactic constructions used
(relative clauses vs. prepositional phrases). We analysed the
proportions of errors in different categories (failed sentence
completions and agreement errors). The results of these analyses
shed a new light on some of the syntactic constraints, which
a language production model has to respect.
Stefan Barattelli, <stefanb@lili.uni-bielefeld.de>
University of Bielefeld, Department of Linguistics,
P.O. Box 10 01 31, 33501 Bielefeld, Germany
Tel.: +49.(0)521.106.5274, Fax.: +49.(0)521.106.6447
G.J. Dalenoort, (Univ. of Groningen, NL),
The correspondence between models for the neural level and for the cognitive level.
In order to understand how the brain can learn, under what conditions
distributed processing can be parallel, and how the brain can so
efficiently "retrieve" information, it is necessary to understand
how functional properties of mental processes, that have been
postulated on the basis of models for cognition, correspond to the
macro- and micro-anatomical properties of the brain. A number of
hypotheses on thesecorrespondences will be discussed.
G.J.Dalenoort <G.J.Dalenoort@ppsw.rug.nl>
Dept. of Psychology (E&A)
University of Groningen
Grote Kruisstraat 2\1, 9712 TS Groningen, The Netherlands
Johannes Fritz & Kurt Kotrschall
J. Fritz and (K. Kotrschal), Konrad-Lorenz Forschungsstelle and Univ. of Vienna,
An adaptive-switch model on the formation of behavioural traditions.
Mainstream biology usually defines tradition as the inter-generation
transmission of behaviour through learning by observation of
other individuals (social learning). By this definition, tradition-
forming has many examples throughout the animal kingdom. Several
defined cognitive mechanisms to acquire information by observation
are known, including the cognitively demanding ability to learn a
task by observation (motor imitation). Another important mechanism
seems to be the spatial orientation of the observers' attention to
a certain stimulus (stimulus enhancement). However, there is a lack
of empirical and theoretical work focusing on the behavioural
transmission in a population based on that mechanisms.
Two independent approaches may be employed in linking mechanisms and
functions: (1) Comparison of the observed transmission curves with
different mathematical models on social and non-social learning.
(2) In most experiments on behavioural transmission naive individuals
try to divert food form a producer (scrounging) instead of learning
to perform the task. Thus, the second approach tries to define the
equilibrium frequency of producers and scroungers, where cultural
transmission should come to an end within a certain population.
By identifying the reason for an individual to choose between social
learning or scrounging and, in relation to that, by defining the
factors which determine the equilibrium frequency of producer and
scrounger, one may characterise the shape of a cultural transmission
curve in a defined population. This requires to combine the two
approaches mentioned above. In our contribution we try that by
presenting the adaptive-switch model, based on data from a three-year
study with a free living, semitame population of greylag geese
(Anser anser). We assume that an inexperienced individual in presence
of a conspecifics demonstrator chooses between scrounging and social
or asocial learning depending on the profitability of either tactic
and on the given social conditions. That allows to predict transmission
pattern in a given social context based on individual decisions.
Furthermore, preliminary data indicate that individuals, which start
with new traditions, can be characterised by certain personality
properties (coping style).
Konrad-Lorenz Forschungsstelle,
4645 Gronau 11, Austria and University of Vienna
P. Gerjets, K. Scheiter, W.H. Tack,
University of the Saarland, Germany
Resource-adapted selection of strategies in learning from worked-out examples.Most tasks can be pursued by using different strategies
(Logan, 1985; Reder & Schunn, 1998). In this paper we focus
on strategies of learning from worked-out examples. Within a
resource-oriented framework these different strategies can be
classified according to their costs and benefits. These features
may determine which strategy will be selected for accomplishing
a task in situations with certain resource limitations. We
investigate specific hypotheses about strategic adaptations to resource
limitations (e.g., time pressure or lack of prior knowledge) within
a hypertext-based learning environment. A comparison of the
strategy selection of good and poor learners is used to assess the
degree of subjects' resource adaptivity. Ideas for modeling
resource-adaptive selection of strategies within the ACT-R
architecture are discussed.
Peter Gerjets <pgerje@cops.uni-sb.dt>,
Katharina Scheiter <katharis@cops.uni-sb.d>,
Werner H. Tack < tack@cops.uni-sb.d>,
Graduate College for Cognitive Science
Collaborative Research Center 378 Department of Psychology
University of Saarland University of Saarland Georg-August-University
D-66041 Saarbruecken/Germany D-66041 Saarbruecken/Germany D-37073 Goettingen/Germany
Petros A.M. Gelepithis,
Kingston University, England
Complex Intelligent Systems:
Juxtaposition of Foundational notions and a Research Agenda.
The cardinality of the class, C, of complex intelligent systems, i.e.,
systems of intelligent systems and their resources, is steadily increasing.
Such an increase, whether designed, sometimes changes significantly and
fundamentally, the structure of C.
Recently, the study of members of C and its structure comes under a
variety of multidisciplinary headings the most prominent of which include
General Systems Theory, Complexity Science, Artificial Life, and
Cybernetics. Their common characteristic is the quest for a unified theory
of a certain class of systems like complex or living systems. So far, the
only candidate for a general theory of intelligent systems is Newell’s Soar.
To my knowledge there is presently no candidate theory of C except Newell’s
claimed extensibility of Soar.
This paper juxtaposes the elements of Newell’s conceptual basis with
those of an alternative conceptual framework based on the thesis that
communication and understanding are the primary processes shaping the
structure of C and its members.
It is patently obvious that a research agenda for the study of C can be
extremely varied and long. The third section of this paper presents a
highly selective research agenda aimed to be modified by the workshop
proceedings so that research collaboration and further steps in advancing
the theory and design of complex intelligent systems may be taken.
Keywords: communication, understanding, non-linear dynamical systems,
complexity, intelligence, system, representation, meaning, design.
E-mail: <Petros@kingston.ac.uk>
N. Hayes, Huddersfield, United Kingdom
Cognition in its social context: social identification and social representation.Modelling human cognition is made uniquely challenging by the way that
the same human being may evaluate the same information quite differently
at different times. These different evaluations depend on a number of
factors, not least of which are social identity processes and social
representations. Social identity refers to the "them and us" aspects of
group membership, and in this context has particular relevance in the
way that different social identifications become salient on different
occasions; and in doing so, bring about different cognitive appraisals.
Social representations are the shared beliefs held by social groups and
by societies as a whole, which are often maintained in the face of
directly conflicting personal experience.
Attempts to simulate human cognition which deal with information
processing as a consistent or cumulative process are therefore
intrinsically misleading. Nonetheless, researchers in the fields have
identified a number of systematic mechanisms involved in the acquisition
and maintenance of social identification and social representations.
This paper examines those mechanisms in terms of the challenges they
present for cognitive modelling.
Nicky Hayes <nicky@nickyhayes.co.uk>
http://www.nickyhayes.co.uk/nicky/
K. Kessler, University of Bielefeld, Germany
Models of the interaction of spatial and linguistic representations in the brain.In this paper we present our work on the study of spatial
representation and processing in connectionist networks. This
includes integrating a network to model the interpretation of
spatial prepositions (Kessler, 2000) with our work on the
representation of multiple objects in a connectionist model of
the visuo-spatial sketch pad (Kessler & Rickheit, 2000).
Consequently we will first ask how spatial information is
extracted and processed in different hierarchical and/or parallel
neural structures. We will then discuss some possible underlying
principles like combined representations of different proprioceptive
channels, vector coding and vector rotation in neural populations
and shifter circuits (for a review see Kessler, 2000). Other empirical
data suggest more general features of the underlying structures in
cognitive systems such as the parallel processing of the what,
the where and of proprioceptive information. Considering the
language about space ñ in particular, localisations that make use
of 'in front', 'behind', 'left' and 'right' (i.e. 'take the cup
at the right of the screen'), one can see that in most situations
one has to choose between at least two possible referential systems.
The choice of a referential system is mediated by specific contextual
factors like the social relation between the communication partners
and the cognitive effort for the alignment of differing perspectives
with the own orientation (for a review see Kessler, 2000). Bringing
together the physiological and cognitive data we emphasise a
dynamic view of the 'meaning' of spatial prepositions, in which
linguistic information is dynamically bound to the situation-
specific (re)construction of the spatial constellation. That binding is
mediated by social context which interacts with the cognitive effort
that is associated with each perspective. This dynamic view is
supported by the data in hand and has been evaluated as to
consistency and as to empirical fruitfulness by means of a running
connectionist network model (Kessler, 2000). Another cognitive
function essential for spatial cognition and manipulation is the
possibility for temporary maintaining multiple object representations.
Especially in those spatial constellations where at least two persons
interact and many objects are part of the visual scene, multiple
representations have to be maintained for further processing and
have to be protected from interference. The last point is also known
as the 'feature integration problem' or more recently as the 'binding
problem'. We will present our solution of this problem using control
structures and contrast it with the possibility to use phase
synchronisation along oscillating units (Kessler & Rickheit, 2000).
Different empirical predictions of these two possibilities will be
discussed and a series of studies will be proposed for their
evaluation using analysis of coherence in EEG. Finally, the impact
on other cognitive processes like the generation of cognitive routes
and maps and the production of localisations of the model for
visuo-spatial short-term memory in combination with the general
principles of the model for the interpretation of localisations will be
discussed.
- Kessler, K. (2000). Raumkognition und Lokalisationsaeusserungen:
Ein konnektionistisches Modell des Verstehens von Richtungspraepositionen.
Wiesbaden: Deutscher Universit‰tsverlag.
- Kessler, K. & Rickheit, G. (2000). Dynamic binding and context
influences during scene perception: Simulations and experimental results.
In N. Taatgen & J. Aasman (Eds.), Proceedings of the Third International
Conference on Cognitive Modeling, 23-25.03.2000 in Groningen (pp. 177-184).
Veenendaal: Universal Press.
Dr. Klaus Kessler <kessler@lili.uni-bielefeld.de>
Dept. of Linguistics, SFB 360, University of Bielefeld
Tel: 0049-(0)521-1065268; Fax: 0049-(0)521-1062996
Postfach 100131, 33501 Bielefeld
Homepage: http://www.lili.uni-bielefeld.de/~kessler/
S. Kobori, D. Fujii and (T. Nakamura), Ryukoku University, Shiga, Japan.
Problem Solving Processes in a Card Game.'Calculation' is a game that can be played by one person (a solitaire),
based on incomplete information. We performed experiments on human
subjects, and compared problem solving by an expert and 12 novices.
The data were recorded on videotape in an experimental design based
on verbal protocols.
Chunks were defined as configurations of tableau cards which
corresponded to patterns of cards the subjects were presumed to perceive or
recognize. The data were also analyzed by means of an evaluation function
of a computer program that can play 'Calculation'.
The expert appeared to be excellent in the ability of composing complex
chunks and recognizing them. This can be explained by the fact that
structured knowledge is essential in order to become an expert,
but also by the necessity to construct an interactive mapping between the
external world and the inner world. We clarified the
relation between the evaluation function and the characteristics of the
expert player's strategy.
We suggest that problem solving processes of experts involve the cycle of
composition and recognition of flexible chunks, and that the strategies
derive from the knowledge which controls such a cycle. We guess
that experts solve the problem by making use of the character of the game,
in spite of the limitation of capacity of working memory.
Address first author: Satoshi Kobori <kobori@rins.st.ryukoku.ac.jp>
Department of Electronics and Informatics,
Faculty of Science and Technology,
Ryukoku University, Shiga, Otsu, 520-2194, Japan.
http://milan.elec.ryukoku.ac.jp/
K.B. Koster, University of Groningen, The Netherlands
Piaget rediscovered: A new metatheory for constructive learning ?
The central question in my presentation is if and how the most
recent interpretation of the concept of equilibration by Piaget and
his co-workers still can contribute to a sound metatheory of
development and evolution and can function as a paradigm for future
research in cognitive development. For more than sixty years Piaget
and his coworkers tried to unravel the mysteries of ontogenetic
change, but this heritage has vanished in modern cognitive,
developmental, and instructional psychology. Most current models of
developmental mechanism are either based on very general evolutionary
theories, or on the study of the acquisition of task-specific rules in
one domain.
The richness of Piaget's epistemological studies in the sixties and
seventies about the acquisition of fundamental notions about time,
speed, number are still a good starting point for modeling the process
of reflective abstraction in children. According to Piaget learning
consists in drawing from a system of actions and operation at a lower
level certain characteristics which are then reflected (in the almost
physical meaning of the term ) onto actions or operations at the
higher level. Reflective abstraction operates through reconstructions
which transcend, and at the same time integrate, previous
constructions.
Klaas B. Koster <k.b.koster@ppsw.rug.nl>,
Department of Education, University of Groningen
Grote Rozenstraat 38, 9712 TJ Groningen
Tel: (31)(0)50-3636542
E. Laurent, (H. Ripoll & J. Baratgin), University of the Mediterranean
(Faculty of Sport Sciences), Marseille, France
Visuo-spatial cognition in sport: an approach based on the study of expertise
The domain of sport is a privileged context to study behaviours
and cognitive processes of experts. Physical and sport activities
imply high levels of constraints for which experts are very well
adapted. As investigations in other domains of expertise (for
example chess, Chase & Simon, 1973), many studies in sport have
demonstrated the superiority of experts to recall or recognise
specific material organised according to the logic of their domain
of expertise [(e.g., Allard, Graham, & Paarsalu, 1980; Garland &
Barry, 1991)].
In recent studies we [(Baratgin, Ripoll, Ripoll, Courrieu, &
Laurent, submitted; Courrieu, Ripoll, Ripoll, Baratgin, & Laurent,
submitted; Laurent, Ripoll, Baratgin, & Kehlhoffner, 2000)] have
investigated experts' cognitive processes in a more strictly
perceptual task (a similarity judgement task of basket ball play
configurations) in which we reproduced the classical effect of
expertise, demonstrating better performance for coherent
(structured) configurations.
Beyond this classical result, we have questioned the problem
of the format of the knowledge basis and the cognitive processes
activated in the performance of such tasks. Do experts use
semantic rules, or do they analyse only the perceptual features
of the stimuli? What is the level of control of these processes?
Results have shown that these two processes (perceptual and semantic)
could be involved in the similarity judgement task. A second aspect
of the data (namely, the faster response making for experts even
when material is not structured) was considered. We did an experiment
(Poplu, Laurent, Baratgin, & Ripoll, 2000) in which we showed that
experts probably encode information in an automatic way.
Finally, we will show that one of our perspectives of research
stands in an interdisciplinary approach to cognitive sciences that
connects cognitive psychology, artificial intelligence, and
robotics.
References:
Chase, W.G., & Simon, H.A. (1973). Perception in chess.
Cognitive Psychology, 4, 55-81.
Poplu, G., Laurent, E., Baratgin, J., & Ripoll, H. (2000, July).
Modalit/es d'encodage perceptif de configurations de jeux.
Paper presented at the International Congress of the French
Society of Sport Psychology, Paris, France.
I. Lemmens & P. Braspenning, University of Maastricht, NL
A Formal Analysis of Smithsonian Computational Reflection.As human beings, we are capable of thinking about the world that surrounds
us, as well as about our (own) possible thoughts, feelings, and actions in
that world. In some crucial sense, we are capable of reflective thinking.
This ability is essential for learning, reacting to unexpected
circumstances, recovering from mistakes, and so forth. In Artificial
Intelligence, computational forms of the reflective process have been
explored. This has led to the development of so-called reflective systems.
The topic of computational reflection is attractive, even fascinating,
although, in some sense, it all comes down to applying the symbolic
processing paradigm to the inner workings of computational engines,
resulting from previous design activities. Since every computational
engine interacts with its environment, computational reflection minimally
concerns the design of a meta-engine. This meta-engine interacts with a
previously designed engine, which in this interaction plays the role of
environment. Yet, as fascinating as the topic of computational reflection
may be, the interest in this topic within Computer Science and Artificial
Intelligence is not constant, albeit rather persistent. Not constant,
because viewing relevant work in the literature of Artificial Intelligence
and Object-Oriented Software Engineering shows quite important
contributions widely scattered in time . However, there are also periods
in which nearly no research activities concerning computational reflection
seem to happen. We think that this situation is likely caused by the
inherent cognitive complexity of the topic of computational reflection.
It is our belief that the use of standard modelling methods in order to
tackle the issue of computational reflection can help in decreasing the
cognitive complexity of the topic. Therefore, we reconstructed the theory
about computational reflection, developed by B.C. Smith (Smi82), and
applied the Unified Modeling Language to this theory, which results in
more formal class- and object models for computational reflection. The
used representation, is likely to decrease the cognitive complexity of
computational reflection. The representation also aids in alleviating much
of the muddled interpretation of the researchers interested in the topic
of computational reflection.
Lemmens, I.M.C., Braspenning, P.J.:Lemmens, I.M.C., Braspenning,
P.J.: A Reconstruction of Smithsonian Computational Reflection. Report
CR&S00-01, University Maastricht (2000).
Smith, B.C.: Reflection and Semantics in Lisp. MIT Technical
Report MIT/LCS/TRT-272, Massachusetts Institute of Technology (1982).
Inge Lemmens & Peter Braspenning
Section Communications Research & Semiotics (CR&S)
Faculty of General Sciences
University Maastricht, The Netherlands
M. Pichat et G. Ricco Universite Paris 8, Saint-Denis (93), France
A multi-contextual study of advanced mathematical thinking: The case of the derivatives of functions
The present research aims at linking epistemological and didactic
factors to the construction of mathematical knowledge by high school
students. It focuses on the mathematical concept of derivative
because this notion turns out to be particularly complicated (Tall,
1991) given its complex roots in mathematical calculus (limits,
neighbourhoods, infinitesimals, etc.).
Students understanding of one given mathematical concept may
fruitfully be studied by taking into account the contexts in which
this concept occurs and lives (Brousseau, 2000; Chevallard, 1992). In
the present work, three contexts are invested, each of them being a
precious source of information in the comprehension of students'
elaboration of a given notion:
(i) Historical context (epistemological background of the concept, main
difficulties encountered, epistemological obstacles) (Bachelard, 1938)
(ii) Social context (respective roles of the teacher and of the students
during the learning of the notion, such as interventionism, maieutics
or co-operation)
(iii) Didactic context (tasks proposed to students, didactic
transposition of the concept). (Vergnaud, 1997)
In order to combine the different contexts we mentioned to cognitive
aspects of the acquisition of the concept of derivative by students,
four studies are effectuated. (1) An epistemological-historical
investigation outlines the main conceptual obstacles encountered by
the principal mathematicians involved in the construction of the
notion of derivative ; (2) An analysis of three main french school
books of mathematics emphasises interventionism and hiding of
analytical difficulties ; (3) An examination of didactic episodes (6th
lower grade) introducing the notion of derivative stresses teacher's
avoidance of the complex calculus-related difficulties of the notion
by an omnipresent focus on an algebraic point of view and an intense
interventionism ; (4) A main study of students' understanding of the
notion from a questionnaire (studying algorithm learning / conceptual
development dialectic) points many deep and handicapping conceptual
difficulties.
It is showed that the results gained from these different studies
are largely convergent and tend to indicate that there is a relative
historical/didactical/cognitive continuity with regard to the difficulty
to elaborate calculus based references to derivatives. It is also
pointed out that these lacks of mathematical constructions (that are
analysed and presented in this research) prevent students from
being able to use correctly and fruitfully the mathematical algorithms
they learned about derivatives (procedural knowledge is strongly
limited by weakness of conceptual knowledge).
Keywords : advanced mathematical thinking, learning, conceptualisation, derivative,
didactic context, epistemology, obstacles.Michael Pichat <michael.pichat@univ-paris8.fr>
UPRES EA2305 Cognition et Didactique
UFR 7, Universite Paris 8, Saint-Denis (93), France
Kim Plunkett, Oxford University
Acquisition and Atrophy in Connectionist Models of InflectionA wealth of empirical data from children, adults and aphasics
attests to qualitative and quantitative differences in the
behaviour of regular and irregular words in the inflectional
systems of different languages. These differences have been
traditionally viewed as support for the theory that inflectional
processes require different types of mental computation. I will
offer an alternative account that uses purely associative processes
to account for the acquisition and atrophy of inflectional processes.
It would appear that this account has reasonable cross-linguistic
validity.
P. Saariluoma, University of Helsinki, Finland
Current address: International institute for applied systems analysis Laxenburg, Austria
From perception to apperception.It seems today rather evident that human mind constructs representations
with specific mental contents. Yet, relatively little attention has been
paid to the fact that many individual contents in mental representations
cannot be perceived but they are rather complex constructs of perceivable
and non-perceivable information. We can represent numerous contents such as
electron, infinity, eternity, possible, justice, tomorrow and past, which
cannot be reduced to perception for the reasons that perception is about
processing a stimulus, which is physically present. One can perceive a
closed door, but one can also represent the room behind it. The first
content is perceived while the second is conceptually constructed or
apperceived.
In this paper, a number of the preconditions and properties of apperception
will be discussed in the light of empirical evidence collected from chess,
architecture, economic thinking and religious cognition. Such problems will
be the functional organization or representations, their coherence and
consistency. Finally, the conceptual limits of explanatory models, e.g.,
neural, logical and content-oriented, in apperception research will be
addressed.
Pertti Saariluoma < saari@iiasa.ac.at>,
IIASA, International Institute for Applied Systems Analysis
A-2361 Laxenburg, Austria
Phone: +43 2236 807-0, Fax: +43 2236 71313
K. Scheiter, P. Gerjets, University of the Saarland, Germany,
E. Heise, Georg-August Univ. Goettingen, Germany
Hypertext navigation and conflicting goal intentions:
Using log files to study distraction and volitional protection in learning and problem-solving.
This paper describes a theoretical analysis and experimental
investigation of difficulty related distraction by conflicting
goal intentions in learning and problem solving with hypertext.
Log files are used to capture hypertext navigation in the face
of opportunities to implement competing goal intentions. We study
how differences in task difficulty influence the volitional
protection of the current goal intention. First attempts to
integrate volitional processes of action control into cognitive
architectures are presented.
Katharina Scheiter <katharis@cops.uni-sb.de>,
Peter Gerjets <pgerjet@cops.uni-sb.de>,
Elke Heise <eheise@gwdg.de>
Graduate College for Cognitive Science
Collaborative Research Center 378
Department of Psychology University of Saarland
University of Saarland Georg-August-University
D-66041 Saarbruecken/Germany, D-66041 Saarbruecken/Germany, D-37073 Goettingen/Germany
P.H. de Vries (Univ. of Groningen, NL)
Performance of cognitive tasks in conceptual networksCognitive tasks can be represented at different levels of
description. At the functional level a representation is based on
the principles of present-day computer programs, in particular those
of production systems, or semantic networks. At the structural level
the self-organizing interactions of neuronal excitation patterns
represent the execution of tasks. Because of the procedural
knowledge and structured objects involved, the logistics of these
tasks are too complex to be programmed at the neuronal level nor do
there exist learning mechanisms adequate for the autonomous
development of the required architectures. In order to find these
requirements the representation of tasks in conceptual networks is a
necessary step. By means of the selective propagation of excitation
loops along permanent and temporary connections the performance of
tasks can be represented as process of collective control. Essential
aspects of these processes are the distinct representation of
location, identity, and procedural knowledge, combined with
context-dependent binding and the role of the critical threshold of
a cell-assembly. Simulations of conceptual networks will be
discussed in relation with experiments from cognitive psychology.
P.H. de Vries <P.H.de.Vries@ppsw.rug.nl>
Dept. of Psychology (E&A)
University of Groningen
Grote Kruisstraat 2\1, 9712 TS Groningen, The Netherlands
ESSCS european society for the study of cognitive systems