The symposium is part of the opening celebrations for the new Institute of Cognitive Science at the University of Osnabrück. It is dedicated specifically to the inauguration of the two Graduate Schools in the Department of Human Sciences that start on 1st October 2002: The Graduate School “Integrative Competencies”, which looks at somatic, psychological, social and cultural parameters of well-being, and the Cognitive Science Graduate School, which is focussed on “Cognitive Architectures” and the integration of category-based vs. pattern-based models of cognition.

The symposium will be concerned with the opposition between two types of models for human behaviour that have often shown up as adversaries in methodological discussion in the human sciences over the past decades: One that takes cognition to be in principle a rule-governed and category-based enterprise, whose creative an innovate aspects as well as the potential for adjusting to new situations is primarily a matter of recursive rules. This approach has probably figured most prominently in the human sciences in the linguistic theories inspired by Noam Chomsky. These are fundamentally deductive theories and their categories are taken to be genetically determined, while the experience of the subject plays only a minor role.
The other type of model starts more typically from empirical regularities in cognitive behaviour that are regarded as recurring patterns and are not necessarily based on deductive categories. Creativity and productivity, i.e. the innovative aspects of behaviour and its contextual adjustability, are here taken to result from the transfer and the combination of established patterns previously experienced by mechanisms based on relative and relational similarity, like analogy.
Although at present both types of models are each characteristic for different and opposed methodologies in the human sciences, one might sensibly explore the hypothesis that, at least in some forms of human behaviour, there is an interaction of mechanisms some of which are better characterised according to one of the two models and others that are more fruitfully seen as functioning according to the other model. This hypothesis - call it the Duality Hypothesis - is of course faced with the question of how such different mechanisms could interact.

The intention of the symposium is to discuss the opposition as well as the perspectives, foundations, and possible implementations of the Duality Hypothesis.
 

Place: University of Osnabrück, Neuer Hörsaalkomplex, Kolpingstr.7, Osnabrück

Programme Chairs: Peter Bosch and Julius Kuhl

Poster Session: The symposium will be accompanied by a poster session (approx. 10-15 Posters)

Contact: Marion Schmitz (maschmit@uos.de)
 

Thursday, 21st Nov, 2002

9:00-9:30 Opening

9:30-10:30 “Rules and patterns: Integrating different perspectives of cognitive representations”
Helmar Gust, Artificial Intelligence and Cognitive Science, University of Osnabrück

Coffee

11:00-12:00 “Creativity and brain laterality: poly- vs. mono-semantic ways of thinking”
Vadim Rotenberg, Abarbanel Medical Health Center, Tel Aviv University, Israel

12:00-13:00 “The brain basis of language processing and language learning”
Angela Friederici, MPI of Cognitive Neuroscience, Leipzig

Lunch

14:30-15:30 “Personality and information processing: Affect-based modulation of integrative versus analytical styles”
Julius Kuhl, Psychology, University of Osnabrück

15:30-16:30 “Dilemmas of the mind: Antagonistic constraints and complementary systems in cognition and action”
Thomas Goschke, Psychology, University of Dresden

Coffee

17:00-18:00 General discussion

Friday, 22nd Nov, 2002

8:30-9:30 “Structure-mapping in learning and reasoning”
Dedre Gentner, Cognitive Science, Northwestern University, Evanston

9:30-10:30 “Concept learning in geometrical spaces”
Peter Gärdenfors, Cognitive Science, Lund University

Coffee

11:00-12:00 “The computational theory of mind”
Brian McLaughlin, Philosophy, Rutgers University, New Brunswik

12:00-13:00 “Explaining semantic productivity”
Peter Bosch, Computational Linguistics and Cognitive Science, University of Osnabrück

Lunch

14:30-15:30 “A philosophical look at compositionality”
Jeff Pelletier, Computing Science, University of Alberta, Edmonton

15:30-16:30 “An abuse of context in semantics for natural language”
Ernest Lepore, Cognitive Science, Rutgers Center for Cognitive Science, New Brunswik

Coffee

17:00-18:00 General discussion
 

Explaining Semantic Productivity
Peter Bosch

What I mean by productivity is linguistic productivity: roughly, the fact that a finite human mind is capable of producing and understanding potentially infinitely many different linguistic expressions on equally many different occasions. Ordinary speakers perform the miracles of productivity on the fly and without even being aware of it: they produce new expressions not encountered before and use old expressions for new purposes.

The received wisdom of the last forty or so years is that recursive syntactic rules and semantic compositionality may be the core of an explanation for productivity. However, infinity of the set of expressions really is not the point, and that’s why recursion can at most be part of the story and is in fact a little misleading as a showpiece of an explanation of linguistic productivity. The point, I want to suggest, is projectibility: projecting properties from the known to the unknown, and in a fashion that other speakers can follow.

One may reasonably doubt that the familiar category-based rules will suffice to account for processes of word formation, there are problems about the compositionality of certain syntactic constructions and, of course, pragmatics isn’t really compositional either. I will not touch upon most of these problems though, but I will concentrate on problems of the context dependence of the semantic interpretation of linguistic utterances. I shall argue that there is no way of accounting for sentence contents, i.e. for the truth conditions sentences have under particular circumstances of utterance, in a compositional a priori fashion. I shall argue that, in addition to familiar category-based compositional mechanisms, representations of linguistic experience are required that cannot be formulated in terms of a denumerable set of categories. Such representations are mapped onto other situations of language use via relations of similarity, and I will argue that category-based rules are here no serious alternative – even though  they are still required for other purposes in a theory of projectibility.

 
The Brain Basis of Language Processing and Language Learning
Angela D. Friederici

The brain areas involved in the processing of semantic and syntactic information during sentence comprehension in the adult native speaker will be specified on the basis of data from functional magnetic resonance imaging (fMRI). Semantic and syntactic processes are supported by separable temporo-frontal networks with a left hemispheric dominance. Second language learners of German show an activation pattern similar to natives for semantic, but not for syntactic processes. The investigation of the temporal course of brain activation by means of event related brain potentials (ERP) indicates that native speakers of German elicit an early left anterior negativity (ELAN) and a late centro-parietal positivity (P600) for the processing of syntactic violations and a so-called N400 component for semantic processes. Again, second language learners demonstrate an ERP pattern similar to natives for the semantic condition, but not for the syntactic condition. The combined data suggest that syntactic processes, in particular, are supported by different brain mechanisms in the adult second language learner compared to the native. This may be due to either the fact that language is learned after the "critical period" for language learning or the fact that second language speakers and natives are not at the same level of language proficiency. To evaluate these two options in a strictly controlled manner an ERP and an fMRI experiment were conducted with adults learning a miniature artificial grammar. ERP indicated that proficient second language learners demonstrate a biphasic early negativity – late positivity pattern suggesting that second language processing is not processed differently from native language processing once high proficiency is reached. An additional fMRI experiment of artificial grammar learning indicated that the initial learning stages are supported by the hippocampus, a brain system known to be involved in general learning but that the late proficiency stage is subserved by a brain area known to be involved in native syntax processing.
 

Concept Learning in Geometrical Spaces
Peter Gärdenfors

Traditional theories of concept formation, such as symbolic or associationist representations, have problems explaining the quick learning exhibited by humans. In contrast to such models, I advocate a third form of representing information that employs geometrical structures. I argue that this form is appropriate for modelling concept learning. By using the geometrical structures of what I call conceptual spaces, one obtains a general way of modelling similarity judgments. I define properties and concepts in term of conceptual spaces. A learning model that shows how properties and concepts can be learned in a simple but naturalistic way is then presented. I also discuss the advantages of the geometric approach over the symbolic and associationist traditions.

References:
P. Gärdenfors: Conceptual Spaces: The Geometry of Thought, MIT Press, Cambridge, MA, 2000.
P. Gärdenfors: "Concept learning: A geometrical model", Proceedings of the Aristotelian Society, vol. 101 (2001), pp. 163-183.

 
Structure-mapping in Learning and Reasoning
Dedre Gentner

Two contributors to human learning and reasoning ability are (1) analogical processes and (2) language and other representational systems. Analogy and similarity can be explained as structure-mapping processes. These play a key role in higher-order learning and reasoning. Carrying out an analogical comparison (1) highlights common structure (thus fostering the extraction of abstract rules and categories); (2) delineates a set of structural correspondences (thus facilitating introspective analysis); (3) focuses attention on alignable differences (thus revealing pertinent distinctions); and (4) invites the projection of inferences between situations (thus permitting rapid change of knowledge).

Analogy – or more generally, structure-mapping – is a bridge between abstract rules and categories on the one hand and experiential patterns on the other. The alignment of common structure is instrumental both in the bottom-up process of induction from instances and in the top-down process of applying a rule to a specific instance. Structural processes are most clearly seen in the formation and application of relational categories such as bisector and carnivore. Two  further implications of this framework are the importance of relational language in human learning and reasoning; and the necessity of structured representations for higher-order cognition.
 

Dilemmas of the Mind:  Antagonistic Constraints and Complementary Systems in Cognition and Action
Thomas Goschke

Throughout the history of psychology and cognitive science, theoretical controversies about how the mind works have often been centered around fundamental oppositions between qualitatively different cognitive architectures and modes of mental functioning. Among the dichotomies that are the focus of ongoing debates are those between associative and rule-based, analytic versus holistic, rational versus experiential, logical versus intuitive, verbal versus imaginal, propositional versus analog, symbolic versus subsymbolic, or explicit versus implicit processing. While such dichotomies have often been interpreted in terms of mutually incompatible views about the nature of cognition, I propose that they rather reflect complementary cognitive systems and processing modes, which have evolved in response to antagonistic functional requirements in the control of adaptive behavior. For instance, in the domain of learning and memory, the antagonistic requirements to gradually extract invariant regularities across a large number of processing episodes on the one hand, and to store specific episodes on the other, have presumably promoted the evolution of multiple memory systems with different computational properties and neural substrates. Likewise, in the domain of action control, the requirement to maintain goals in the face of distractions on the one hand, and the requirement to flexibly switch between goals in response to changing task demands on the other, have presumably promoted the evolution of complementary control systems. I discuss evidence from experimental psychology and cognitive neuroscience for the existence of complementary processing systems and address the question of how the balance between such complementary systems may be regulated.

 
Rules and Patterns: Integrating Different Perspectives of Cognitive Representations
Helmar Gust

In most cases rules and patterns are viewed as an opposition. We will try to overcome this opposition. Starting with a general more philosophical view on the cognitive modeling task, we will sketch a formal framework, where rules and patterns can be treated in a uniform way. This framework uses ideas from topolopy, rough sets, coalgebras, ontologies, and classification methods like support vector machines. We will interpret patterns essentially as bundles of experienced state changes establishing a topology on a subspace of states. On the other hand, rules define classes of predicted state changes in this framework, where both the preconditions and the relations between the current and the target state are specified in a categorical way. Integrating both leads to consistency constraints between the pattern driven and the categorical perspectives.

Contents (preliminary):

           0.   The Cognitive Modeling Task
           1.   The Concepts of Rules and Patterns
           1.1 What are Rules
           1.2 What are Patterns
           2.   Ontologies from a topological point of view
           2.1 Connecting Intension and Extension
           2.2 Relevant Instances and Closure Operations
           3.   Coalgebras and Bisimulations
           4.   Conclusions and Future Work

 
Personality and Information Processing: Affect-based Modulation of Integrative versus Analytical Styles
Julius Kuhl

A neurobiologically informed functional-design approach to personality is presented. In contrast to cognitive approaches, personality functioning focuses on effects of cognitive systems that cannot be understood unless interactions among cognitive macrosystems are taken into account. Interactions between parallel-intuitive and sequential-analytical processing systems are modulated by emotional states. Personality systems interaction theory (PSI theory) specifies the way emotion and motivation modulate communication among parallel and sequential cognitive systems. Moreover, PSI theory distinguishes between two analytical and two intuitive systems: In addition to the familiar low-level (irrational) form of intuition, a high-level (rational) form of intuitive processing is described. In a similar vein, analytical processing is not confined to the familiar high-level form of sequential processing (e. g.,  analytical thinking), but is extended by a low level form of analytical processing. Experimental evidence is presented that demonstrates advances in the level of processing accomplishable by a cognitive architecture that consist of two (or even four) processing system that differ substantially in their processing characteristics.

 
An Abuse of Context in Semantics for Natural Language
Ernest Lepore

Indexicals are linguistic expressions whose meaning remains stable while their reference shifts from utterance to utterance.  Paradigmatic cases are ‘I’, ‘here’, and ‘now’.  Recently, a number of linguists/philosophers/cognitive scientists have argued that various constructions in our language harbor hidden indexicals. I say 'hidden' because these indexicals are unpronounced, even though they are alleged to be real linguistic components.  Constructions taken by some authors to be associated, or to ‘co-habit’, with hidden indexicals include: definite descriptions and quantifiers more generally (hidden indexical refers to a domain – Davies, Westerstahl, Soames, Higginbotham, Stanley and Williamson), propositional attitude verbs (hidden indexical refers to a mode of presentation – Richard), comparative adjectives (hidden indexical refers to comparison classes – Partee, Kamp, Ludlow); subjunctive conditions (hidden indexical refers to world – Kratzer, Thomason).
I am skeptical of all such posits, first, because evidence typically proffered in support of their existence I believe is better accounted for in other ways; and secondly, because every alleged case familiar to me is flawed, or so I will argue.  I begin by concentrating on a recent, influential argument for hidden indexicals – the so-called Argument from Binding.  I present a reductio of it.  I then present two requirements any indexical – hidden or otherwise – should satisfy, illustrating how various alleged hidden indexicals fail to do so. It's my working hypothesis that all alleged hidden indexicals fail these tests.

 
The Computational Theory of Mind
Brian McLaughlin

The central questions of the philosophy of mind are the nature of mental phenomena, and how mental phenomena fit into the causal structure of reality.  The computational theory of mind aims to answer these questions.  The central tenet of the theory is that a mind is a computer.  According to the theory, mental states and events enter into causal relations via operations of the computer. The main aim of the theory is to say what kind of computer – what kind of computational mechanism – a mind is.  The answer is still unknown.  Pursuing it is the main research program of the theory.
In the most general sense, a computer is, roughly, a system of structures functionally organized in such a way as to be able to compute. These structures, their functional organization, and the basic modes of operation of the system when it computes comprise the functional architecture of the computer. The two tasks of the computational theory of mind are: (1) to identify the functional architecture of the computing system that grounds our mental abilities and (2) to explain how those abilities are exercised via operations of the system.  The tasks are related.  The explanation of how operations of the system constitute exercises of our mental abilities will justify the claim that our possession of those abilities consists in our being at least partly constituted by the system.
Computationalists hold that the functional architecture of the computing system that grounds our mental abilities resides in our brains.  There is, however, no consensus as to what even the general character of that architecture is. The symbol system paradigm and the connectionist paradigm are the two dominant research paradigms within the computational theory of mind.  They differ primarily in what kind of computer the mind is assumed to be, and thus in the kinds of functional architectures explored.  The symbol system paradigm presupposes that the mind is a kind of automatic formal system, while the connectionist paradigm presupposes that it is a system of connectionist networks.  In my talk, I will first discuss some common ground between these two paradigms, and then I will discuss the issue of whether the connectionist paradigm can explain the systematicity of thought without implementing a symbolic architecture.
 

 
A Philosophical Look at Compositionality
Jeff Pelletier

Although it sounds like the start of a bad joke, I think it is true to say that there are two types of people in the (academic) world: those who look to the 'parts' of objects/phenomena they wish to explain or understand, and those who look to the way an object/phenomenon 'fits in with' other aspects of the world.  Let's call these two groups "atomists" and "contextualists".  There are various subtypes within each of the groups, and I wish to pick out the 'compositionalists' from within the atomistic group for further discussion.
My interest is with their views on the explanation of "meaning" in the realm of theories of natural language understanding.  As a first pass, their view is that "the meaning of a complex expression is a function of the meanings of the parts, plus the mode of combination of those parts".  This formulation leaves a number of issues open, and gives rise to differing accounts of the issue of semantic compositionality.
Besides trying to get straight on just what semantic compositionality is, I intend to discuss certain linguistic phenomena that have been brought forward as examples that show it to be "empirically false".  I will consider what compositionalists might say about such examples, and I will consider whether the strategies they might invoke amount to showing that there is no empirical issue involved at all.  In this realm, I will also look at certain formal accounts of semantics that purport to show that "any semantics can be converted to a compositional semantics"... and that therefore there is no empirical content to the compositional/non-compositional debate. In the end I will advocate a certain attitude toward semantic phenomena: they are best explained by an atomistic, but non-compositional theory.

 
Creativity and brain laterality: poly- vs. mono-semantic ways of thinking
Vadim S.Rotenberg

After a short critical review of the modern concepts of brain hemisphere functions a theory will be proposed that explains the difference between hemisphere functions in terms of the opposite types of the organization of relationships between objects and events. Left hemisphere is suggested to be responsible for the organization of monosemantic context while right hemisphere is responsible for the organization of the polysemantic context. This approach helps to solve the contradictions related to the previous concepts. The difference between Self-Image and Self-concept as well as the role of both hemispheres in different psychological functions will be displayed in the frame of this approach. In details will be discussed the role of both hemispheres in the process of creativity including also reasons of failures in this process. It will be shown that the organization of the polysemantic context in contrast to the monosemantic context does not require physiological brain activation. It will be emphasized that the development of human left hemispheric logical thinking and consciousness determines the corresponding development of the equally high right hemispheric function and that the formation of the polysemantic context is a unique human function as important as consciousness. Finally, it will be discussed the role of its functional insufficiency in the pathogenesis of mental and psychosomatic disorders.