Herbstakademie Ascona 2003 (16.10.03/Tsch)

 ABSTRACTS (oral presentations) (move forward to poster presentations)

SATURDAY, October 18th

Hermann HAKEN Stuttgart (D)
In how far can Dynamic Systems Theory, Synergetics and Related Approaches Contribute to Cognitive Science?
As it is witnessed, for instance, by the recent book "The Dynamical Systems Approach to Cognition: Concepts and Empirical Paradigms Based on Self-Organization, Embodiment, and Coordination Dynamicsî edited by W. Tschacher and P. Dauwalder the dynamic systems approach is more and more invoked to tackle cognition. Therefore it might be worth while to discuss dynamic systems theory and related approaches with respect to their potentialities and limitations concerning cognition. Thus I will discuss fields such as dynamic systems theory, bifurcation theory, chaos theory, general system theory, synergetics, the theory of stochastic processes and some more.
One may safely say that dynamic systems theory originated from the mathematical treatment of the motion of celestial bodies, in particular the motion of the planets around the sun. Basically the motion, i.e. the dynamics, is described by differential equations that determine the velocities and positions of the individual bodies because of the forces exerted on them. The equations are deterministic. The development of dynamic systems theory includes both quantitative and qualitative approaches and incorporates branches such as chaos theory and bifurcation theory. Among the concepts are stability, instability, and attractors. This approach ignores, however, the role of fluctuations that may be quite important. In modern science, whenever a mathematical treatment of systems whose states change in the course of time is asked for, dynamic systems theory is taken into consideration.
The interdisciplinary field of synergetics originated from laser physics. It is based on a mathematical approach that, taking into account deterministic and stochastic processes, deals quite generally with the formation and functioning of structures, irrespective of the nature of the material substrate and thus applies to a great variety of disciplines. The methodology of synergetics allows us to unearth far reaching analogies between otherwise quite different systems.
Finally, I will discuss neural network approaches. They are mainly based on comparatively simple rules about the action of the individual elements (model neurons), but because of the network character require computers. More recently, analytical treatments were developed in particular for studying effects of synchronization between groups of neurons. Without going into the technical details of all these approaches I will discuss their applicability to various aspects of cognitive science, in particular to the problem of emergence.

SUNDAY, October 19th

Andy CLARK Bloomington (USA)
Dynamics and De-Coupled Representation: Balancing Words, Tools and Inner State
How, if at all, can embodied and dynamical approaches deal with traditionally ëcognitiveí phenomena? One key move may be to take very seriously the role of human activity and human-built structures in altering the way difficult problems are presented and solved. Our best practices and artifacts, on this view, enable the same basic adaptive strategies to tackle perception, action and high-level reason (Clark 1997, 2001). An important challenge to this view depicts these practices as themselves requiring the use of new forms of ëde-coupledí internal representation (Sterelny, in Press). I critically assess this claim, and attempt to sketch an alternative, integrative model of the relation between high-level reason and sensori-motor performance systems. In this sketch, I pay special attention to the cognitive role of words and public symbols. Fully de-coupled ways of knowing, I conclude, need not demand strong forms of de-coupled internal representation in addition to the new resources provided by public language symbols such as words and numerals.

Monica COWART & Michael MASCOLO North Andover (USA)
Open Constraints: Varieties of Scaffolding in the Emergence of Higher-Order Action
Dynamic systems and embodied models of cognition have been successful in providing explanations of how lower-order actions (e.g., reaching) emerge through dynamic coupling between individual and environment. A central concern, however, is how such models explain the emergence of higher cognitive capacities. Clark (2001) has suggested three ways to address this problem. One can (a) propose hybrid models mixing principles of classical and embodied approaches, (b) bootstrap models of higher-order cognition by extending existing models of embodied cognition, or (c) elaborate models of embodied cognition through analyses of how external cognitive aids mediate the emergence of higher thinking. To increase the viability of this third option, we analyze how the concept of scaffolding can explain how individual-environment coupling prompts the formation of higher-order action. We begin with the assertion that action, the objects of action, and context form an integral system. As part of the functioning of this system, scaffolding refers to the ways in which context and the use of objects support the emergence of novel and more complex modes of acting. Drawing upon studies on the microdevelopment of everyday acting and thinking (i.e., tying shoes; adding integers), we elaborate a taxonomy embracing three categories of scaffolding: social scaffolding (e.g., direction from adults), self-scaffolding (e.g., cognitive scaffolding, bridging, analogical mappings), and ecological scaffolding (e.g., contextual, positional, task-scaffolding). In so doing, we provide precise examples of each form of scaffolding and explain ways in which they contribute to the formation of higher-order action within dynamic and embodied models of cognition.

Wayne CHRISTENSEN Wien (A)
Self-Directedness, Integration and the Foundations of High Order Cognition
Situated cognition has provided important advances in understanding the control of adaptive behaviour, but the architecture and evolutionary appearance of high order cognition remains a difficult outstanding problem. In this paper I will present an account of self-directed agents that explains the appearance of higher order forms of cognition through the evolutionary elaboration of integrative neural systems involved in producing flexible, coordinated behaviour in complex task environments. I will suggest that the appropriate focus for understanding high order cognition should be on situational awareness rather than representations or concepts per se. Standard approaches to animal cognition draw a distinction between associative learning mechanisms and behaviour mediated by representations/ concepts, with the label ëcognitioní usually ascribed exclusively to the latter. The background assumption is that cognitive processes take the form of some kind of propositional inference. However, whilst there is good empirical evidence for representational ability in animals, propositional inference is a poor model for the multidimensional relational processing involved in situational awareness. It is therefore more plausible to view representations and concepts as emerging from, and embedded in, the multifaceted situational modelling processes I describe. In turn this means that the view of high order cognition as representationally driven is misconceived, and should be replaced with a representational-involvement paradigm that recognizes representation as acting in conjunction with a spectrum of non-representational processes.

Rolf PFEIFER Zürich (CH)
The Dynamics of Embodiment in the Emergence of Cognition
The classical approach to artificial intelligence, also called the cognitivistic approach, views intelligence at the level of algorithms or computer programs. Algorithms, by definition, do not have any interesting kinds of dynamics. With the advent of embodiment ó the incorporation of morphology, materials, control, and interaction with the environment into a theory of intelligence ó the situation has changed dramatically. If we are to build embodied agents we must understand the interaction and tradeoffs between these aspects, which is the essence of the "principle of ecological balanceî, one of a set of design principles for intelligent systems (see below).
Rather than focusing on the neural substrate only, in the embodied approach the focus is on the complete organism which includes morphology (shape, distribution and physical characteristics of sensors and actuators, limbs, etc.) and materials. Often, given a particular task environment, if the morphology and the materials are right, the amount of neural processing (or more generally, control) required may be dramatically reduced. Stated differently, we not only have to take the neural dynamics into account but also the physical dynamics of the agent and how it interacts with the neural one.
The synthetic methodology, i.e. the methodology of "understanding by buildingî, which serves to understand natural phenomena such as walking, perception, or memory, proceeds by building physical systems, and then tries to abstract general principles of intelligent behavior. In a number of scientific disciplines ó biology, neuroscience, psychology, and robotics ó there has lately been a surge of interest in human development, from infants to adults. In developmental robotics, where robots are employed to elucidate developmental processes, robots interact with their environments over extended periods of time, mimicking a developmental process, which enables the robot to acquire itís own "personal historyî. In contrast to humans, this "personal historyî can be completely recorded into time series files and subjected to statistical and information theoretic analyses. This opens up entirely novel possibilities for cognitive science and provides, for example, the basis for studying the so-called "symbol grounding problemî, how symbols come to be used to convey meaning. It turns out that the ability to move and interact with the real world, is an essential enabler of meaning acquisition.
In the presentation, I will introduce a number of abstract principles that characterize intelligent embodied systems, i.e. the principle of "cheap designî, the principle of "sensory-motor coordinationî, and the principle of "ecological balanceî. They will be illustrated with examples from our own research and from work by other research laboratories. I will speculate about how embodied artificial intelligence might contribute to a dynamical theory of cognition and outline a few issues that I consider essential for future research.

Gabriel GOMEZ Zürich (CH)
Simulating Development in Real Robots
In order to understand the emergence of cognition, we cannot limit our investigations to control architectures only, but we must include considerations on physical growth, change of shape, and body composition, which are salient characteristics of maturation. Due to the current state of technology we cannot have physically growing robots, therefore we propose a method to Çsimulateë development at the level of the sensory, motor, and neural systems.
We present a quantitative investigation of the effect of developmental progressions on sensory-motor coordinated behaviors. The development is Çsimulatedë by increasing the resolution of the cameras, freezing and releasing degrees of freedom, and adding nodes to the neural control architecture.
An initial experimental setup with reduced vision, noisy motor commands, low number of degrees of freedom, a few pressure sensors and a neural control architecture with a reduced number of nodes gradually becomes an experimental setup with good vision, Çdenoisedë motor commands, large number of degrees of freedom, a larger number of pressure sensors and a neural control architecture with a sufficient number of nodes.
The preliminary results show that a system following this developmental approach, learns faster than a system starting out from the full high resolution - high precision system and shows a more robust behavioral performance against environmental interaction.
Our experiments are performed using an industrial robot manipulator (Mitsubishi MELFA RV-2AJ) with six degrees of freedom, a stereo color active vision system, and a set of tactile sensors mounted in the robotës gripper.

Roul Sebastian JOHN Osnabrück (D)
EROSAL ó An Empirical Robot Study on Animal Learning
At the last, tenth, Herbstakademie in Kloster Seeon, we presented a talk on our ideas for a project that was aimed at avoiding the "categorization trap in self-organized learningî. Our presentation centered around the idea to use an autonomous agent (mobile robot) as a model to explain the peculiar results of our visual discrimination learning experiments with chickens (Gallus gallus fd), and to evaluate this model empirically by testing it in exactly the same experiments within the same experimentally controlled environment that was used for the animal subjects before. This new empirical methodology was called "comparative cognitive roboticsî. As we pointed out, an important requirement for such a real-world modelling approach would be to enable the robot model to develop its own categorizations. Whereas current simulation- and even robot-based approaches easily get caught in the "categorization trapî by equipping their models with the features and categories that seem adequate from the point of view of the human engineers, we wanted our model to operate without any pre-coded human categories. At this coming Herbstakademie at Monte Verità, we would now like to present the results we were able to achieve in our project so far. We constructed a robot model that is able to operate autonomously under the same experimental conditions as the animal subjects. The control architecture was kept as parsimonious as possible. Most importantly, an exemplar-based learning mechanism similar to the ones presented by Richard Shiffrin at the 9th Herbstakademie at Monte Verità was used. But unlike previous exemplar-based models, and in line with our earlier criticism of these models, our model operates directly on the sensor data, without assuming a process of matching this data on pre-coded, human-like categories. Actually, our robot model is responding differently to different stimulus categories although no category representation at all is being formed within the model. Categories only exist on the level of description by an outside observer. We call this approach "categorization without categoriesî. In the experiments replicated so far, the performance of the robot model generated by this mechanism did match with the empirical data gathered from the animal subjects not only qualitatively, but even quantitatively. Taking this concrete project as an example, we will discuss the pros and cons of using autonomous agents as empirical models of self-organized learning in animals.

MONDAY, October 20th

Jürgen KRIZ Osnabrück (D)
Imagination ó The Teleological Principle of Attractors in Psychotherapy and Human Activities
The central term in systems theory ó Çattractorë ó can be understood as a rediscovery of the old concept Çteleologyë. This concept, however, does not only make sense in the natural sciences. Moreover, especially in the area of human cognitive and interactive processes a famous phenomenon is related to attracting and teleological forces: imagination. Similar to the difference between play and game, imagination is in contrast to Çgoal directed planningë more open towards new aspects in the development of the whole system-environment-dynamics. This can especially be used in psychotherapy and other human activities.

Tony PRESCOTT Sheffield (UK)
Stability and Variability in Young Childrens' Drawings of the Human Figure
Most children between the ages of 3- and 5- years old undergo a transition in their production of human figure drawings from depicting a 'tadpole' figure (a figure lacking a body/torso as a distinct contour) to drawing a more conventional figure in which a body/torso is included. Past studies of 'tadpole' drawers have found that they (i) draw the tadpole form with reasonable consistency; (ii) can often assemble a conventional figure from parts, (iii) can often draw a conventional figure if given instructions, but will usually revert back to tadpole drawing afterwards; (iv) know about bodies ('tummys') and can often point to them on their drawings; and (v) may think that 'tadpole' figures are better representations of people than conventional figures. This paper will argue that 'tadpole' drawing shows many of the properties of a stable attractor and that the tadpole/conventional shift therefore provides a valuable setting in which to investigate phase transitions in cognitive development. I will review recent data collected in Sheffield concerning the variability of childrens' figure drawing activity close to the tadpole/conventional transition. These results show (i) that figure drawing is sensitive to interventions that raise or lower the production difficulties of the drawing task (ii) that drawing near transition also responds to interventions that raise the 'salience', or subjective importance, of the torso whilst holding production difficulty constant; and (iii) that a child's understanding of the biological functions of the body may act as an important control parameter for their figure drawing behaviour.

Wolfgang TSCHACHER Bern (CH)
Gestalt Formation Processes in Cognition
The notion of cognition as emerging from a self-organizing complex system is advocated here. I argue that the relationship between a complex cognitive system and certain properties of the environment (affordances) is analogous to the relationship between any self-organizing complex system and its non-equilibrium control parameters (gradients) as described within the framework of Haken's synergetics. Pattern formation in a self-organizing system tends to reduce the systemís distance from equilibrium (i.e., decrease the gradients). Thus, self-organization is an optimizing process in that it chooses the best ways to decrease gradients (see Tschacher et al., 2003, in the Ascona volume).
In the context of cognition this property ñ the "intention" to achieve optimality ñ can be regarded as the foundation of Gestalt formation and, generally, intentionality. I will illustrate these ideas by results of a study focussing on Gestalt perception.

Rafael NUÑEZ San Diego (USA)
Mind, Motion, and Abstraction: The Embodied Cognitive Foundations of Mathematics
Mathematics is a highly technical domain, developed over millenia, and characterized by the fact that the very entities that constitute what Mathematics is are idealized mental abstractions. These entities cannot be perceived directly through the senses. Even the simplest entity in, say, Euclidean geometry (i.e., a point, which is dimensionless) can't be actually perceived. This is obvious when the entities in question are related to infinity (e.g., limits, least upper bounds, mathematical induction, infinite sets, points at infinity in projective geometry and so on) where, by definition, no direct experience can exist with the infinite itself (Nuñez, 2000). Lakoff and Nuñez (2000) showed that most of these idealized abstract technical entities in Mathematics are created by human cognitive mechanisms such as conceptual metaphors, conceptual blends, fictive motion, aspectual schemas, and so on. In this talk I will discuss how fictive motion (Talmy, 1998) is essential in the creation and conceptualization of many mathematical entities, such as functions, limits, series, convergence, continuity, real numbers, least upper bounds, and fractals. Fictive motion is a fundamental embodied cognitive mechanism through which humans conceptualize static entities in dynamic terms, as when we say "the road goes along the coast". In classic examples like this one often the Figure has the potential to enact movement (e.g., a car moving along that road). In Mathematics proper, however, the Figure has always a metaphorical component (i.e., the Figure as such can't be literally capable or incapable of enacting movement). Through the study of classic texts in Mathematics, as well as of detailed video analysis of spontaneous gestures synchronous to speech (McNeill, 1992), I will analyze three main issues:
1) How fictive and metaphorical motion play an essential role in bringing many mathematical entities into being.
2) How modern Mathematics (especially after the 19th century), when dealing with important mathematical concepts, is at odds with cognitive mechanisms such as fictive and metaphorical motion that helped creating these very concepts.
3) The importance of micro-analysis of gesture production in the study of human abstraction.

Annette HOHENBERGER Frankfurt (D)
The Emergence of Linguistic Structure through Language Processing and Acquisition
Language is a cognitive module with stunning structural complexity and therefore an excellent object for studying the emergence of patterns from a dynamical point of view. In a dynamical perspective, linguistic structure is not innate but emerges as the result of the interaction of task requirements, processing and acquisitional constraints. Linguistic structures are recursive, asymmetric, binary-branching, and projective. This form is the optimal solution given the task to convey information in real time, given a one-dimensional output channel, and given a cognitively immature subject that acquires structures under the constraints of memory and other limitations acting as control parameters.
The structural units of language - segments, morphemes/words, phrases - naturally fall out as order parameters in processing and acquisition. They are the result of recursive embedding operations on various levels of processing and acquisition: phonological, morphological/lexical, and syntactic.
I will provide evidence from language processing and acquisition that corroborates a fully-fledged dynamical account of language and point out convergent lines of development in recent generative linguistics and dynamical systems theory. With his Minimalist Program (Chomsky 1995), Chomsky has recently initiated a major re-conceptualization of the language faculty. Constraints on the grammar have been relegated to its interfaces, namely Phonetic Form (PF), i.e. the Articulatory-Perceptual (AP-) System, and Logical Form (LF), i.e. the Conceptual-Intentional (CI-) System. Especially the shift of focus on the AP performance systems makes Minimalism an attractive linguistic theory that can account for linguistic structure in terms of procedural constraints.

Jürgen BROSCHART Hamburg (D)
The Dynamics of Grammatical Case Marking
Any conceivable state-of-affairs to be expressed by language must be categorized in terms of a comparatively small inventory of relations such as "agent", "patient", "subject", "object", "topic", "comment" or the like. Because of this fundamental set of relations, the study of case marking in the languages of the world is suspected to yield important insights into the nature of human cognition.
However, despite almost 3000 years of research, even basic facts of case marking still remain unsolved. This is due to the fact that a traditional "structuralist" approach is unable to handle variable and dynamic features such as the great variety of uses of the same sign, the use of different case markings for the "same" meaning, and the shift of meaning of the same sign over time. As a result, completely different denominations such as "subject", "ergative" and "instrumental", or "ablative" and "genitive" may or may not be considered identical in case theory, and form and function seem to fall apart.
An "interactivist" approach specifically designed for the description of the dynamics of case marking is able to solve these problems. The complexity of case marking appears to result from rather simple principles of "changing perspective" with respect to "normal" and "idiosyncratic" preferences of interaction.

Annemarie PELTZER-KARPF Graz (A)
The Interplay of Non-Linear Processes in Early Language Development
This paper features temporal asynchrony in language development. The non-linear approach proposed unites developmental cognitive neuroscience and dynamic systems theory. We start from the assumption that maturational factors and experience play complementary roles in forming specialized systems which display different degrees of experience-dependent modification and operate at different time scales. In this particular context language development is seen as the self-organization of dynamic systems showing degrees of persistent order at first dominated by the search for coherence, followed by the reorganization into different clusters and finally the emergence of coherent patterns with large internal coupling strength and stability. The basic assumption is that the time-course of these behavioural phase-shifts is linked to system-specific developmental changes in the central nervous system. We shall follow the chaotic itinerary of individual systems up to the age of three and try to locate their neurobiological correlates. Of prime interest are the following postnatal cellular events: a dramatic increase in glucose metabolism which exceeds adult values during childhood; an overshoot phase of synapses and neurotransmitter receptors around age 2 followed by a system-specific decrease; the scaled myelination of nerve sheaths and the growth of long-range fibres from age 2 which are essential for figure-ground segregation and pattern formation in various domains. Psycholinguistic charts drawn from long term studies will illustrate how the joint activity of biological and general cognitive factors affects the dynamics of language development in ordinary and exceptional circumstances. (The data were collected in the course of two projects funded by the Austrian Nationalbank, 4244 & 6179 and a FWF-project P 10250-SPR)

Kerstin SANDER Magdeburg (D)
Emotional Preexperience Influences Semantic Specificity of Human Amygdala Activation ó An fMRI Study
The role of the amygdala is not restricted to the processing of fear as a biologically relevant signal. As we recently showed, listening to laughing and crying activates the human amygdala bilaterally and to a similar amount. In this fMRI study, we investigated the semantic specificity of amygdala activation by presenting laughing and crying either in original form or in a reversed manner. The latter also served as an acoustical control because original and reversed stimuli only differed in their temporal structure, while their spectral content was identical.
Twenty right-handers were scanned in a 3T MR scanner while listening to laughing and crying presented alternatingly in an fMRI block-design. Stimulus files were counterbalanced for emotional quality (laughing, crying), gender of voice (female, male), and sequence of presentation mode (original followed by reversed or vice versa). Subjects had to detect randomly introduced upward pitch-shifts.
The perception of laughing and crying in both presentation modes activated the amygdala bilaterally. Interestingly, the activation was modulated by the sequence of presentation mode. Hearing first the original laughing and crying lead to similar, bilateral activation by both original and reversed stimuli. In contrast, hearing reversed stimuli first, caused stronger right amygdala activation by these, while stronger left amygdala activation was obtained for the original stimuli. This shift in lateralization might depend on the familiarity with emotional stimuli, whereby familiar stimuli engage primarily the left amygdala, but unfamiliar stimuli produce an orienting response in the right amygdala. However, this pattern itself seems to depend on emotional preexperience.

Márk MOLNAR, Zsófia GAAL, Zoltán NAGY, Cornelis STAM Budapest/Amsterdam (HUN/NL)
Nonlinear Complexity Analysis of the EEG and Cognitive Evoked Potentials in Stroke and Vascular Dementia
Introduction: The aim of the study was to evaluate the scalp distribution of cognitive evoked potential components and EEG-complexity measures in cerebrovascular disease, i.e. stroke and vascular dementia, for the assessment of the functional condition of the patients and compare it with normal controls.
Materials and methods: EEG and cognitive evoked potentials were recorded in patients with unilateral ischemic stroke (IS) and in those with vascular dementia (VD). Of the cognitive potentials the P3 wave and the CNV were recorded. For the complexity analysis of the EEG the methods used include the point-correlation dimension (quantifying dimensional complexity) nonlinear cross-prediction (quantification of nonlinearity based on amplitude and time asymmetry), synchronization likelihood (measurement of nonlinear and linear components of synchronization), and Omega-complexity (quantification of spatial synchrony based on covariance matrix analysis).
Results: It was found that the effect of white matter lesions on synchronization depended upon the actual frequency band investigated. No consistent relationship was observed between amplitude-, and latency changes of cognitive evoked potentials, caused by the lesions, and the degree of the above changes of complexity.
Conclusions: Changes of the indices of nonlinear complexity could more reliably used for the assessment of pathological changes, caused by cerebrovascular disease, than those of the cognitive evoked potentials. Synchronization in the beta and gamma range seems to be the most likely to be affected in IS and VD. Nonlinear EEG indices appear to be a powerful tool in the assessment of brain function.

Steve BOKER Notre Dame (USA)
Latent Differential Equations Applied to Dynamical Systems Models of Behavior
While dynamical systems theory has provided appealing process--oriented metaphors for continuously evolving and adapting behavior, statistical methods for testing and potentially falsifying these theories has only recently become available. Psychometric measurement has developed powerful techniques such as confirmatory factor analysis in order to extract latent constructs from observed multivariate measures. But such models have tended to either use interindividual differences or time independent intraindividual differences to identify latent constructs. If the psychological construct of interest is a process, then the structure of its dynamics as manifested from repeated observations of many individuals may be used to identify latent variables. Models of process dynamics can be built by formalizing the relationships between derivatives of these variables as differential equations. These differential equations models can then be fit to repeated observations data using structural equations statistical modeling software. We present some methods used to construct and fit such models as well as an application in which the coordinative structure of motion captured interpersonal movement is used to test a cognitive theory concerning the formation of segmentation from repeating auditory stimuli.

Karl-Heinz LEIST München (D)
Riding a Pedalo and Learning to Ride it: Comparisons of Hakenís Synergetic Approach and Action Theory
Haken (1996) has shown that pedalo riding and learning to ride a pedalo can be explained in the framework of his synergetic theory of motor coordination and motor learning: At the beginning of the learning process, riding e.g. is governed by three order parameters, as soon as the process is performed only one order parameter is involved. From the perspective of action theory, Leist (1984, 1998) found comparable results. When riding is learned feedforward control is characterized by one overall anticipation "Making round to drive continouslyî. Learners are able to demonstrate what this means when they perform the transition from riding forwards to riding backwards. The ensuing question is how results of research from these two perspectives can be linked in order to interpret Hakenís synergetic approach in the framework of action theory (and vice versa). First results will be presented.

TUESDAY, October 21st

Mark BICKHARD Bethlehem (USA)
The Emergence of Intensionality in Certain Far From Equilibrium Systems
The realm of intensionality, normativity, and modality seems cut off from the world of basic facts and substances. This has posed a perplexing problem since the Pre-Socratics. I will first address why this is such a perplexing problem, then the general nature of the solution, and then proceed to some particular models of the emergence of normative function and representation ñ and the involvement of modality in representation.

Harald ATMANSPACHER Freiburg (D)
Acategorial Mental States
Mental representations are based on categories in which the state of a mental system is stable. Acategorial states, on the other hand, are characterized by the fact that instabilities play an important role for their understanding. A precise and refined terminology for the description of categorial and acategorial mental states and their stability properties is introduced within the framework of the theory of dynamical systems. These concepts are illustrated by selected empirical observations in neuropsychology and neurophysiology.

Hans DIEBNER Karlsruhe (D)
A Model of how Brains Model the World
We introduce and discuss within the framework that we call operational hermeneutics an adaptive cognitive model that allows for the simulation of the perceived part of the world. Very much like the hermeneutic circle the model has a bootstrapping character concerning the update of knowledge. The model as well as the link to hermeneutics are motivated through a historical back view to relevant features of dynamical system theory and to the Bayesean inference principle in statistics. We also discuss the relevance of the model for socio-biological behavior through the simulation of dynamical states of other cognitive systems which introduces an interface problem.

Karl GRAMMER Wien (A)
Dynamic Systems and Inferential Information Processing in Human Communication
Research in human communication on an ethological basis is almost obsolete. The reasons for this are manifold and lie partially in methodological problems connected to the observation and description of behavior, as well as the nature of human behavior itself. In this talk, we present a new, non-intrusive, technical approach to the analysis of human non-verbal behavior, which could help to solve the problem of categorization that plagues the traditional approaches. We utilize evolutionary theory to propose a new theory-driven methodological approach to the "multi-unit multi-channel modulationî problem of human nonverbal communication. Within this concept, communication is seen as context- dependent (the meaning of a signal is adapted to the situation), as a multichannel and a multi-unit process (a string of many events interrelated in "communicativeî space and time), and as related to the function it serves. Such an approach can be utilized to successfully bridge the gap between evolutionary psychological research, which focuses on social cognition adaptations, and human ethology, which describes every day behavior in an objective, systematic way.

Hani MONK-VITELSON, Ilan GOLANI, Chaim PICK, Juval PORTUGALI Tel Aviv (ISR)
Spatial Behavior of Prewalking Infants: Preliminary Findings, Preliminary Questions
This presentation reports on preliminary results from a set of experiments conducted within the frame of an ongoing research concerning infantsí locomotion prior to independent walking. The paradigm under which the study is designed conceives of the infant, his/her mother and the surrounding space as elements of a single dynamic system. The conceptual framework is related to Attachment Theory and the empirical framework to ethology. That is, to observations and recording of unrestrained behavior, in this case, exploratory behavior of infants exposed to unfamiliar environment at the presence of their mothers. 30 mother-infant pairs were video-recorded in an empty gym (130m2). Mothers are being asked to stay seated at gymís center, while infants are free to move. Infantsí excursions throughout the hall are tracked from the videocassettes by means of the Noldus Ethovision® system, and momentary mother-infant distance is computed. In our presentation we will present preliminary results from the above experiments, discuss questions concerning the dynamics of forces acting upon the infant and mother during the experimental exposure, and suggest a preliminary interpretation based on a comparison between experimented and real life situations.

Juval PORTUGALI Tel Aviv (ISR)
Cognitive Maps and Order Parameters: What can they tell about each other?
A common approach in the study of cognitive maps is to examine cases where the proper operation of the mind/brain entails distortions between the cognitive and the real map. Such cases have been termed systematic distortions in cognitive maps. Studies in this domain have identified distortions that are due to hierarchy, alignment, rotation, anchoring and few others. It is common to see these distortions as decision heuristics employed by the mind/brain in situations of partial or insufficient information. The paper presents new experimental results suggesting that such heuristic play a role similar to the role of attention parameters in the synergetic paradigm of pattern recognition. This interpretation brings to the fore an interesting question, namely, ëwhere do order parameters come fromí? The answer suggested by the paper is that order parameters emerge out of the interaction between internally represented heuristics and the externally represented information afforded by the environment under consideration.

Luc CIOMPI Lausanne (CH)
Dynamic Effects of Emotions in Social Stability and Creativity According to the Theory of Fractal Affect-Logic
The theory of fractal affect-logic (that has been presented at the Herbstakademie 1997 in Gstaad) postulates that interactions between emotions and cognition are self-similar on different individual, microsocial and macrosocial levels. Linear and nonlinear conditions of both social stability and social innovation are analysed from this perspective, on the basis of recent neurobiological findings and system-theoretic conceptualisations including Hermann Hakenís synergetics and the concept of autopoiesis of self-referential psychosocial systems developed by Niklas Luhmann.
At variance with Luhmannís approach, emotions play a crucial role in social dynamics according to fractal affect-logic (Ciompi 1982, 1997, 1999, 2002). Emotions are understood as directed energetic states of evolutionary origin that exert numerous general and specific operator-effects on collective thought and behaviour. Directed emotional energies mobilise and organise social dynamics by the mechanism of emotional contagion. Under normal conditions, most operator-effects of emotions reinforce social conformity and stability. When system-internal emotional tensions increase to a critical level, they can however provoke sudden nonlinear modifications of the global patterns of social functioning. The level of emotional tension thus functions as control parameter for social innovation and creativity, whereas formerly marginal patterns of feeling and thinking become new order parameters or affective-cognitive attractors that "enslaveî collective behavior. The influence of the specific emotional quality of such order parameters is further analysed.

Peter UHLHAAS Frankfurt (D)
Gestalt Perception in Schizophrenia Spectrum Disorders
Gestalt perception is a paradigmatic example of self-organisation, where, in computational terms, novel input produces novel output as the result of the interaction between organisational processes (Watt & Phillips, 2001). This view is supported by recent research in neuroscience in which Gestalt perception has been related to the dynamic formation of cell assemblies based on synchronization of neural activity (Phillips & Singer, 1997). The study of Gestalt perception in schizophrenia spectrum disorders may, therefore, contribute to the conceptualization of the pathophysiology of such disorders as involving deficiencies in the self-organisation of cognitive and neural processes.
Phillips & Silverstein (in press) have proposed that dysfunctional Gestalt perception in schizophrenia spectrum disorders is representative of a wider disturbance in grouping of contextually related information which manifests itself in the characteristic disorganisation of mental processes. From this viewpoint, dysfunctional Gestalt perception in schizophrenia spectrum disorders involves both reduced synchronized oscillations in the gamma-band range and NMDA-receptor hypofunction.
The purpose of the present research was to examine Gestalt perception in schizophrenia spectrum disorders in clinical and non-clinical subject populations to identify the mechanisms, clinical and cognitive correlates of dysfunctional Gestalt perception. Three tasks were selected that represent robust examples of Gestalt perception. The results can be summarized as follows: 1) dysfunctional Gestalt perception is specifically related to disorganised forms of schizotypy, chronic and acute schizophrenia; 2) dysfunctional Gestalt perception involves deficits in both the processing of current and preceding ëcontextí; and 3) dysfunctional Gestalt perception is related to deficits in social cognition, specifically Theory of Mind.

Günter SCHIEPEK München (D)
Emotions, Cognitions, and the Self-Organizing Brain
Cognitive processes are tightly related to emotions and feelings. In synergetic terms, we can hypothesize that emotions are the driving forces or control parameters of all our cognitions. This well-known idea, worked out in Ciompiís concept of the Affect-logic, can now be made visible by methods of functional brain imaging (fMRI). As demonstrated in a study by Sanfey et al. (2003), even economic decisions are guided by (or at least correlated to) the activation of emotion-related brain regions. Not only decisions, but all other change processes (order transitions) are triggered and accompanied by emotions. They enhance and support or inhibit learning processes, depending on their quality and intensity. Consequently, continuous feedback about relevant emotions and other dynamic features of learning or developmental processes will be important in order to support change processes. How this real-time monitoring can be realized by computer-aided strategies will be outlined in this contribution. Nonlinear dynamics goes into practice now.

Fabrice LEROY & Laurent PEZARD Lille/Paris (F)
An Attractor Neural Network Model of Story Recall in Schizophrenia
The recall of a story can be modelled as a dynamical cognitive process, oscillating between essential propositions (the macro-level of discourse) and unessential ones (micro-level). Cognitive dysfunctions in schizophrenia, as expressed in the symptoms of loose associations and tangentiality, can affect the transition between these two proposition levels. In a previous study, we observed an over-activation of the micro-level and an under-activation of the macro-level. We propose here an attractor neural network model to simulate typical transitions from micro-proposition to macro-propositions and their modifications in schizophrenia.
In this study, we first use an Hopfield neural network to test connectivity and noise accounting for dysfunctional transitions. A small decrease of connectivity, coupled with a minimal threshold of noise, appears to be sufficient to simulate a more difficult transition from the micro-level to the macro-level, as revealed by our previous works with schizophrenic patients. Secondly, we try to answer a more technical question introducing a new parameter (asymmetry), so that the network can correctly recall a entire sequence. Further directions are then proposed to refine these simulations and their links with language production disorders in schizophrenia.

Ivelisse LAZZARINI St. Louis (USA)
Nonlinear Dynamics of Occupation: A Case in Point
This poster presentation discusses a case study from the perspective of nonlinear dynamics as it pertains to the practice of occupational therapy in an acute care psychiatric setting. Through a conceptual framework of brain dynamics, treatment interventions will be described and explained to elucidate the complexity of self-organized systems.

Annika LATZEL, André BÜSSING, Britta HERBIG München (D)
Phenomenology of Reflection - Ideas for Reflection as a Technique in Ongoing Education
Reflection is discussed in various research areas as a main effecting mechanism in behavioral and cognitive change. Here, reflection is conceptualized in terms of Fergus and Reids (2001) reflexivity entailing cognitive and experiential/affective components. Empirical investigation is difficult considering reflection a complex and dynamic skill with interacting subcomponents. A conceptual approach is presented for measuring reflection on action relying on verbal data. Verbal action descriptions come from an experimental laboratory study with pre- and post-test-design (n=37 nurses) comparing an experimental group with a control group. Treatment is a repertory grid interview intended for explication of implicit knowledge. Control group only undergoes action and control treatment. Action measure utilizes "real" standardized patients embodied by actors. Data indicate high ecological validity, while ensuring maximum experimental control. Analysis shows performance improvement and reflection in experimental and - unexpectedly - control group (e.g. Büssing, Herbig & Latzel, 2003).
Analysis of reflection components is done by content analysis with quantitative and qualitative elements. Two coding systems for cognitive and affective components are used. Coding units are prerated thought episodes (n=1615). Underlying logic of the cognitive coding system e.g. is based on a temporal criterion. Assumption is, that speaker conveys a higher level of cognitive complexity when refering in one thought episode to more than one time period (e.g. category "mental simulation of critical situation action with earlier experience" gets higher score than category "simply mentioning performed action"). Results suggest reliable (Kappa measures) and valid (criterion validity using category correlation with scales for relevant personality traits) measuring of reflection components. Results are discussed for significance in ongoing education. They support potential of the treatment "explication of implicit knowledge" for activating considerable cognitive and experiential processes suggesting its use as a training method in ongoing education.

Stacey POPONAK Notre Dame (USA)
Estimating Dynamical Systems Model Parameters using Local Regression Equations
Continuous time modeling can be achieved through dynamical systems analysis where differential equations are utilized to model the observed data. In practice, an impediment of obtaining accurate parameter estimates for a dynamical systems analysis occurs in the estimation of derivatives from observed data. A proposed solution to this problem was tested by fitting a novel modification of the local regression (loess) model to repeated observations and obtaining analytical exact derivatives of the resulting smoothing function. Specifically, using a moving window, local regression equations can be fit and then differentiated within small increments of the original time series in order to obtain estimates of the derivatives. Factors affecting the accuracy of the resulting parameter estimates are discussed and general guidelines for conditions where this procedure is applicable are given. A practical application of the loess derivative estimation method is provided, analyzing a motion-capture data set pertaining to the breaking and building of spatial and temporal symmetry between dyads of participants during dance. The results of a dynamical systems model fit to the motion-captured dance data are presented and discussed in the context of interpersonal synchronization.

Thomas REISCH, Kathrin ENDTNER, Martin THOMMEN, Wolfgang TSCHACHER Bern (CH)
Cognitive Patterns and Cognitive Incongruence in Borderline Personality Disorder Patients
Borderline personality disorder (BPD) leads more than any other personality disorder to inpatient or outpatient treatment. According to M. Linehan cognitive dysfunction can be seen as an important component of patients with BPD. However, the cognitive patterns of this group of patients have been derived from theoretical considerations or clinical single case observations, but have rarely been assessed systematically. In the study presented here we investigate the cognitive patterns (FAMOS questionnaire) and the subjective incongruence of these cognitions (INK questionnaire) of BPD patients and of patients without a diagnosis of BPD. All data sets included in the study stem from a pre treatment sample of the Bern Psychotherapeutic Day Treatment Program. According to psychopathology criteria (DSM), 5 of 9 listed symptoms have to be present for the diagnosis of BPD. As a result of this definition it can be stated that the group of patients suffering from BPD generally is inhomogeneous. It can be assumed that the cognitive impairments are linked to specific BPD symptoms rather than to the presence of the diagnosis itself. Therefore, we will present the analyses of the cognitive patterns on a diagnostic level as well as on a symptom level. Results will be linked to potential consequences for the treatment of BPD.

Andrée SCHIRTZ Celerina (CH)
Self-Efficacy and Mood Process Data as Predictors of Reduction in Heroin Consumption after Detoxification
Self-efficacy and mood of 90 patients were surveyed during a five-day heroin detoxification treatment with buprenorphine. The mean age of the study group was 30 years (SD=7), the majority were men (77%). They all consumed heroin regularly and most of them showed elevated psychological distress (SCL-90-R mean 0.89, SD 0.34) before entering treatment. Their feelings of control measured with the FKK (Krampen, 1991), a questionnaire assessing feelings of control as a trait, were situated in the normal value area. Regarding the process variables at entry and during detoxification, we found mood worse and self-efficacy lower than in the control group. These process variables were predicitve of the reduction in heroin consumption 90 days after detoxification, while the comparatively stabler patient characteristics measured once at entry (e.g. SCL-90-R, FKK) were not. To be more precise, we found that low average mood with low extreme values (i.e. not too bad) and high average self-efficacy with low extreme values (i.e. not too good) were related to a greater reduction in heroin consumption.
The implications of these findings are of methodological and practical order. First, our results seem to speak in favour of process research, rather than of a single cross-section measure. Second they suggest, that there exist cut-off values, from which on high self-efficacy and low mood turn into a negative outcome predictor.

Toru YAZAWA Hachioji (JPN)
Scaling Analysis of Neurodynamical Control Systems of the Heartbeat in the Spiny Lobster, Panulirus japonicus
Both in vertebrates and invertebrates, heartbeat is controlled by the balance of activities of acceleration and inhibition which are mediated by the amount of neuronal impulse discharges. Since crustacean animals use small numbers of neurons for the control of the heart and since the neuro-anatomy of heart control system is well documented in crustaceans, we adopted crustaceans for the analysis of central nervous function. The heart of the crustaceans receives innervation of only three kinds of axons originating from the central nervous system; one pair of inhibitory and two pairs of acceleratory axons. This simplicity has a great advantage for the physiological and physical study of neurodynamical control systems of the heartbeat. Thus, we performed time series analysis of ECG data of Japanese spiny lobster, Panulirus Japonicus. The ECGs were recorded in three conditions of the heart. The isolated hearts, which are obviously disconnected from the center but are well maintained at a steady condition, and the intact hearts, heartbeats of which were recorded at either a freely moving condition or a stressful condition receiving severe immobilizing stress. DFA (Detrended Fluctuation Analysis) revealed a difference of operation of the CNS. The three conditions of the hearts were found to have different scaling properties. Focusing on the difference between the scaling exponents of the isolated heart and the stressful heart, both of which had, intriguingly, almost identical power spectra, we will discuss the possibility to index the CNS operation by means of the scaling exponents.

back to program