Madison Chaos and Complex Systems Seminar

Spring 2003 Seminars

All seminars are Tuesday at 12:05 pm in 4274 Chamberlin except as noted.

Short List


January 28, 2003

Dreams and change

Boris Matthews, Harmonia Center for Psychotherapy

Dreams reveal to us what is going on outside our limited waking consciousness. The range of what dreams show us is vast, extending from individual consciousness, to perceptions of others, to soundings of the general human condition and the shifting forces in the Zeitgeist. Example dreams from clinical practice raise tantalizing theoretical questions about causality, and the relativity of time.

The presenter, Boris Matthews, Ph.D., was trained at the C.G. Jung Institute of Chicago. He has practiced analytical (Jungian) psychology since 1987 in Milwaukee, and just opened a practice in Madison. In addition to his clinical work, he has translated and edited extensively.

February 4, 2003

Sociophysiology of Bohm's Dialog

Russell Gardner, Medical College of Wisconsin

The core values of Bohm's dialoging value for people interacting can be subjected to an analysis that hinges on social rank hierarchy, a system of interaction with ancient evolutionary roots. Sociophysiology depends on across-species comparisons (similarities between people and non-human animals) and contrasts. This presentation will address Bohm's distinctive approach with respect to how his scheme benefits from the ally-seeking and story-using capacities of humans. These contrast to other animals to major extents (dogs are ally-seeking too, and other animals use rudimentary stories). But humans extend these to pervasive extents and Bohm's work illustrates their action.

February 11, 2003

Chaos and Self-Organization in Spatiotemporal Models of Ecology

Clint Sprott, UW Department of Physics

When you walk in the woods, you cannot help but notice the complexity of biological systems.  Traditionally, scientists have presumed that such complexity requires complex models or models in which chance (randomness) plays a role.  Thus most spatiotemporal models of biological systems have been either very complex or stochastic or both.  It has previously been shown that a simple stochastic cellular automaton can self-organize into a realistic spatial pattern that resembles a forest landscape, and that this pattern evolves in time in a plausible way.  This talk describes how similar behavior can arise in a simple deterministic cellular automaton, provided the rules are carefully chosen.  A genetic algorithm is used to search the large space of rules for ones that give realistic behavior.  Such a deterministic model is essential for studying the sensitivity to initial condition (chaos), and the rules can provide insight into the underlying biological mechanisms.  The model can also be used to supply or correct missing or corrupted patches of data in spatial patterns or photographic images.


February 18, 2003

Are humans really are what they eat? (factors contributing to a complexity of human drug metabolism)

Olga Trubetskoy, PanVera, UW Research Park

We are continually under threat from chemical warfare in the guise of foreign compounds present in our food and environment, odorants, drugs and our bodies' own waste chemicals. An individual exposure to these compounds leaves a "metabolic signature" in our body, changing the way we are interacting with our environment. The discussion will include a contribution of many factors(food and eating habits, genetic, environmental, pre-natal exposure, etc.) to the creation of a complex "metabolic fingerprint map"  unique for each individual organism. This "metabolic signature" is a current basis for the individualized medical treatment plan allowing future prescription of the specific drugs to patients based on their genetic background and personal life experiences.


February 25, 2003

The "I" Illusion: How and why does a human brain construct a self?

Deric Bownds, UW Departments of Molecular Biology and Zoology

This talk presents a series of simple experiments documenting the presence - as we sense, act, emote, think - of something rather different from the "I" that our common sense usually tells us is running the show.  The confabulator that generates an "I" can go awry in cases of brain damage and also in ordinary experience. While it is sobering and ultimately accurate to call our sense of having an "I" and conscious will an illusion, it is incorrect to call the illusion a trivial one. It is a basic building block of human psychology and social life.  An obvious evolutionary rationale is provided by the consequences of this illusion:  our dominance as a species on this planet.


March 4, 2003

Love at Goon Park

Deborah Blum, UW Department of Journalism and Mass Communications

Why an alcoholic, sarcastic, poetry-writing, work-obsessed primatologist from the University of Wisconsin was exactly the right person to convince the world that love really mattered.

March 11, 2003

Weather Forecasting:  Behind the Scenes

Daryl Kleist, Steve Decker, and Justin McLay, UW Department of Atmosphseric and Oceanic Sciences

Various factors have lead to a change in the scope of modern weather forecasting.  These factors include more computing power, better basic understanding of our atmosphere, and a recognition that probabilistic forecasts may be more useful than deterministic ones.  In addition, significant advances have been made in observing systems, data assimilation, and numerical weather prediction, which have led to improvements in short to medium range numerical forecasts.

 Despite all the improvements made, forecasting is arguably becoming increasingly complex.  Weather forecasters today are responsible for sorting through an enormous amount of information to tailor a weather forecast.  Each time a forecast is prepared, forecasters are challenged with interpreting potentially conflicting output from several different models, which are routinely run multiple times daily.  In addition to traditional model output, the availability of model output statistics and ensemble forecasts makes the forecasting process even more complicated.  Lastly, there is an enormous amount of observational data available to a forecaster, including satellite, radar, and high resolution surface data, which can be used for validation of model performance or aid in short term forecast preparation.

This presentation will provide a brief historical perspective on weather forecasting, followed by examples of real-time forecast preparation for several different time ranges.  The utility, strengths, weaknesses, and motivation for use of each type of data (i.e., model output and observations) will be described.

March 25, 2003

Parsing a bacterial genome

Mark Craven, UW Department of Biostatistics and Medical Informatics

A central challenge in biology is to uncover the complete gene-regulation network of an organism.  This challenge can now be profitably attacked given the availability of complete genomes and high-throughput technologies for interrogating the states of cells.  A key step in addressing the challenge is to assemble a "parts list" of the regulatory elements for a given genome.  We have been developing an approach, based on probabilistic language models, that uses DNA-sequence and gene-expression data to predict a variety of regulatory elements in bacterial genomes.  Given experimentally verified instances of certain regulatory elements, our approach learns models that can be used to predict other instances in a genome.  We have applied this approach to the task of predicting a nearly complete map of promoters, terminators and operons in the genome of E. coli.

April 1, 2003

Memory across eye-movements: 1/f dynamic in Visual Search

Deborah J. Aks, Department of Psychology, UW-Whitewater

The ubiquity of apparently random behavior in visual search (e.g., Horowitz & Wolfe, 1998) has led to our proposal that the human oculomotor system has subtle deterministic properties that underlie its complex behavior.  We report the results of one subject’s performance in a challenging search task in which 10,215 fixations were accumulated.  A number of statistical and spectral tests revealed both fractal and 1/f structure. First, scaling properties emerged in differences across eye positions and their relative dispersion (SD/M)—both decreasing over time. Fractal microstructure also emerged in an Iterated Function Systems test and delay plot. Power spectra obtained from the Fourier analysis of fixations produced brown (1/f2) noise and the spectra of differences across eye positions showed 1/f (pink) noise. Thus, while the sequence of absolute eye positions resembles a random walk, the differences in fixations reflect a longer-term dynamic of 1/f pink noise. These results suggest that memory across eye-movements may serve to facilitate our ability to select out useful information from the environment.  The 1/f patterns in relative eye positions together with models of complex systems (e.g., Bak, Tang & Wiesenfeld, 1987) suggest that our oculomotor system may produce a complex and self-organizing search pattern providing maximum coverage with minimal effort.


April 8, 2003

Elite workers and the colony-level pattern of labor division in the yellowjacket wasp, Vespula germanica

Christine R. Hurd, UW Department of Zoology

Measurements of carbohydrate foraging behavior of Vespula germanica yellowjackets show that the distribution of the number of foragers over the number of trips is highly skewed with a few foragers making a disproportionate number of trips.  We tested several empirical models based on different biological assumptions to see which model best described the distribution.  For all periods of observation, the data are well fitted by a straight line on a log-log plot.  This fit indicates that the distribution of labor is non-increasing monotonic; i.e. continually decreasing, and follows a power law.  Stochasticity and self-organization are two possible explanations for the power law distribution.  As an alternative approach, cluster analysis of various foraging characteristics of individual foragers clearly separated foragers into two groups and is consistent with a bimodal model for the division of foraging labor.  Based on these cluster results, we operationally defined workers as either ‘elite’ or ‘non-elite’.  We found that elite foragers are not more likely than non-elites to be task specialists.  The data show that workers develop into elites but do not support the hypothesis of self-reinforcement as the mechanism.

April 15, 2003

Quantifying landscape pattern : What can "Superfrag" tell us?

Jeff Cardille, UW Department of Zoology

Despite the widespread use of landscape metrics in the field of landscape ecology, basic questions remain about our ability to relate indices of landscape pattern to real-world ecological processes. A large part of this uncertainty is our lack of understanding of the statistical properties of landscape metrics themselves—though we can derive the range through their formulas, much less is known about the actual distributions of their values in real-world landscapes. With the widespread availability of satellite-based land cover classifications, and the ability of desktop computers to compute landscape metric, our ability to produce values far outstrips our ability to understand them.

We are undertaking a project to understand the statistical and spatial distribution of common landscape metrics using real-world land cover classifications. The approach centers on dissecting the land cover of the conterminous United States into tens of thousands of square mini-landscapes, computing landscape metric values, and interpreting the results both statistically and graphically. With this new approach, we are attempting to answer the most basic questions for the ecological community: What are the statistical properties of common landscape metrics? How sensitive are these distributions to changes in the grain, extent, and classification scheme used to represent the spatial data? Across space, how are metrics related to human and biophysical patterns? Developing the answers to these questions, and providing ways for ecologists to access these results in real time, is of critical importance if we are to understand the relationship between spatial pattern and ecological processes.

April 22, 2003

Health care dynamics: Prices and policy initiatives, 1950 - 1999

George Pasdirtz, UW Division of Information Technology

One element of the health care crisis added on to comparatively high levels of spending, inadequate insurance coverage, limited distribution of services and labor shortage is the high price of health care goods and services. In the 1980's, U.S. health care prices began diverging from the aggregate price level, meaning that health care was becoming more costly relative to other products and services. Since the 1970's, the United States has experimented with various regimes of cost control. Specific policy measures such as the Nixon Economic Stabilization Program (ESP) 1971-1975, the 1973 HMO Act, voluntary hospital regulation or the Voluntary Effort (VE) of 1977-1979, and Medicare's prospective payment system (PPS) enacted in 1983 had little effect, although  general policy measures such as the Volcker deflation of 1979 did have an impact on overall inflation. In this seminar we will use a simulation model to investigate not only the dynamics of health care prices from 1950-1999 but also why health care costs have been resistant to policy initiatives?

To sort out the forces acting on health care prices we need a formal model that will allow us to run policy experiments while controlling other macroeconomic factors. Over a specific period of time, different administrations have engaged in many policy initiatives with a wide range of intended and unintended consequences. A simulation model allows us to explore price dynamics while controlling for different policy regimes. We also need a model that can distinguish normal inflation from cyclical episodes. For example, results from the USE20 simulation model (1900-1949), presented at last semester's seminar, showed that although health care prices during the Inter War years did follow the normal course of inflation they did not match the large cyclical swings displayed by other prices. Health care prices neither inflated as much as the rest of the economy during the Roaring Twenties nor deflated as much during the Great Depression. And, it was at the point that the economy started deflating that the Hoover administration started recommending that something be done about the health care problem. Health care became expensive relative to the rest of the economy because of macroeconomic forces not because of a health care crisis.

Are the same macro-economic dynamics operating in the Post War era? Could the health care crisis that started in the 1980's be once again the result of cyclical macroeconomic factors that have little to do with health care? We will present results from the USL20 simulation model showing price dynamics and policy impacts in the Post War period (1950-1999) and also look at other factors usually thought to explain the health care crisis: technology, population aging, chronic disease trends, insurance, etc.

April 29, 2003

Apples and eco-labels

Michelle Miller, UW Center for Integrated Agricultural Systems

The Wisconsin Apple Growers Association, in partnership with the University of Wisconsin, is exploring various eco-labeling approaches to better communicate to consumers about their efforts to reduce pesticides that pose a health risk.  The systems feedback modeling approach applied to farmer/consumer relationships in the food system suggests potential action scenarios to address industry concerns.  Work in the first phase of the project has been to clarify industry interest and needs, work with growers and others to compare various communications strategies employed by different eco-label programs, develop a dynamic hypothesis that addresses the situation faced by WI apple growers in particular, and then craft a strategy to improve the quality of relationships between targeted system components.  Gathering baseline data on pesticides and IPM practices used by growers is critical in this first phase.  The second phase of the project will explore ways to strengthen desirable feedback between farmers and consumers and then test various scenarios to insure a desired result.

May 6, 2003

Measuring the complexity of currency markets by fractal dimension analysis

Abdol S. Soofi, Department of Economics, UW-Platteville

We use the theory of nonlinear dynamical systems to measure the complexity of currency markets by estimating the correlation dimension of the returns of the Dollar/Pound and Dollar/Yen daily exchange rates (the spot rates).

Given a set of n vectors in the m-dimensional state space, the correlation dimension estimates the probability of finding two vectors in the set that are separated by a distance not larger than a radius r. The correlation dimension could be helpful in searching for attractors in the time series.

We test the significance of the results by comparing them to correlation dimension estimates for surrogate time series, i.e. stochastic linear time series with the same power spectrum and amplitude distribution as given by the original data. We find discernible nonlinear structure in the returns of the Dollar/Pound daily rate.