Log in
Pitt Logo

Font Size

Cpanel

Epistemology of Modeling & Simulation

Epistemology of Modeling & Simulation Conference

The Epistemology of Modeling & Simulation conference took place at the University of Pittsburgh April 1st through 3rd, sponsored by the Center for Philosophy of Science at the University of Pittsburgh and the MIDAS Center for Excellence in the Graduate School of Public Health.  The conference included keynote addresses by Paul Thagard, Marc Bedau, Nicholas Rescher, Ian Lustick, Marc Lipsitch and Wendy Parker, with 40 paper and 30 poster presentations (further details available at www.modelingepistemology.pitt.edu).

The purpose of the conference was to bring together serious work in philosophy of science with active and ongoing efforts in modeling across a range of disciplines: epidemiology, physics, biology, climate change, cognitive science and neuroscience.  Many of the speakers raised issues of difficulty in evaluation of predictive models, with Rescher, Lustick's, and Lipsitch and Parker's keynotes cases in point. But a main emphasis throughout the conference was on the different purposes to which models are put beyond point prediction, with different forms of evaluation appropriate to those purposes.  Models can serve purposes of explanation and illuminate potential explanatory mechanisms, can stimulate hypothesis formation and eliminate early non-contenders, can guide data collection, reveal core uncertainties, make clear robustness and fragility of a phenomenon with regard to different inputs, indicate insufficiency of proposed accounts, format and guide policy decision, indicate efficiency of alternative interventions, reveal the simplicity in complex phenomena and the complexity behind apparently simple phenomena. Model building and use in none of these cases reduces to simple prediction, nor is evaluation in any of these cases entirely aligned with evaluation appropriate to point prediction.  The emphasis on varieties of models, varieties of model use, and varieties of appropriate evaluation came out particularly clearly in open interdisciplinary exchanges in break-out session on three classic models: the ancient Anasazi, the CCSM3 climate change model, and the Epstein-Burke model of smallpox and bioterrorism.  All participants in the conference were invited to join a 'pro' group on the virtues or the 'con' group on the vices of one of these models, with access to an expert on the model but with someone else taking notes and reporting back to the group as a whole.  It was here that inevitably conflicting virtues of simplicity and full realism, explanatory generality and predictive detail, scientific accuracy and policy accessibility came to the fore, along with a range of questions regarding necessary calibration and adjustment on the one hand and artificial tweaking and overfitting on the other.

It is hoped that the conference will mark a new step toward collaboration between philosophers of science and ongoing scientific and policy efforts in modeling and epistemology.  There were certainly cases in which the modelers may have found philosophers' concerns, or their way of phrasing them, remote and isolated from the everyday demands of modeling.  There were undoubtedly also cases in which the philosophers may have found the modelers' concerns, or their way of phrasing them, artificially restricted to a particular area or as yet unclearly articulated.  But there were also many cases in which immediate methodological issues were seen to reflect larger epistemic concerns suitable for philosophical analysis, and in which claims in philosophy of science found instantiation, application, and occasionally refutation in the details of contemporary modeling.  As Josh Epstein summarized the conference, 'there were many fruitful collisions.'  The hope is that there will be many more.

Keynote Presentations 

mab_sdu

 

Mark Bedeau
Reed College 

"How Computer Models Test the Arrow of Complexity Thesis and Reveal Darwin’s Dirty Secret"

(Introduction by Kevin Zollman, Carnegie Mellon University)

 

 


Publicity 2010


Nicholas Rescher 
University of Pittsburgh

"How Modeling Can Go Wrong"

(Introduction by James G. Lennox, University of Pittsburgh) 

 

 

 


paulthagard_001

 

Paul Thagard
University of Waterloo 

"Modeling Neurons, Minds, and Groups: The Epistemology of Multilevel Simulations"

(Introduction by Susan Sterrett, Carnegie Mellon University) 

 

 

 


 

lustick 2

 

Ian Lustick
University of Pennsylvania

"Using Agent-Based Modeling to Forecast and Analyze Rare Political Events: Illustrations from Pakistan and Thailand"

(Introduction by Patrick Grim, SUNY Stonybrook)

 

 

 

 

 

You are here: Home Epistemology of Modeling & Simulation