ABSTRACT
Inthisstudyofnonlineardynamicalsystems(NDS),aswithanyother approachtobuildingstatisticalmodels,wehavetobemindfulofthreesorts ofquestions.Whataretheeventsinthephysicalworldandinourmental covariatesofthatworldweseektomodel?Whatcanwechooseasrepresentationalstructuresinsymbols,algebra,orlogicalrelationsthatareintrinsicallytractable?Andwhatrulescanweadoptforevaluatingtherelative worthofalternativemodelsoftherealworld?Itispossibletobuildgeneral theoriesabout‡ndingahierarchicalorderforalternativemodels,butthese areincomplete(Speekenbrink,2003).Thesethreeareasofdiscourse,data, models,andmappingbetweendataandmodels,allinvolvedecisionsabout whatisthoughttobescienti‡callyrelevantandpotentiallyfruitfultopredict in a wider range of apparently related scenarios.
