ABSTRACT
Therationaleforusingnonlineardynamicalsystems(NDS)toolsisoften tiedtoevidenceshowingtemporaldependenciesandnonlinearitiesin humanbehavior.Aswesamplebroaderrangesofourteststimuli,whatwe ‡nd,ratherthanproportionalscaling,arebehavioralchangesresembling powerlawsandcontainingsequentialdependencies;forareviewofstudies,seeGuastello,Koopmans,andPincus(2009);Holden,VanOrden,and Turvey(2009);Shelhamer(2007);andWard(2002).Unfortunately,thesepropertiescanviolateassumptionsofpopulargenerallinearmodeling(GLM) statistics.Thus,tolearnwhetherrecurringpatternsmaybehiddenamidst behavioralnoise, we need to use analyses designed for this purpose. In Aks (2009), I summarizedsomeNDStechniquesandshowedapplications toeyemovementsinvisualsearch.Variousanalysesshowedlong-range1/f correlationsacrossthesequenceofeye‡xations(whenhorizontalandverticaleyemovementswereanalyzedseparately).Theresultssuggestedthat thegeneratingmechanismcanbeunderstoodfurtherthroughalternative analyses,forwhichrecurrencequanti‡cationanalysis(RQA)andrecurrence plots(RP)arewellsuited.Withtheirhighsensitivitytodataorder,RQAcan
CONTENTS
State-Space Analysis and Time-Delay Reconstruction Using RQA and RP ...................................................................................... 232 RQAMethods ......................................................................................................239
