Perceptualbistabilityariseswhentwoconflictinginterpretationsofanambiguousstimulusorimagesinbinocularrivalry(BR)competeforperceptualdominance.Fromacomputa-tionalpointofview,competitionmodelsbasedoncross-inhibitionandadaptationhaveshownthatnoiseisacrucialforceforrivalry,andoperatesinbalancewithadaptation.Inparticular,noise-driventransitionsandadaptation-drivenoscillationsdefinetwodynamicalregimesandthesystemexplainstheobservedalternationsinperceptionwhenitoperatesneartheirbounday.Inordertogaininsightsintothemicrocircuitdynamicsmediatingspon-taneousperceptualalternations,weusedareducedrecurrentattractor-basedbiophysicallyrealisticspikingnetwork,wellknownforworkingmemory,attention,anddecisionmaking,whereaspike-frequencyadaptationmechanismisimplementedtoaccountforpercep-tualbistability.Wethusderivedaconsistentlyreducedfour-variablepopulationratemodelusingmean-fieldtechniques,andwetesteditonBRdatacollectedfromhumansubjects.Ourmodelaccountsforexperimentaldataparameterssuchasmeantimedominance,coefficientofvariation,andgammadistributionfit.Inaddition,weshowthatourmodeloperatesnearthebifurcationthatseparatesthenoise-driventransitionsregimefromtheadaptation-drivenoscillationsregime,andagreeswithLevelt’ssecondrevisedandfourthpropositions.Theseresultsdemonstrateforthefirsttimethataconsistentreductionofabiophysicallyrealisticspikingnetworkofleakyintegrate-and-fireneuronswithspike-frequencyadaptationcouldaccountforBR.Moreover,wedemonstratethatBRcanbeexplainedonlythroughthedynamicsofcompetingneuronalpools,withouttakingintoaccounttheadaptationofinhibitoryinterneurons.However,theadaptationofinterneuronsaffectstheoptimalparametricspaceofthesystembydecreasingtheoveralladaptationnecessaryforthebifurcatiotooccur,andintroducesoscillationsinthespontaneousstate.