AnimalLearning& Behavior 1996,24(4),423-436 Reacquisition following extinction in appetitive conditioning SEANT. RICKERandMARKE.BOUTON UniversityofVermont, Burlington, Vermont Infourexperimentsutilizinganappetitiveconditioningpreparation,reacquisitionofconditioned respondingwasfound to occurboth rapidlyandslowlyfollowing extinction. InExperiment1,ac quisitionofrespondingtoatonethathadbeenconditionedandextinguishedoccurredmorerapidly thanacquisitionineitheragroupthatreceivedequivalentexposuretothefoodunconditionedstim ulusora "rest"controlgroup thatreceivedonlyexposuretothe apparatusinthefirsttwophases. However, reacquisitionwas impairedrelativeto acquisitionina "learning-experienced"group that hadpreviouslyreceivedconditioningandextinctionwithadifferentstimulus. Experiments2and3 producedsimilarresults, butalsofound thathighresponding duringreacquisitionwasconfinedto trialsthatfollowedreinforced,ratherthannonreinforced,trials.Experiment4,inwhichveryfewini tial conditioning trials were used, produced reacquisition that was slow compared with both learning-experiencedandrestcontrols.Theresultsareconsistentwitharoleforsequentiallearning: Reacquisitionisrapidwhenanimalshavelearnedthatreinforcedtrialssignalotherreinforcedtrials. It is often assumed that reacquisition ofconditioned spontaneousrecoverytobepresentattheoutset.Itisalso responding following extinction occurs more rapidly possiblethatreinstatement, whereunconditionedstimu than acquisition witha novel conditionedstimulus (CS; lus (US) presentations in the samecontextas extinction e.g., McAllister& McAllister, 1994).Theassumptionis restore responding to an extinguished CS (Bouton & based, in large part, upon a number ofearly studies Bolles, 1979b; Bouton & Peck, 1989; Rescorla& Heth, (Brogden, Lipman, & Culler, 1938; Finch & Culler, 1975),orevenrenewal,wheretestinganextinguishedCS 1935; Frey& Butler, 1977; Frey & Ross, 1968; Hilgard inadifferentcontextthanthatused inextinctionresults & Marquis, 1935; Hoehler, Kirschenbaum, & Leonard, in increasedresponding (Bouton& Bolles, 1979a;Bou 1973; Konorski & Szwejkowska, 1950, 1952a, 1952b; ton&Ricker, 1994),couldhavebeenresponsibleforthe Smith & Gormezano, 1965; Szwejkowska, 1950). In high levelofrespondingobserved.Earlystudiesofreac these experiments, responding during reconditioning of quisition often used massed conditioning and recondi the CS was comparedwith responding during the origi tioning trials. With trials spaced closely together, it is nalconditioningphaseinthesameanimals. Rapidreap possible that stimulus aftereffects ofthe US could have pearance ofthe conditioned response after extinction providedacontextforrenewalofconditionedresponding wastakenasevidencethattheCSretainssomeexcitatory (Bouton,Rosengard,Achenbach,Peck,&Brooks, 1993). strengthfollowingextinction,andthatthisstrengthtrans Early studies had not controlled for any ofthese effects fers positively to the reacquisition phase. The CS was (forargument, seeBouton, 1986). seenasstartingwithagreaterassociativestrengthduring In contrast, clear evidence ofslow reacquisition has reacquisitionthanduring conditioning. been obtained in the conditioned emotional response TheideathattheCSsurvivesextinctionwithsomeex (CER) procedure (Bouton, 1986; Bouton & Swartzen citation is clearly consistent with a variety ofevidence truber, 1989). Intheseexperiments,attemptsweremade (e.g., Bouton, 199I, 1993). However, as Bouton (1986) to minimize recovery effects that might complicate in pointed out, numerous difficulties exist in interpreting terpretationofreconditioning.Thegeneralprocedurein theliteratureonreacquisition.Thehighlevelofrespond volvedgivingonegroupconditioning,extinction,andre ingobtainedduringreacquisitioninstudiespriorto 1986 conditioningphaseswithoneCS,andcomparingtherate oftencouldhavebeenduemerelytospontaneousrecov ofacquisition in the final phase with that ofa control ery. That is, in many early experiments, reacquisition group that received conditioningwith a novel CS atthe began after a time interval that would have permitted sametime.(Thesewerethefirstexperimentstocompare acquisition in an extinguished CS with acquisition ofa novel CS pairedwith the US atasimilarpointintheex ThisresearchwassupportedbyGrantIBN-9209454fromtheNa periment.)Bouton(1986)useda"rest"control,whichre tionalScienceFoundation.Experiments Iand2werepartofathesis ceivedexposuretotheapparatusbutnoPavlovianevents submittedbythefirstauthorinpartialfulfillmentoftherequirements during the first two phases. Bouton and Swartzentruber ofthemaster'sdegreeattheUniversityofVermont.Correspondence (1989) useda"learning-experienced"control,whichre shouldbedirectedtoM. E. Bouton,DepartmentofPsychology,Uni versityofVermont, Burlington,VT05405. ceivedconditioningandextinctionofadifferentCSdur- 423 Copyright 1996Psychonomic Society, Inc. 424 RICKERAND BOUTON ing the first two phases. The basic finding was that, moreextinctionshouldproducetheresultwhen lessex givenenoughextinctiontrials, conditioning in the third tinctionshouldbemorelikelytodoso(e.g.,Kehoe, 1988). phaseproceededmoreslowlyforthegroupthatreceived Given that reacquisition can occur both rapidly and conditioning with the extinguished CS than for groups slowly,ouroverallgoalwastodeterminewhatconditions thatreceivedanovelCS. Boutoninterpretedthesefind produce the two results. One possibility, suggested by ings as indicating that, following extinction, a memory Napieretal. (1992),is thatdifferentconditioningprepa ofthat phase exists which interferes with subsequent rationshaveintrinsicdifferencesthatyieldrapidorslow conditioning. reacquisition. Ourinitial goal, then, was to characterize Slow,ratherthan fast, reacquisitionhasalsobeenob reacquisition in a preparation that had not been studied tained in taste-aversion learning. Forexample, Danguir before.Wechoseanappetitiveconditioningmethodthat and Nicolaidis (1977) found extremely slow reacquisi is widely used in the conditioning literature (e.g., Ball tion;theirresultssuggestedthatanaversiontosaccharin eine& Dickinson, 1991; Bouton& Ricker, 1994; Dela couldnotberelearnedfollowing extinction. Incontrast, mater, 1995; Farwell & Ayres, 1979; Hall & Channell, Revusky and Coombes (1979) found that reacquisition 1985; Kaye & Mackintosh, 1990; Pearce & Redhead, .did occur. Itwas rapidcomparedtoacquisition inarest 1995):Food pellets served as the US, and the rat's en control group butcomparable in rate to acquisition in a tries into the magazine to which the pellets were deliv learning-experiencedcontrolgroup(Revusky& Coombes, eredservedasthe conditionedresponse. Themagazine 1979,Experiment1).Morerecentstudies(Hart,Bourne, entryresponseisoftendescribedasaPavlovianCR,but & Schachtman, 1995)have foundreacquisitionto occur sinceobtainingthepelletisdirectlycontingentonthere more slowlythan acquisition in a learning-experienced sponse,theresponsepresumablyalsohas an instrumen group. tal component. Indeed, when magazine approach is put. The results obtained with CER and taste-aversion on an omission schedule (so that responding on a trial learning contrast sharply with data recently reported in leadstotheomissionofareinforcerthatisotherwisede the rabbit nictitating membrane response (NMR) para livered; Holland, 1979), the response is acquired and digm (Napier, Macrae, & Kehoe, 1992). In a series of maintained (suggesting some Pavlovian control) but is four experiments, Napieret al. addressed the criticisms alsoweakerthanthatinacontrolgroupthatreceivesthe offeredbyBouton(1986) concerningpreviousresearch sameCS-USpairingswithouttheomissioncontingency withthismethod. However,evenaftercontrollingforre (suggesting an instrumental component). As an instru instatement, renewal, and spontaneous recovery, they mental response, magazine entry is probably best de consistently found reacquisition to occur rapidly when scribed as a partially reinforced discriminated operant; compared with a rest control. Rapid reacquisition oc manymagazineentriesgo unreinforcedduring boththe curredwhetherextinctionconsistedofCS-alonepresen CS and the intertrial interval. The complexity ofthe tations or explicitly unpaired CS and US presentations magazine-entryresponsewillrequiresomecautioninin (rulingoutrenewalbyUSpresentationsasamechanism terpretation. Butthere now appearto be many similari in the final phase). Rapid reacquisition also occurred ties between instrumental and Pavlovian learning (e.g., whentheCSwassubjectedtoproceduresthatcouldhave Mackintosh, 1983; Rescorla, 1987). And reacquisition givenitan inhibitoryvalueinthe secondphase, suchas after instrumental extinction, like Pavlovian extinction, feature-negativeanddifferentialconditioning.Theeffect isoftenassumedtooccurrapidly(e.g.,Bullock& Smith, persisted despite a large number ofextinction trials 1953). (three times the number required to eliminate sponta The present experiments obtained results suggesting neous recovery). Reinstatement tests also produced no that both slow and rapid reacquisition can occur in the evidence that US-alone presentations reinstated extin magazine-entry preparation. They also identified se guishedrespondingintheNMRparadigm(NapieretaI., quentiallearning(e.g., Capaldi, 1994)as avariablethat 1992, Experiment 1). Napier et a!. interpreted their re contributestowhichoftheseresults isobtained. During sults with a model that formally captured the idea that conditioning, reinforced trials may be associated with residual excitatorystrengthwasavailableto groups that other reinforced trials. When reinforced trials are pre had been through conditioning and extinction (Kehoe, sentedfollowing extinction,theymayrenewresponding 1988). onsubsequenttrials. . A recent fear-conditioning study has also found evi dence consistentwithrapidreacquisition(McAllister& EXPERIMENT1 McAllister, 1994).Usingtheacquireddrivemethod(ac quisition ofa hurdle-jumping response to escape fear We designed Experiment 1as an initial investigation conditionedto contextual cues), McAllisterand McAl ofreacquisition in our appetitive conditioning prepara listerfoundthatreconditioningoffearfollowing 30hof tion. In Experiment 1A, we examined the rate ofreac extinctionexposuretothecontextoccurredmorerapidly quisitioninagroupthathadreceivedconditioningandex thandidconditioninginagroupanalogoustoarestcon tinctionwithatoneCS(GroupR)relativetoconditioning tro!. However,theresultwasnotobtainedinthreeearlier inseveralcontrolgroups.Thefirstofthesewasalearning experiments that used less extinction (McAllister & experienced control similar to the one used by Bouton McAllister, 1994, Experiments 1-3). Itisnot clearwhy andSwartzentruber(1989); thisgroupreceivedaniden- REACQUISITIONAFTEREXTINCTION 425 tical treatment with a different CS (a light-offstimulus) againbaitedwith4pelletseach.Duringthesession,thesubjects during conditioningandextinction(Group L). Asecond weretrainedto approachandeatfromthe foodcupuponactiva controlreceivedequivalentexposuretotheUSduringthe tionofthefeedermechanism.Approximately25pelletswerede liveredduringeachsession. conditioning phase but no pairings ofthe US with any Conditioning. Thesubjectsthenreceived6daysofcondition stimulus(GroupU). Thelastgroup(GroupC)wasarest ing.Oneachday,therewasone90-minsession.ForGroupsRand control that received no stimulus presentations during L, thissessionconsistedofeightpairingsofthe 30-secCS with theconditioningorextinctionphases(cf. Bouton, 1986; thefoodUS,withameanintertrialinterval(IT!)of10min(short NapieretaI., 1992). Duringa final reacquisitionphase, estIT!= 9min).Throughout,theoffsetoftheCScoincidedwith allgroupsreceivedtone-foodpairings.InExperiment1B, the onsetoftheUS. ForGroup Rthe CS wasthetone (T),while wedoubledthenumberofextinctiontrialsandcompared for Group L the CS was the lightoff(L). Group Ureceivedthe same eight US presentations but no CS. Group C was merely areacquisitiongroupwithalearning-experiencedcontrol. placedintheapparatusfortheentire90min;nostimuliwerepre sented.OnDayI,thefirstsessionforeachgroupstartedwithtwo ExperimentlA nonreinforcedpresentationseachofTandLaloneinthesequence TLTL,withameanIT!ofImin. Method Extinction. Followingconditioning, the rats received 10ses Subjects. The subjects were 32 male Wistar rats bredat the sionsofextinction. Allsessionswere90min indurationandoc UniversityofYermont.Theywereapproximately 120daysoldat curredonconsecutivedays.ForGroupsRandL,the30-secCSwas the startofthe experimentand were individuallyhoused insus presentedeighttimesalonewithameanIT!of10minduringeach pended stainless steel cages in a room maintainedon an 18:6-h session. The CSs presentedwere the same as those used during light:dark cycle. The experimentwas conductedon consecutive conditioning. GroupsUandCwerebothplacedintheapparatus daysduringthelightportionofthecycle.Theratswerefood de for the 90-min session, but no stimuli were presented. Nofood privedto80%oftheirfree-feedingweightsandmaintainedatthat waspresentedduringthisphase. levelthroughouttheexperiment. Reacquisition. Followingextinction, the rats received four Apparatus.TwosetsoffourSkinnerboxes, housed insound moresessionsofconditioning. All fourgroups receivedone90 attenuation chambers and located in two separate rooms, were min session aday in which the 30-sec tone CS was pairedwith used. Thesetwosetsofboxeshaveprovideddifferentcontextsin foodeighttimes,withameanITIof10min. otherstudies,butwerenotsousedinthepresentexperiment.Each boxinone setmeasured 26 x25 x 19cm. Thefront, back,and Dataanalysis.Conditionedrespondingwasmeasuredbymeans ofan elevation score ofthe form e= c- p, where c represents onesidewallweremadeofaluminum;theceilingandotherside thenumberofmagazineentriesmadeduringthe30-secCSandp wall were made ofclear plastic. The floor consisted oftubular representsthenumberofmagazineentriesmadeduringthe30sec steelbars 16mmindiameter,spaced3.2cmcentertocenter,and immediatelyprecedingthe CS. Elevationscoreswere calculated mountedperpendiculartothefrontwall.Onthefrontwallofeach foreachtrialandthenconvertedtofour-trialblockaveragesprior box, Icmabovethe floor, wasarecessed4 x 4cmfoodcup. A toanalysis.Thesescoreswerethenanalyzedbymeansofamixed 2.5 x 2.5cmleverprotrudedfrom the front wall5cmabovethe design analysis ofvariance (ANOYA), with group as a between floor and I cm to the right ofthe food cup. The subjects were subjects factor and blockas a repeated measure. An identical placedintheboxesthroughadoorintherightwall. ANOYAwasusedtoanalyzerespondingduringthepre-CSperiod Each box in the second set measured 24 x 22 x 18 cm. The (prescores).Arejectioncriterionofp<.05wasusedthroughout. front and back walls were aluminum, while the ceiling andside Posthoccomparisonswereconductedusingttestswithapooled wallswereclearplasticwithverticalblackstripes2cmwideand errortermandaBonferroniadjustmentforTypeIerrorrate. 2.5 cmapart. Thefloorconsistedofstainlesssteel bars3mm in diameter, spaced 1.5cmfrom centertocenter,andmountedpar alleltothe front wall. Onthe frontwallofeachbox, Icmabove Results thefloorandcentered3.5cmfromtherightwall,wasarecessed Conditioningproceededuneventfully, withGroups R 4 X 4 cm food cup. A4 x I cm lever protruded from the front and L reaching asymptote by Day 5. The data are sum wall5cmabovethefloorand6cmtotheleftofthefoodcup.The marized on the left side ofFigure 1. A group X block subjectswereplacedintheboxesthroughtheceiling. ANaYAontheconditioningdatarevealedonlyasignif Inbothsetsofboxes,illuminationwasprovidedbytwo7.5-W icantmaineffectofblock [F(II,154) =7.11]; respond white incandescent bulbs mounted on the ceilingsofthe sound ing tendedto increaseoverblocks. Neitherthe mainef attenuationchambers,25cmabovethefloor. TwoCSswereused inthisstudy. Onewasthe 30-secpresentationofa3000-Hztone fect ofgroup [F(l,14) = 1.22] nor the group X block (80dBre20,uN/M2[AJ)providedbyageneratorwiredtoidenti interaction[F(11,154) = 1.16]approachedsignificance. calspeakersmountedineachchamber25cmabovetheboxfloor. Extinction proceededuneventfullyas well (rightpor Background noise was 65 dB. The other CS consisted ofthe tionofFigure 1).Agroup XblockANaYAcarriedouton 30-secoffsetofthehouselights, whichproducedcompletedark the datarevealedonlyasignificantmaineffectofblock ness.TheUSconsistedoftwo45-mgfoodpellets(Traditionalfor mula, P.1.Noyes, Lancaster,NH) delivered0.2secapart. Maga [F(l9,266) = 15.43].Neitherthemaineffectofgroupnor zine entries were detected by photocells mounted within the the group X blockinteractionweresignificant(Fs < 1). magazines,justbehindtheplaneofthewalloftheSkinnerboxes. Groupcomparisonsduringreacquisition.Thedata Theapparatuswas controlledbycomputerequipmentlocatedin for each group during the reacquisition phase are pre anadjacentroom. sented in the top panel ofFigure 2. All groups acquired Procedure.Pretraining.Eachsubjectfirstreceived30minof conditionedrespondingatthistime.GroupsRandLap exposuretotheboxitwasassignedtoforthedurationoftheex peared to differ, with Group R responding somewhat periment. Foodcups were baitedwith4food pelletspriorto the startofeachexposuresession.Onthe followingday, subjectsre more on early trial blocks and somewhat less on later ceivedone30-minsessionofmagazinetraining. Foodcupswere ones. Groups Uand Cwere both somewhat slowerthan 426 RICKER ANDBOVTON 6 • GroupR ---0-- GroupL 5 4 ~ 0 u (/) c:: 3 0 ';j '" :> ~0 2 0+---"-""---"-""--"T-""--"T-..,.---:;:'--.!j'---r-D--rLJ-r-"'--r--;'" o 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 4-TrialBlocks Figure 1.Meanelevationscoresduringconditioningand extinctionin Experi mentIA. Groups RandLinacquiring responding, with GroupV Conditioningversusreconditioning.Theupperpanel beingslowest. Agroup X blockANOVArevealedasig ofFigure 3 shows Group R's performance in both the nificant main effect ofblock [F(7,196) = 12.28] and a original conditioning phase (Phase 1) and during re significantgroup X blockinteraction[F(21,196) = 2.88]. acquisition(Phase3).Awithin-subjectcomparisonofper Themaineffectofgroupwasnotsignificant[F(3,28) = formance inthesephasescancomplementourbetween 2.78]. subjects analysis of reacquisition, although it does The interaction suggests that group differences ex confoundconditioninghistoryoftheCSwithothervari isted only on certain trial blocks. Therefore, a series of ables, such as prior experience with the reinforcer, the fiveunplannedcomparisonswereconductedwithanad apparatus, and the food-deprivation schedule. As Fig justmentforerrorrate(which seta =.01). Thesecom ure3suggests,respondingwasclearlygreaterduringre parisons compared Group R with each ofthe control conditioning. A within-subject phase X block ANOVA groups separatelyontheblockswhere the largestgroup revealed a significant main effect ofphase [F(l,7) = differences appeared, and similarly compared Group L 16.73].TheANOVAalsorevealedasignificantmainef with each ofthe other controls. Group R did not differ fectofblock[F(7,49) = 4.86]andasignificantphase X significantlyfromGroupLonBlock7[t(78) = - 2.11], blockinteraction [F(7,49) = 5.41]. where the largest difference between these two groups AsimilarANOVAonprescoresrevealedasignificant appeared. However, Group R responded significantly maineffectofblock[F(7,49) =2.53].Prescoresincreased morethanGroupV onBlock4 [t(78) = 3.13]andmore slightlyoverblocks,regardlessofphase. The main effect than Group C on Block 3 [t(78) = 2.77]. Group L also ofphase approached significance, but was not reliable respondedmorethanGroupV [t(78) = 2.88]andGroupC [F(I,7) =5.15,p = .06], and the phase X block inter [t(78) = 2.68] onBlock5. action was not significant [F(7,49) = 1.00]. Mean pre AlthoughGroupsRandLfailedtodifferintheprevi scoreswere 1.34and 1.71 duringconditioningandreac ous analysis, previous CER and taste-aversion experi quisition, respectively. mentsthatfound slowreacquisitionhadcomparedanal ogous groups by themselves. Therefore, an additional ExperimentIB analysis was conductedusing onlythese two groups. A Method group X block ANOVA revealed a significant main ef Subjectsand Apparatus. Thesubjectswere 16male Wistar fectofblock[F(7,98) = 6.98]andasignificantgroup X ratsofthesamestockasthoseusedinExperiment IA.Theywere block interaction [F(7,98) =2.55]. The main effect of approximately 160 days old at the start ofthe experiment, and group was not significant (F< 1). The interaction sug werehousedandmaintainedasinExperiment IA.Onesubjectin gests thatacquisitionproceededdifferentlyinGroupsR GroupLbecamesickduringthecourseoftheexperimentandwas andL.However,thelargestdifferencebetweenthegroups removed, leaving7subjectsinthatgroup. Theapparatuswasthe (Block 7)was notreliable [t(42) = -2.00]. sameasthatusedinExperiment IA. Procedure.Pretraining. Boxexposureandmagazinetraining TheANOVAcarriedoutonprescoresrevealednosig werecarriedoutas inExperiment IA. Approximately28pellets nificant effects [largest F(7,196) = 1.87]. Mean pre weredeliveredduringeachmagazinetrainingsession. scores forthereacquisitionphasewere 1.71, 1.33, 1.38, Conditioning. Following magazine training, the subjects re and 1.29forGroups R, L,V, andC,respectively. ceived eight sessions ofconditioning. Each rat received two REACQUISITIONAFTEREXTINCTION 427 • Group R; the mean elevation scores for the phase were GroupR Experiment IA ---0--- GroupL 0.63 and0.33 forGroupsLandR,respectively. Agroup 10 ---e-- GroupU X block ANOVA revealed a significant main effect of 9 ---l:J.--- Groupe group [F(1,13) =4.72] and a significantmain effectof block [F(39,507) = 6.42]. The interaction was not sig 8 nificant (F < 1). Although Group L appeared to extin guishmoreslowly,thisdifferencehaddisappearedbythe lastfour sessions, whereGroups LandRhadmeanele vation scores of0.14 and 0.17, respectively. Four un planned comparisons (at a = .0125) between the two groupsonthefirstblockofeachofthelastfoursessions revealedno significantdifferences [ts(405)< 1]. Groupcomparisonduringreacquisition.Thefocal datafrom reacquisitionare presentedinthe lowerpanel O+---r--r----,;----,.---r--,---,----, o 2 3 4 5 6 7 8 ofFigure2.Duringthisphase,bothgroupsreceivedcon 4-Trial Blocks ditioningwiththe tone. Thegroups initiallydidnotdif Experiment IB fer, but Group Rshowed less responding than Group L 10 overblocks. This descriptionwasconfirmedbystatisti 9 cal analysis. A group X block ANOVA revealed a sig 8 nificantmaineffectofblock [F(ll,143) = 11.93]anda significant group X block interaction [F(11,143) = 3.33]. The main effect ofgroup was not significant [F(1,13) =2.33]. Comparisons(ata = .0167) between groupsonBlocks7, 11,and 12,wherethelargestgroup Experiment IA • GroupR 10 ---0-- GroupL o~-r--r--r-.--r--r-!;:=;::=;::=;::=;==;I"' 9 o 2 3 4 5 6 7 8 9 10 11 12 8 4-Trial Blocks Figure2. Meanelevation scoresduringreacquisition in Ex perimentslA(upperpanel)and1B(lowerpanel). 90-minsessionsoneachday. Eachsessioncontainedeight pair • ingsofthe30-secCSwiththefoodUS,withameanITIof10min. Phase I • ForGroupR,theCS wasT;forGroupL,itwas L.OnDay 1,the Phase3 firstsessionforeachgroupstartedwithtwopresentationseachof o+--,--.---r----,-=r==;:==::;==;_-' Tand Lalone in thesequenceTLTL, withamean ITIof1min. o 2 3 4 5 6 7 8 Extinction.Followingconditioning,theratsreceived20sessions 4-Trial Blocks ofextinction.Asinconditioning,thereweretwo90-minsessions oneachday. In each session, the 30-secCS was presentedeight Experiment IB timesalonewithameanITIof10min. Atotalof160extinction 10 trialsweregiven.TheCSspresentedwerethesameasthoseused 9 during conditioning. No food was presented during this phase. 8 Reacquisition.Followingextinction,theratsreceivedsixmore 7 sessions ofconditioning. Bothgroups receivedtwo 90-min ses 0 Cl6 sionsadayinwhichthe30-sectoneCSwaspairedwithfoodeight u times, withamean ITI of10min. Dataanalysis wascarriedout "'5 .gC asinExperiment I. 4 ~3 Results ioil2 Conditioningproceededuneventfully;duringthefinal 1 • Phase I four-trialblockofacquisition,GroupsRandLhadmean o • Phase3 elevationscoresof5.28and5.14,respectively.Agroup X -1 +--,--.--,-r--r-..,...--r'==r=r==;==;==;---' blockANOVArevealedonlyasignificantmaineffectof o 2 3 4 5 6 7 8 9 10 11 12 block [F(15,195) = 10.55]. Neither the main effect of 4-Trial Blocks group[F(l,13) = 2.56]northeinteraction[F(15,195) = Figure3.AcomparisonofmeanelevationscoresforGroupR 1.52] were significant. During extinction, responding duringconditioningandreacquisitioninExperimentslA(upper appeared to extinguish more slowly in Group Lthan in panel)and1B(lowerpanel). 428 RICKERAND BOUTON differencesappeared,revealedlessrespondinginGroupR AccordingtoBouton's(1993)retrievalview,presentation thaninGroupLonBlock7[t(35) = -2.68]andBlock II ofnonreinforced trials during reacquisition might help [t(35) = -2.53] but not on Block 12 [t(35) = -2.50]. retrieveamemoryofextinctionforareacquisitiongroup TheANOVA onprescoresrevealedonlyasignificant andthus causereacquisitionto occurslowly. main effect ofblock [F(11,143) =3.67]; responding This analysis resembles Capaldi's (e.g., 1994) well during the pre-CS period increased somewhat and then developedsequentialtheory. Itimpliesthatthe outcome decreasedoverblocks.Allothereffectswerenotsignifi ofagiventrialcanserveasasignal(orretrievalcue)for cant(Fs< 1).Meanprescoresforthereacquisitionphase theoutcomeofthenexttrial. Thus,duringconditioning, were 1.27and0.90 for Groups RandL, respectively. reinforcedtrials mightcometo signalthatthe next trial Conditioningversusreconditioning.Thelowerpanel will be reinforced, while during extinction nonrein ofFigure 3presents a comparison ofGroup R's perfor forced trialsmightcometosignalthatthe nexttrial will manceduringreacquisitionandtheoriginalconditioning benonreinforced. This viewimpliesaparticularpattern phase. A phase X block ANOVA revealed a significant ofresponding during reacquisition with partial rein maineffectofphase[F(l,7) = 12.09],asignificantmain forcement. Following conditioning and extinction, re effect of block [F(11,77) = 7.80], and a significant spondingonatrialafterareinforcedtrialshouldbehigh, phase X block interaction [F(11,77) = 2.46]. Again, whilerespondingfollowing anonreinforcedtrial should conditioning occurred more rapidly in the reacquisition below. Sincereacquisitionwasconductedwithapartial phasethanintheoriginalconditioningphase. reinforcement procedure in this experiment, we could A similar ANOVA on prescores revealed no signifi test this prediction. A retrieval view emphasizing the canteffects [largestF(lI,77) = 1.19]. roleofprevioustrialspredictsmorerespondingafterre inforcedthanafternonreinforcedtrials. Discussion We used a learning-naive control similarto Group U The results ofExperiments lA and IB indicate that in Experiment lA (as opposed to Group C) so that the reacquisitionintheappetitivemethodcanberapidwhen control would have had equal exposure to the US. This comparedwith"learning-naive"controls(GroupsUand groupalsoreceivedconditioninginthethirdphasewith C inExperiment lA)andwiththe originalconditioning a partial reinforcement procedure. Here one might also phase. However,inExperiment 1B,performanceduring expecthigherresponding on trials following reinforced reacquisitionwasimpairedrelativetothatinGroup L, a trials,ifonlybecauseassociativestrengthmightincrease learning-experiencedcontrol(arelatedgroup X session aftereachCS-USpairing.Clearly,anysequentiallearn interactionwasobtainedinExperiment lA).GroupLit ing effect in a reacquisition group would have to be selfappearedtoshowa"learning-to-Iearn"effectanalo greaterthanthatobservedinGroupU. gous to that previously observed in NMR conditioning Wealsoperformedacontextswitchfollowing USex (e.g.,Kehoe& Holt, 1984;Kehoe,Morrow,& Holt, 1984). posure in order to reduce possible blocking by context That is, previous conditioning (andextinction)with the conditioning (e.g., Randich, 1981). There is some evi light-offCS facilitated conditioning with the tone rela dence to suggest that this occurred in Experiment 1A, tivetolearning-naivecontrols. Somewhatparadoxically, because inthe final phase Group Uappeared to acquire alearning-to-learneffectwaslessevidentinareacquisi conditioned responding more slowly than Group C, the tiongroupthathadreceivedallthreephaseswiththesame controlthathadnotreceivedUS exposure inPhase I. CS. That result might implicate an interference mecha nism inadditiontothe learning-to-Iearnmechanism. Method AlthoughthefactthatGroupRwaslowerinrespond Subjects ingthanGroupLlaterinthephaseindicatessomeinter Thesubjectswere 16male Wistarrats from thesamestockas ference by prior conditioning and extinction with the inthepreviousexperiments,andwere maintainedandhousedas sameCS,itisworthnotingthat, inCER,theinterference before.Theywere approximately 100days oldatthestartofthe experiment. effectisobservedstartingearlyinreacquisition. Instead ofacquiringrespondingmoreslowly,GroupRappeared Apparatus to approach a lower asymptote (see especially Experi Theapparatuswasthesameas in Experiments IAand IB,ex ment IB).Theoriesthatpredictslowreacquisitionwould ceptthatadishcontaining 10mlof'Heinzdistilledwhitevinegar actually expect group differences to occur early, rather (H.1.HeinzCo.,Pittsburgh)wasplacedineachsound-attenuation than late, in the phase (Bouton, 1993; Pearce & Hall, chamberofthefirst setofboxesinordertoprovideadistinctive 1980;Wagner, 1981). scentcue; adish containing IgofVick's Vaporub(Richardson Vicks, Inc., Shelton, CT) was similarlyplaced in each sound attenuationchamberofthesecondsetofboxes.Inthepresentex EXPERIMENT2 periment, the two sets ofboxes, which provided the different contexts,werefullycounterbalanced. In Experiment 2, we conducted reacquisition with a Procedure partialreinforcementprocedurethatintermixednonrein Pretraining.Boxexposureandmagazinetrainingwerecarried forcedtrialsamongthereinforcedtrials.Thissortofpro outas in Experiment I, with the exceptionthat each subject re cedure was used by Bouton (1986) and Bouton and ceivedone session in Context A(the conditioning context) and Swartzentruber(1989),whoobtainedslowreacquisition. oneinContextB(theUSexposurecontext).Sessionsindifferent REACQUISITIONAFTEREXTINCTION 429 contextswereconductedonthesameday,althoughboxexposure main effect ofgroup [F(1,14) < 1] nor the interaction and magazine training were conductedon separate days, as be [F(lI,154) = 1.10]wassignificant.Prescoresincreased fore. Approximately20pelletsweredeliveredduringeachofthe somewhat and then decreased over blocks, but did not magazinetrainingsessions. differ between groups. Mean prescores were 1.64 and Conditioning.Thesubjectsthenreceived8daysofcondition 1.55 forGroups RandU, respectively. ing. On each day, there were two 90-min sessions, one in Con Figure 5presents the responding by Groups Rand U textAandoneinContextB.ForGroupR, sessionsinContextA consistedofeight pairingsofthe 30-sectone CS with the food on trials that followed reinforced versus nonreinforced US, with a mean ITI of10min. Sessions in Context B for this trials. Agroup X trialtype X sessionANOVAwascon groupconsistedonlyofexposuretotheapparatus. ForGroupU, ducted on the elevation scores. The ANOVA revealeda sessions in Context A consisted ofexposure to the apparatus, significantmaineffectofgroup [F(l,14) = 7.20],asig whilesessionsinContextBconsistedofeightpresentationsofthe nificant maineffectofsession [F(5,70) = 12.59],anda USalone,withamean IT!of10min. As in theprecedingexper iments,onDay I allsessionsstartedwithtwopresentationsofthe significantgroup X sessioninteraction[F(5,70) =3.23] tonealone,withameanIT!ofImin,sothatconditioningbegan (these effects correspond to the preceding analysis of withanonnovelCSinallgroups.GroupRreceivedsessionsinthe reacquisition). In addition, there was asignificantmain orderABBA, while Group U receivedthem in the orderBAAB. effectoftrialtype[F(l,14) =5.48]and,mostimportantly, Extinction. Followingconditioning,the rats received 20ses a significant trial type X group interaction [F(I,14) = sionsofextinctioninContextA.Thereweretwo90-minsessions 17.62]. While responding overall was higher following in Context A each day. ForGroup R, each session consistedof reinforced than following nonreinforced trials, this ef eightpresentationsofthetonealone, withamean IT!of10min. ForGroupU,thesesessionsconsistedonlyofexposuretotheap fect was confined to Group R. The session X trial type paratus.Nofoodwaspresentedduringthisphase. interaction was significant [F(5,70) =2.70], but more Reacquisition. Followingextinction,theratsreceived3more importantly,thesession X trialtype X groupinteraction days(sixsessions)ofconditioninginContextA. Bothgroupsre was also significant [F(5,70) = 2.37]. Initially, for ceivedtwo90-minsessionsadayinwhichthe30-sectoneCSwas Group R only, responding was higher following rein pairedwithfood fourtimesandpresentedalonefourtimes,with forcedtrialsthanfollowingnonreinforcedtrials,but,over ameanITIof10min. Twotrialsequenceswereusedoverthesix sessions, this difference disappeared. This was con sessions. Inone sequence,reinforced(R)andnonreinforced(N) trialsalternated(i.e.,RNRNRNRN); intheothersequence,trials firmed with simple comparisons (at ex = .0125). In semialternated(i.e., RNRRNNRN). Theorderofpresentationof GroupRthereweresignificantdifferencesbetweentrial the two sequences was ASSAAS, where A refers to alternating types duringthefirsttwosessions [ts(82)>4.44],butin andSreferstosemialternating. Group Uthere werenosuchdifferences(ts < 1). A similar analysis was conducted on prescores. The Results ANOVA revealed significant main effects only ofses By the endofconditioning, rats in Group Rwere re sion [F(5,70) = 5.63J and trial type [F(I,14) =5.77J. spondingwell tothe CS;themeanelevationscoreinthe Responding during the preperiodwas somewhathigher lastfour-trial blockwas 7.72. Aone-wayrepeatedmea followingreinforcedtrialsthanfollowingnonreinforced sures ANOVA was conducted for Group R, and it re trials,wheretherespectivemeanswere1.74and1.54for vealed a significant main effect of four-trial block Group Rand 1.81 and 1.28 for Group U. The fact that [F(l5,105) = 4.21]. Extinction also proceeded without thiseffectdidnotvarywithgroup[group X trialtypein incident,withrats inGroupRshowinglittleornospon teraction, F(l,14) = 1.15] is interesting, given that the taneous recovery by Day 4. By the last session ofex tinction, Groups Rand Uhad meanelevation scores of 0.00 and -1.88, respectively. Atrial-block ANOVA re 10 vealed a significant main effect ofblock [F(39,273) = • GroupR 9 7.71]. ---.-- GroupU Reacquisition data are presented in Figure 4. During 8 reacquisition,GroupsRandUinitiallydidnotdiffer,but 7 overtrials, Rresponded more than U. Agroup X block .<.U. . ANOVA carried out on elevation scores revealed a sig u0 6 nificantmaineffectofgroup [F(l,14) =7.20],asignif C'" 5 ,"'-- .g niciafinctamntaignroeuffpecXtobflbolcokckin[tFer(alcIt,i1o5n4)[F=(II8,.11504],)an=d3a.4si5g] . ~<>U 4 ,.--.-... ",, ~ AlthoughrespondingoverallwashigherinGroup R,the 3 ... I " groupsdidnotdifferinitiallyandtowardtheendofreac 2 .--.....-." ,'eI quisition. This was confirmed with four post hoc com parisons (at ex = .0125). On Block 1, Groups Rand U didnot differ [t(44) =0.53]. However, on Blocks6and 0 10,theydid [ts(44) ~ 3.44]. ByBlock 12,the groups no 0 2 3 4 6 8 9 10 11 12 longerdiffered [t(44) = 1.86]. 4-Trial Blocks TheANOVAonprescoresrevealedonlyasignificant Figure4. Meanelevationscoresduring reacquisition in Ex main effect ofblock [F(lI,154) = 3.13]. Neither the periment2. 430 RICKERAND BOUTON this idea, itdoes appearthattherapid reacquisition ob • GroupR(r) 10 served in the present method is initially restricted to --0-- GroupR(n) trials thatfollow reinforcedtrials, as a retrieval view or 9 ---e-- GroupD(r) sequentialtheorypredicts. 8 ---0-- GroupD(n) EXPERIMENT3 OnepurposeofExperiment3was to replicate these quential learning effect observed in Experiment 2. An other purpose was to further explore the learning-to learneffectthatwasobservedinExperiment 1A. Inthat experiment, conditioning and extinction with the light 2 offCSfacilitatedconditioningofthetoneCSinthefinal phase. This sort ofresult is captured by the model pre sentedby Kehoe (1988), which states that conditioning o-t----r---,---,,....---.,.----.----. andextinctionwithanystimulusresultsinresidualexci o 2 3 4 5 6 tation which may transfer to reconditioning or condi Session tioning ofanother stimulus. An alternative possibility, however, is thatthe sequential learningeffectidentified Figure5.Meanelevationscoresduringreacquisition in Ex periment2.HeredataforGroupsRandUarebrokendowninto in Experiment 2 may contribute to this result as well. trialsfoUowingreinforcedtrials(r)andtrialsfoUowingnonrein Followingconditioning,theoccurrenceoftheUS might forcedtrials(n). potentiallyserveasageneralsignalofreinforcement,re gardless ofthe identity ofthe CS. Ifso, we might like wiseobserveenhancedrespondingfollowing reinforced elevationscoresdidvarywithgrouponthesetrials. This trialsinalearning-experiencedcontrolgroup. InExper suggests that the aftereffects ofa reinforced trial were iment 3, we therefore compared responding on trials experiencedinbothgroups (higherprescores), butonly afterreinforced andnonreinforced trials in Phase 3 ina inGroupRdidareinforcedtrialsignalreinforcementon learning-experiencedcontrol (Group L),a reacquisition thenexttrial. group,andarestcontrol. Discussion Method Reacquisitionagainoccurredrapidlycomparedwitha Subjects learning-naive control, even when a partial, ratherthan Thesubjectswere24femaleWistarratsobtainedfromCharles continuous, reinforcement procedure was used. How River,Canada.(TheratsinExperiments IA-Band2werederived ever, closer analysis revealed that high responding in fromthisstock.)Theywereapproximately90daysoldatthestart oftheexperimentandwerehousedandmaintainedasbefore.Dur Group R was initially confinedto trials following rein ingthecourseoftheexperiment,3oftheanimalsbecamesick(I forced, rather than nonreinforced, trials. This result is from eachgroup) and were removed, leaving 7 animals in each consistent with a retrieval view, especially one empha group. sizingtheretrievalroleofindividualtrials(e.g.,Capaldi, 1994). The fact thatthese differences in responding did Apparatus not lastbeyondthe first two sessions maybe due to the Theapparatus was the same as that in Experiments I and 2, withtheexceptionthatthetwoNoyesfoodpelletsthatservedas change incontingenciesbetweenreinforced andnonre theUSweredelivered0.4secapart. inforced trials and the trial types that followed them. That is, during reacquisition, nonreinforced trials now Procedure tended to signal reinforced trials and reinforced trials Pretraining.Boxexposureandmagazinetrainingwerecarried tendedtosignalnonreinforcedtrials,reversingwhatmight outasinExperiments Iand2.Approximately 18pelletswerede have been learned during extinction and conditioning. liveredduringeachofthemagazinetrainingsessions. Responding following nonreinforced trials was not Conditioning. Thesubjectsreceivedeightsessionsofcondi tioning,oneperday.ForGroupR,thesessionsconsistedofeight lowerinGroup Rthan inGroup U. Thiscouldpresenta pairingsofthe30-sectonewiththe(oodUS,withamean ITIof problemfortheretrievalview. If, forGroup R,nonrein 10min. ForGroupL, thesessionsconsistedofeightpairingsof forced trials were signaling nonreinforcement on the the 30-seclightoffwiththeUS, with amean ITIof10min. For nexttrial,thenrespondingfollowingnonreinforcedtrials GroupC,thesesessionsconsistedonlyofexposuretotheappara shouldhavebeenlowerinthisgroupthaninGroupU,for tus; no stimuliwere presented. Forall groups, the first session whomnosuchsignalingshouldoccur.Onepossibilityis startedwithtwononreinforcedpresentationseachofthetoneand lightoff,intheorderTLTL. thatifretrievalofanextinctionmemoryismoresuscep Extinction. Following conditioning,the rats received 20 ses tible to retrieval failure (Bouton, 1993), itmay bemore sions ofextinction, I perday. ForGroup R, these sessions con difficult to retrieve the memory ofnonreinforcement sistedofeightpresentationsofthetonealonewithameanITIof thanofreinforcementinGroupR.Whateverthemeritof 10min.ForGroupL,thesessionsconsistedofeightpresentations REACQUISITION AFTEREXTINCTION 431 ofthelightoffalonewithameanIT!of10min.GroupCreceived • GroupR onlyexposuretotheapparatusduringthesesessions.Nofoodwas 10 presentedduringthisphase. ---0-- GroupL Reacquisition. Followingextinction, the rats received three 9 ---/).-- Groupe ~~....o 0, moresessionsofconditioning,again oneperday. Forallgroups, 8 I , ," thetonewaspairedwithfoodfourtimesandwaspresentedalone I , I ' ,..(]-----o' four times with amean IT! of 10 min during these sessions. In 7 II " , eachsession, reinforcedandnonreinforcedtrials alternated(i.e., 8~ II RNRNRNRN). '" 6 II § 5 II I Results .~ I i) 4 oI Conditioningproceededuneventfully, with Groups R G3 and Lreaching mean elevation scores of5.93 and 5.75, 3 respectively,duringthefinal four-trial block.Agroup X 2 four-trial block ANOVA revealeda significantmain ef fect onlyofblock [F(15,210) == 2.09]. Neitherthemain effect of group nor the interaction were significant o-i---,---,---,---..,.---..,.----, [F(1,14) == 1.97andF(15,21O) == 0.93, respectively]. o 2 3 4 5 6 4-Trial Blocks Extinction proceeded uneventfully as well, with rats showinglittleornorespondingbySession7. Agroup X Figure6. Meanelevationscoresduring reacquisition in Ex blockANOVArevealedasignificantmaineffectonlyof periment3. block [F(39,468) == 1.58]. Neither the main effect of group northe interactionweresignificant(Fs < I). Thedatafrom reacquisitionarepresentedinFigure6. separately,usingthe errorterms from the overallanaly During reacquisition, Group Ronce again acquired re sis.ForGroupR,thisanalysisrevealedbothmaineffects sponding more rapidly than Group C but more slowly ofsession [F(2,36) == 9.60] and trial type [F(1,18) == thanGroupL.Agroup X blockANOVArevealedamain 7.14]. ForGroups LandC, however, there were no sig effect ofgroup [F(2,18) == 5.17] and a main effect of nificant effects in these analyses [largest F(1,18) == block [F(5,90) == 9.57]. The interactionwas notsignifi 3.89]. Furthercontrasts inGroup Lon Sessions I and2 cant [F(10,90) == 1.15]. The main effect ofgroup was revealed no differences in responding following rein explored with unplanned comparisons (at a == .017). forced versusnonreinforcedtrials [ts(24) ~ 1.36].Thus, These comparisons revealed that Group R showed less GroupR,butneitherGroupLnorGroupC,gaveevidence overall responding than Group L [t(18) == -3.19], that ofthesequentiallearningeffect.Theideathatfacilitated Group R showed more responding than Group C responding in Group L was more than a result ofse [t(18) == 4.64], andthatGroup Lshowedmorerespond quential learning was further supported by ANOVAs ing than Group C [t(18) == 7.83]. Although responding comparing the groups' responding after reinforced and inGroup Lseemsunusually high atthe outset, the first nonreinforcedtrials. RespondingwasgreaterinGroupL trial elevation score in this group wasonly -0.29. thaninGroupCfollowingbothreinforced[t(18) == 3.15] TheANOVAonprescoresrevealedasignificantmain and nonreinforced [t(18) > 2.74] trials; facilitation was effect only of block [F(5,90) == 3.82]. Prescores in not dependenton trial type. Incontrast, there was more creasedslightlyandthendecreasedoverblocks.Neither responding inGroup Rthan inGroup Cfollowing rein themaineffectofgroupnorthe interactionweresignifi forcedtrials[t(18) == 2.14,.025<p<.05]butnotfollow cant[F(2,18) == 1.24andF(10,90) == 0.96,respectively]. ing nonreinforced trials [t(18) == 1.34]. The facilitation MeanprescoresforGroups R, L, andCwere 1.86,2.91, effect for Group R, but not Group L, was thus demon and 2.15,respectively. strablyconfinedtotrialsfollowing reinforcedtrials. Figure 7 presents responding ofeach group on the A group X trial type X session ANOVA on the pre trialsfollowingreinforcedversusnonreinforcedtrials. A scores revealed a significant main effect of session group X trial type X session ANOVA revealed signifi [F(2,36) == 4.75]. However, the three-way interaction cantmaineffectsofgroupandsessionas intheprevious was also significant [F(4,36) == 2.80]. No other effects analysis [F(2,18) == 5.17 and F(2,36) == 11.06, respec weresignificantinthisanalysis[largestF(4,36) == 1.56]. tively]. The interaction ofthese factors was not signifi Following reinforced trials, mean prescores for Groups cant [F(4,36) == 1.39]. There was a significantmainef R, L, andCwere 1.64,2.63,and 1.75,respectively; fol fect of trial type [F(l,18) == 11.46], indicating that lowing nonreinforced trials, they were 2.07, 3.19, and' responding was higher overall on trials following rein 2.15, respectively. The three-way interaction suggests forcedtrials.Noneoftheinteractionsinvolvingtrialtype that differences in trial types existed for some groups weresignificant(Fs ~ 1.22). onlyon some sessions. Therefore, separatetrial type X Because we had an a priori interest in which ofthe session ANOVAs were carriedout for eachgroup sepa groups showed differentiation between trial types, trial rately,asintheanalysisofelevationscores.Theonlysig type X sessionANOVAswereconductedforeachgroup nificanteffectswereamaineffectofsessionforGroupL 432 RICKERAND BOUTON 10 9 • GroupR(r) 8 --i:}- GroupR(n) 7 o ---e-- GroupL (r) o15 6 ---0-- GroupL (n) '" § 5 ..................... GroupC(r) .~ > 4 ...._...f:s.......- GroupC(n) .,!:l. >Ll 3 2 O+----""T""----,----.......----, o 2 3 Session Figure7.MeanelevationscoresduringreacquisitioninExperiment3.Heredata arebrokendownintotrialsfollowingreinforcedtrials(r)andtrialsfollowingnon reinforcedtrials(n). [F(2,36) = 4.64] and a trial type X session interaction bothnonreinforcedandreinforcedtrialsclearlyindicates forGroupC[F(2,36) = 4.95].TheinteractioninGroupC that the learning-to-Iearn effect was not restricted to was further exploredusing simple comparisons (ata = trialsfollowingreinforcedtrials.Therefore,thelearning .017). Theserevealedonlya smallnonsignificanteffect to-learneffectevidentinthispreparationcannotbecom oftrialtypeonSession3[t(31) = -2.05,p =.05].There pletelyattributedto sequentiallearning. fore, itcanbeconcludedthatthere were noproblematic differences inprescores. EXPERIMENT4 Discussion Thepurposeofthe final experimentwastotakealast As in the previous experiments, reacquisition again lookatwhetherreacquisitioncouldoccurslowlyrelative occurred rapidly as compared with the learning-naive to alearning-naive control inthis conditioningprepara.. control (Group C) but was slow as compared with the tion. We decided to take advantage ofan implication of learning-experiencedcontrol (Group L). In this experi thesequentiallearningviewsuggestedbyExperiments2 ment, the interference effectappearedearly in the third and 3. Specifically, ifrapidreacquisition was due inpart phase,aswouldbeexpectedgiventheresultsofprevious to reinforced trials' signaling further reinforced trials, studies(Bouton, 1986;Bouton&Swartzentruber, 1989). thenonewaytoreducethefacilitativeeffect,andthuscre This result is again consistent with the idea that two ateanopportunitytoviewtheinterferenceeffect,mightbe mechanisms influence reacquisition rate: a savings to drastically reduce the number ofconditioning trials. mechanism that allows groups that have been through With fewer conditioning trials, there would be fewer conditioningandextinctionwithanyCStorespondmore occasions for the animals to associate reinforced trials inPhase3thangroupsthathavenot,andaninterference withotherreinforcedtrials,andthusreacquisitionmight mechanismthatproduceslessofthiseffectinareacqui actuallyoccurslowlyascomparedwithalearning-naive sition group than in a learning-experienced control control. group. It is interesting to note that previous experiments While responding was greater on trials following re showing slow reacquisition in CER and taste aversion inforced trials than on trials following nonreinforced haveinvolvedveryfewconditioningtrials(eightinCER, trials, further analyses determined that this effect was and one-two in taste aversion). In Experiment 4 we confinedtoGroupR.Thisresultreplicatesthemainresult therefore conducted conditioning with only eight trials. ofExperiment 2, and continues to suggestthat sequen Toobtainconditioningwiththismethodinsofewtrials, tial learning plays a role in producing rapid reacquisi we increasedthe magnitudeofthe food US from two to tion. Incontrast,Group L didnotshowareliablediffer five 45-mg food pellets and increased the ITIs from 10 entiation. Itis possible that the present experimentwas to 20 min. Trial-spacing parameters were identical to notpowerful enough to detect an effectthat might have those ofBouton and Swartzentruber(1989) in the CER existed in Group L; there was a clearnumerical trend in preparation. We hoped that by using only eight condi thedata. However,thefactthatGroupI:srespondingwas tioning trials, acquisition ofsequential learning would significantly facilitated relative to Group C's following be minimized for Group R and reacquisition would
Description: