Clues To The Relative Intensity Of The Effect Of Recognition Occasions

Context effects in recognition refer to such a phenomenon that acue-to-memory match in the context infOrmation facilitates recognition judgmefltof an episode. The systematic laboratory research could be traced to Geise1manand Glenny (1977 ) .After that, the reseaxchers have put forward many hypothesesto account fOr the prerequisites of the phenomenon, in which the most imPortamwere the integration hypothesis (Baddeleyl982), the outshining hypothesis(Smith,1986 ) and the menta1 reinstatement hypothesis (Bjork, kichardson-K1avehn, 1989).In preseni laboratory stlldy the researchers could cofltrol thephenomenon more ably based on the previous experimefltal stlldies. HOwever,there were sti1l many unceftainiies remaining in the field, the most colltroversialof Which was the basic theoretical problemwthe relationship betWeen the relativecue strength and the magnitllde of context effects. In answer to the question, thereexisted thxee main theories f the olltshining hypothesis predicts that colltext effectswill diminish as the memory strength of an item increases; the additive g1obalmatching models predict that the magnitude of colltext effects wi1l be unthectedby the memory strength of the item; and the multiplicative global matchingmodels predict that the magnitllde of the coniext effect will increase with thestrength of the item. Directed at the three Predictions,' Mamane & Phelps(1995)have advanced the Genera1 Context Model under the framework of globalactivation. Yet, facing with so many contToversia1 experimeni materials (Da1ton,1993; Smith, 1986; MUrnane & Phelps, 1995: Dougal & Caren, l999), themodel needed further developmeni in order to explain more exactly. The presentstudy aimed to imProve experimental design under the guidance of the GeneralContext Mode1 to provide more evidence, and also expected to detect someunderlying variables to stimulate further reseaxch.The research included three experiments, using the Chinese characters asmateria1s and the graduates wtth normal color sense as subjects. A1l the threeexperiment proceduxes programmed wtth Bor1and Delphi and C++ weredisplayed on the computer screen in a standardized fOrm. The contexts were:IIImanipu1ated by changing the physica1 characteristics of the screen; and the itemstrength was manipu1ated thxough a1tering presentation frequency (Experiment land 2) and inducing different levels of processing (Experiment 3). A1l theexperiments selected the hit rate as the dependent variable. Based on the aboveconsideration, the research aimed to exp1ore the following questions: 1 \ When therecognition judgment is not very urgent, how about the effects of the itempresentation frequency in the size of coniext effects (Experimeni l); 2' When therecognition judgment is urgent, then how about the effects (Experiment 2);3 \ howaboot the effects of different leve1s of processing in the size of conteXt effects(using incidental ledring paxadigm)(Experiment 3).The experiment results indicated:1' The magnitude of the colltext effects did not change significamly as afunction of the item presentation frequency regardless of the ntae of therecognition j udgment,2. The magnitude of the cofltext effects did change significam1y as afunction of the levels-of processing, and the activation of high-leve1-processingitems oatshined the context effects.In the research, experiment 1 and eXPeriment 2 supported the additive globa1matching model, while experimeni 3 conformed to the prediction of theoutshining hypothesis. Reviewing the experimental design, we can figore outthree under1ying factors: 1 \ the methods used for manipu1ating cue strength, 2.the difference of task arrangemeni; 3 \ the methods used for manipu1ating context.Clarifying the effects of the three factors will be conducive to further research. Asto the theoretical consideration, the bounded-activation hyPothesis prompted byMurnane and Phelps, and the familiaxity-bas...