perceptual match effects in direct tests of memory: the

Copyright 2002 Psychonomic Society Inc 312

Memory amp Cognition2002 30 (2) 312-323

There has been considerable interest in understandingthe conditions under which perceptual similarity betweenthe encoding stimulus and the test probe plays a criticalrole in subsequent memory performance (eg Cave Bostamp Cobb 1996 Curran Schacter amp Bessenoff 1996 Grafamp Ryan 1990 Light LaVoie Valencia-Laver Owens ampMead 1992) Part of this interest may be attributable to theconjecture that perceptual similarity matters only for im-plicit memory (for reviews see Richardson-Klavehn ampBjork 1988 Roediger amp McDermott 1993 chap 2) Forexample both Jacoby (1983) and Rajaram and Roediger(1993) have shown that the closer the physical representa-tion of the test probe to the representation of the target ex-perienced during initial encoding the better the perfor-mance on implicit (also known as indirect) tests of memory

Repetition priming effects sometimes appear to be mod-ulated by changes in type font (or changes between hand-written presentation of stimuli during study and typed pre-

sentation of stimuli at test eg Jacoby amp Hayman 1987Jacoby amp Witherspoon 1982 Roediger amp Blaxton 1987a1987b) whereas other times level of facilitation in primingappears unaffected by such changes (eg Carr Brown ampCharalambous 1989 Clarke amp Morton 1983 FeustelShiffrin amp Salasoo 1983 Graf amp Levy 1984) Roediger andMcDermott (1993 chap 2) after reviewing physical surfacevariations including typography concluded that ldquoalterationsin typography have much less effect than do changes inmodality or symbolic form (pictures to words)rdquo (p 89)

In contrast explicit (direct) tests of memory such asrecognition performance did not seem to be enhanced byclose physical matches between the item encoded duringstudy and the test probe Explicit memory tests such asfree recall were helped more by prior exposure to a pictureof the target than by seeing the printed word (Durguno Iumlgluamp Roediger 1987 Rajaram amp Roediger 1993 Weldon ampRoediger 1987) Overall there has been less evidence thatperceptual similarity between encoding and probe itemsaffects explicit memory tests (eg Craik Moscovitch ampMcDowd 1994 Murdock amp Walker 1969 Rajaram amp Roe-diger 1993) although the results of Graf and Ryan (1990)are a notable exception

EXPLANATORY FRAMEWORKS

Why does physical similarity sometimes seem to helpexplicit memory Understanding when and why physicalsimilarity affects access to a memory trace has importanttheoretical implications for our understanding of the rep-resentation and processing of memory

These experiments were described in a talk at the 40th Annual Meet-ing of the Psychonomic Society November 1999 and also as part of anaddress at the Annual Meeting of the American Psychological SocietyJune 1999 This research was supported by Grants 1-R01-MH52808 and2-R01-MH52808 from the National Institute of Mental Health Wethank Jason Arndt Melanie Cary Margaret Peterson Yue Shang andMichelle Simmons for their comments on an earlier draft of this paperand Katerina Avramides for help in running subjects Correspondenc econcerning this article should be addressed to L M Reder Departmentof Psychology Carnegie Mellon University Pittsburgh PA 15213 (e-mailredercmuedu ericksoncmuedu or dkd2pittedu)

mdashAccepted by previous editorial team

Perceptual match effects in direct tests of memory

The role of contextual fan

LYNNE M REDER DIMITRIOS K DONAVOS and MICHAEL A ERICKSONCarnegie Mellon University Pittsburgh Pennsylvania

The aim of the present study was to determine whether physical attributes of a memory representa-tion would affect explicit memory performance and if so what type of factors would affect the size ofa perceptual match effect Subjects studied words in different uncommon fonts and were later askedwhether the word had been studied earlier Words could be re-presented in the original font a fontstudied with another word or a font not seen earlier In two additional experiments we varied thenumbers of words studied in the same unusual font (1 vs 12 words per font) Recognition memory forthe words was better if the test and study fonts matched and this effect was larger for fonts not sharedwith other words Moreover old judgments were most likely to be classified as remember responseswhen words were re-presented in the same font when it had not been studied with other words Althoughwe found a significant effect of levels of processing this factor did not interact with whether the fontmatched between study and test These results are consistent with the predictions of the source of ac-tivation confusion model of memory and suggest that perceptual information operates according to thesame memory principles as conceptual information

FONTS ARE CONTEXT CUES IN MEMORY 313

There has been considerable debate as to the appropri-ate explanatory framework to account for these memory dissociationsmdashnamely that physical similarity seems somuch more important for indirect than for direct tests ofmemory On the one hand some theorists have postulatedthat there exist qualitatively different memory systemsmdashthat is an implicit and an explicit memory systemmdashwithdifferent properties and mechanisms so that the samevariables could affect these two systems differentially(eg Curran et al 1996 Schacter amp Tulving 1994 Tul-ving amp Schacter 1990) Perhaps the best known of theseproposals involves the perceptual representation system(PRS) of Schacter and Tulving (1994) which states thatthe perceptual aspects of a memory experience are storedin the implicit memory system known as PRS whereasthe semantic aspects of an experience are stored in a dif-ferent memory system

On the other hand another group of researchers hastaken this same memory dissociation (role of physicalsimilarity in the two types of memory tests) and arguedthat it is unnecessary to posit different memory systemsrather they argue that these effects can be understood interms of whether the encoding and the test processes engagethe same set of cognitive operations (Kolers 1975 1979Kolers amp Ostry 1974) This view referred to as transferappropriate processing (TAP) has been supported by re-sults described in Blaxton (1989) Roediger Srinivas andWeldon (1989) and Graf and Ryan (1990) The crux of thischallenge has been based on (1) finding dissociationsamong implicit memory tasks (eg Roediger et al 1989)and (2) finding perceptual similarity effects in explicitmemory experiments (eg Light et al 1992)

Graf and Ryan (1990) have found that matching thefont between encoding and test facilitated memory forboth direct and indirect tests but that this effect was mod-ulated by whether the encoding task focused on perceptualfeatures (enhancing the effect) or conceptual features(minimizing the effect) It is important to note that Curranet al (1996) found that perceptual specificity effects wereunaffected by encoding task thereby questioning the TAPaccount of these perceptual similarity effects In sum thereappears to be considerable uncertainty in terms of whenand why visual specificity effects occur in some explicitmemory situations and not in others

OUR PROPOSAL

In this paper we will offer a new account of when andwhy perceptual match of features can enhance memoryperformance Unlike the PRS view our postulation is thatperceptual information is represented and processed in afashion analogous to that for conceptual informationmdashforexample they strengthen and decay at the same rate andspread activation in the same way Like the TAP frameworkour belief is that allocation of attention affects how muchinformation of a particular type is represented (eg percep-tual vs elaborative) however we posit that the perceptual

match effects described above depend on the ability to ac-cess a stored memory trace not to match encoding and re-trieval processes More specifically we propose that theextent of the benefit of a perceptual match in a memorytask depends on the relative distinctiveness of the percep-tual features It is important to operationalize distinctive-ness and that is done as part of the theoretical frameworkdescribed below

Distinctiveness in the Source of Activation Confusion Model of Memory

Before defining distinctiveness within the source of ac-tivation confusion (SAC) model it is important to explainhow recognition is assumed to operate SAC adopts thenotion adopted by a number of memory theorists (eg Ja-coby amp Dallas 1981 Mandler 1980) that recognition judg-ments can be based on either of two processes familiarityor recollection Details and formal specification of therepresentation and processing assumptions of this dualprocess theory can be found elsewhere (eg Reder Ang-stadt Cary Ayers amp Erickson 2002 Reder et al 2000Reder amp Schunn 1996 Schunn Reder NhouyvanisvongRichards amp Stroffolino 1997)

According to SAC during the study phase of an exper-iment a subject may lay down a trace encoding the fact thata particular word was studied in the experimental context(consult Figure 1) Later when asked to recognize whethera word had been studied the subject will respond affir-matively either by retrieving the trace that the word wasstudied in the experimental context or by inferring that theword was studied owing to the sense of familiarity asso-ciated with the queried term Familiarity is affected by thecurrent level of activation of the concept node Recollectionrequires that the episode node associated with the acti-vated concept receive sufficient activation from the conceptand context nodes to pass over threshold The amount of ac-tivation sent from a source depends on the competitionmdashthat is the number and strength of competing associationsfrom the sending concept

We posit that distinctiveness is inversely related to thenumber of contextual associations of the concept in ques-tion In other words concepts with fewer contextual asso-ciations are more distinctive When there are fewer com-peting contextual associations it is easier to retrieve theencoding context because more of the source activationwill flow back to the associated event node Using imple-mented computational models Reder et al (2000) demon-strated that the recognition advantage of words of lowernormative frequency can be explained as being a result oftheir fewer contextual associations

Representing the Font of the Letters That Encode the Word

Figure 1 presents a schematic illustration of how we as-sume that font is represented in a visual presentation of astudy word A given font is represented as a single conceptnode that is linked to the episode node that encodes seeing

314 REDER DONAVOS AND ERICKSON

a particular word in that font The dashed line from thatfont node to the node for the word represents the processof perceiving the word in that font not an associativememory connection That process link is also schematic inthat the connection from the font node must pass betweenthe letters that are orthographic constituents of the lexicalentry associated with that concept to the concept node forthe word

Our assertion of a generalized font concept has empir-ical support from Hofstadter (Hofstadter 1985 Hofstadt-er amp McGraw 1995) who showed that people are able togeneralize the form of other letters intended by the de-signer by a sample of the letters in that font We assume that afont is represented as a node just as one can represent se-mantic conceptsmdashfor example a conceptual node for dogsWe hypothesize that these font nodes behave according tothe same memory principles as all other types of infor-mation associated with a memory trace They are strength-

ened in exactly the same way as any other type of conceptwith associated benefits and are vulnerable to the sameprinciples of interference inherent in other types of com-peting associations

It is our claim that a common font such as the one thatis used to print this document has a strong node to repre-sent the font but is associated with so many other contextsbesides this document that it is not distinctive (ie it hasa very high contextual fan) Figure 1 also illustrates the hy-pothesized memory representation for words studied in afont that is used to present many words in the experimentFor space reasons we give an example of two words stud-ied with the same font (ldquoGadzooxrdquo) but have many linesemanating from the font node to indicate that more wordswere studied with it

According to SAC when a word is presented in a fontat test activation will spread to associated contexts fromboth sources (ie the concept and font nodes) as well as

Figure 1 A schematic illustration of the proposed memory representation of several wordsin an experiment that varies font fan The word desk is illustrated as being encoded in aunique font (on the left) and two other words as encoded in a high-fan font

Experimental List Context

ldquoFarmrdquo Encoded in Experiment

Farm

ldquoDeskrdquo Encoded in Experiment

Desk

Font Font

ldquoCourtrdquo Encoded in Experiment

Court

Font

Low Font Fan High Font Fan


from the context node If the font is the same as the oneused during study there will be a link from the font nodeto the relevant episode node The fewer contextual associ-ations out of the font node (ie the less fan) the more ac-tivation can spread to the relevant episode node making itmore likely to get over threshold to enable a recollectionThe phenomenon that fan or competing associations willaffect memory performance has been demonstrated manytimes (eg Anderson 1974 Cantor amp Engle 1993 Kim-berg 1994 Lewis amp Anderson 1976 S B Peterson ampPotts 1982 Reder amp Ross 1983 Reder amp Wible 1984Zbrodoff 1995) and has been demonstrated in this theo-retical framework with words of different normative andexperimental frequency (Reder et al 2000) as well aswith pseudowords (Reder et al 2002) The idea that thefan of the font should matter has not been tested before

The two fonts used by Graf and Ryan (1990) were moredistinctive than Helvetica or Palatino but their fonts wereeach used with half of the words studied during the ex-periment so the fan out of each was very high We hypoth-esize that if each font is relatively novel and is associatedwith only one word during study the effect of matchingfont between study and test will be greater In our first ex-periment we sought to conduct a partial replication ofGraf and Ryan Like Graf and Ryan we hoped to showthat perceptual match effects can be obtained in directtests of memory They concluded that the orienting taskmodulated whether or not a perceptual match effect oc-curred We predicted that when the effect of perceptualmatch was allowed to be stronger by using fonts with less

contextual fan the perceptual match effect would obtainregardless of the orienting task during encoding

We employed the same two orienting tasks as those usedby Graf and Ryan (1990)mdashnamely judging legibility orpleasantness A noncontroversial prediction in this partialreplication is to find a main effect of orienting task so thatperformance will be better if the words are encoded usingconceptual processing such as judging pleasantness thanif the encoding involves only a superficial task such asjudging legibility

EXPERIMENT 1

MethodSubjects The subjects were 30 Carnegie Mellon undergraduate s

enrolled in psychology courses who participated as partial fulfill-ment of a course requirement Fifteen were randomly assigned to thelegibility group and 15 to the pleasantness group

Design and Materials The design consisted of three factors ori-enting task during encoding (judging legibility or pleasantness of theword) word status (judged or not during encoding) and font-matchingcondition at test (original font swapped font or novel font) The firstfactor was manipulated between subjects and the other two factorswere manipulated within subjects The third factor could not be crossedwith both levels of the second because original font was not definedfor nonstudied words During encoding a font was seen with onlyone word and each word was seen only once During test one halfof the studied words were re-presented with their original encodingfont and half were different one fourth of the words presented witha swapped font and one fourth with a novel (not previously used) fontIn order to maintain the same proportion of old to new fonts for newitems as that used for old items three quarters of the new words werepresented in an swapped font and one quarter in a novel font

Figure 2 Examples of fonts used to present words at encoding and at test


The design involved a total of 144 words and 108 distinctive fontsOf the 144 words one half were randomly sampled to make up theencoding list and all were presented at test All the words were con-crete nouns 3ndash14 letters in length (M = 54) and of moderate frequency(M = 75 range 40ndash120 occurrences per million according to KuIumlceraamp Francis 1967)

The 108 fonts were collected from various Internet sources andwere selected on the basis of the following criteria The fonts seemeddistinctive (ie seemed dissimilar to other fonts selected for the ex-periment as well as to common fonts such as Helvetica) were easilyreadable and were similar in size to the other fonts used in the ex-periment Examples of fonts used in the experiment are shown inFigure 2

For each subject fonts were randomly assigned to words and bothwords and fonts were randomly assigned to conditions Since thisrandom assignment was performed separately for each subject anystimulus effects would become part of the subject error term

Procedure The subjects were tested individually in sessions thatlasted approximately 30 min The session consisted of two phasesan encoding phase and a word recognition test phase Prior to theencoding phase the subjects were told that the purpose of the ex-periment was to explore what attributes of a word make it eithermore or less legible (for the legibility group) or more or less pleasant(for the pleasantness group) They were not told that they were to re-member the words for a subsequent test

During the encoding phase a series of 72 words each presentedin a different unique font was presented to the subjects who wereasked to judge them In the legibility group they were instructed toldquoignore the meaning of the word and to only pay attention to the fontthat the word was presented inrdquo The subjects in the pleasantness groupwere told to ldquoignore the font that the word was presented in and toonly pay attention to the meaning of the wordrdquo The subjects madetheir judgments on a 4-point Likert scale in which 1 indicated verylegiblepleasant and 4 indicated very illegibleunpleasan t Wordswere displayed for 2 sec before the Likert scale was displayed thatprompted the subjects to make a rating Responses were entered onthe computer keyboard with the keys labeled for easy identification

In the subsequent test phase individual words were displayed onthe computer screen and the subjects were instructed either to pressa key labeled ldquoNrdquo if they believed the displayed word (irrespectiveof the font) had not been presented during the encoding phase or topress ldquoOrdquo if they remembered seeing the displayed word during theencoding phase This procedure continued until all 144 words (72 stud-ied and 72 new words) had been judged

Results and DiscussionTable 1 presents the proportions of hits and false alarms

and mean d9 as a function of encoding task (pleasantnessor legibility) The proportions are also partitioned by fontmatch condition in the case of old words whether the fontused at test matched the font used during encoding (orig-inal) had been used with another word (swapped) or waspresented in a font not used during encoding (novel) Inthe case of new words only the latter two categories ap-plied Given that the false alarm rate for the original fontcondition is undefined d9 for the original font conditionwas calculated using the false alarm rate of the swappedcondition We conducted analyses of variance (ANOVAs)on hits and d9 scores using encoding task and font-matching condition as factors (We also analyzed falsealarms as a function of orienting task) The analyses wereconducted both with and without the novel condition Thecomparison of our results to those of Graf and Ryan

(1990) is more straightforward when the novel conditionis omitted but both analyses are listed for completeness(the F statistic that includes the novel condition is listedsecond)

As one would expect from manipulations of levels ofprocessing (Craik amp Lockhart 1972) subjects who ratedpleasantness correctly recognized more old words thandid the subjects judging legibility [F(128) = 1134 MSe =0021 p lt 01 F(128) = 1358 MSe = 0039 p lt 01]There were also fewer false alarms in the pleasantnesscondition than in the legibility condition but this differ-ence was not reliable The signal detection analysis sup-ports the same conclusion The subjects were reliablymore accurate in the pleasantness condition [F(128) =991 MSe = 0866 p lt 01 F(128) = 1257 MSe = 1467p lt 01]

Comparison of d9 for just the swapped and the novelfonts reveals an interaction of encoding task and font type[F(128) = 450 MSe = 0141 p lt 05] so that the novelfont yielded a higher d9 for the words in the pleasantnessjudgment task and the swapped font yielded a higher d9 forthe subjects in the legibility judgment task however nei-ther specific contrast (within encoding task) was reliableand the pattern was not replicated in the experiments to bedescribed later

Of greater interest is the effect of font matching As waspredicted there was a main effect on hits [F(128) = 1828MSe = 0010 p lt 001 F(256) = 2782 MSe = 0010 p lt001] reflecting the better performance when the font ofthe word at test matched the font used to present the wordduring encoding The signal detection analysis reinforcedthese conclusions showing a reliable effect of matchingfont on d9 [F(128) = 1417 MSe = 0123 p lt 001 F(256) =907 MSe = 0139 p lt 001]

Figure 3 compares the results of Graf and Ryan (1990in the right panel) and of our analogous conditions (in theleft panel) Although Graf and Ryan did not report an in-teraction between encoding task and font match condition(original vs swapped) their conclusions were based on a

Table 1 Proportion of Hits and False Alarms and Mean d9 Scores as aFunction of Font-Matching Condition and Encoding Task in

Experiment 1

Font-Matching Condition

GroupDV Original Swapped Novel

PleasantnessHit 87 78 74FA NA 29 22d9 189 159 176

LegibilityHit 77 64 53FA NA 35 33d9 117 079 056

NotemdashFont-matching indicates what condition the test word was seenin NA means not applicable since the original font is undefined fornonstudied items Consult the text for definition of conditions DV de-pendent variable


significant difference owing to font matching in the legi-bility condition and the absence of a difference in thepleasantness condition In our experiment the same con-ditions showed significant effects on d9 of both encodingtask and font matching condition (reported above) and theinteraction was not reliable (F lt 1)

In sum like Graf and Ryan (1990) we found a main effectof orienting task however we also found a main effect offont-matching condition but no interaction of orientingtask with the font-matching effect It is worth noting thata proponent of the TAP framework might argue that ourfailure to find an interaction was due to lack of power de-spite our two main effects Thus the question naturallyarises of whether our main effect of font matching was dri-ven totally by the legibility orienting task condition orwhether unlike Graf and Ryan we found a reliable effectof font type in the pleasantness orienting task conditionEven in the pleasantness condition the original font con-dition produced better performance than did the swappedfont condition [F(114) = 458 MSe = 0152 p = 05]

Given that we obtained a sizeable main effect of orient-ing task the difference between our results (ie finding aneffect of matching font for both orienting tasks) and Grafand Ryanrsquos (1990 ie finding it only for the perceptualorienting task) should not be attributed to a weakness in ourorienting task manipulation So why was there a differencebetween our results and theirs Two explanations occur tous As was explained earlier we believe that Graf and Ryanrsquosfont manipulation was weaker than ours because they usedonly two different fonts whereas we used a different un-usual font for each study word Another explanation al-though less interesting is that their font-matching effectwas masked by a ceiling effect in the pleasantness orient-

ing task condition Not surprisingly according to levels ofprocessing (Craik amp Lockhart 1972) subjectsrsquo perfor-mance was much better in the pleasantness judgment taskthan in the legibility judgment task condition in both ourstudy and Graf and Ryanrsquos Given that Graf and Ryanrsquos sub-jects outperformed ours in the pleasantness condition per-haps our subjects were not vulnerable to the ceiling effectproblem whereas theirs were Experiment 2 was designedto test both the font fan and the ceiling effect explanations

EXPERIMENT 2

The same two orienting tasks were used as those in Ex-periment 1 but in several other respects Experiment 2 dif-fered from Experiment 1 Words were again presented inunusual fonts however for half of the words that werejudged during the orienting task the font used to displaythat word was a font that also was used to present 11 otherwords Words displayed in fonts shared by 12 words dur-ing encoding were described as being in the high-fan fontcondition We consider the high-fan font condition to beroughly analogous to the Graf and Ryan (1990) paradigmFor the remainder of the words each presentation fontused during encoding was unique to that word Thesewords were thus assigned to the low-fan font conditionthat replicates the design used in Experiment 1

The other difference from Experiment 1 was the addi-tion of a second task whose purpose with respect to thisstudy was to vary the delay between encoding and test Atissue was whether the delay between study and test wouldinfluence recognition accuracy and whether that variablewould differentially affect the impact of font fan on fontmatching depending on orienting task

00

05

10

15

20

25

dcent

PleasantnessLegibility

Original SwappedFont Match Condition

Experiment 1

00

05

10

15

20

25


Experiment 3Graf amp Ryan (1990)

dcent


Figure 3 Comparison of Graf and Ryanrsquos (1990) data (from their Experiment 3 Table 9 on p 986) withcorresponding data from our Experiment 1 Figures plot mean d 9 scores as a function of encoding task andfont-matching condition


MethodSubjects The subjects were 28 Carnegie Mellon undergraduates

who were paid $8 for their participation The pleasantness group in-cluded a total of 13 subjects (6 in the immediate and 7 in the delayedrecognition conditions) The legibility group included a total of 15 sub-jects (8 in the immediate and 7 in the delayed recognition conditions)

Design and Materials The design consisted of five factors thefirst three being the same as those in Experiment 1 orienting task(judging legibility vs pleasantness) word status (judged vs not seenduring encoding) and font-matching condition (original vs swappedvs novel) Two new factors were introduced font fan (high-fan fontvs low-fan font) and delay (immediate vs delayed test) Orientingtask and delay were the only between-subjects factors The level offont fan for a given swapped font matched the level of font fan usedwith that word during encoding

The words were not identical to those used in Experiment 1 but weredrawn from the same database using the same criteria The fontswere from the same pool as that generated for Experiment 1 Con-struction of lists was quite similar to Experiment 1 except for the in-corporation of the font fan factor

Procedure The subjects were tested individually in sessions thatlasted approximately 80 min The session consisted of three phasesan encoding task a filler task and the recognition task The lattertwo tasks were counterbalanced across subjects for each encodinggroup The filler task was word identification of briefly presentedwords The stimuli included words judged during the encodingphase as well as words that had not been rated Words and fontswere randomly assigned to appear in either the identification or therecognition taskmdashthat is they did not appear in both tasks The en-coding task instructions and the procedure during the recognitionphase were the same as those in Experiment 1 Assignment to judg-ment task and delay condition was randomly determined with theconstraint that an equal number of subjects was assigned to eachcondition

Results and DiscussionTable 2 presents the mean proportions of hits false

alarms and d9 scores as a function of encoding task font-matching condition and font fan and by whether recogni-tion immediately followed encoding or was delayed by anintervening task As before d9 was calculated so that thefalse alarms for the original font condition were based onswapped fonts because original fonts do not exist for newwords For novel fonts high versus low fan is not definedso there is only one entry for that factor

As with Experiment 1 we conducted analyses over allthree levels of matching (including the novel font condi-tion) but for many analyses it does not make sense to in-clude that level inasmuch as fan is not defined for novelfont (F statistics from analyses that are possible with thenovel font condition are listed second) The four-wayANOVAs on d9 used encoding task font-matching condi-tion fan of font and delay of test as factors

Not surprisingly there was a main effect of encodingtask on d9 [F(124) = 6893 MSe = 1567 p lt 0001F(124) = 7826 MSe = 1093 p lt 0001] so that judgingpleasantness produced better recognition performancethan did judging legibility Also as was expected therewas a main effect of delay [F(124) = 627 MSe = 1567p lt 05 F(124) = 726 MSe = 1093 p lt 05] so that per-formance declined with delay The interaction of these twofactors was not reliable (Fs lt 1) We again demonstrated

a perceptual match effect in an explicit memory task Thesubjects performed best if the font at test matched the fontof the word during encoding [F(124) = 1087 MSe =0199 p lt 01 F(248) = 291 MSe = 0203 p = 06] Also

Table 2 Proportions of Hits and False Alarms (FAs) and Mean d9 Scores

as a Function of Font Fan Font Match Encoding Task andTest Delay in Experiment 2

EncodingFan Font-Matching Condition

Condition Original Swapped Novel

Immediate Test ConditionPleasantness

High FanHits 94 87FA NA 06d9 343 300

Low FanHits 92 91FA NA 10d9 304 301

NovelHits 86FA 05d9 323

LegibilityHigh Fan

Hits 63 63FA NA 33d9 088 085



Delayed Test ConditionPleasantness




LegibilityHigh Fan

Hits 58 60FA NA 40d9 054 063




replicating Experiment 1 the effect of matching font didnot interact with encoding task (Fs lt 1)

The font fan manipulation did not produce a main ef-fect nor did it interact with encoding task (all Fs lt 2) Im-portantly we did find the predicted interaction betweenthe font fan and the font-matching conditions [F(124) =434 MSe = 0235 p lt 05] A low-fan font facilitated per-formance more than did a high-fan font when the font wasthe original encoding font This prediction derives fromthe assumption that there exists a link from the originalfont node to the episodic node that encodes the studyevent Activation will spread from a font node to theepisodic node only if the font at test is the same as the oneused during the encoding phase How much will spreadfrom the font node to any episodic node depends on thenumber (fan) of competing associationsmdashthat is howmany other words have been seen with that font

Although there is no link from a swapped font to theepisode node it is worth noting that there was a suggestionof a greater tendency to say old for both new and old wordsin the swapped font condition when the font was high fanWe believe that the greater familiarity of the high-fanfonts causes this spurious feeling of familiarity Indeedrecently we have replicated that finding (M J Peterson ampReder 2002)

Just as we had failed to find an interaction of the font-matching condition and orienting task the three-way in-teraction of font fan font-matching condition and orient-ing task (on d9) also was not reliable [F(124) = 267MSe = 0235 p gt 05] On the other hand the three-way in-teraction of font fan font-matching condition and delaywas reliable [F(124) = 449 MSe = 0235 p lt 05] Fig-ure 4 shows an interaction of font fan and font matchingbut only in the delay condition Earlier we speculated thatGraf and Ryan (1990) might have failed to find an effect

of font matching in the pleasantness judgment conditionowing to a ceiling on performance with that orienting taskIn this experiment our subjects performed near ceiling inthe immediate pleasantness condition and near floor in thedelayed legibility condition Our failure to find sensibleresults in those two conditionsmdashnamely no advantage oforiginal over swapped font for low fan in the immediatecondition for the pleasantness task or for high fan in thelegibility task at a delaymdashmay have been the result of ceil-ing and floor effects respectively

One might question the robustness of the (significant)interaction of font fan with font matching given that therewas a higher three-way interaction with delay One of theobjectives of the next experiment was to keep performanceoff both the ceiling and the floor and to clearly establishthe result that the font-matching effect is modulated byfont fan so that it is larger in the low font fan condition butexists in the high-fan condition too

Stronger test of our theoretical position Accordingto SAC the reason why low-fan fonts are at an advantagein the matching font condition is that the link from the fontnode to the episode node will send more activation whenthere are not many competing contextual links from thenode to siphon off activation (refer to Figure 1) We alsopostulate that recollections derive from sufficient activa-tionrsquos reaching the node that represents the trace of havingstudied the word in the experimental context If there isinsufficient activation at the episode node subjects willguess whether the word was studied on the basis of a fa-miliarity assessment derived from the activation level ofthe concept node or in some cases on the basis of the fa-miliarity of the font node If our account is correct notonly should there be more old responses when there is alink from the font node to the episode node (ie in thematch condition) and when that link has fewer competi-

00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan

Figure 4 Mean d 9 scores as a function of font-matching condition and fan font for the immediate anddelayed tests collapsed over encoding task for Experiment 2


tors (the low font fan condition) but this advantage shouldbe manifested exclusively in the recollective responsesExperiment 3 employs the rememberknow paradigm totest these predictions

EXPERIMENT 3

RememberKnow ParadigmThe rememberknow paradigm (Gardiner 1988 Tul-

ving 1985) requires subjects to assess the type of memo-rial information that led to each old judgment In this par-adigm after a subject responds old the next judgment isto assess whether or not the old judgment was based on arecollection of experiencing the item on the study list Ifso the subject is to respond remember When subjects donot specifically recollect the experience of studying theitem but instead feel that the item is sufficiently familiarthat it must have been on the study list the appropriate re-sponse is know

Although it might seem surprising that subjects can ac-tually make remember versus know discriminations theprocedure has been used many times with considerablesuccess in terms of separating these two types of processes(Gardiner 1988 Gardiner amp Java 1990 Gardiner amp Parkin1990 Rajaram 1993 Tulving 1985) It is worth notingthat there exist signal detection accounts of rememberknow judgments (eg Donaldson 1996 Hirshman ampHenzler 1998 Hirshman amp Master 1997 Inoue amp Bellezza1998) suggesting that the different responses merely re-flect a shift in criterion However there is ample evidencethat such an account is insufficient to explain the data (seeReder et al 2000 for a more detailed discussion)

Given the representational assumptions of the SACmodel it has been useful to employ this procedure alsobecause remember maps naturally onto recollectionsSubjects should give more remember responses to studiedwords tested in the font used at encoding and more if thatfont had been studied with only one word rather than witha dozen

MethodSubjects The subjects were 64 Carnegie Mellon Undergraduate s

enrolled in psychology courses who participated as a partial fulfill-ment of their course requirements

Procedure Design and Materials The materials procedure anddesign of Experiment 3 differed little from those in Experiment 2Given that the orienting task factor did not interact with the factorsof interest in either Experiment 1 or 2 and seemed only to producelarge differences in performance level this factor was dropped In-stead all the subjects were asked to judge the appropriateness of thefont used to present the word thereby encouraging them to attend toboth the meaning and the perceptual aspects of the words presented The appropriateness of the font to the wordrsquos meaning was rated ona Likert scale with four values

The other factor that was no longer varied was delay between en-coding and test Instead all the subjects performed a distractor task(arbitrarily one of two psychometric tests one measuring nonverba lfluency and the other focal visual attention neither used words or thetype of fonts used in the study list) between the encoding task and therecognition test The filler test was administered for exactly 10 min

and was included to minimize ceiling effects We chose a differenttype of filler task than the one used in Experiment 2 to ensure thatour results were not dependent on a filler task that involved similartypes of materials

The other difference between this experiment and the previousones was that we asked the subjects to judge whether their old re-sponses were remember responses or know responses After com-pleting the filler task the subjects were instructed to respond in oneof three ways When the subjects believed that the displayed wordhad not been presented earlier they were to press a key labeled ldquoNrdquowhen the subjects had a conscious recollection of having seen thedisplayed word in the study phase they were instructed to press ldquoRrdquowhen the subjects knew that the displayed word had been presentedearlier but did not have a conscious recollection of having seen itthey were instructed to press ldquoKrdquo Before the subjects were allowedto begin the recognition phase of the experiment they were carefullyinstructed in how to distinguish remember from know memorieswere given everyday examples and were asked to provide examplesof their own that illustrated that they understood the distinction be-tween responding remember and responding know

The same fonts were used as those in Experiment 2 Likewise thepool of words used for this experiment was sampled from the samedatabase as that used in Experiment 2 Stimuli were randomly as-signed to condition for each subject

Results and DiscussionTable 3 presents the mean proportions of hits and false

alarms and mean d9 scores as a function of font-matchingcondition and font fan The proportion of remember re-sponses is listed parenthetically beside the proportion ofhits The remember values are absolute proportions andthus the number will always be smaller than the corre-sponding hit proportion (The difference between the twovalues represents the proportion of know responses forthat condition)

As in Experiments 1 and 2 we conducted ANOVAs ond9 performance with and without the novel font conditionplacing the statistics that include the novel font conditionsecond Again there was a main effect of font-matchingcondition [F(163) = 3862 MSe = 013 p lt 001 F(2124) =2056 MSe = 010 p lt 001] so that recognition accuracy

Table 3 Proportion of Hits and False Alarms (FAs) and

Mean d9 Scores as a Function of Font-Matching and Font Fan Conditions in Experiment 3



High fanHits 85 (58) 80 (45)FA NA 33d9 169 152

Low fanHits 90 (68) 81 (50)FA NA 30d9 202 163

NovelHits 77 (44)FA 29d9 149

NotemdashThe proportions of remember judgments for hits are given inparentheses


was best when the fonts at encoding and test matchedThere was also a significant effect of font fan [F(163) =909 MSe = 035 p lt 005] so that recognition accuracywas better when the test font was low fan As was antici-pated the interaction between font-matching conditionand font fan was statistically reliable [F(163) = 727MSe = 011 p lt 01] The left panel of Figure 5 illustratesthis interaction between font fan and whether the fontmatched between encoding and test (novel fonts are unde-fined for fan font and hence are excluded from the figure)

Remember judgments The unique predictions forExperiment 3 involve the proportion of remember responsesin the various conditions As was predicted there was amain effect of font-matching condition [F(163) = 8718MSe = 002 p lt 0001 F(2124) = 6529 MSe = 001 p lt0001] so that there were more remember hits when thefonts matched from encoding to test Also there was amain effect of font fan [F(163) = 2041 MSe = 0017 p lt0001] so that there were more remember hits when thefont was unique than when the font had been studied witha dozen words This greater tendency to give a rememberresponse for low- than for high-fan fonts was twice aslarge in the original font condition than in the swappedfont condition (12 vs 6 respectively) and this inter-action between fan and font-matching condition was reli-able [F(163) = 405 MSe = 001 p lt 05] This interactionis illustrated in the right panel of Figure 5

GENERAL DISCUSSION

In this article our goal has been to explore the mecha-nisms that drive the perceptual match effect in recognitionmemory In particular we have sought to test an accountfor these effects that does not posit different mechanismsthan one would posit to explain semantic (ie conceptual)memory effects We have modeled the effects describedin this paper with a simulation that uses the same param-

eter values (eg for spread of activation decay strength-ening) for the perceptual information as for the semanticinformation (Cary amp Reder 2000) These parameter val-ues were taken from earlier modeling efforts in other do-mains (eg Reder et al 2000)

Although a levels of processing manipulation affectedmemory performance as has traditionally been found thismanipulation did not interact with the matching font ef-fect The variable that modulated the matching font effectwas the fan of the font A matching font that was uniqueproduced a greater memory benefit and more recollectionresponses than did a matching font that had been associ-ated with a dozen words Our results do not rule out thepossibility that a portion of the effect is modulated by amatch in processing as is proposed by the TAP frame-work however the perceptual matching effects seemmore dependent on a memory representation and retrievalmechanismmdashspecifically the font fan effect as specifiedin SAC the degree of interference from the variable num-ber of associations competing for activation from the fontAlthough it might be possible to develop an explanationfor these effects within the TAP framework it is importantto emphasize that these effects were predicted by SAC notpostdicted

The perceptual matching effects in our experiment canbe explained without positing separate memory systemssuch as the ones offered in the PRS of Schacter and Tul-ving (1994) According to that view perceptual memoryeffects are due to an independent processing system thatpresumably operates in a qualitatively different way thandoes conceptual memory processes However the natureof these results seems quite consistent with an accountthat treats perceptual information exactly the same as con-ceptual information Not only have we demonstrated thatperceptual matching effects can be obtained for explicitmemories we have also demonstrated these same font-matching effects for implicit memory tasks (Cary amp Reder

00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan

Figure 5 Mean d9 scores (left panel) and proportions of remember hits (right panel) as a function of font-matching condition and fan font for Experiment 3


2000) Here too we used the same parameter values aswell as assuming the same memory representation forboth implicit and explicit memory for both perceptual andsemantic information

ConclusionIn summary we have offered a partial explanation for

when and why one obtains perceptual similarity effects inmemory We suspect that the modality effects are a gener-alization of these findings and that those effects in audi-tion (eg Clarke amp Morton 1983 Graf Shimamura ampSquire 1985 Jacoby amp Witherspoon 1982 KirsnerMilech amp Standen 1983 Light et al 1992 Roediger ampBlaxton 1987a 1987b Schacter amp Graf 1989) would havebeen stronger if a different voice had been used for eachstimulus item Our interpretation also implies that thememory advantage associated with pictures over wordsmight be attributable in part to the greater distinctivenessof the pictures After all any given picture like an unusualfont has been seen in relatively few contexts and will haveless contextual fan to interfere with episodic recognitionIn fact in recent work in our laboratory (Arndt amp Reder2001) we have found that distinctiveness ratings for fontsare affected by the randomly determined contextual variabilitymdashthat is the number of words with which eachfont was paired during the study phase

It is not necessary to assume a multiple systems view(ie that perceptual and conceptual information are storedin and processed by qualitatively different systems) to ex-plain our results Likewise these data do not require the as-sumptions of the multiple processes view that perceptualmemory effects are somehow separate from conceptualmemory effects and that these effects are only invoked whenan encoding task orients processing in one way or theother This is not to say that the way people process infor-mation does not have an impact on their subsequent pro-cessing of the same material (Franks Bilbrey Lien ampMcNamara 2000) Our data suggest that performance de-pends on more than recreating the same processing at testthat occurred during encoding Performance appears to beaffected by a representation and processing pair in such away that if the representation has a greater number ofcompeting associations performance will be degraded

REFERENCES

Anderson J R (1974) Retrieval of propositional information fromlong-term memory Cognitive Psychology 6 451-474

Arndt J amp Reder L M (2001) The role of contextual competitionon judgments of distinctiveness Manuscript in preparation

Blaxton T A (1989) Investigating dissociations among memorymeasures Support for a transfer-appropriate processing frameworkJournal of Experimental Psychology Learning Memory amp Cogni-tion 15 657-668

Cantor J amp Engle R W (1993) Working-memory capacity as long-term memory activation An individual-differences approach Journalof Experimental Psychology Learning Memory amp Cognition 191101-1114

Carr T H Brown J S amp Charalambous A (1989) Repetition

and reading Perceptual encoding mechanisms are very abstract butnot very interactive Journal of Experimental Psychology LearningMemory amp Cognition 15 763-778

Cary M amp Reder L M (2000 November) Modeling the role of per-ceptual features in word recognition and fragment completion Paperpresented at the 41st Annual Meeting of the Psychonomic SocietyNew Orleans

Cave C B Bost P R amp Cobb R E (1996) Effects of color and pat-tern on implicit and explicit picture memory Journal of Experimen-tal Psychology Learning Memory amp Cognition 22 639-653

Clarke R amp Morton J (1983) Cross modality facilitation in tachis-toscopic word recognition Quarterly Journal of Experimental Psychol-ogy 35A 79-96

Craik F I M amp Lockhart R S (1972) Levels of processing Aframework for memory research Journal of Verbal Learning amp Ver-bal Behavior 11 671-684

Craik F I M Moscovitch M amp McDowd J M (1994) Contribu-tions of surface and conceptual information to performance on im-plicit and explicit memory tasks Journal of Experimental Psychol-ogy Learning Memory amp Cognition 20 864-875

Curran T Schacter D L amp Bessenoff G (1996) Visual speci-ficity effects on memory Beyond transfer appropriate processingCanadian Journal of Experimental Psychology 50 22-33

Donaldson W (1996) The role of decision processes in rememberingand knowing Memory amp Cognition 24 523-533

Durguno Iumlglu A Y amp Roediger H L III (1987) Test differences inaccessing bilingual memory Journal of Memory amp Language 26377-391

Feustel T C Shiffrin R M amp Salasoo A (1983) Episodic andlexical contributions to the repetition effect in word identificationJournal of Experimental Psychology General 112 309-346

Franks J J Bilbrey C W Lien K G amp McNamara T P (2000)Transfer-appropriate processing (TAP) and repetition priming Mem-ory amp Cognition 28 1140-1151

Gardiner J M (1988) Functional aspects of recollective experienceMemory amp Cognition 16 309-313

Gardiner J M amp Java R I (1990) Recollective experience in wordand nonword recognition Memory amp Cognition 18 23-30

Gardiner J M amp Parkin A J (1990) Attention and recollective ex-perience in recognition memory Memory amp Cognition 18 579-583

Graf P amp Levy B A (1984) Reading and remembering Conceptualand perceptual processing involved in reading rotated passages Jour-nal of Verbal Learning amp Verbal Behavior 23 405-424

Graf P amp Ryan L (1990) Transfer-appropriate processing for im-plicit and explicit memory Journal of Experimental PsychologyLearning Memory amp Cognition 16 978-992

Graf P Shimamura A P amp Squire L R (1985) Priming acrossmodalities and priming across category levels Extending the domainof preserved function in amnesia Journal of Experimental Psychol-ogy Learning Memory amp Cognition 11 386-396

Hirshman E amp Henzler A (1998) The role of decision processesin conscious recollection Psychological Science 9 61-65

Hirshman E amp Master S (1997) Modeling the conscious correlatesof recognition memory Reflections on the rememberndashknow para-digm Memory amp Cognition 25 345-351

Hofstadter D (1985) Metafont metamathematics and metaphysicsIn D Hofstadter (Ed) Metamagical themas Questing for the essenceof mind and pattern (pp 260-296) New York Basic Books

Hofstadter D amp McGraw G (1995) Letter spirit Esthetic percep-tion and creative play in the rich microcosm of the Roman alphabetIn D Hofstadter amp the Fluid Analogies Research Group Fluid con-cepts and creative analogies (pp 405-466) New York Basic Books

Inoue C amp Bellezza F S (1998) The detection model of recogni-tion using know and remember judgments Memory amp Cognition 26299-308

Jacoby L L (1983) Remembering the data Analyzing interactiveprocesses in reading Journal of Verbal Learning amp Verbal Behavior22 485-508

Jacoby L L amp Dallas M (1981) On the relationship between auto-


biographical memory and perceptual learning Journal of Experimen-tal Psychology General 110 306-340

Jacoby L L amp Hayman C A (1987) Specific visual transfer in wordidentification Journal of Experimental Psychology Learning Mem-ory amp Cognition 13 456-463

Jacoby L L amp Witherspoon D (1982) Remembering withoutawareness Canadian Journal of Psychology 36 300-324

Kimberg D Y (1994) Executive functions working memory andfrontal lobe function Unpublished doctoral dissertation CarnegieMellon University

Kirsner K Milech D amp Standen P (1983) Common and modality-specific processes in the mental lexicon Memory amp Cognition 11621-630

Kolers P A (1975) Specificity of operations in sentence recognitionCognitive Psychology 7 289-306

Kolers P A (1979) Reading and knowing Canadian Journal of Psy-chology 33 106-117

Kolers P A amp Ostry D J (1974) Time course of loss of informa-tion regarding pattern analyzing operations Journal of Verbal Learn-ing amp Verbal Behavior 13 599-612

Ku Iumlcera H amp Francis W N (1967) Computational analysis of present-day American English Providence RI Brown University Press

Lewis C H amp Anderson J R (1976) Interference with real worldknowledge Cognitive Psychology 8 311-335

Light L L LaVoie D Valencia-Laver D Owens S A ampMead G (1992) Direct and indirect measures of memory for modal-ity in young and older adults Journal of Experimental PsychologyLearning Memory amp Cognition 18 1284-1297

Mandler G (1980) Recognizing The judgment of previous occur-rence Psychological Review 87 252-271

Murdock B B amp Walker K D (1969) Modality effects in free re-call Journal of Verbal Learning amp Verbal Behavior 8 665-676

Peterson M J amp Reder L M (2002) The role of spurious featurefamiliarity in recognition memory Manuscript in preparation

Peterson S B amp Potts G R (1982) Global and specific compo-nents of information interpretation Journal of Verbal Learning amp Ver-bal Behavior 21 403-420

Rajaram S (1993) Remembering and knowing Two means of accessto the personal past Memory amp Cognition 21 89-102

Rajaram S amp Roediger H L III (1993) Direct comparison of fourimplicit memory tests Journal of Experimental Psychology Learn-ing Memory amp Cognition 19 765-776

Reder L M Angstadt P Cary M Ayers M A amp Erickson M A(2002) A reexamination of stimulus-frequency effects in recognitionTwo mirrors for low- and high-frequency pseudowords Journal of Ex-perimental Psychology Learning Memory amp Cognition 28 138-152

Reder L M Nhouyvanisvong A Schunn C D Ayers M SAngstadt P amp Hiraki K (2000) A mechanistic account of the mir-ror effect for word frequency A computational model of rememberndashknow judgments in a continuous recognition paradigm Journal of Ex-perimental Psychology Learning Memory amp Cognition 26 294-320

Reder L M amp Ross B H (1983) Integrated knowledge of differenttasks The role of retrieval strategy on fan effects Journal of Experi-mental Psychology Learning Memory amp Cognition 9 55-72

Reder L M amp Schunn C D (1996) Metacognition does not implyawareness Strategy choice is governed by implicit learning and mem-ory In L M Reder (Ed) Implicit memory and metacognition (pp 45-77) Mahwah NJ Erlbaum

Reder L M amp Wible C (1984) Strategy use in question-answering Memory strength and task constraints on fan effects Memory amp Cog-nition 12 411-419

Richardson-Klavehn A amp Bjork R A (1998) Measures of mem-ory Annual Review of Psychology 36 475-543

Roediger H L III amp Blaxton T A (1987a) Effects of varyingmodality surface features and retention interval on priming in word-fragment completion Memory amp Cognition 15 379-388

Roediger H L III amp Blaxton T A (1987b) Retrieval modes pro-duce dissociations in memory for surface information In D S Gor-fein amp R R Hoffman (Eds) The Ebbinghaus Centennial Conference(pp 349-379) Hillsdale NJ Erlbaum

Roediger H L III amp McDermott K B (1993) Implicit memory innormal human subjects In H Spinnler amp F Boller (Eds) Handbookof neuropsycholog y (pp 64-131) Amsterdam Elsevier

Roediger H L III Srinivas K amp Weldon M S (1989) Dissoci-ations between implicit measures of retention In S S LewandowskyJ C Dunn amp K Kirsner (Eds) Implicit memory Theoretical issues(pp 67-84) Hillsdale NJ Erlbaum

Schacter D L amp Graf P (1989) Modality specificity of implicitmemory for new associations Journal of Experimental PsychologyLearning Memory amp Cognition 15 3-12

Schacter D L amp Tulving E (1994) What are the memory systemsof 1994 In D L Schacter amp E Tulving (Eds) Memory systems(pp 1-38) Cambridge MA MIT Press

Schunn C D Reder L M Nhouyvanisvong A Richards D Ramp Stroffolino P J (1997) To calculate or not to calculate A sourceactivation confusion model of problem familiarityrsquos role in strategyselection Journal of Experimental Psychology Learning Memoryamp Cognition 23 3-29

Tulving E (1985) Memory and consciousness Canadian Psycholo-gist 26 1-12

Tulving E amp Schacter D L (1990) Priming and human memorysystems Science 247 301-306

Weldon M S amp Roediger H L III (1987) Altering retrieval de-mands reverses the picture superiority effect Memory amp Cognition15 269-280

Zbrodoff N J (1995) Why is 9 + 7 harder than 2 + 3 Strength and in-terference as explanations of the problem-size effect Memory amp Cog-nition 23 689-700

(Manuscript received August 3 2000revision accepted for publication November 19 2001)


There has been considerable debate as to the appropri-ate explanatory framework to account for these memory dissociationsmdashnamely that physical similarity seems somuch more important for indirect than for direct tests ofmemory On the one hand some theorists have postulatedthat there exist qualitatively different memory systemsmdashthat is an implicit and an explicit memory systemmdashwithdifferent properties and mechanisms so that the samevariables could affect these two systems differentially(eg Curran et al 1996 Schacter amp Tulving 1994 Tul-ving amp Schacter 1990) Perhaps the best known of theseproposals involves the perceptual representation system(PRS) of Schacter and Tulving (1994) which states thatthe perceptual aspects of a memory experience are storedin the implicit memory system known as PRS whereasthe semantic aspects of an experience are stored in a dif-ferent memory system

On the other hand another group of researchers hastaken this same memory dissociation (role of physicalsimilarity in the two types of memory tests) and arguedthat it is unnecessary to posit different memory systemsrather they argue that these effects can be understood interms of whether the encoding and the test processes engagethe same set of cognitive operations (Kolers 1975 1979Kolers amp Ostry 1974) This view referred to as transferappropriate processing (TAP) has been supported by re-sults described in Blaxton (1989) Roediger Srinivas andWeldon (1989) and Graf and Ryan (1990) The crux of thischallenge has been based on (1) finding dissociationsamong implicit memory tasks (eg Roediger et al 1989)and (2) finding perceptual similarity effects in explicitmemory experiments (eg Light et al 1992)

Graf and Ryan (1990) have found that matching thefont between encoding and test facilitated memory forboth direct and indirect tests but that this effect was mod-ulated by whether the encoding task focused on perceptualfeatures (enhancing the effect) or conceptual features(minimizing the effect) It is important to note that Curranet al (1996) found that perceptual specificity effects wereunaffected by encoding task thereby questioning the TAPaccount of these perceptual similarity effects In sum thereappears to be considerable uncertainty in terms of whenand why visual specificity effects occur in some explicitmemory situations and not in others

OUR PROPOSAL

In this paper we will offer a new account of when andwhy perceptual match of features can enhance memoryperformance Unlike the PRS view our postulation is thatperceptual information is represented and processed in afashion analogous to that for conceptual informationmdashforexample they strengthen and decay at the same rate andspread activation in the same way Like the TAP frameworkour belief is that allocation of attention affects how muchinformation of a particular type is represented (eg percep-tual vs elaborative) however we posit that the perceptual

match effects described above depend on the ability to ac-cess a stored memory trace not to match encoding and re-trieval processes More specifically we propose that theextent of the benefit of a perceptual match in a memorytask depends on the relative distinctiveness of the percep-tual features It is important to operationalize distinctive-ness and that is done as part of the theoretical frameworkdescribed below

Distinctiveness in the Source of Activation Confusion Model of Memory

Before defining distinctiveness within the source of ac-tivation confusion (SAC) model it is important to explainhow recognition is assumed to operate SAC adopts thenotion adopted by a number of memory theorists (eg Ja-coby amp Dallas 1981 Mandler 1980) that recognition judg-ments can be based on either of two processes familiarityor recollection Details and formal specification of therepresentation and processing assumptions of this dualprocess theory can be found elsewhere (eg Reder Ang-stadt Cary Ayers amp Erickson 2002 Reder et al 2000Reder amp Schunn 1996 Schunn Reder NhouyvanisvongRichards amp Stroffolino 1997)

According to SAC during the study phase of an exper-iment a subject may lay down a trace encoding the fact thata particular word was studied in the experimental context(consult Figure 1) Later when asked to recognize whethera word had been studied the subject will respond affir-matively either by retrieving the trace that the word wasstudied in the experimental context or by inferring that theword was studied owing to the sense of familiarity asso-ciated with the queried term Familiarity is affected by thecurrent level of activation of the concept node Recollectionrequires that the episode node associated with the acti-vated concept receive sufficient activation from the conceptand context nodes to pass over threshold The amount of ac-tivation sent from a source depends on the competitionmdashthat is the number and strength of competing associationsfrom the sending concept

We posit that distinctiveness is inversely related to thenumber of contextual associations of the concept in ques-tion In other words concepts with fewer contextual asso-ciations are more distinctive When there are fewer com-peting contextual associations it is easier to retrieve theencoding context because more of the source activationwill flow back to the associated event node Using imple-mented computational models Reder et al (2000) demon-strated that the recognition advantage of words of lowernormative frequency can be explained as being a result oftheir fewer contextual associations

Representing the Font of the Letters That Encode the Word

Figure 1 presents a schematic illustration of how we as-sume that font is represented in a visual presentation of astudy word A given font is represented as a single conceptnode that is linked to the episode node that encodes seeing










Farm


Desk

Font Font


Court

Font







EXPERIMENT 1




















Experiment 1











EXPERIMENT 2



00

05

10

15

20

25

dcent



Experiment 1

00

05

10

15

20

25



dcent






















LegibilityHigh Fan

Hits 63 63FA NA 33d9 088 085







LegibilityHigh Fan

Hits 58 60FA NA 40d9 054 063











00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan




EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES











































































Farm


Desk

Font Font


Court

Font







EXPERIMENT 1




















Experiment 1











EXPERIMENT 2



00

05

10

15

20

25

dcent



Experiment 1

00

05

10

15

20

25



dcent






















LegibilityHigh Fan

Hits 63 63FA NA 33d9 088 085







LegibilityHigh Fan

Hits 58 60FA NA 40d9 054 063











00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan




EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES







































































EXPERIMENT 1




















Experiment 1











EXPERIMENT 2



00

05

10

15

20

25

dcent



Experiment 1

00

05

10

15

20

25



dcent






















LegibilityHigh Fan

Hits 63 63FA NA 33d9 088 085







LegibilityHigh Fan

Hits 58 60FA NA 40d9 054 063











00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan




EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES

















































































Experiment 1











EXPERIMENT 2



00

05

10

15

20

25

dcent



Experiment 1

00

05

10

15

20

25



dcent






















LegibilityHigh Fan

Hits 63 63FA NA 33d9 088 085







LegibilityHigh Fan

Hits 58 60FA NA 40d9 054 063











00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan




EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES







































































EXPERIMENT 2



00

05

10

15

20

25

dcent



Experiment 1

00

05

10

15

20

25



dcent






















LegibilityHigh Fan

Hits 63 63FA NA 33d9 088 085







LegibilityHigh Fan

Hits 58 60FA NA 40d9 054 063











00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan




EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES





















































































LegibilityHigh Fan

Hits 63 63FA NA 33d9 088 085







LegibilityHigh Fan

Hits 58 60FA NA 40d9 054 063











00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan




EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES










































































00

05

10

15

20

25

dcent

High FanLow Fan


Immediate

00

05

10

15

20

25


Delay

dcent

High FanLow Fan




EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES




































































EXPERIMENT 3


























GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES





































































GENERAL DISCUSSION





00

05

10

15

20

25

dcent

High FanLow Fan


0

2

4

6

8

Pro

port

ion

Rem

embe

r Hits


High FanLow Fan







REFERENCES







































































REFERENCES


































































perceptual match effects in direct tests of memory: the

Documents