tactile recognition m3

8/17/2019 Tactile Recognition m3

1/14

Recognition memory for tactile sequences

Paul Mahrer and Christopher Miles

Cardiff University, UK

In three experiments participants were required to compare the similarity in item order for two temporallyseparated sequences of tactile stimuli presented to the fingers of the hand. Between-sequenc e articulatory

suppression but not tactile interference impaired recognition accuracy (Experiment 1), and the null effectof tactile interference was not due to the second tactile sequence overwriting the sensory record of the firstsequence (Experiment 2). Experiment 3 showed that compared to a condition where the second sequence

was presented in the tactile modality only, recognition was enhanced when the second sequence was seen

presented either to the hand or on a diagrammatic representation of a hand. A final experiment showedthat the effects of Experiment 1 were replicated when the underside of the forearm was used for stimuluspresentation, suggesting that the data are not idiosyncratic to the first method of presentation. The pattern

of results suggests memory for a sequence of tactile stimuli involves the deployment of strategies utilisinga combination of verbal rehearsal and visuo-spatial recoding rather than relying solely on the retention of

sensory traces. This is taken to reflect limitations in both the capacity and duration of tactile sensorymemory.

INTRODUCTION

An influential account of short-term memory(STM) is provided by Baddeley and Hitch’s (1974;Baddeley, 1986) working memory model. Itsemphasis on independent slave systems controlledby a general attentional component has tended toconcentrate STM research on the modalities of vision and audition. Nonetheless, while not com-parable in depth to either the auditory or visualliterature, that concerning the tactile modality iscurrently quite substantial and can be dis-tinguished along two empirical themes. The firstof these is concerned with fundamental psycho-physiological issues and their application to therefinement of haptic communication techniquesamong the visually impaired. The other comprisesa relatively small number of studies that haveemployed established experimental paradigms inorder to determine the short-term retentioncharacteristics of tactile stimuli. The results of

these studies suggest that tactile span, the rate of forgetting of tactile material, and tactile serial

recall profiles resemble those found for the audi-tory modality in a number of respects.

Tactile span and single-stimulusstudies

An early study (Bliss, Crane, Mansfield, &Townsend, 1966) reported a tactile span of fiveitems when jets of air (varying in number from twoto twelve) were applied simultaneously to the

labelled three main joints of all eight fingers.Using a different procedure, Heller (1987) tracednumber outlines onto the palms of both hands andfound tactile span to be a function of presentationrate. A span of four items was obtained with a 1-second presentation rate, but this increased tonearly seven items with a slower, 5-second rate.Heller (1987) argued that persisting after-

MEMORY, 2002, 10 (1), 7–20

# 2002 Psychology Press Ltdhttp://www.tandf.co.uk/journals/pp/09658211.html DOI:10.1080/09658210143000128

Requests for reprints should be sent to Christopher Miles, School of Psychology, Cardiff University, PO Box 901, Cardiff CF1

3YG, UK. Email: [email protected] are due to two anonymous referees for their comments on an earlier version of this paper and to Tom Fowler for help with

data collection. This work was partially supported by an undergraduate grant to the first named author from the States of Jersey(Channel Islands, United Kingdom) Education Committee.


2/14

sensations impair performance at fast presenta-tion rates, whereas a slower rate allows any sen-sations to fade, thus increasing the perceiveddistinctiveness of successive items.

With respect to those attempts to estimate ratesof forgetting of tactile stimuli, most authors haveadopted a tactile analogue of the Brown-Petersondistractor task (Brown, 1958; Peterson & Peter-son, 1959). Typically, participants are presented,unseen, with a discrete tactile stimulus to a rangeof possible locations on their forearms and arethen required to indicate its original position aftervarious delays. A deduction in location accuracyas a function of delay suggests the decay of asensory trace from tactile STM. Using this para-digm, Gilson and Baddeley (1969) examined theeffects of both silence and articulatory suppres-

sion on the delayed recall (varying between 0 and60 seconds) of a single tactile stimulus. Althoughrecall accuracy declined as a function of delay forboth conditions, an activity by delay interactionwas found. For articulatory suppression, thedecline in accuracy was linear and reachedasymptote after 45 seconds, whereas in the silentcondition accuracy was maintained with delays of up to 15 seconds and thereafter declined withasymptote being reached at 60 seconds. Gilsonand Baddeley (1969) interpreted their data as

evidence for a dual-process model of tactilememory, involving a sensory trace immune toarticulatory suppression and a sensory tracemaintained by an abstract rehearsal process.

In contrast to these data, Sullivan and Turvey(1972) found that the decline in accuracy withinthe tactile Brown-Peterson paradigm reachedasymptote after only 5 seconds for both a silentcondition and a condition in which participantsundertook the completion of written arithmetic

tasks. Although overall performance was poorerin the arithmetic condition, there was no activityby delay interaction. Sullivan and Turvey’s (1972)data therefore suggest a tactile memory char-acterised by the decay of a single, transient,unrehearsable sensory trace.

More recently, Miles and Borthwick (1996)employed both silence and articulatory suppres-sion as filler activities within the tactile Brown-Peterson paradigm. Although both conditionsresulted in a systematic decline in performance,

with overall accuracy being poorer in the articu-latory suppression condition, no filler activity bydelay interaction was found, yet in neither condi-tion was asymptote reached even following a 20-second delay. As such, their data are consistent

with Sullivan and Turvey’s (1972) decay model(with the caveat that a tactile memorial trace maypersist for up to 20 seconds after presentation). Insummary, although all three studies support theexistence of a tactile STM, there is no firmagreement regarding either the precise rate of decay or the nature of possible rehearsal pro-cesses.

Tactile serial recall paradigms

Notwithstanding the assumption that recency andsuffix effects are purely auditory phenomena (seeCrowder & Morton, 1969), Watkins and Watkins(1974) presented eight-item tactile sequences tothe labelled fingers of both hands with each finger

being touched once. Half the sequences werefollowed by an auditory suffix and half by a tactilesuffix (a brisk stroke across all eight fingers).Regardless of whether participants recalledsequences verbally or by pointing out thesequence on a diagram, reliable recency and suffixeffects were found. Using a different procedure,Nairne and McNabb (1985) had participants lowertheir palms onto wooden blocks each of which hada different number of protruding pegs. Comparedto a visual condition in which participants merely

saw the blocks, tactile presentation resulted in apronounced recency effect with written serialrecall. More recently, Mahrer and Miles (1999)employed a variant of Watkins and Watkins’(1974) paradigm, such that rather than recallingsequences either verbally or through the use of adiagram, participants recalled by moving theirfingers in a sequence consistent with the pre-sentation sequence. ‘‘Normal’’ presentation con-ditions (silent presentation, and a 1-second

presentation rate) resulted in a strong recencyeffect that was attenuated by both a delay in recalland a same-modality suffix. However, recencyfollowing presentation conditions that impairedstimulus encoding opportunities (concurrent ver-balisation, and a fast, 0.5-second presentationrate) was resistant to both a delay in recall and atactile suffix.

Although the tactile serial recall literatureprovides additional evidence that modality andsuffix effects can arise in conditions where both

speech and visual or gestural information nor-mally associated with language is absent (see alsoCampbell & Dodd, 1980; Shand & Klima, 1981),there is little agreement among authors withrespect to the exact processes responsible for the

8 MAHRER AND MILES


3/14

recall of the final sequence item. For instance,Watkins and Watkins (1974) argue that bothrecency and suffix effects within the tactile mod-ality reflect the functioning of a modality-specificsensory memory (but see Bloom & Watkins,1999). In contrast, Nairne and McNabb (1985)suggest that the primary determinant of recency,regardless of presentation modality, is the extentto which the final item is discriminable both fromother sequence items and from the concurrentactivities of STM. In an attempt to reconcile bothsensory and discriminability theories of tactilerecency, Mahrer and Miles (1999) propose a dual-process model that can account for recall patternsfollowing both normal and difficult presentationconditions. Whatever the theoretical under-pinnings of tactile recency, both the Brown-

Peterson and serial recall paradigms are essen-tially concerned only with memory for a singletactile stimulus. Although most serial recall stu-dies report the pattern of pre-terminal recall, littletheoretical significance is placed on the retrievalof these earlier items in comparison to the finalitem. For instance, although Watkins and Watkins(1974) argue that the primacy and asymptotecomponents of a tactile serial recall curve reflectthe retrieval of verbal codes, Mahrer and Miles(1999) suggest, in contrast, that for some condi-

tions recency can be the result of spatial encoding,and that such encoding can extend to includeseveral pre-terminal items. However, in neitherthe Watkins and Watkins (1974) or the Mahrerand Miles (1999) studies are these speculationsaddressed empirically. Even span studies havelittle to say regarding the manner in which asequence of tactile stimuli is encoded. Heller(1987) emphasises the viability of ‘‘print-on-palm’’ as a method of tactile communication but

does not consider the improved span found with aslow, 5-second presentation rate to be a functionof ‘‘higher cognitive activities’’. Rather, he sug-gests it is the result of potentially confusing sen-sory traces from previous items fading during therelatively long inter-stimulus interval.

A further issue concerning span and serial recallstudies relates to the type of errors that occurduring sequence retrieval. With span studies, andNairne and McNabb’s (1985) serial recall para-digm, errors may arise from both false item recall,

and the loss of item order information; that is,correct items being recalled in their incorrect serialposition. However, for those serial recall studies inwhich tactile sequences are presented to a corre-sponding number of fingers (Mahrer & Miles,

1999; Watkins & Watkins, 1974), errors can resultonly from the loss of item order information. Thus,the current series of experiments aims to examinemore closely the memorial processes underpinningthe retention of a complete tactile sequence withparticular regard to the way in which item order ismaintained. A delayed forced-choice recognitionapproach is adopted whereby participants arerequired to estimate how many items in the secondof two tactile sequences retain their order of presentation compared to their order in the firstsequence. Using this procedure, participants arecompelled to attend to the item order within eachsequence, rather than merely judge whether thesecond sequence is the same as or different fromthe first. A 10-second delay introduced betweenthe presentation of the first and second sequence

enables us to compare the extent to which differ-ent filler activities interfere with the representa-tions of those items comprising the first sequence.

EXPERIMENT 1

The first experiment employed silence, articu-latory suppression (repeating the word ‘‘the’’),and tactile interference as filler activities. If tactilesequences are encoded verbally (Watkins &

Watkins, 1974), then the rehearsal of such codesshould be impaired by articulatory suppressionbut not by tactile interference. In contrast, if tac-tile sequences are represented by a sensory code(Heller, 1987), then such a code ought to be par-ticularly vulnerable to tactile interference.Although some fading of sensory traces is expec-ted during the 10-second delay (Heller, 1987;Sullivan & Turvey, 1972), the data of both Gilsonand Baddeley (1969) and Miles and Borthwick(1996) suggest that the tactile sensory trace will

persist for periods in excess of 10 seconds.

Method

Participants. Participants were 24 volunteerundergraduates (18 females, 6 males; mean age =23 years) from the School of Psychology, Cardiff University. All participants received course creditfor their time.

Materials. Materials comprised three blocksof 25 pairs of eight-item sequences. The firstsequence in each pair comprised a randomordering of the digits 1–8. The second sequencealso comprised the digits 1–8, however the

TACTILE RECOGNITION 9


4/14

ordering of the digits in the second sequence wasrelated to that of the first in one of five possiblecombinations or similarity gradients. The secondsequence was either identical to the first (a simi-larity gradient of eight), or it was completelyasynchronous (a similarity gradient of zero).Between these extremes, the second sequencecould share two, four, or six items with the firstsequence with respect to item order. Examples of similarity gradients are as follows:

First sequence Second sequence Similarity gradient

2 3 1 7 4 8 6 5 8 1 6 5 7 2 4 3 Zero

2 3 1 7 4 8 6 5 1 3 8 7 6 4 5 2 Two

2 3 1 7 4 8 6 5 2 8 1 7 5 4 6 3 Four

2 3 1 7 4 8 6 5 2 3 1 8 4 7 6 5 Six

2 3 1 7 4 8 6 5 2 3 1 7 4 8 6 5 Eight

For the similarity gradients of ‘‘eight’’ and‘‘zero’’, all eight digits in the second sequence wereeither in the same order as, or were asynchronousto, those of the first sequence. However, for theremaining gradients, the location of correspondingdigits in the second sequence could vary con-siderably. Thus, in the example provided here, thesecond sequence has a similarity gradient of four,as the location of its first, third, fourth, and seventh

digits correspond to those of the first sequence.During stimulus presentation, the sequence of

digits corresponds to the participants’ fingers suchthat the digit 1 corresponds to the left hand littlefinger, the digit 2 corresponded to the left handring finger, and so on through to the digit 8 thatcorresponded to the little finger of the partici-pant’s right hand (the thumb on each hand wasignored). For each block of 25 sequence pairs,there was an equal yet randomised distribution of

the five second-sequence similarity gradients. Thethree blocks of sequence pairs were rotatedaround the filler activities.

Design. The experiment employed a 2-factor(filler activity by similarity gradient) repeatedmeasures design. Participants identified the num-ber of items retaining their order of presentationin a second tactile sequence compared to the itemorder of an initial sequence. Sequences wereseparated temporally by one of three possible fil-

ler activities (silence, articulatory suppression, ortactile interference). Each filler activity wasblocked across 25 trials. The three filler activitiescombined to give six orderings, and four partici-pants were allocated to each ordering.

Procedure. Participants were tested individu-ally in a sound-proofed experimental chamber.The nature of the task was conveyed verballyand the requirements of item order were empha-sised. At all times, participants’ fingers wereheld clear of, and parallel to, the table. For eachtrial, a sequence of eight tactile stimuli (in theform of a tap from the non-writing end of a pen,and at approximately the rate of one tap per 750ms) was presented to the midphalangeal upper-side region of each finger. A 10-second intervalthen ensued, during which time an appropriatefiller activity occurred. This was followed imme-diately by the second tactile sequence that waspresented at the same rate as the first. Partici-pants were permitted to open their eyes onlybetween trials. Three different filler activities

occurred during the 10-second interval betweensequences. For the articulatory suppression con-dition, participants were cued (with a tap on thetable) to begin repeating aloud the word ‘‘the’’immediately following presentation of the firstsequence. After 10 seconds had elapsed, partici-pants were cued to cease articulation, at whichpoint the second sequence was presented. Forthe tactile interference condition, the experi-menter dragged the end of a pen back and forthacross the midphalangeal region of both the par-

ticipants’ hands for 10 seconds before presentingthe second sequence. For the silent filler condi-tion, participants between sequences experi-enced a 10-second period without tactileinterference or articulatory suppression. Afterhaving received the second sequence, partici-pants were required to report verbally the num-ber of items (zero, two, four, six, or eight) whichhad retained their order of presentation whencompared to the item order of the first sequence.

Although participants were instructed to makesimilarity judgements, they were not encouragedto engage in any specific strategy. In total, parti-cipants completed three blocks of 25 trials andwere allowed a 2-minute rest interval betweenblocks.

Results

Participants received a score of one mark for each

correct assessment (a maximum of five per simi-larity gradient); a mark of zero was awarded forany other response. Mean recognition scores foreach filler activity as a function of similarity gra-dient are shown in Figure 1.

10 MAHRER AND MILES


5/14

Recognition scores were subjected to a 2-factor(365) repeated measures Analysis of Variance

(ANOVA), with filler activity (silence, articu-latory suppression, tactile interference) and simi-larity gradient (zero, two, four, six, eight) asfactors. There was a significant main effect of filleractivity, F (2, 46) = 5.38, MSe = 1.00, p < .01;means: silence = 1.67, articulatory suppression =1.31, tactile interference = 1.69. Further analysis(Newman-Keuls; p < .05) revealed that recogni-tion was poorer following articulatory suppressioncompared to that following either silence or tactileinterference. The main effect of similarity gra-

dient was also significant, F (4, 92) = 5.96, MSe =2.26, p < .001; means: zero = 1.97, two = 1.86, four= 1.68, six = 0.88, eight = 1.40. Further analysis(Newman-Keuls) showed that recognition wassignificantly poorer when the second-sequencesimilarity gradient was six compared to when thesimilarity gradient was either eight ( p < .05), zero,two or four ( p < .01). The filler activity by simi-larity gradient interaction was non-significant (F =1.72).

Discussion

The results are unequivocal: the ability to assessthe similarity of two tactile sequences is impaired

by between-sequenc e articulatory suppression butnot by either silence or tactile interference. Thisfinding is consistent with Watkins and Watkins’(1974) suggestion that tactile sequences areencoded verbally and subsequently rehearsed.The lack of an effect of tactile interference onperformance argues against the persistence of first-sequence sensory information (Heller, 1987).Likewise, the analysis of similarity gradientreveals that neither disparate nor identical secondsequences were more easily recognised than thosesecond sequences with ambiguous similarity gra-dients of two or four.

The results of Experiment 1 thus offer littlesupport for the suggestion that sensory informa-tion is necessary for the retention of a tactilesequence: post-presentation tactile interference

did not impair performance in comparison to thecontrol condition. A plausible candidate expla-nation for the lack of a tactile interference effectin Experiment 1 is that the second stimulussequence acted to overwrite the sensory record of the first. By this account, similar levels of perfor-mance will be observed in both the control andtactile interference conditions because the secondtactile sequence exerts an equivalent retroactiveinterference effect regardless of the filler activity.Support for this explanation comes from Heller

(1980, 1987) who reported a span of three itemswhen tactile sequences were traced onto onepalm, but a span of six items when both palmswere employed. This finding was interpreted to bethe result of sensory aftersensations (or over-writing) acting to disrupt tactile perception. Hellersuggests further that such disruption can beminimised by ensuring spatial separation duringstimulus presentation. In Experiment 2, therefore,we manipulate the spatial separation of each tac-

tile sequence constituting a pair.

EXPERIMENT 2

This experiment was designed to examine theprediction that sensory traces of the first tactilesequence are overwritten by the second tactilesequence. For half the trials, the second sequencewas presented to the hand that received the firstsequence, whereas for the remaining trials the

second sequence was presented to the oppositehand. Both tactile interference and silence wereemployed as filler activities. Tactile interferencewas always presented to the hand that received thefirst sequence. For both conditions (tactile and

Figure 1. Mean recognition score for each filler activity as a

function of similarity gradient.



6/14

silence) articulatory suppression was undertakenduring presentation of the first sequence in orderto minimise reliance on verbal coding. Predictionsare clear: to the extent that the task requiresretention of the sensory representations from thefirst sequence, tactile interference betweensequences should impair recognition accuracy,regardless of whether the second sequence ispresented to the same or opposite hand to thatwhich received the first sequence. However, if thesecond sequence acts by overwriting sensoryinformation from the first sequence, then in thesilent control condition recognition will beimpaired when the second sequence is presentedto the same hand that received the first sequence.Considering the generally low performance levelsobserved in Experiment 1, and on the basis of a

brief pilot study, the number of tactile stimuli ineach sequence was reduced from eight to five toavoid floor effects in this and all subsequentexperiments.

Method

Participants. Participants were 24 volunteerundergraduates (19 females, 5 males; mean age =22 years) from the School of Psychology, Cardiff

University. All participants received payment andnone had taken part in the previous experiment.

Method. Materials comprised four blocks of 25 pairs of five-item sequences. The first sequencein each pair comprised a random ordering of digits1–5. The similarity gradients were such that thesecond sequence in each pair shared zero, one,two, three, or five digits in terms of maintainedpresentation order in comparison to the first

sequence. There was an equal and randomiseddistribution of second-sequence similarity gra-dients within each block, and the sequence blockswere rotated around the filler activities. The digitstranslated onto the participants’ right hand suchthat the digit 1 corresponded to the thumb,through to the digit 5 which corresponded to thelittle finger. For other hand trials, the digit 1 in thesecond sequence corresponded to the participants’left hand little finger, through to the digit 5 whichcorresponded to the participants’ left hand thumb.

The sequence blocks were rotated around eachfiller activity for each hand of presentation.

Design. The experiment employed a 3-factor(hand of presentation by filler activity by simi-

larity gradient) repeated measures design. Parti-cipants undertook a total of 100 trials: one block of 50 trials in which both sequences were presentedto the same (right) hand, and another block of 50trials in which the second sequence was presentedto the other (left) hand. Half the participantsreceived 50 same hand trials followed by 50 other hand trials, whereas for the remaining partici-pants, this ordering was reversed. Within each 50-trial block, both silence and tactile interferencefiller activities were presented in two counter-balanced 25-trial blocks.

Procedure. The procedure was similar to thatreported for Experiment 1 with the exception thatduring other hand trials, the second sequence waspresented to the participants’ other (left) hand

using the digit-to-finger correspondence justdescribed. Each participant was cued verbally atthe start of each trial to commence overt repeti-tion of the word ‘‘the’’ 2 seconds prior to receivingthe first sequence. After presentation of the firstsequence, the participant was cued non-verballyto cease articulation, at which point the appro-priate filler activity occurred. If the filler activityinvolved articulatory suppression then the parti-cipant was not cued to stop articulation until justprior to presentation of the second sequence.

Results

Data were scored as described for Experiment 1.Mean recognition scores for each hand of pre-sentation are shown as a function of filler activityand similarity gradient in Figures 2a and 2b.

Recognition scores were subjected to a 3-factor(26265) repeated measures ANOVA with hand

of presentation (same, other), filler activity(silence, tactile interference), and similarity gra-dient (zero, one, two, three, five) as factors. Themain effects of both hand of presentation andfiller activity were non-significant (both F s < 1).However, the main effect of similarity gradientwas significant, F (2, 92) = 5.97, MSe = 1.71, p <.001; means: zero = 1.41, one = 1.93, two = 1.66,three = 1.68, five = 2.28. Further analysis (New-man-Keuls) showed that recognition scores weresignificantly higher when the similarity gradient

was five compared to gradients of zero, two, orthree ( p < .01) and, additionally when the simi-larity gradient was one compared to zero ( p < .05).Of the main effect interactions, only that betweenfiller activity and similarity gradient was reliable,

12 MAHRER AND MILES


7/14

F (4, 92) = 3.76, MSe = 1.04, p < .01. Further ana-lysis (Newman-Keuls; p < .05) showed that for thesilent filler condition, recognition scores weresignificantly higher when the similarity gradientwas five than for any other similarity gradientregardless of filler activity. For the tactile inter-ference condition, recognition scores were sig-nificantly higher when the similarity gradient wasone compared to gradients of zero or three. Noneof the other main effect interactions, including the3-way hand of presentation by filler activity bysimilarity gradient interaction reached sig-nificance.

Discussion

The results show that limiting sensory interferenceby presenting the second tactile sequence to thehand opposite to the one that received the firstsequence does not improve performance relativeto a condition where both sequences are presentedto the same hand, thereby ruling out the over-writing hypothesis. Further, consistent withExperiment 1, there is no evidence that tactileinterference presented as a filler activity impairsrecognition performance. Although Heller (1987)has argued that performance on some tactile STM

tasks improves when sensory perception is opti-mised, the results of both this and the previousexperiment suggest that the predominant encod-ing format and representation in the currentparadigm is non-sensory in character.

An important observation with respect to theseresults is that although the opportunity for verbalencoding of the first sequence was severelyrestricted by requiring participants to undertakearticulatory suppression during its presentation,

recognition performance nevertheless exceededthe 20% chance level. It seems likely that suchperformance was achieved by the deployment of encoding strategies other than those that are pri-marily verbally based. One such possible encodingstrategy is the use of spatial imagery. Indeed, therelationship between touch and imagery has beenwell documented by several authors (see, e.g.,Heller, 1991; Warren & Rossano, 1991), and manyparticipants in Experiment 2 reported referring toa mental image of their hands when comparing

sequences. Visuo-spatial codes are immune toconcurrent verbalisation (Logie, 1986) and thuscan be generated and sustained during an articu-latory suppression condition (Carpenter &Eisenberg, 1978; Baddeley, 1993). The active use

Figure 2a. Mean recognition score for each filler activity as a

function of similarity gradient (‘‘same hand’’ trials).

Figure 2b. Mean recognition score for each filler activity as a

function of similarity gradient (‘‘other hand’’ trials).



8/14

of spatial imagery, therefore, maybe the preferredencoding strategy when opportunities for verbalencoding are particularly difficult (Experiment 2),and a subsidiary strategy when verbal encoding ispossible (Experiment 1).

The extent to which spatial imagery is relied onwithin the current paradigm can be determined byvarying the presentation medium of the secondtactile sequence. If the first stimulus sequence isencoded (at least partially) as a visuo-spatialrepresentation, then presenting the secondsequence visually should improve recognitioncompared to a condition where the secondsequence is presented in the tactile modality only.Access to the visuo-spatial scratch pad (VSSP) iseither via a process of recoding non-visual infor-mation or is obligatory and automatic for visual

stimuli (Baddeley, 1986). Visual presentation of the second sequence should therefore negate thenecessity of actively encoding it into a formatcongruent with that of the first sequence. How-ever, if the second sequence is presented in thetactile modality only, effortful and attentionalresources would need to be allocated to therecoding process, and such a diversion of resour-ces should impair the memorisation of the firstsequence.

EXPERIMENT 3

This experiment contrasted both tactile and visualsecond-sequence presentation mediums. As forExperiment 2, articulatory suppression wasundertaken during presentation of the firstsequence, and both silence and tactile interferencewere used as filler activities. For one third of thetrials the procedure followed that described for

Experiments 1 and 2, in that both sequences werepresented to the hand while participants had theireyes closed. However, for an additional third of the trials, the first sequence was presented to thehand while participants had their eyes closed,whereas the second sequence was presented on adiagrammatic representation of a hand whileparticipants had their eyes open. For the remain-ing trials, both sequences were presented to thehand, but participants were able to see the pre-sentation of the second sequence. Thus, if parti-

cipants resort to the use of spatial imagery whenverbal encoding is denied through articulatorysuppression, then recognition should be enhancedin each of the second-sequence visual presentationconditions relative to the standard presentation

condition. In addition, any reliance on residualsensory information should be apparent for thatcondition in which the second sequence is seenbeing presented to the hand. That is, a cumulative,beneficial effect of sensory and spatial codingshould be evident compared to the other twopresentation conditions in which second-sequencesensory information is absent (diagram condi-tion), or the benefits of second-sequence spatialinformation are at the cost of recoding sensoryinformation (standard condition).

Method

Participants. Participants were 24 volunteerundergraduates (16 females, 8 males; mean age =

20 years) from the School of Psychology, Cardiff University. All participants received course creditand none had taken part in the previous experi-ments.

Materials. Material comprised six blocks of 25pairs of five-item sequences constructed in thesame manner as those sequences used in Experi-ments 1 and 2. The correspondence of sequencedigits to participants’ fingers was the same asdescribed for Experiment 2. There was an equal

and randomised distribution of second-sequencesimilarity gradients within each sequence block. Inaddition to sequence blocks, a full-size diagram-matic representation of a right hand wasemployed and was drawn by tracing an adult handonto an A4 sheet of paper. The fingers of the handwere spread apart to reflect the manner in whichparticipants were required to hold their handsduring stimulus presentation.

Design. The experiment employed a 3-factor(second-sequence presentation medium by filleractivity by similarity gradient) repeated measuresdesign. Participants undertook 150 trials. Thethree second-sequence presentation mediums(hand/not seen; hand/seen; diagram/seen) wereeach blocked across 50 trials and combined to givesix presentation orderings. Four participants wereallocated to each of these presentation orderings.Within each 50-trial block, silent and tactileinterference filler activities were presented in two

counterbalanced 25-trial blocks.

Procedure. The procedure corresponded tothat of Experiment 2 for those trials where bothsequences were presented unseen and to the hand.

14 MAHRER AND MILES


9/14

For those trials involving visual presentation of the second sequence, participants were cued non-verbally with a pen tap to open their eyes at theend of the 10-second filler activity. Participantsthen saw the second sequence being presented oneither their own hand or the diagrammaticrepresentation, using a digit-to-finger correspon-dence and presentation rate equivalent to thatemployed when items were presented to the hand.The orientation of the diagrammatic hand was asfor the participants’ own hand; i.e., the fingersfaced away from the participant.

Results

Data were scored as for the previous experiments.Mean recognition scores for each filler activityand similarity gradient under each secondsequence presentation medium are shown in Fig-ures 3a-c.

Recognition scores were subjected to a 3-factor(36265) repeated measures ANOVA with sec-ond-sequence presentation medium (hand/notseen, hand/seen, diagram/seen), filler activity(silence, tactile interference), and similarity gra-dient (zero, one, two, three, five) as factors. There


function of similarity gradient (second sequence ‘‘tactile pre-

sentation, eyes closed’’ trials).


function of similarity gradient (second sequence ‘‘tactile pre-sentation, eyes open’’ trials).

Figure 3c. Mean recognition score for each filler activity as a

function of similarity gradient (second sequence ‘‘diagram-

matic presentation, eyes open’’ trials).



10/14

was a significant main effect of second-sequencepresentation medium, F (2, 46) = 5.49, MSe = 2.09,

p < .01; means: hand/not seen = 1.63, hand/seen =2.05, diagram/seen = 1.97. Further analysis(Newman-Keuls) showed that compared to thehand/not seen condition, recognition was sig-nificantly enhanced in both the hand/seen ( p <.01) and diagram/seen ( p < .05) conditions. Therewas no difference in recognition scores betweenthe hand/seen and diagram/seen conditions. Themain effect of similarity gradient was also sig-nificant, F (4, 92) = 4.13, MSe = 1.94, p < .01;means: zero = 1.65, one = 1.93, two = 1.79, three =1.78, five = 2.27. Further analysis (Newman-Keuls) revealed that performance was sig-nificantly better when the gradient was five com-pared to gradients of zero ( p < .01), one, two, or

three ( p < .05). Neither the main effect of filleractivity nor any of the main effect interactionsproved significant.

Discussion

The results of Experiment 3 indicate that seeingthe second sequence presented either to the handor on a diagrammatic representation of a handimproves recognition compared to a condition

where the second sequence is presented in thetactile modality only. Such findings suggest areliance on spatial imagery when opportunities forverbal encoding are denied. Presenting the secondsequence visually enables visuo-spatial codes to beformed and retained automatically within theVSSP (Logie, 1986). Thus, comparing a spatialrepresentation of the second sequence with aspatially encoded representation of the firstsequence avoids the necessity of overcoming an

additional recoding process when the secondsequence is presented unseen. The absence of aneffect of filler activity, and in particular the lack of a cumulative effect of visual and tactile presenta-tion (hands/seen condition), again suggests that,for this paradigm at least, tactile recognition is notbased predominantly on sensory information.

It was suggested during the review process thatthe pattern of results across these experimentsmight be particular to the paradigm adopted toexamine tactile sensory recognition memory. In

particular, it was suggested that because the fin-gers were laid out in a row in order to present thetactile sequences, this provided spatial cues to theparticipant and/or facilitated the recoding of eachtactile stimulus into an appropriately named digit.

From our perspective it seems unlikely that par-ticipants recoded each tactile stimulus verbally atencoding in Experiments 2 and 3 because articu-latory suppression was required during presenta-tion of the first sequence in both. However, inExperiment 1 the first sequence was presented insilence, thereby allowing verbal recoding of eachstimulus, and it is therefore plausible that theresults of that study are attributable to the specificparadigm employed. In order to address this pos-sibility empirically Experiment 1 was repeated,and the paradigm described by Miles and Borth-wick (1996) was adapted. In that study a singlediscreet tactile stimulus was applied, unseen, tothe underside of the forearm on each trial. Aftervarious delays the participant was required toindicate the location of the stimulus. Recall

accuracy was impaired independently by botharticulatory suppression and tactile interferencepresented during the delay period, suggesting thatthe paradigm requires both the retention of sen-sory tactile codes and the development of articu-latory codes.

EXPERIMENT 4

Any paradigm that involves the comparison of

two tactile sequences will, necessarily, require thatstimuli within a sequence be spatially separated atpresentation. It follows that both spatial andarticulatory recoding opportunities will be avail-able to the participant. In Experiment 4, there-fore, we manipulated the availability of suchrecoding opportunities by contrasting the degreeof spatial separation between successive stimuli ina sequence. Thus, using the underside of the par-ticipant’s forearm as the presentation medium, in

one condition stimuli in both sequences werepresented 0.5 inch apart (the ‘‘near’’ condition)and in the other, stimuli were presented 1 inchapart (the ‘‘far’’ condition). We predict that in the‘‘near’’ condition, due to the spatial proximity of successive stimuli, both spatial and articulatoryrecoding opportunities will be restricted. To theextent that performance of the task is thendependent on the retention of sensory tactilecodes, tactile interference presented betweensuccessive sequences will impair recognition

accuracy. In contrast, in the ‘‘far’’ condition, dueto the decrease in spatial proximity between suc-cessive stimuli, articulatory recoding of successivestimuli will be possible (see Miles & Borthwick,1996). Therefore, articulatory suppression per-

16 MAHRER AND MILES


11/14

formed between sequences should impair recog-nition accuracy in the ‘‘far’’ condition.

Method

Participants. Participants were 24 volunteerundergraduates (10 male and 14 female; mean age22 years) from the School of Psychology, Cardiff University. All participants were paid a smallhonorarium and none had taken part in the earlierexperiments.

Materials. Materials were prepared in thesame way as described for Experiment 1 with theexception that each sequence comprised the digits1–5 rather than 1–8. The digit 1 corresponded to

the location closest to the wrist and the digit 5corresponded to the location closest to the elbow.As in Experiment 1 the ordering of the digits inthe second sequence was related to that of the firstin one of five possible similarity gradients.

Design. The design of the experiment fol-lowed that described for Experiment 1 with theexception that 12 participants were assigned atrandom to the ‘‘near’’ condition and 12 to the‘‘far’’ condition. The experiment thus comprised a

3-factor (condition by filler activity by similaritygradient) design.

Procedure. The general procedure was thesame as that described for Experiment 1. Fol-lowing Miles and Borthwick (1996) a scale in 1-inch units was drawn on the underside of theparticipant’s right forearm, extending from theelbow to the wrist. The precise number of 1-inchdivisions differed between participants because of

individual differences in forearm length. For allparticipants only the five most central locationswere used. Each participant was tested individu-ally in a sound-proofed air-conditioned experi-mental chamber. The participant sat at a rightangle to the table with his or her arm lying flat onthe table at a right angle to the body, with theforearm facing uppermost and with eyes shut. Foreach trial the first sequence of tactile stimuli wasapplied at the rate of one stimulus per 750 ms.Each location was stimulated once only during a

sequence via a ballpoint pen applied briefly andgently. The second sequence was applied in thesame way. The participant’s task was to make asimilarity judgement (zero, one, two, three, five)as described for Experiments 2 and 3. In the

‘‘near’’ condition, stimulus locations were 0.5 inchapart, thus the middle 2.5 inches only of theforearm were employed. In the ‘‘far’’ condition,stimulus locations were 1 inch apart, thus the full5-inch scale was employed. In the control condi-tion, the participant sat quietly throughout the 10-second interval between sequence pairs. In thearticulatory suppression condition, the participantwas required to repeat the word ‘‘the’’ rapidly andcontinuously throughout the interval. In the tac-tile interference condition the ballpaint pen wasdragged gently back and forth over the five loca-tions on the forearm at the rate of approximatelytwo complete drags per second throughout theinterval. In total each participant completed threeblocks of 25 trials and was allowed a 2-minute restperiod between blocks.

Results

Data were scored as described for the earlierexperiments. Mean recognition scores for eachfiller activity as a function of similarity gradientfor the ‘‘near’’ and ‘‘far’’ conditions are shown inFigures 4a and 4b.

Recognition scores were subjected to a 3-factor(26365) mixed ANOVA with condition, filler

activity, and similarity gradient as factors. The


function of similarity gradient (‘‘near’’ trials).



12/14

main effects of condition, F (1, 22) = 5.0, Mse =3.65, p < .04; means: ‘‘near’’ = 1.44; ‘‘far’’ = 1.89,

and similarity gradient, F (4, 44) = 12.5, Mse = 2.07, p < .0001; means: zero = 1.2, one = 1.4, two = 1.5,three = 1.7, five = 2.5, were significant. Furtheranalysis (Newman-Keuls) of the latter effectshowed recognition accuracy at similarity gradientfive to be superior to each of the others ( p < .01).The predicted interaction between condition andfiller activity was significant, F (2, 44) = 3.8, Mse =1.22, p = .03. Further analysis (Newman-Keuls)showed recognition accuracy to be higher ( p < .01)in the ‘‘far’’ condition for both the control and

tactile interference conditions compared to the‘‘near’’ condition. Articulatory suppressionimpaired recognition in the ‘‘far’’ condition ( p <.01). Compared to the control condition, recogni-tion accuracy did not vary as a function of eithertactile interference or articulatory suppression inthe ‘‘near’’ condition. No further main effects orinteractions achieved significance.

Discussion

Manipulation of the spatial proximity of tactilestimuli within a sequence resulted in higherrecognition accuracy in the ‘‘far’’ condition.However, this effect was modified such that the

benefit existed only for the control and tactileinterference conditions: in the articulatory sup-pression condition accuracy for the ‘‘near’’ and‘‘far’’ conditions was equivalent. The result sug-gests, consistent with predictions, that participantsverbally recoded the sequences in the ‘‘far’’ con-dition. Once again thee is no direct evidencesupporting a prominent role for tactile sensorymemory: in the ‘‘near’’ condition (where oppor-tunities for both articulatory and visuo-spatialrecoding were restricted) there was no differencein recognition accuracy between the three condi-tions. Nevertheless, even tactile interference inthe ‘‘near’’ condition did not reduce performanceto the 20% chance level, suggesting perhaps aprocess of abstract rehearsal (see Gilson & Bad-deley, 1969). In these respects the data mirror

those obtained in Experiment 1, tending therebyto argue against the view that the findings for theearlier studies are paradigm-specific.

GENERAL DISCUSSION

In contrast to our understanding of tactile recencyeffects and decay rates for a single stimulus, littleis known of the memorial underpinnings for asequence of tactile stimuli. The present study

aimed to partially redress this current omission inthe literature by employing a recognition para-digm that compelled participants to estimate thesimilarity of two temporally separated tactilesequences in terms of maintained item order. Lossof item order information is a possible source of error within all experimental designs that incor-porate tactile sequences as the to-be-rememberedstimulus material. The principal findings of thefour experiments can be summarised as follows:

articulatory suppression but not tactile inter-ference impaired tactile recognition in two ver-sions of the task (Experiments 1 and 4). Theabsence of an effect of tactile interference was notdue to the second tactile sequence overwriting thesensory record of the first sequence (Experiment2). Recognition was enhanced when the secondtactile sequence was presented visually, therebyimplying that the first sequence was subject to aform of visuo-spatial encoding (Experiment 3).

The findings of the current study broadly agree

with those speculations concerning the retentionof pre-terminal items within the tactile serialrecall paradigm. Whereas Watkins and Watkins(1974) argue that the final item in a tactilesequence is afforded special memorial status


function of similarity gradient (‘‘far’’ trials).

18 MAHRER AND MILES


13/14

within a modality-specific sensory store, pre-terminal recall is seen as reflecting the retrieval of verbal labels generated through a recoding pro-cess. In contrast to this view, Mahrer and Miles(1999) argued that tactile recency can result fromspatial encoding, and that such imagery mayextend to encompass earlier sequence items.Although the present data are consistent withboth Watkins and Watkins’ (1974) and Mahrerand Miles’ (1999) positions, they are clearlyinconsistent with Heller’s (1987) suggestion thatrecall for a sequence of tactile stimuli does notreflect ‘‘higher cognitive activities’’ but instead isdependent on adequate sensory perception. Thelack of a direct effect of tactile interference(Experiments 1–4) or of hand of presentation(Experiment 2) suggests at best a limited role for

sensory information within the paradigms repor-ted here. Although the ease with which identicalsequence pairs were recognised implies a revivi-fication of sensory traces, it may well be that thecomparison of identical sequences is undemand-ing regardless of encoding format. That therecognition of tactile sequences does not rely onsensory information reinforces Watkins andWatkins’ (1974) claim that memory for tactilesensory events is limited both in capacity andduration.

In some respects, the data patterns of the pre-sent study resemble those of Keller, Cowan, andSaults (1995) who examined recognition memoryfor temporally separated pairs of auditory tones.For instance, as for the recognition of tactilesequences, recognition for auditory tones isimpaired by a verbal distractor task; althoughKeller et al. (1995) doubt that extensive verbalrecoding is employed in tone recognition. None-theless, Keller et al. (1995) report an additional

distractor condition in which a series of four tonesis presented before the first target tone but isrecalled before the second target tone. Theharmful effects of this auditory distraction onsubsequent recognition are described in terms of interference to a tone rehearsal process based on‘‘auditory imagery’’. Thus, both memory for tac-tile sequences and that for non-verbal auditorystimuli demonstrate functional parallelism: eachmay be dependent on the recoding of sensoryevents either into a verbal or imageable format

(with the caveat that auditory imagery is unlikelyto be ‘‘spatial’’ in nature). However, Keller et al.(1995) doubt that memory for tones is as accurateas that for auditory-verbal stimuli, as the formerwill not have access to echoic memory. Similarly,

the retention of tactile sequences is less faithfulthan that for a single tactile item. Whereas a dis-crete tactile stimulus may reside within a mod-ality-specific sensory store, a sequence of tactilestimuli exceeds such a store’s capacity and there-fore can be retained only through attention-demanding and vulnerable recoding processes.

Manuscript received 19 January 2000Manuscript accepted 2 March 2001

REFERENCES

Baddeley, A.D. (1986). Working memory. Oxford:Oxford University Press.

Baddeley, A.D. (1993). Human memory: Theory and practice. Hove, UK: Lawrence Erlbaum Associates

Ltd.Baddeley, A.D., & Hitch, G.J. (1974). Working mem-

ory. In G. Bower (Ed.), Attention and performanceVII . Hillsdale, NJ: Lawrence Erlbaum AssociatesInc.

Bliss, J.C., Crane, H.D., Mansfield, P.K., & Townsend,J.T. (1966). Information available in brief tactilepresentations. Perception & Psychophysics, 1, 273–283.

Bloom, L.C., & Watkins, M.J. (1999). Two-componenttheory of the suffix effect: Contrary Findings. Jour-nal of Experimental Psychology: Learning, Memory,& Cognition, 25, 1452–1474.

Brown, J. (1958). Some tests of the decay theory of immediate memory. Quarterly Journal of Experi-mental Psychology, 10, 12–21.

Campbell, R., & Dodd, B. (1980). Hearing by eye.Quarterly Journal of Experimental Psychology, 32,85–99.

Carpenter, P.A., & Eisenberg, P. (1978). Mental rota-tion and the frame of reference in blind and sightedindividuals. Perception and Psychophysics, 23, 117–124.

Crowder, R.G., & Morton, J. (1969). Precategoricalacoustic storage (PAS). Perception and Psycho-

physics, 5 , 365–373.Gilson, E.Q., & Baddeley, A.D. (1969). Tactile short-term memory. Quarterly Journal of Experimental Psychology, 21, 180–184.

Heller, M.A. (1980). Tactile retention: Reading with theskin. Perception & Psychophysics, 27 , 125–130.

Heller, M.A. (1987). Improving the tactile digit span.Bulletin of the Psychonomic Society, 25(4), 257–258.

Heller, M.A. (1991). Haptic perception in blind people.In M.A. Heller & W. Schiff (Eds.), The psychologyof touch. Hillsdale, NJ: Lawrence Erlbaum Associ-ates Inc.

Keller, T.A., Cowan, N., & Saults, J.S. (1995). Can

auditory memory for tone pitch be rehearsed? Journal of Experimental Psychology: Learning,Memory, and Cognition, 21(3), 635–645.

Logie, R.H. (1986). Visuo-spatial processes in workingmemory. Quarterly Journal of Experimental Psy-chology, 38A, 229–247.



14/14

Mahrer, P., & Miles C. (1999). Memorial and strategicdeterminants of tactile recency. Journal of Experi-mental Psychology: Learning, memory, and Cogni-tion, 25, 630–643.

Miles, C., & Borthwick, H. (1996). Tactile memoryrevisited. Memory, 4(6), 655–668.

Nairne, J.S., & McNabb, W.L. (1985). More modalityeffects in the absence of sound. Journal of Experi-

mental Psychology: Learning, Memory and Cogni-tion, 1(3), 596–604.

Peterson, L.R., & Peterson, M.J. (1959). Short-termretention of individual verbal items. Journal of Experimental Psychology, 58 , 193–198.

Shand, M.A., & Klima, E.S. (1981). Non-auditory suffixeffects in congenitally deaf signers of American SignLanguage. Journal of Experimental Psychology:Human Learning and Memory, 7 , 464–474.

Sullivan, E.V., & Turvey, M.T. (1972). Short-termretention of tactile stimulation. Quarterly Journal of Experimental Psychology, 24, 253–261.

Warren, D.H., & Rossano, M.J. (1991). Inter modality

relations: Vision and touch. In M.A. Heller & W.Schiff (Eds.), The psychology of touch. Hillsdale, NJ:Lawrence Erlbaum Associates Inc.

Watkins, M.J., & Watkins, O.C. (1974). A tactile suffixeffect. Memory and Cognition, 2, 176–180.

20 MAHRER AND MILES

tactile recognition m3

Documents