-
Journal of Experimental Psychology:Learning, Memory, and
Cognition19%, Vol. 22, No. 6,1354-1363
Copyright 19% by the American Psychological Association,
Inc.0278-7393/96/S3.00
Music-Dependent Memory:The Roles of Tempo Change and Mood
MediationWilliam R. Balch
The Pennsylvania State University, AltoonaBenjamin S. Lewis
The Pennsylvania State University, University Park
Music-dependent memory was obtained in previous literature by
changing from 1 musical piece toanother. Here, the phenomenon was
induced by changing only the tempo of the same musicalselection.
After being presented with a list of words, along with a piece of
background music,listeners recalled more words when the selection
was played at the same tempo than when it wasplayed at a different
tempo. However, no significant reduction in memory was produced by
recallcontexts with a changed timbre, a different musical
selection, or no music (Experiments 1 and 2).Tempo was found to
influence the arousal dimension of mood (Experiment 3), and recall
washigher in a mood context consistent (as compared with
inconsistent) with a given tempo(Experiment 4). The results support
the mood-mediation hypothesis of music-dependent memory.
Context-dependent memory (CDM) refers to a change incontext or
environment that causes some of the materiallearned in the original
context to be forgotten. McGeoch(1932) first formally proposed the
principle, calling it alteredstimulus conditions. The CDM effect
has been obtained withvarious context manipulations, including
place (e.g., Godden& Baddely, 1975; Smith, 1979; Smith,
Glenberg, & Bjork,1978), olfactory cues (Cann & Ross, 1989;
Schab, 1990), timeof day (Holloway, 1978); gender of speaking voice
(Geiselman& Glenny, 1977), alcohol or drug states (e.g., Eich,
1980), andmood states (e.g., Bower, Monteiro, & Gilligan, 1978;
Eich,1995b; Eich & Metcalfe, 1989; Lewis & Williams,
1989).Despite the variety and number of studies reporting
CDM,however, the reliability of these effects has been
questioned(Bjork & Richardson-Klavehn, 1989). For instance,
Fernandezand Glenberg (1985) did not find place-dependent
memory,and Bower and Mayer (1989) and Mueller, Grove, andThompson
(1991) did not report reliable mood-dependentmemory effects.
Background music has recently been found to affect memory,thus
joining the list of CDM contexts (Balch, Bowman, &Mohler, 1992;
Smith, 1985). In these studies, performed withtypical CDM
procedures, a series of words was presentedalong with a particular
selection of music. Later, recall for the
William R. Balch, Department of Psychology, The
PennsylvaniaState University, Altoona; Benjamin S. Lewis,
Department of Psychol-ogy, The Pennsylvania State University,
University Park.
We thank George Balch for assistance in the digitization
andrecording of the musical contexts used in this study; Paul
Cornwell,Kevin Murnane, Lance Shotland, Mark Lafer, and Elaine
Prestia fortheir help in securing student volunteers; John Johnson
for helpfuladvice concerning the mood induction procedures of
Experiment 4;Steven Smith, for useful suggestions regarding the
manuscript; andDave Myers, Amanda Yost, and Chris Papotto for
conducting some ofthe experimental sessions.
Correspondence concerning this article should be addressed
toWilliam R. Balch, Department of Psychology, The Pennsylvania
StateUniversity, 3000 Ivyside Park, Altoona, Pennsylvania
16601-3760.Electronic mail may be sent via Internet to
[email protected].
words was tested. Participants recalled fewer words when
adifferent piece as compared with the same piece was played.
This type of CDM (i.e., music-dependent memory) is theparticular
concern of the present research. Our central pur-pose is to explain
why music-dependent memory occurs.However, the key issues addressed
in this study are alsorelevant to CDM effects obtained by the
manipulation of othercontexts, such as mood (e.g., Eich, 1995b) or
place (e.g., Smithet al., 1978).
The first issue addressed here is the identification andtesting
of specific contextual changes that might induce music-dependent
memory. In previous studies, the CDM effect hasgenerally been
obtained with procedures involving a number ofsimultaneous changes
in context. For instance, Balch et al.(1992) and Smith (1985)
induced the effect by playing onemusical selectioncalled the
presentation or learning contextduring presentation of the words
and a different selectionduring recall. The presentation and recall
contexts thus dif-fered in the sense of their overall identities
and presumably ina multiplicity of specific musical features or
dimensions. Someof these changes include differences in the tempo,
the timbresof the instruments playing the music, the melodic
phrases, theharmonic sequences, and probably the moods induced by
themusic.
Compound differences have also been used in obtainingCDM effects
with other contexts: In place-dependent memorystudies, researchers
have used different rooms with differentfurnishings (e.g., Smith,
1979), differently furnished roomscontaining different olfactory
cues (Dalton, 1993), or differentgeographical environments such as
land versus underwater(e.g., Godden & Baddely, 1975), and in
mood-dependentmemory studies, researchers have used different
pieces ofmusic expressing different moods and accompanied by
differ-ent mood instructions (e.g., Eich, Macaulay, & Ryan,
1994;Eich & Metcalfe, 1989).1
1 Regarding the mood-dependent memory studies cited,
multiple-feature differences are involved with respect to the
procedures andstimuli used in establishing the moods, though not
necessarily withrespect to the internal mood states themselves.
1354
-
MUSIC-DEPENDENT MEMORY 1355
The use of these multiple-feature changes in CDM studies
isconsistent with the theoretical view of context as the
combinedcontributions of several sources of information (e.g.,
Bower,1972; Tulving, 1983). This view has recently been called
themental-context hypothesis (Smith, 1995). One implication of
thishypothesis is that the more contextual changes that are
made,the greater the CDM effect. In regard to this point, Eich et
al.(1994, p. 203) have suggested that simultaneous changes in
twoeffective dimensions of mood (e.g., pleasantness and arousal)may
increase mood-dependent memory, as compared with theeffect induced
by either single-dimensional change. However,they note that so far
this hypothesis has been based mainly oncorrelational evidence
(Eich & Metcalfe, 1989; Eich et al.,1994).
Although varying several aspects of context simultaneouslyis
possibly the most effective method of inducing CDM, thisprocedure
does not clarify what manipulations might besufficient causes of
CDM. One such manipulation might be achange in tempo. For instance,
Balch et al. (1992) found thatchanging to a different piece having
a different tempo loweredrecall (compared with using the same
musical selection).However, changing to a different piece having
the same tempodid not induce this effect.
Therefore, tempo may be a specific dimension closelyassociated
with music-dependent memory. However, moredirect evidence for this
hypothesis would be obtained if CDMcould be induced by an
experimentally controlled tempochange. To this end, Experiments 1
and 2 both include (a) asame-context condition, in which the same
piece is played at thesame tempo during recall and (b) a
different-tempo context, inwhich the identical selection is played
at a faster or slowertempo. A recall difference between these two
conditions wouldshow that a controlled tempo change is sufficient
to inducemusic-dependent memory. In other words,
tempo-dependentmemory would be demonstrated.
Several other contextual manipulations are also tested fortheir
effects on recall. Experiment 1 includes a different-selection
context, in which a different piece is played at the sametempo
during recall. If recall is lower in this context than in
thesame-context condition, then a change in overall
musicalcontextwithout altering tempocan induce music-depen-dent
memory. This type of change was not found to be effectiveby Balch
et al. (1992) but is tested again here.
In Experiment 2, a different-timbre condition is introduced
toisolate and test the effect of a change in timbre on
music-dependent memory. For instance, a selection originally
playedwith a piano timbre is played with a brass timbre. As in
thedifferent-tempo condition, the different-timbre context
repre-sents a controlled change in a single musical
dimension.However, tempo and timbre changes need not influence
recallin the same way. Therefore, both tempo-dependent
andtimbre-dependent memory are tested in Experiment 2.
In addition, a no-context condition is included in this
experi-ment. This type of context change is a test for what
Smith(198S) has called contextual cuing, referring to superior
recallobtained with a same-context condition as opposed to a
nullrecall context. Although most CDM studies have reportedbetter
memory for the same (as compared with a different)context, it has
been more difficult to show that recalling in the
same context serves as a facilitative recall cue and
actuallyimproves memory relative to a no-context control
(Rovee-Collier, Earley, & Stafford, 1989, p. 149; Ucros, 1989).
Forbackground-music contexts, contextual cuing was found bySmith
(1985) but not by Balch et al. (1992) or by Thaut and del'Etoile
(1993).
A second issue was motivated by our finding tempo-dependent
memory in Experiments 1 and 2. This issue relatesto the mechanisms
that might underlie the effect of tempochanges on recall. We focus
on the hypothesis that tempochanges may induce mood alterations and
that these mooddifferences may in turn induce CDM. A similar
viewpoint hasbeen expressed in connection with other CDM effects
and isknown as the mood-mediation hypothesis (Bower, 1981;
Eich,1995a; Eich & Birnbaum, 1988). This hypothesis, along
withother possible mechanisms, is tested in Experiments 3 and
4.
Experiment 1Experiment 1 compared word recall after a 1-min
retention
interval under three recall-context conditions: same
context(identical to the presentation context), different tempo
(thesame piece played at a different tempo than during
presenta-tion), and different selection (a different piece played
at thesame tempo as during presentation). If fewer words
wererecalled with the different-tempo context than with the
samecontext, this result would demonstrate tempo-dependentmemory
and indicate that changing tempo is a sufficientcondition for CDM.
If fewer words were recalled with thedifferent-selection context
than with same context, this findingwould suggest that changing a
variety of other musical features(i.e., musical selection) can
induce music-dependent memory.
MethodParticipants. Volunteers for this experiment were 168
undergradu-
ates enrolled in introductory psychology courses at the Altoona
andUniversity Park campuses of The Pennsylvania State University.
Equalnumbers of students from each campus made up each
experimentalgroup, and all participants received a small amount of
academic creditfor their services.
Materials and apparatus. The words used for all conditions of
thisexperiment were 24 common (A-frequency) two- and
three-syllablenouns taken from Spreen and Schulz's (1966) norms
(e.g., furniture,cousin, etc.). All words were originally presented
together with one oftwo selections of background music: an excerpt
(Mozart, 1953) fromthe Rondo (third movement) of a Mozart piano
sonata in C major (K.309) or "Jazz Holiday," a piano composition
(Nevin, 1957). Thesepieces are designated here as the classical and
jazz selections,respectively. We chose pieces representing two
different genresbecause we wished to use selections that were not
likely to soundsimilar to participants in the different-selection
condition. In addition,neither piece was considered likely to have
been heard frequently bymost of the participants in the
experiment.
These selections were originally played by William R. Balch on
aKorg SG-1D keyboard and digitized on a 386-SX computer
withsequencing software (Master Tracks Pro 5). For both selections,
apiano timbre was simulated by means of a Roland U-220 SoundModule
(timbre program 001 on that unit). Each piece was played in Cmajor
(requiring no transcription of the classical [C] piece butrequiring
the jazz [J] piece to be transcribed one whole tone higher by
-
1356 BALCH AND LEWIS
computer). Two audiocassette recordings were made from
eachselection, one at a slow (S) tempo (60 quarter notes per
minute) andthe other at a fast (F) tempo (140 quarter notes per
minute). Allrecordings had a duration slightly longer than the time
required forword presentation (250 s). If the end of a selection
occurred duringrecording, the computer continued playing from the
beginning.
Thus, four recordings were used in this experiment: SC, FC, SJ,
andFJ. In comparing the fast and slow versions of either the
classical (SCvs. FC) or the jazz (SJ vs. FJ) selection, note that
each version wasidentical except for the tempo.
Design. The first two independent variables were selection
(classi-cal [C] or jazz [J]) and tempo (slow [S] or fast [F]).
Together, thesevariables generated the four possible presentation
contexts: SC, FC,SJ, or FJ. In other words, each presentation
context was based oneither the classical or the jazz selection,
played at either the slow or thefast tempo (60 or 140 quarter notes
per minute, respectively).
The third independent variable was recall context. This
variableconsisted of three conditions: same context, different
tempo, ordifferent selection. For the same-context condition, the
same record-ing was used for both the presentation and recall
contexts. For thedifferent-tempo condition, the recordings used for
these contexts werebased on the same selection but different
tempos. For the different-selection condition, on the other hand,
presentation and recallcontexts were based on a different piece
played at the same tempo.
Fourteen different participants were randomly assigned to each
ofthe 12 experimental groups. In terms of presentation and
recallcontexts, the four same-context groups can be designated as
SC-SC,FC-FC, SJ-SJ, and FJ-FJ. The four different-tempo groups
wereSC-FC, FC-SC, SJ-FJ, and FJ-SJ. Finally, the four
different-selectiongroups were SC-SJ, FC-FJ, SJ-SC, and FJ-FC.
The dependent variable was the number of words (out of
24)correctly recalled during the recall phase of the
experiment.
Procedure. An individual session was conducted with each
partici-pant, who was told that the purpose of the experiment was
to ratewords for pleasantness. Participants were then given a
pencil and abooklet for rating the words on a 6-point scale: 1 for
very unpleasant, 2for moderately unpleasant, 3 for slightly
unpleasant, 4 for slightlypleasant, 5 for moderately pleasant, and
6 for very pleasant. To make therating task more enjoyable,
participants were told, background musicwould be played while the
words were shown.
Typed words were presented visually on index cards, 1 word
percard. To help give the participants sufficient exposure to the
material,two different random orders of the 24-word list were run
consecutivelyto generate a complete 48-word sequence. During the
instructions,participants had been informed that each word would be
repeatedsomewhere in the sequence and that they should rate each
presentedword according to their impression of its pleasantness at
the moment.One of two different 48-word sequences was assigned to
half theparticipants in each group.
After the instructions, the experimenter started the recording
of theassigned musical context. After 10 s, he or she began showing
words atthe rate of one every S s. This presentation phase lasted
about 250 s.The musical recording was stopped at the end of
presentation, and theparticipant's rating booklet was taken.
Next, a piece of "distraction music" was played: a
techniqueadopted from Balch et al.'s (1992) study (their Experiment
3). Thisprocedure was intended to help equate the otherwise
differentdistraction levels that might be involved in the various
recall-contextconditions. Immediately changing to a different
context (as in thedifferent-tempo and different-selection
conditions) might have startledor distracted participants more than
playing the identical recordingagain (as in the same-context
condition). This extra distraction mighthave disrupted attention
and impaired recall performance in a manneruninteresting from a
memory standpoint. An intentionally distractingpiece inserted
between presentation and recall, however, should have
helped eliminate such differences in the degrees of distraction
pro-duced by the recall contexts.
The distraction piece was an excerpt from Shirabe-Sagaribe
(1980).(The title is translated, "The Sound of Wind Through
BambooLeaves.") This essentially atonal music was played on several
orientalinstruments, including a bamboo flute, and was chosen to be
distract-ingly different from both the classical and jazz
contexts.
After 30 s of distraction music, the recall-context recording
wasstarted. Participants were handed a piece of paper and asked to
writedown, in any order, as many of the presented words as they
couldrecall. The interval between the end of presentation and the
beginningof recall was about 1 min. Two minutes were allowed for
the recallsession itself, and then participants were debriefed.
Results and DiscussionAn alpha level of .05 was used for the
statistical tests
reported in all the present experiments.The results of
Experiment 1 are illustrated in Table 1, which
gives the mean recall scores and standard deviations for
thevarious conditions. The table shows the scores for the
threerecall contexts: the same context, the different-tempo
context,and the different-selection context. Average recall across
thepresentation conditions was highest for the same context(12.38),
lowest for the different tempo (10.98), and intermedi-ate for the
different selection (11.75). This outcome appears tobe consistent
with the hypothesis that a tempo change per se issufficient to
induce music-dependent memory.
To check the statistical significance of these results, 2 x 2 x3
(Selection x Tempo x Recall Context) independent-groupsanalysis of
variance (ANOVA) was performed. As expected,the main effect of
recall context was significant, F(2, 156) =6.83, MSE = 3.99, p <
.01. Neither of the other main effectsand none of the interactions
were significant (ps > .10).
The music-dependent memory effects can be assessed bymultiple
Bonferroni comparisons between the same-contextmean and each of the
other recall-context means. The 1.40
Table 1Recall Scores for the Contexts in Experiment 1
Presentationcontext"
SlowMSD
FastMSD
SlowMSD
FastMSD
Samecontext
Recall contextDifferent
tempoClassical selection
12.501.95
13.071.69
11.211.72
10.351.45
Jazz selection
12.352.06
11.571.60
11.211.72
11.072.13
Differentselection
11.641.60
12.791.67
11.861.84
10.712.76
Note. Each participant's recall score was the number of
correctwords out of 24; n = 14 for each entry.aAll presentation
contexts were played with the piano timbre.
-
MUSIC-DEPENDENT MEMORY 1357
difference in recall between the same-context (12.38)
anddifferent-tempo conditions (10.98) was significant, '(165)
=3.73, p < .002. However, the 0.63 difference between recall
inthe former-context and different-selection conditions (11.75)was
not significant, f (165) = 1.73,p > .10.
Thus, changing tempo induced music-dependent memory.However,
changing to a different selection played at the sametempo did not.
Tempo, in particular, appears to be a dimen-sion of musical context
that has significant memory conse-quences. However, because
Experiment 1 was designed to testthe individual effects of changing
tempo or selection, nostatistical inference is made concerning the
comparison be-tween the tempo-dependent and selection-dependent
effects.
Experiment 2In Experiment 2, word recall in the same-context
condition
of Experiment 1 was compared with recall under each of
threechanged contexts. First, the different-tempo context, whichwas
found to reduce recall in Experiment 1, was tested again.This time,
however, a second controlled and single-dimen-sional change was
also tested: different timbre (in which onlytimbre was altered). To
test for contextual cuing, a furthercontext manipulation was
introduced: a no-context condition(in which no musical selection
was played during recall).
A significant drop in recall under the different-tempocontext
would replicate the finding obtained in Experiment 1that changing
tempo is a sufficient cause for music-dependentmemory. Decreased
recall under the different-timbre contextwould show an analogous
effect for a timbre change. Finally, amemory decrement under the
no-context condition woulddemonstrate contextual cuing.
MethodParticipants. Participants were 128 undergraduates who
took part
in this experiment for extra credit. As in Experiment 1, half of
theparticipants in each condition were from the Altoona campus of
ThePennsylvania State University, and half were from the University
Parkcampus.
Materials and apparatus. The same word list, musical
selections,and tempos used in Experiment 1 were applied again in
Experiment 2.Likewise, the same four recorded contexts used in the
previousexperimentall in the piano timbrewere used here. These
contextswere designated SC-P, SJ-P, and so forth (P indicates the
pianotimbre). An additional set of four comparable contexts was
recorded ina brass timbre created by the Roland U-220 Sound Module
(timbreprogram 042 on that unit): SC-B, SJ-B, and so forth (B
indicates thebrass timbre).
Design. Because of the introduction of the brass timbre,
thepresentation-context variables were expanded from two in
Experiment1 to three in Experiment 2: selection (classical [C] or
jazz [J]), tempo(slow [S] or fast [F]), and timbre (piano [P] or
brass [B]). Thus, eightpresentation contexts were generated: SC-P,
SC-B, FC-P, FC-B,SJ-P, SJ-B, FJ-P, and FJ-B.
The recall-context variable included two contexts from
Experiment1 (same context and different tempo) and two additional
contexts(different timbre and no context). These conditions were
explained inthe introduction to Experiment 2 and are briefly
illustrated now. Thepresentation-context/recall-context sequences
were as follows: for theeight same-context groups: SC-P/SC-P,
SC-B/SC-B, and so forth; forthe different-tempo context: SC-P/FC-P,
SC-B/FC-B, and so forth;
for the different-timbre context: SC-P/SC-B, SC-B/SC-P, and
soforth; and for the no-context condition: SC-P/N, SC-B/N, and
soforth (N designates no musical selection during recall).
Four different participants were randomly assigned to each of
the 32combinations of the eight presentation contexts and four
recallcontexts.
Procedure. All of the same procedures used for conducting
experi-mental sessions in Experiment 1 were applied in Experiment
2. In theno-context condition, though, no selection was played
during recall.The retention interval for all conditions was again 1
min, and the samedistraction music was used.
Results and DiscussionA preliminary 2 x 2 x 2 x 4 (Selection x
Tempo x
Timbre x Recall Context) independent-groups ANOVA re-vealed that
musical selection was not significant as a maineffect, nor was it
included in any significant interactions withthe other variables
(ps > .10). Therefore, the results werecollapsed across musical
selections.
Table 2 gives the mean recall scores and standard deviationsfor
the conditions in Experiment 2. The table shows the scoresfor the
four recall contexts: the same-context, the different-tempo
context, the different-timbre context, and the no-context
conditions. Average recall across the presentationconditions was
higher for the same context (12.06) than for thedifferent-tempo
context (10.19). These scores appear to bereasonably similar to the
comparable numbers found in Experi-ment 1 (12.38 and 10.98,
respectively). Again, changing onlytempo seems to be sufficient to
induce music-dependentmemory. However, average recall in the
different-timbre(12.59) and no-context conditions (12.53) was about
the sameas that obtained for the same context (12.06). Therefore,
thereseems to be no evidence for the timbre-dependent memory
orcontextual-cuing effects.
To test the statistical significance of the results, a 2 x 2 x
4(Tempo x Timbre x Recall Context) independent-groupsANOVA was
performed. As expected, the main effect of recallcontext was
significant, F(3, 112) = 10.49, MSE = 3.89, p < .01. However,
recall in theactive mood was significantly better for the fast
(10.75) than forthe slow (9.08) presentation tempo, F(l, 22) =
7.45, MSE =2.24,p < .05. (See Table 4.)
This pattern of results is consistent with the hypothesis
thattempo-dependent memory is mediated by mood. Apparently,
Table 4Recall Scores for the Tempo and Mood Contexts in
Experiment 4
MoodRelaxed
MSD
ActiveMSD
Slow
12.411.989.081.08
TempoFast
9.332.61
10.751.82
Note. Words were presented in a musical-tempo context and
recalledin a mood context; n = 12 for each condition.
-
MUSIC-DEPENDENT MEMORY 1361participants recall more when they
are put into a mood thatmatches, rather than mismatches, the mood
induced by thetempo of the musical presentation context. This
experimentshows that the recall context need not be musical, as
inExperiments 1 and 2. Thus, what underlies the CDM effectfound in
Experiment 4 is mood, rather than musical variablessuch as phrase
repetition or beats per minute. Note that thelatter was specified
as the operative variable in the temporal-frame hypothesis.
General DiscussionConsidered together, the present experiments
support the
mood-mediation hypothesis of music-dependent memory.
Ex-periments 1 and 2 show that changing tempo, but not
musicalselection (Experiment 1) or timbre (Experiment 2),
signifi-cantly reduces recall. In Experiment 3, tempobut again,
notselection or timbrewas found to influence the arousaldimension
of mood. Experiment 4 linked tempo and moodtogether in terms of
memory. Tempo was manipulated in themusical presentation context;
however, in recall, a moodcontext was manipulated strictly by
verbal means. Duringrecall, more was remembered in mood contexts
that matchedthe moods of the original presentation tempos.
Therefore, thisCDM effect was mediated by the one presumably
commonproperty of the matching presentation and recall
contexts:mood arousal.
In explaining CDM effects, an alternative to the mood-mediation
view is the mental-context hypothesis. According toSmith (1995),
this latter hypothesis considers "mood, place,mental set and other
factors as components of one's mentalcontext, any of which can
serve to cue the representation ofmental context at test" (p. 309).
According to this view, anyone or a combination of many specific
contextual changescould be a sufficient cause of CDM.
The tempo manipulation, tested in Experiments 1 and
2,significantly influenced recall. Thus, tempo change is clearly
asufficient cause of music-dependent memory. However, we didnot
induce CDM with the other contextual changes. Thus, ourresults
provide no evidence that music-dependent memory canhave multiple
causes, an idea expressed by the mental-contexthypothesis.
We acknowledge that there may be causes of this CDMeffectother
than tempothat we simply have not found.Even variations of the
timbre or the selection changes that wetested here might have the
potential to induce music-dependent memory. Though the brass and
piano timbressounded quite different to us, for instance, there may
be othercontrasting timbres that would have induced stronger
CDM.Moreover, the two tempos used here were purposely chosen tobe
very distinct (60 vs. 140 beats per minute), and the
contrastsinvolved in our other musical variables may not have
beenfunctionally equivalent in every relevant way. The 60
beats-per-minute pace may also have rendered the slow-tempo
contextsless discriminable from each other, compared with the
greaterdistinctiveness of each fast context. Note that the fast
tempo iscloser to those tempos originally intended for the jazz
andclassical compositions used here.
However, we can cite one type of functional equivalence. In
Experiment 3, all of the musical contexts were rated about
thesame in the mood pleasantness they induced.
Pleasantnessrepresents one type of encoding that could be relevant
to thepresent memory task, especially because the words to
beremembered were rated for pleasantness during presentation.Yet
there were no obvious differences in the effects of tempo,timbre,
or selection on the mood-pleasantness ratings of themusical
contexts.
Still, we acknowledge that there may be changes in
timbre,selection, or other musical characteristics that would be
morecomparable with the tempo manipulation we used. Therefore,we
conclude only that the tempo change used here is a singleand
sufficient manipulation that induces music-dependentmemory. We do
not claim that this manipulation is necessaryto induce the effect.
By the same token, we do not dismiss thenotion that several
simultaneous contextual changes mightinduce greater CDM than any
single change (Eich et al., 1994,p. 203). More research in the
various CDM areas needs to beaddressed to systematic changes in
context, both single andmultiple.
Though tempo change may be an external cause of music-dependent
memory, Experiments 3 and 4 show that this effectseems to be best
explained in terms of an internal mediator:mood arousal. In
addition, Experiment 4 helps resolve an issueraised by Smith (1995)
concerning the mood-mediation hypoth-esis. He points out an
ambiguity in Eich's (1995a) research onthe mediation of
place-dependent memory by the happy-saddimension of mood (Eich's
Experiment 3). Because Eich usedonly one presentation context (a
pleasant place), recall in themismatched-mood context might have
been worse because ofthe sad-mood procedure rather than because of
the mis-matched mood (Smith, 1995, p. 309). However, in the
presentExperiment 4, presentation tempos and recall moods
werecounterbalanced. Thus, the CDM we found appears to arisefrom
matched or mismatched moods rather than from theparticular recall
contexts used.
In light of the above consideration, our findings clarify
andextend Eich's (1995a) support of the mood-mediation hypoth-esis.
Note that Eich was concerned with place-dependentmemory and an
evaluative dimension of mood (i.e., pleasant-happy vs.
unpleasant-sad). Our own experiments have beenfocused on tempo
change and the arousal dimension of mood.If mood mediation proves
to be the key mechanism for avariety of CDM effects, it seems
plausible that the particularmood dimension underlying any given
effect will depend on thetype of CDM in question.
The present findings also have some implications concerningthe
question of reliability, which has plagued most CDMeffects (e.g.,
Bjork & Richardson-Klavehn, 1989). For in-stance, Eich (1995a)
has summarized the phenomena ofplace-dependent and mood-dependent
memory as having had"mercurial histories, with the mostly positive
results reportedin the 1970s giving way to mostly negative results
in the 1980s,leading theorists in the 1990s to wonder whether
either PDM[place-dependent memory] or MDM [mood-dependentmemory]
even exists" (p. 305).
The mood-mediation view, which is supported by the pre-sent
research, helps explain such reliability problems. Forinstance,
several studies have not found place-dependent
-
1362 BALCH AND LEWIS
memory (e.g., Fernandez & Glenberg, 1985). However, in
anearly study by Godden and Baddely (1975), a 46% difference
inrecall between same- and different-place contexts was ob-tained
with scuba divers as participants. This large effect maywell have
been due to the choice of contexts: on land orunderwater. Because
these two locations were likely to haveevoked very different moods,
place-dependent memory couldhave been mediated by this difference.
(See Eich, 1995a, for amore detailed discussion of this point.)
Certain reliability problems have also surfaced in
music-dependent memory research. For instance, contextual
cuing(i.e., a recall difference between same-music and
no-musiccontexts) was found by Smith (1985). However, no
contextualcuing for background music was found in the present
Experi-ment 2 or in the studies by Balch et al. (1992) and Thaut
and del'Etoile (1993).
Mood mediation may explain why contextual cuing has
beendifficult to demonstrate with background music. Although
adifferent-music condition could disrupt mood and induceCDM, the
no-music recall context should not prevent partici-pants from
mentally reinstating the mood they had experi-enced during the
presentation music. Therefore, this nullcontextunlike a
different-music contextshould not inter-fere with recall. In other
words, contextual cuing would not beexpected. This interpretation
seems plausible in light of recenttheoretical discussions of
spontaneous context reinstatement(e.g., Bjork &
Richardson-Klavehn, 1989).
Reliability has generally not been a problem for music-dependent
memory induced by different-music contexts. Thiseffect, obtained in
the present Experiments 1 and 2, reinforcessimilar effects found by
both Smith (1985) and Balch et al.(1992). Different-music contexts
have apparently not inducedCDM only when the tempo was purposely
kept the same asthat of the presentation music, as in the following
cases: thedifferent-selection context of Experiment 1, the
different-timbre context of Experiment 2, and the same-tempo
contextin Balch et al.'s (1992) study in which a different
selection wasplayed at the original presentation tempo (their
Experiment2). The preceding observations are consistent with the
viewthat tempo change and mood mediation figure importantly
inmusic-dependent memory.
To what degree is the present research congruent witheveryday
examples of music-dependent memory? Our resultsseem generally
consistent with several aspects of practicalmusical experience. For
instance, consider the familiar notionthat old songs brings back
associated memories. In this case,musical context appears to have
memory consequences thatoccur in everyday life as well as in
experimental studies. Insongs, music can also influence the
recognition of lyrics(Serafine, Crowder, & Repp, 1984;
Serafine, Davidson, Crow-der, & Repp, 1986), illustrating a
related type of memoryconsequence.
One aspect of musical experience is more difficult to explainin
terms of experimental research. In some naturally occurringcases,
music-dependent memory may involve the kind ofcontextual cuing not
found in the present Experiment 2 or inmost CDM studies. As already
mentioned, decreases inmemory due to an altered context have been
easier todemonstrate than facilitation due to the same context.
Yet
sometimes a few bars of an old melody appear to cue memoriesthat
might have been difficult to retrieve otherwise. If so,several
influences might contribute to this phenomenon: manypairings of the
musical context and the associated material tobe remembered, for
instance, or the personal significance ofthe music and material to
the individual involved. The range ofconditions possible in natural
occurrences of CDM far exceedseven the broad variety now
encountered in laboratory studies.
ReferencesBalch, W. R., Bowman, K., & Mohler, L. A. (1992).
Music-dependent
memory in immediate and delayed recall. Memory & Cognition,
20,21-28.
Bjork, R. A., & Richardson-Klavehn, A. (1989). On the
puzzlingrelationship between environmental context and human
memory. InC. Izawa (Ed.), Current issues in cognitive processes
(pp. 313-344).Hillsdale, NJ: Erlbaum.
Bower, G. H. (1972). Stimulus-sampling theory of encoding
variability.In A. W. Melton & E. Martin (Eds.), Coding
processes in humanmemory (pp. 85-124). Washington, DC: Winston.
Bower, G. H. (1981). Mood and memory. American Psychologist,
36,129-148.
Bower, G. H., & Mayer, J. D. (1989). In search of
mood-dependentretrieval. Journal of Social Behavior and
Personality, 4, 133-168.
Bower, G. H., Monteiro, K. P., & Gilligan, S. G. (1978).
Emotionalmood as a context for learning and recall. Journal of
Verbal Learningand Verbal Behavior, 17, 408-417.
Cann, A., & Ross, D. A. (1989). Olfactory stimuli as context
cues inhuman memory. American Journal of Psychology, 102,
91-102.
Dalton, P. (1993). The role of stimulus familiarity in
context-dependent recognition. Memory & Cognition, 21,
223-234.
Eich, E. (1980). The cue-dependent nature of state-dependent
re-trieval. Memory & Cognition, 8, 157-173.
Eich, E. (1995a). Mood as a mediator of place dependent
memory.Journal of Experimental Psychology: General, 124,
293-308.
Eich, E. (1995b). Searching for mood dependent memory.
Psychologi-cal Science, 6, 67-75.
Eich, E., & Birnbaum, I. M. (1988). On the relationship
between thedissociative and evaluative properties of drugs. In G.
M. Davies &D. M. Thomson (Eds.), Memory in context: Context in
memory (pp.81-93). Chichester, England: Wiley.
Eich, E., Macaulay, D., & Ryan, L. (1994). Mood dependent
memoryfor events of the personal past. Journal of Experimental
Psychology:General, 123, 201-215.
Eich, E., & Metcalfe, J. (1989). Mood-dependent memory for
internalvs. external events. Journal of Experimental Psychology:
Learning,Memory, and Cognition, 15, 433-455.
Fernandez, A., & Glenberg, A. M. (1985). Changing
environmentalcontext does not reliably affect memory. Memory &
Cognition, 13,333-345.
Geiselman, R. E., & Glenny, J. (1977). Effects of imagining
speakers'voices on the retention of words presented visually.
Memory &Cognition, 5, 499-504.
Godden, D. R., & Baddely, A. D. (1975). Context-dependent
memoryin two natural environments: On land or underwater. British
Journalof Psychology, 66, 325-331.
Holloway, F. A. (1978). State-dependent retrieval based on time
ofday. In B. T. Ho, D. W. Richards III, & D. L. Chute (Eds.),
Drugdiscrimination and state-dependent learning (pp. 319-343). New
York:Academic Press.
Lewis, V. E., & Williams, R. N. (1989) Mood-congruent vs.
mood-statedependent learning: Implications for a view of emotion.
Journal ofSocial Behavior and Personality, 4, 157-171.
-
MUSIC-DEPENDENT MEMORY 1363
McGeoch, J. A. (1932). Forgetting and the law of disuse.
PsychologicalReview, 39, 352-370.
Mozart, W. A. (1953). Rondo [Sheet Music]. Secaucus, NJ:
WarnerBrothers Publications.
Mueller, J. H., Grove, T. R., & Thompson, W. B. (1991).
Mood-dependent retrieval and mood awareness. Cognition and Emotion,
5,331-349.
Nevin, M. (1957). Jazz holiday [Sheet Music]. Paramount,
CA:Schroeder and Gunther.
Rovee-Collier, C , Earley, L., & Stafford, S. (1989).
Ontogeny of earlyevent memory: III. Attentional determinants of
retrieval at 2 and 3months. Infant Behavior & Development, 12,
147-161.
Schab, F. R. (1990). Odors and the remembrance of things
past.Journal of Experimental Psychology: Learning, Memory, and
Cognition,16, 648-655.
Serafine, M. L., Crowder, R. G., & Repp, B. H. (1984).
Integration ofmelody and text in memory for song. Cognition, 16,
285-303.
Serafine, M. L., Davidson, J., Crowder, R. G., & Repp, B. H.
(1986).On the nature of melody-text integration in memory for
songs.Journal of Memory & Language, 25, 123-135.
Shirabe-Sagaribe. (1980). On Japan Kabuki and other traditional
music[Cassette]. New York: Elektra/Asylum/Nonesuch Records.
Smith, S. M. (1979). Remembering in and out of context. Journal
ofExperimental Psychology: Human Learning and Memory, 5,
460-471.
Smith, S. M. (1985). Background music and context-dependent
memory.American Journal of Psychology, 98, 591-603.
Smith, S. M. (1995). Mood is a component of mental
context:Comment on Eich (1995). Journal of Experimental
Psychology:General, 124, 309-310.
Smith, S. M., Glenberg, A., & Bjork, R. A. (1978).
Environmentalcontext and human memory. Memory & Cognition, 6,
342-353.
Spreen, O., & Schulz, R. W. (1966). Parameters of
abstraction,meaningfulness, and pronunciability for 329 nouns.
Journal of VerbalLearning & Verbal Behavior, 5, 459-468.
Thaut, M. H., & de l'Etoile, S. K. (1993). The effects of
music on moodstate-dependent recall. Journal of Musk Therapy, 30,
70-80.
Tulving, E. (1983). Elements of episodic memory. Oxford,
England:Clarendon Press.
Ucros, C. G. (1989). Mood state-dependent memory: A
meta-anarysis.Cognition and Emotion, 3, 139-167.
Received July 14,1995Revision received February 21,1996
Accepted February 28,1996
Low Publication Prices for APA Members and AffiliatesKeeping you
up-to-date. All APA Fellows, Members, Associates, and Student
Affiliatesreceiveas part of their annual duessubscriptions to
theAmerican Psychologist aadAPAMonitor. High School Teacher and
International Affiliates receive subscriptions to the.i4.R4Monitor,
and they may subscribe to the American Psychologist at a
significantly reducedrate. In addition, all Members and Student
Affiliates are eligible for savings of up to 60%(plus a journal
credit) on all other APA journals, as well as significant discounts
onsubscriptions from cooperating societies and publishers (e.g.,
the American Association forCounseling and Development, Academic
Press, and Human Sciences Press).
Essential resources. APA members and affiliates receive special
rates for purchases ofAPA books, including the/>aZ)/;caftoMiwa/
ofthe American PsychologicalAssociation,and on dozens of new
topical books each year.
Other benefits of membership. Membership in APA also provides
eligibility forcompetitive insuranceplans, continuing education
programs, reduced APAconventionfees,and specialty divisions.
More information. Write to American Psychological Association,
Membership Services,750 First Street, NE, Washington, DC
20002-4242.