366 Copyright 1999 American Psychological Society VOL. 10, NO.

4, JULY 1999
PSYCHOLOGICAL SCIENCE
Research Report
THE MYSTERY OF THE MOZART EFFECT:
Failure to Replicate
Kenneth M. Steele, Karen E. Bass, and Melissa D. Crook
Appalachian State University
AbstractThe Mozart effect is the purported increase in spatial-
reasoning performance immediately after exposure to a Mozart piano
sonata. Several laboratories have been unable to confirm the existence
of the effect despite two positive reports from the original laboratory.
The authors of the original studies have provided a list of key proce-
dural components to produce the effect. This experiment attempted to
produce a Mozart effect by following those procedural instructions
and replicating the procedure of one of the original positive reports.
The experiment failed to produce either a statistically significant
Mozart effect or an effect size suggesting practical significance. This
general lack of effect is consistent with previous work by other inves-
tigators. We conclude that there is little evidence to support basing
intellectual intervention programs on the existence of the Mozart
effect.
Rauscher, Shaw, and Ky (1993) reported that 36 undergraduates
increased their mean spatial-reasoning scores the equivalent of 8 to 9
IQ points on portions of the Stanford-Binet Intelligence Scale: Fourth
Edition (Thorndike, Hagen, & Sattler, 1986) after listening to 10 min
of Mozarts Sonata for Two Pianos in D Major, K. 448. (Hereafter, we
refer to this effect as the Mozart effect.) The Mozart effect was tem-
porary, having disappeared within 10 to 15 min. Subsequently,
Rauscher, Shaw, and Ky (1995) reported a replication of the Mozart
effect, using elaborations of the Stanford-Binet Paper Folding and
Cutting (S-B PF&C) task as the dependent measure.
The hypothesis that musical experiences of short duration can have
a direct causal influence on spatial reasoning on both a short-term and
a long-term basis (Rauscher, 1997) is important for both practical and
theoretical reasons. However, several attempts by other laboratories to
confirm the existence of a Mozart effect have been unsuccessful
(Carstens, Huskins, & Hounshell, 1995; Dalla Bella, Dunlop, Dawe,
Humphrey, & Peretz, 1999; Kenealy & Monsef, 1994; Newman et al.,
1995; Steele, Ball, & Runk, 1997; Stough, Kerkin, Bates, & Mangan,
1994; Weeks, 1996).
Rauscher and Shaw (1998) reviewed some of the negative results
and described key components necessary to produce a Mozart effect.
Negative results in some studies were explained by the choice of an
inappropriate dependent measure, and the use of a PF&C task was
endorsed. Rauscher and Shaw warned investigators to attend to issues
concerning the order of presentation of the listening and task condi-
tions. A major concern was that a pretest immediately before the treat-
ment may produce a carryover effect that obscures enhancement by
the music.
The purpose of this experiment was to confirm the existence of the
Mozart effect by following the advice of Rauscher and Shaw (1998).
The experiment was designed to be a faithful replication of the central
conditions of the Rauscher et al. (1995) experiment. Rauscher et al.
tested the effect of exposure to Mozarts Sonata for Two Pianos (K.
448) against silence and against highly repetitive music (Philip Glasss
Music With Changing Parts) in a mixed-groups design. We chose to
replicate the 1995 experiment because more procedural details were
available for that experiment than for the earlier experiment, because
the overall magnitude of the Mozart effect was stronger in the 1995
experiment, and because the procedure was consistent with the rec-
ommendations of Rauscher and Shaw (1998).
METHOD
Participants
Rauscher et al. (1995) employed 79 participants, distributed among
three conditions. This study involved 125 participants (42 male and 83
female) distributed among three conditions. Participants came from
introductory psychology courses and received credit for participation.
Materials and Apparatus
Rauscher et al. (1995) used several sets of 16 PF&C items that
were derived from the Stanford-Binet IQ measure. (The actual test
contains 2 practice items and 18 test items.) We used two sets of 16
PF&C items; each set was composed approximately of half derived
items and half true items from the Stanford-Binet measure. The PF&C
sets were of approximately equal difficulty and had been developed in
another laboratory (Rideout & Laubach, 1996).
1
Each PF&C item was
adjusted in size to occupy the center portion of an overhead trans-
parency measuring 21.5 by 28 cm. Figure 1 shows an example PF&C
item.
Stimulus tapes were created from the CD performances used in
Rauscher et al. (1995). Mozarts Sonata for Two Pianos, K. 448
(Mozart, 1985, track 1) is a lively three-movement piece. Although
invariably described as 10 min in duration, the first section (Allegro
con spirito) is actually 8 min, 24 s in duration in this performance.
Philip Glasss (1973, track 1) Music With Changing Parts is more than
60 min in duration without break; the first 8 min, 24 s was recorded
for use in the experiment. The Glass composition eschews a tradition-
al melody and uses repetition of units. The units are often only a few
seconds long, and the repetition often lasts for several minutes.
Exposure to music is an established mood-induction technique, and
musical selections by Mozart have been used to induce a mood of ela-
tion (Kenealy, 1997; Westerman, Spies, Stahl, & Hesse, 1996). It has
been suggested that the performance difference interpreted as evidence
of a Mozart effect could be produced indirectly through differences in
Address correspondence to Kenneth M. Steele, Department of Psychology,
Appalachian State University, Boone, NC 28608; e-mail: steelekm@
appstate.edu. 1. We thank Bruce Rideout for permission to use the sets.
PSYCHOLOGICAL SCIENCE
Kenneth M. Steele, Karen E. Bass, and Melissa D. Crook
mood or arousal among treatments (Steele et al., 1997). We therefore
used a mood measure to investigate this possibility. Individuals
described the Mozart selection as lively, bouncy, and happy; the
same individuals described the Glass selection as repetitive, obnox-
ious, and grating. We hypothesized that such different reactions
would be captured in a measure of mood. The 65-question Profile of
Mood States (POMS; Educational & Industrial Testing Service, 1971)
was adapted for use in this experiment by drawing 3 questions from
each of the six mood factors reported by the measure. The factors are
Depression, Tension, Anger, Vigor, Fatigue, and Confusion. Two ques-
tions unrelated to the POMS were added, 1 to begin and 1 to end our
measure, for a total of 20 questions on mood, all answered using a 5-
point ordinal scale. Our prediction was that the Glass selection would
produce stronger indications of unpleasant mood relative to the Mozart
sonata.
Tapes for the Mozart and the Glass selections were prepared from
the CD recordings and were played on a Sony CFD-545 unit.
Procedure
Rauscher et al. (1995) used the following experimental procedure.
First, all subjects were administered 16 PF&C items as a pretest. On
the basis of their performance, they were assigned to create three
groups of equivalent capabilities. The next day, the three groups
were exposed to 10 min of a stimulus condition and immediately test-
ed with 16 PF&C items. The stimulus conditions consisted of listen-
ing to either the Mozart or the Glass selection or sitting in silence.
Each PF&C item was shown via an overhead projector for 1 min, with
a 5-s warning of the end of that trial. The three groups repeated this
daily procedure for 3 additional days, with the exception that the
group that heard the Glass selection heard other material on succeed-
ing days.
Our procedure was slightly different. Because the Mozart effect
was significant only on the 1st posttreatment day in the study by
Rauscher et al. (1995), we restricted our experiment to the one post-
treatment assessment. Performance on the PF&C pretest was not
scored prior to group assignment. Instead, we assumed that random
assignment would create equivalent groups. For schedule reasons, a
time period of 48 hr elapsed between sessions for our participants.
Rauscher and Shaws (1998) concern was that the pretest should not
occur too close in time to the treatment condition; consequently, our
lengthening of the interval by an additional 24 hr between pretest and
treatment condition should not have affected the results adversely.
Sessions were conducted in the early evening when the psycholo-
gy building was quiet. Participants were recruited to create a group of
15 students per session. Participants arrived at the first session and
were informed that they would be participating in a puzzle experi-
ment. The two sample PF&C items from the Stanford-Binet IQ mea-
sure were used to explain the task. The first sample item was projected
onto a white screen measuring 3.5 m by 3.5 m. The experimenter
explained the task using instructions slightly modified and abbreviat-
ed from the instructions provided by the Stanford-Binet measure. The
second sample S-B PF&C item was then presented and explained.
After the participants were given the opportunity to ask questions,
answer sheets were provided, and 16 PF&C items were projected for
up to 1 min each (depending on how quickly all subjects had made
their choice), with a 5-s warning of the end of each 1-min period. The
number of participants was limited to approximately 15 per session to
ensure that distance from the screen and visibility of the projected
PF&C item were equal for all participants.
The second session began 48 hr later. Participants were reminded
of the nature of the task, and answer sheets were distributed. Follow-
ing exposure to the scheduled stimulus condition, participants were
immediately tested on a new set of 16 PF&C items. The two sets of
PF&C items were used in counterbalanced order across sessions and
groups. After completing the PF&C task, the participants were imme-
diately given the mood assessment instrument and instructed to indi-
cate their mood when the PF&C task began. Performance on the
PF&C items and answers to the mood questions were analyzed at a
later date.
RESULTS
Table 1 shows the results of the experiment. The results are
grouped according to subjects assignment to treatment condition, but
pretest results show performance on PF&C items prior to the treatment
condition. The pretest results indicate that random assignment was
VOL. 10, NO. 4, JULY 1999 367
Fig. 1. Practice Stanford-Binet Paper Folding and Cutting item. The
top row shows a piece of paper undergoing a series of transformations
from left to right. The dotted line indicates the location of a fold; dou-
ble lines indicate the location of cuts. The task is to pick the illustra-
tion in the bottom row that represents the appearance of the
transformed paper when it is unfolded. For the item illustrated here,
the correct answer is C.
Table 1. Mean number of Paper Folding and Cutting items
answered correctly
Pretest Posttreatment
Listening condition N M SE M SE
Mozart 44 9.66 0.56 11.77 0.48
Silence 42 9.88 0.47 11.60 0.43
Glass 39 9.90 0.70 12.15 0.62
PSYCHOLOGICAL SCIENCE
Mystery of the Mozart Effect
368 VOL. 10, NO. 4, JULY 1999
successful in creating groups not significantly different in initial task
performance, F(2, 122) = 0.05, p = .95.
The posttreatment results, the mean number of PF&C items chosen
correctly after exposure to the scheduled treatment condition, indicate
that improvement in performance was approximately equal for all
groups. This interpretation is supported by the results of analysis of
variance: There was a significant main effect of session, F(1, 122) =
76.1, p < .001, but no significant effect of treatment, F(2, 122) = 0.11,
p = .89, and no significant interaction of treatment by session, F(2,
122) = 0.48, p = .62.
A more sensitive measure is to assess relative change at the level of
the individual. An analysis of covariance was used to examine post-
treatment performance, adjusted for an individuals initial perfor-
mance on the PF&C task. There was no significant treatment effect
observed with this procedure either, F(2, 121) = 0.61, p = .55.
Treatment condition did not influence cognitive task scores but did
influence mood scores in the anticipated manner. Significant differ-
ences among the groups were seen for both the Tension factor, F(2,
122) = 6.32, p = .002, and the Anger factor, F(2, 122) = 7.21, p = .001.
Tension scores were highest for the Glass group, intermediate for the
silence group, and lowest for the Mozart group (Tukey HSD pair-wise
probabilities: Glass vs. Mozart, p = .001; Glass vs. silence, p = .075).
Anger scores were also highest for the Glass group, intermediate for the
silence group, and lowest for the Mozart group (Tukey HSD pair-wise
probabilities: Glass vs. Mozart, p = .001; Glass vs. silence, p = .019).
DISCUSSION
The main result was that no significant Mozart effect was found
despite replication of the procedure used by Rauscher et al. (1995).
There is a large discrepancy between the results of the two studies.
The effect size for the contrast of Mozart versus silence in Rauscher et
al. (1995) was substantial (d = 0.72), whereas the effect size for the
same contrast in this study was d = 0.06. One would have expected
that fidelity to the procedure of Rauscher et al. would produce a simi-
lar effect size. Chabris (1998) calculated an average effect size of d =
0.16 for all 15 Mozart-versus-silence comparisons published, or sub-
mitted, to date.
Although there was no Mozart effect on cognitive task perfor-
mance, there was an effect on mood scores. Participants did attend to
the music, and were significantly less happy listening to the amelodic,
repetitive selection from Glass than listening to silence or the Mozart
sonata. Because differences in mood have been shown to affect per-
formance on other cognitive tasks (Kenealy, 1997; Spies, Hesse, &
Hummitzsch, 1996), these results suggest that production of a perfor-
mance difference indirectly through differences in mood or arousal
must be differentiated from the direct neurophysiological priming
effect hypothesized in Rauscher et al. (1993).
If mood or arousal differences are important, then the presence of
other subjects in a group testing procedure may have effects. In a repli-
cation of the present experiment with the modification that only one
individual was tested in a session, the results were the same (Steele,
Shannon, Kirby, & Olmstead, 1998). There was no differential
improvement on the spatial-reasoning task, although treatments did
affect mood, as in the present study.
There are issues beyond negative empirical results that confront the
investigator of the Mozart effect. One issue concerns the lack of spec-
ification of the class of music selections that are supposed to have the
hypothesized effect. Rauscher and Shaw (1998) have not advanced
beyond descriptions such as complexly structured music (p. 839).
Rauscher and Shaw cited Nantais (1997); Rideout, Dougherty, and
Wernert (1998); and Wilson and Brown (1997) as providing confirma-
tion of the Mozart effect. However, all three experiments found that
music selections other than the Mozart sonata produced significant
improvement relative to the control condition. For example, Rideout et
al. found improvement after subjects listened to a performance by
Yanni. It is difficult to determine whether this musical selection is or
is not consistent with the complex structure criterion of Rauscher
and Shaw. Additionally, Nantais found that the choice of control con-
dition was a critical issue. Nantais reported a significant difference
when the control condition was sitting alone in silence, but not when
the control condition was listening to a story. This pattern of results is
consistent with an account focused on mood or arousal differences
among conditions.
Another issue is uncertainty regarding the suitability of dependent
measures. In the original report, Rauscher et al. (1993) presented only
the combined performance from the three Stanford-Binet subtests
(PF&C, Matrices, and Pattern Analysis). In that report, they stated,
For our sample, these three tasks correlated at the .01 level of signif-
icance. We were thus able to treat them as equal measures of abstract
reasoning ability (p. 611).
The results of the 1993 experiment were recounted differently in
Rauscher and Shaw (1998). Rauscher and Shaw conjectured that other
investigators have had difficulties in producing a Mozart effect
because the investigators have not distinguished between spatial-
temporal tasks (like the PF&C task) and spatial pattern-recognition
tasks (like the Raven Progressive Matrices task). Rauscher and Shaw
indicated that their 1993 effect occurred only with the S-B PF&C task,
not with the Matrices or Pattern Analysis tasks. This improvement dif-
ference among the three Stanford-Binet tasks was used to account for
the negative results obtained from studies that used the Raven Pro-
gressive Matrices task. However, no such difference in Stanford-Binet
task performance was found by Dalla Bella et al. (1999) and Kenealy
and Monsef (1994), who used both the S-B PF&C and the Matrices
tasks, or by Weeks (1996), who used all three tasks reported in
Rauscher et al. (1993).
The popular excitement about the Mozart effect is due to its claim
to be a brief, easy remedy to improve intellectual skills. However,
previous attempts to increase IQ scores demonstrate how difficult it
is to produce even a small, short-lived gain (Spitz, 1986). Consider-
ing the duration and depth of many intervention programs, an effect
from short periods of listening to music does not seem feasible in
principle. The results of this experiment, and experiments in other
laboratories, are consistent with such an expectation. We conclude
that there is little evidence to support basing intellectual enhance-
ment programs on the existence of the causal relationship termed the
Mozart effect.
AcknowledgmentsThe work reported here was supported by grants
from the Cratis D. Williams Graduate School, Appalachian State Universi-
ty. Portions of these results were submitted by Karen Bass as a thesis in
partial fulfillment of the requirement for graduation with honors in psy-
chology. Portions of these results were presented previously at meetings of
the Southeastern Psychological Association, the North Carolina Cognition
Conference, and the Psychonomic Society. We thank Stan Aeschleman for
comments on the manuscript.
PSYCHOLOGICAL SCIENCE
Kenneth M. Steele, Karen E. Bass, and Melissa D. Crook
VOL. 10, NO. 4, JULY 1999 369
REFERENCES
Carstens, C.B., Huskins, E., & Hounshell, G.W. (1995). Listening to Mozart may not
enhance performance on the revised Minnesota paper form board test. Psychologi-
cal Reports, 77, 111114.
Chabris, C.F. (1998). Brief exposure to music does not increase intelligence. Manuscript
submitted for publication.
Dalla Bella, S., Dunlop, T., Dawe, L., Humphrey, K., & Peretz, I. (1999, April). The
Mozart effect revisited. Paper presented at the annual meeting of the Cognitive Neu-
roscience Society, Washington, DC.
Educational &Industrial Testing Service. (1971). Profile of Mood States. San Diego: Author.
Glass, P. (1973). Music with changing parts [CD]. New York: Elektra Nonesuch.
Kenealy, P. (1997). Mood-state-dependent retrieval: The effects of induced mood on mem-
ory reconsidered. Quarterly Journal of Experimental Psychology, 50A, 290317.
Kenealy, P., & Monsef, A. (1994). Music and IQ tests. The Psychologist, 7, 346.
Mozart, W.A. (1985). Sonata for two pianos, K. 448 [Recorded by M. Perahia & R. Lupu].
On Mozart: Sonata, K.448; Schubert: Fantasia, op. 103, D.940 [CD]. New York:
CBS.
Nantais, K.M. (1997). Spatial-temporal skills and exposure to music: Is there an effect, and
if so, why? Unpublished masters thesis, University of Windsor, Windsor, Canada.
Newman, J., Rosenbach, J.H., Burns, K.L., Latimer, B.C., Matocha, H.R., & Vogt, E.E.
(1995). An experimental test of the Mozart effect: Does listening to his music
improve spatial ability? Perceptual and Motor Skills, 81, 13791387.
Rauscher, F.H. (1997). A cognitive basis for the facilitation of spatial-temporal cognition
through music instruction. In V. Brummett (Ed.), Ithaca Conference 96: Music as
intelligence: A sourcebook (pp. 3144). Ithaca, NY: Ithaca College.
Rauscher, F.H., & Shaw, G.L. (1998). Key components of the Mozart effect. Perceptual
and Motor Skills, 86, 835841.
Rauscher, F.H., Shaw, G.L., & Ky, K.N. (1993). Music and spatial task performance.
Nature, 365, 611.
Rauscher, F.H., Shaw, G.L., & Ky, K.N. (1995). Listening to Mozart enhances spatial-
temporal reasoning: Towards a neurophysiological basis. Neuroscience Letters, 185,
4447.
Rideout, B.E., Dougherty, S., & Wernert, L. (1998). Effect of music on spatial perfor-
mance: A test of generality. Perceptual and Motor Skills, 86, 512514.
Rideout, B.E., & Laubach, C.M. (1996). EEG correlates of enhanced spatial performance
following exposure to music. Perceptual and Motor Skills, 82, 427432.
Spies, K., Hesse, F., & Hummitzsch, C. (1996). Mood and capacity in Baddeleys model
of human memory. Zeitschrift fur Psychologie, 204, 367381.
Spitz, H.H. (1986). The raising of intelligence: A selected history of attempts to raise
retarded intelligence. Hillsdale, NJ: Erlbaum.
Steele, K.M., Ball, T.N., & Runk, R. (1997). Listening to Mozart does not enhance back-
wards digit span performance. Perceptual and Motor Skills, 84, 11791184.
Steele, K.M., Shannon, R.A., Kirby, J.L., Jr., & Olmstead, C.G. (1998). Failure to produce
a Mozart effect with individual testing. Unpublished manuscript, Appalachian State
University, Boone, NC.
Stough, C., Kerkin, B., Bates, T., & Mangan, G. (1994). Music and spatial IQ. Personali-
ty and Individual Differences, 17, 695.
Thorndike, R.L., Hagen, E.P., & Sattler, J.M. (1986). The Stanford-Binet intelligence
scale: Fourth edition. Chicago: Riverside Publishing.
Weeks, S.P. (1996). The effect of music on abstract/visual reasoning performance in high
school music and non-music students. Dissertation Abstracts International, 56(9),
3408A. (UMI Dissertation Services No. 9600110)
Westerman, R., Spies, K., Stahl, G., & Hesse, F.W. (1996). Relative effectiveness and
validity of mood induction procedures: A meta-analysis. European Journal of Social
Psychology, 26, 557580.
Wilson, T.L., & Brown, T.L. (1997). Reexamination of the effect of Mozarts music on
spatial-task performance. The Journal of Psychology, 131, 365370.
(RECEIVED 6/28/98; REVISION ACCEPTED 1/13/99)