The Repeated Recording Illusion: The Effects of Extrinsic and Individual Difference Factors on Musical Judgments

Abstract

The repeated recording illusion refers to the phenomenon in which listeners are under the impression that they hear different musical stimuli while they are in fact identical. This phenomenon has not yet been studied systematically. Thus, the present paper aims to construct an experimental paradigm to enable the systematic measurement of the repeated recording illusion, investigating individual difference factors that contribute to it as well as extrinsic factors responsible for differences in musical judgements when the acoustic input remains the same. Seventy-two participants were misled to think that they had heard three different musical performances of an original piece when in fact they were exposed to the same repeated recording. Each time, the recording was accompanied by a different text suggesting a low, medium or high prestige of the performer. Most participants (75 %) believed that they had heard different musical performances. High levels of neuroticism and openness made it significantly more likely that an individual would fall for the illusion. Musicians were not any more or any less susceptible to the illusion than nonmusicians. For participants who fell for the illusion, the explicit prestige texts influenced evaluations of the music significantly. The effect of repeated exposure was only significant in the more familiar music condition. These results suggest that musical judgements are sometimes not based on musical cues but are influenced by factors that do not depend on the music itself. The repeated recording illusion can constitute a paradigm for investigating psychological biases and individual differences in aesthetic and musical judgements because the illusion allows for the study of their effects while the music remains the same. Results are interpreted within Tversky and Kahneman’s framework of judgements and decision-making.

Pre-press manuscript (accepted in 2017, Music Perception, http://mp.ucpress.edu)

Full text

Note:
This article (low-resolution PDF) was published in Music Perception (September,
2017): http://mp.ucpress.edu/content/35/1/94
By the rights of the University of California all rights reserved. Please direct all
requests for permission to photocopy or reproduce article content through the
University of California Press’s reprints and permissions web page:
http://www.ucpress.edu/journals.php?p=reprints
Published Citation:
Anglada-Tort, M., & Müllensiefen, D. (2017). The repeated recording illusion: The
effects of extrinsic and individual difference factors on musical judgments. Music
Perception, 35(1), 94-117. doi: 10.1525/MP.2017.35.1.94
--
07/09/2017

THE REPEATED RECORDING ILLUSION:THE EFFECTS OF EXTRINSIC
AND INDIVIDUAL DIFFERENCE FACTORS ON MUSICAL JUDGMENTS
MANUEL ANGLADA-TORT &DANIEL MU
¨LLENSIEFEN
Goldsmiths, University of London, London, United
Kingdom
T
HE REPEATED RECORDING ILLUSION REFERS TO
the phenomenon in which listeners believe to hear
different musical stimuli while they are in fact identical.
The present paper aims to construct an experimental
paradigm to enable the systematic measurement of this
phenomenon, investigating potentially related extrinsic
and individual difference factors. Participants were told
to listen to ‘‘different’’ musical performances of an orig-
inal piece when in fact they were exposed to the same
repeated recording. Each time, the recording was
accompanied by a text suggesting a low, medium, or
high prestige of the performer. Most participants
(75%) believed that they had heard different musical
performances. Participants with high levels of neuroti-
cism and openness were significantly more likely to fall
for the illusion. While the explicit information pre-
sented with the music influenced participants’ ratings
significantly, the effect of repeated exposure was only
significant in the more familiar music condition. These
results suggest that like many other human judgments,
evaluations of music also rely on cognitive biases and
heuristics that do not depend on the stimuli themselves.
The repeated recording illusion can constitute a useful
paradigm for investigating nonmusical factors because
it allows for the study of their effects while the music
remains the same.
Received: September 22, 2016, accepted February 24,
2017.
Key words: aesthetics, individual differences, explicit
information, music performance, judgments and
preferences
I
N
1977,
THE
G
ERMAN RADIO STATION
WDR 3
conducted an audience participation experiment
during a live program (see the description in Behne,
1987). The radio broadcaster misled the audience to
think that they would hear three different performances
of the same excerpt of Bruckner’s Symphony No. 4,
providing brief information about three different con-
ductors (Karl Bo¨hm, Leonard Bernstein, and Herbert
von Karajan) just before each recording was played.
However, the radio broadcaster played the same record-
ing three times. The radio station received 536 calls.
81.7%of the callers were misled and reported differ-
ences between the identical music recordings. Only the
remaining 18.3%of the listeners who called in reported
that there were no differences between the three perfor-
mances. Nevertheless, we note that the audience partic-
ipation experiment had several shortcomings, such as
a lack of control over experimental conditions and
a potential sampling bias for those listeners who
believed they had heard different musical performances
to call the radio station. Therefore, one of the main
motivations of the present paper was the replication
of this phenomenon in an experimental setting.
We will refer to this phenomenon, where listeners are
under the impression that they hear different musical
performances while in fact they are identical, as the
repeated recording illusion. Duerksen (1972) was
amongst the first academic studies to use a similar
approach. He played two tape recordings of an identical
piano performance to music major and nonmusic major
students. Participants were told that one performance
was by an eminent professional pianist and the other
one by a student. Both groups rated technical and musi-
cal characteristics of the music recording consistently
lower when told the performance was by a student than
when told it was by a professional. However, Duerksen
(1972) merely attributed the findings to an effect of
expectations and did not investigate whether partici-
pants believed that they had heard the same or different
musical performances.
There are a number of studies that used similar exper-
imental paradigms, presenting participants with identi-
cal recordings in succession (Behne & Wo¨llner, 2011;
Cavitt, 1997, 2002; Elliott, 1995; Griffiths, 2008; Juch-
niewicz, 2008; Radocy, 1976; Silvey, 2009). The main
purpose of these studies was to investigate nonmusical
factors that influence evaluations of musical perfor-
mances, such as the effect of expectations (Cavitt,
1997, 2002; Duerksen, 1972), authority (Radocy,
1976), musicians’ body movements (Behne & Wo¨llner,
Music Perception,VOLUME 35, ISSUE 1, PP.94–117,ISSN 0730-7829, ELECTRONIC ISSN 1533-8312. ©2017 BY THE REGENTS OF THE UNIVERSITY OF CALIFORNIA ALL
RIGHTS RESERVED.PLEASE DIRECT ALL REQUESTS FOR PERMISSION TO PHOTOCOPY OR REPRODUCE ARTICLE CONTENT THROUGH THE UNIVERSITY OF CALIFORNIA PRESS’S
REPRINTS AND PERMISSIONS WEB PAGE,HTTP://WWW.UCPRESS.EDU/JOURNALS.PHP?P¼REPRINTS.DOI:https://doi.org/10.1525/MP.2017.35.1.94
94 Manuel Anglada-Tort & Daniel Mu¨llensiefen

2011; Juchniewicz, 2008), race and gender (Elliott,
1995), concert dress and physical attractiveness (Grif-
fiths, 2008), and band labels (Silvey, 2009). None of
these studies considered the implications of participants
potentially falling for the repeated recording illusion.
Thus, in none of these studies it is possible to determine
whether the illusion occurred in the sample of partici-
pants. We considered the repeated recording illusion to
be a phenomenon that merits further investigation.
Exploring this phenomenon in detail could provide rel-
evant and unique insights to the fields of aesthetics,
music perception, cognition, and choice behavior.
Therefore, the present study attempts to measure sys-
tematically the repeated recording illusion, investigating
individual difference factors that contribute to it as well
as extrinsic factors responsible for differences in musical
judgments when the acoustic input remains the same.
In relation to the individual difference factors, we
suggest that the amount of music training of partici-
pants may play an important role in the repeated
recording illusion. A large number of previous studies
have shown that people with high levels of music
training (i.e., musicians) outperform nonmusicians
on many music-related tasks, indicating that music
training has a positive influence on the efficiency and
accuracy with which characteristics of sounds (e.g.,
pitch and timbre) are encoded in memory (see Pearce,
2015, for a review). For instance, musicians are more
attuned to small variations and nuances in music
(Delie
`ge, 1987) and have better recognition memory
for melody than nonmusicians (Dowling, 1978; Dowl-
ing & Bartlett, 1981; Halpern, Bartlett, & Dowling,
1995; Orsmond & Miller, 1999). We therefore hypoth-
esized that music training would have an effect on the
illusion. However, the tasks involved in the above
research (e.g., to recognize a melody) are very differ-
ent from the task that requires an individual to realize
that the same music recording is played in succession.
Thus, it is difficult to predict the direction in which
music training may affect the repeated recording illu-
sion. The present study only attempts to assess
whether musicians perform differently on this task
compared to nonmusicians.
Arguably, the paradigm used in the repeated record-
ing illusion relies on a judgment bias exerted by a figure
of authority (i.e., participants are told by a researcher in
a lab condition that they will listen to different perfor-
mances). In line with Milgram’s (1963) obedience to
authority experiment, Radocy (1976) found that the bias
exerted by a figure of authority significantly influenced
participants’ evaluations of musical events. We therefore
considered that individual differences on suggestibility
could be an important factor contributing to the illu-
sion. We hypothesized that people with higher levels of
suggestibility would be more likely to fall for the
repeated recording illusion.
The present research also explored music preferences
and personality as possible individual difference factors
related to the illusion. Individuals tend to have stronger
preferences for certain genres of music, becoming more
familiar with the preferred style as a result of repeated
listening. Repeated exposure to a piece of music
increases the liking for it and decreases its subjective
complexity (see North & Hargreaves, 2008, for a review).
In relation to personality, research shows that person-
ality traits relate to specific preferences for music styles
(see Greasley & Lamont, 2016, for a review). For
instance, openness to experience is positively linked to
preference for reflective and complex styles (e.g., classi-
cal music) (Rentfrow & Gosling, 2003). Furthermore,
research on individual differences has found links
between personality and suggestibility, showing, for
example, a positive (but low) relationship between sug-
gestibility and neuroticism (see Gudjonsson, 2003, for
a review). Therefore, we hypothesized that preferences
for music style and personality traits would affect par-
ticipants’ susceptibility to the repeated recording illu-
sion, although we could not specify in which direction.
Extrinsic factors that may be responsible for differ-
ences in musical judgments when the acoustic input is
identical include the effect of explicit information. Pre-
senting music with explicit information has been shown
to be influential in the evaluation of musical perfor-
mances (Cassidy & Sims, 1991; Cavitt, 1997, 2002; Kro-
ger & Margulis, 2016; Margulis, 2010; Margulis, Kisida,
& Greene, 2015; North & Hargreaves, 2005; Silveira &
Diaz, 2014; Silvey, 2009; Vuoskoski & Eerola, 2013). In
an fMRI study, Kirk, Skov, Hulme, Christensen, and
Zeki (2009) presented the same images of artworks with
different contextual information, varying in prestige
(i.e., labeled as ‘‘gallery’’ or ‘‘computer generated’’). The
findings revealed that when the artworks were labeled as
‘‘gallery’’ they were rated higher in an aesthetic value
scale than when labeled as ‘‘computer generated.’’ The
fMRI data showed more activity in the medial orbito-
frontal cortex under the gallery context compared to the
computer one, suggesting a neural system supporting
contextual modulation of aesthetic ratings. In the pres-
ent study, we hypothesized that participants would eval-
uate the same recording more positively when presented
with a text suggesting high prestige of the performer
than when presented with texts of lower prestige levels.
Another important extrinsic factor responsible for
differences in musical judgments when the acoustic
The Repeated Recording Illusion 95

input is identical may be the effect of repeated exposure.
In line with the domain-general mere exposure effect
(Zajonc, 1968), the liking of an initially neutral stimuli
increases with repeated exposure. While the effect of
mere exposure has been extensively studied using
particular pieces of music as stimuli (see North &
Hargreaves, 2008, for a review), only a few studies have
examined this effect on evaluations of performances of
individual pieces. In a recent study, Kroger and Margu-
lis (2016) presented participants with pairs of solo piano
performances and informed them that one was played
by a conservatory student and the other by a world-
renowned professional. After listening to each pair, par-
ticipants had to select which they considered to have
been performed by the professional. The results indi-
cated that participants selected the second performance
as professional more frequently than the first perfor-
mance, although this effect was modulated by the actual
identity of the performer. In relation to the repeated
recording illusion, we hypothesized that participants’
ratings of the same recording would improve with
repeated exposure.
The present research had three main aims. The first
was to construct an experimental paradigm to enable
the systematic measurement of the repeated recording
illusion. The second aim was to investigate possible
individual difference factors that contribute to the
illusion (i.e., music training, suggestibility, music pre-
ferences, and personality). The third aim was to inves-
tigate extrinsic factors responsible for differences in
musical judgments when the acoustic input remains
the same (i.e., explicit information and repeated expo-
sure). In addition, in order to capture higher-order
interactions between the extrinsic and individual dif-
ference factors, an exploratory analysis of the same
data aimed to identify conditions that lead to particu-
larly positive or negative judgments.
In constructing the experimental paradigm of the
repeated recording illusion, participants were misled
to think that they had heard three different perfor-
mances of an original music piece. However, we played
the exact same recording three times in succession.
Each time the recording was accompanied by a text
suggesting low, medium, or high prestige of the per-
former. We repeated this experimental procedure with
two different pieces of music, a piece of classical music
and a piece of popular music for which we assumed
a high stylistic familiarity for most participants. In
order to study the repeated recording illusion without
an effect of explicit information, we examined a non-
prestige group where we did not manipulate prestige of
the performer.
Method
PARTICIPANTS
A sample of seventy-two university students took part
in the experiment (36 male, 36 female), aged 19-39
(M¼24.26, SD ¼3.60). Twenty-nine participants were
considered as trained musicians (M¼45.74, SD ¼5.73
on the Musical Training subscale of the Goldsmiths
Musical Sophistication Index; Mu
¨llensiefen, Gingras,
Musil, & Stewart, 2014; and had 6 to 8 years of formal
music training). Forty-five participants were considered
as nonmusicians (M¼22.71, SD ¼7.34 on the Gold-
MSI; and had 1 year of formal music training on aver-
age). Twelve participants were randomly allocated to
a nonprestige condition (6 male, 6 female), aged 21-29
(M¼24.34, SD ¼3.45). Participation was on a volun-
teer basis and unpaid.
DESIGN
The study employed a 3 x 3 x 2 repeated measures
design. Explicit information (low vs. medium vs. high
prestige text), repeated exposure (first vs. second vs.
third position), and genre of the original music piece
(popular vs. classical music) were the within-participant
factors. The three levels of the explicit information fac-
tor were fully counterbalanced with presentation order
across participants. Half of the participants started with
the popular music piece condition and the other half
started with the piece of classical music. The dependent
variables consisted of a diverse range of musical judg-
ments provided immediately after each listening and at
the end of each music condition. In order to explore the
repeated recording illusion without an effect of explicit
information, we examined a nonprestige group where we
did not manipulate prestige of the performer. In addition,
we measured individual difference factors that were
expected to contribute to the illusion (i.e., music training,
suggestibility, music preferences, and personality).
MATERIALS
In the popular music condition participants listened to
a live recording of ‘‘Jailhouse Rock’’ by Elvis Presley,
recorded in NBC studios in 1968. The length of the
recording was 1 minute and 36 seconds. This piece was
selected because we assumed a high stylistic familiarity
for most participants. In the classical music condition
participants listened to the final part of a live recording
of Bruckner Symphony No. 4 ‘‘Die Romantische,’’ con-
ducted by Gu
¨nter Wand and performed by the Berliner
Philarmonic Orchestra in 1998. The length of the
recording was 2 minutes and 48 seconds. This piece was
selected in order to replicate empirically the experiment
96 Manuel Anglada-Tort & Daniel Mu¨llensiefen

carried out in the German radio station WDR 3 (Behne,
1987). The original recordings were edited and normal-
ized using ableton live computer software. In the popu-
lar live recording we edited the start and end points of
the original recording in order to contain only the musi-
cal performance element of the recording. Similar to the
German radio experiment (Behne, 1987), the start and
end points of the classical music piece were edited to
contain the final part of the performance. We then nor-
malized the volume of the two recordings to be fixed on
the same threshold. Then each recording was duplicated
three times and written to the same compact disc, using
iTunes 12.2.2. Each copy of the music recording was
saved under a different name, which included perfor-
mers’ names as used in the texts suggesting different
levels of prestige. In the nonprestige condition, the
names were Performance 1, Performance 2, and Perfor-
mance 3.
To manipulate the effect of explicit information we
created three texts suggesting low, medium, and high
prestige of the performer. The texts had the same for-
mat, organization and a length of 150 words. In the
popular music condition (‘‘Jailhouse Rock’’), the three
‘‘different’’ performers were presented as different Elvis
impersonators. The prestige texts provided information
about the three impersonators, who differed on skill and
success (Appendix A). In the classical music condition
(Bruckner’s Symphony No. 4), the three ‘different’ per-
formers were presented as different classical conductors.
The prestige texts provided information about the con-
ductors, who differed on skill and success (Appendix B).
Gu
¨nter Wand, the actual conductor of the recording,
was not among these conductors. In the nonprestige
condition, three different texts were created with the
same format, organization and length of 150 words.
While in the popular music condition the three texts
provided neutral information from different parts of
Elvis Presley’s biography, in the classical music condi-
tion the texts provided neutral information from differ-
ent parts of Anton Bruckner’s biography.
In order to evaluate liking as well as more objective
aspects of the performance (e.g., pitch accuracy and
tempo appropriateness), we designed an evaluation
form consisting of ten Likert rating scales and two
open-text boxes. Nine of the rating scales consisted of
sliders ranging from 0 to 100. The rating scales were
provided to evaluate the following dimensions: (1) lik-
ing of the interpretation, (2) timing and rhythm, and (3)
tone quality (from dislike strongly to like strongly), (4)
tempo appropriateness (from very inappropriate to very
appropriate), (5) pitch accuracy (from very inaccurate to
very accurate), (6) emotional quality and (7) overall
quality of the performance (form very bad to very good),
and degree of agreement to two statements: (8) some
aspects regarding the singer’s vocal technique/orchestral
technique could be improved, and (9) some aspects of the
overall interpretation could improve (from strongly dis-
agree to strongly agree). In addition, (10) participants
were asked to rate each recording using a 5-star rating
scale, ranging from 1 star (strongly dislike) to 5 stars
(like strongly). The Likert rating scales were designed
to examine differences in musical judgments when the
acoustic input is the same. After the ten Likert rating
scales, two open-text boxes were provided where partici-
pants could write down anything to describe the perfor-
mance and whether or not they enjoyed it. Answering the
open-text boxes was optional.
At the end of each music condition, participants were
requested to fill out a final evaluation form. In this final
evaluation, participants were asked to rate how much
they liked each recording compared to the others, on
a scale from 0 (much less than the others) to 100 (much
more than the others), where the midpoint of the scale
(‘‘50’’) was labeled as as much as the others. Participants
also had to evaluate the familiarity to the original piece
of music, on a scale from 0 (don’t know at all) to 100
(know very well). In all rating scales, participants were
able to see the number attributed to their specific rating.
We also provided an open-text box where participants
could write down any optional comments regarding the
experience of the experiment. The information from the
open-text boxes was used to determine whether parti-
cipants fell for the illusion or not. When the informa-
tion from the open-text boxes was not sufficient to
make a clear and objective decision, the final compara-
tive rating scales were taken into consideration to deter-
mine whether participants fell for the illusion or not.
The open-text boxes were used in conjunction with the
final comparative rating scales, designed to address
a clear limitation in this experiment: we could not ask
participants explicitly whether the recordings were the
same or different as this would have biased their subse-
quent evaluations and behavior in the experiment.
In order to measure the individual difference factors,
participants filled out different questionnaires corre-
sponding to each factor. To measure participants’ music
training and active engagement with music we used the
Goldsmiths Musical Sophistication self-report question-
naire (Gold-MSI, Mu
¨llensiefen et al., 2014). To measure
participants’ suggestibility, we used the Social Desirabil-
ity Scale (SDS-17) (Sto¨ber, 2001) and 8 items adopted
from the Susceptibility Persuasive Strategies Scale
(STPS) (Kaptein, Ruyter, Markopoulos, & Aarts,
2012), which measured bias to authority, consensus, and
The Repeated Recording Illusion 97

persuadability, used in a previous study (Unal, Temizel,
& Eren, 2014). To assess music preferences and stylistic
familiarity, we used the Short Test of Music Preferences
revised (STOMP-R, Rentfrow & Gosling, 2003). To
measure personality, we used the Big Five Inventory
(BFI) (John & Srivastava, 1999).
PROCEDURE
Participants were tested individually in small cubicle
rooms. They listened to the music recordings using
professional headphones (KNS 8400 Studio Head-
phones, KRK systems) and at a comfortable listening
level that could be adjusted by the individual partici-
pants prior to the actual experiment. Participants were
told that the main purpose of the study was to measure
people’s skills in evaluating technical and musical
aspects of different musical performances of the same
original piece. After filling out the Gold-MSI question-
naire, participants were instructed to listen to three
different interpretations of the same piece of music
and to evaluate them as accurately as possible. Before
listening to each recording, participants were pre-
sented with the corresponding text suggesting different
levels of prestige. Immediately after reading the text
participants listened to the recording. Immediately
after listening to each recording, participants com-
pleted the evaluation form, where they were presented
with the ten Likert rating scales and two open-text
boxes. The experiment had two parts with exactly the
same procedure and experimental instructions, but
using popular music (‘‘Jailhouse Rock’’) and classical
music (Bruckner’s Symphony No. 4) respectively.
Immediately after listening to the three recordings of
each part, participants filled the final evaluation form
consisting in the final comparative rating scales and
the open-text box. Between completing the two parts
of the experiment participants were asked to fill out the
STOMP-R questionnaire. In the nonprestige condition
the procedure was the same. Participants were also
instructed that they would listen to three different per-
formances of the same piece, but the texts presented
with the music did not induce any kind of prestige. The
three recordings were presented as Performer 1, Per-
former 2, and Performer 3. Two weeks after the exper-
iment, participants were asked via email to fill out the
BFI, SDS-17, and the 8 items measuring suggestibility.
The experiment and questionnaires were implemented
in Qualtrics software (Qualtrics, Provo, UT). This
research was granted ethical approval by the Ethics
Committee of the Department of Psychology of Gold-
smiths College, University of London.
Results
THE REPEATED RECORDING ILLUSION
In order to determine whether participants fell for the
repeated recording illusion or not we used the following
procedure: We first assessed the information provided
in the open-text boxes. From a total of 14 open-text
boxes (7 in the popular music condition and 7 in the
classical music conditions), on average participants pro-
vided information in 87.35%of the boxes (93.7%in the
popular music condition and 93.68%in the classical
music condition). By using the information provided
in the open-text boxes we were able to identify 48 par-
ticipants out of 72 (66.67%) in the popular music con-
dition and 50 participants out of 72 (69.45%) in the
classical music condition, who provided specific infor-
mation either reporting differences between perfor-
mances or reporting that the recordings were the same.
There were cases wherein the information from the
open-text boxes was not sufficient to make a clear and
objective decision but suggested a direction: either that
the participant was not aware that the recordings were
identical or that the participant suspected that they were
the same. In these cases, we took into consideration
the scores from the final comparative rating scales
where participants had to compare how much did they
like each recording in comparison to the others, on
a scale from 0 (much less than the others) to 100 (much
more than the others), where the midpoint of the scale
(‘‘50’’) was labeled as as much as the others. We only
classified the participant when the scores from the final
comparative ratings confirmed the suggested direction
from the text boxes. It is important to note that we never
took into consideration the scores form the final com-
parative ratings on its own.
When the information from the open-text boxes was
not sufficient and/or too ambiguous to make a clear and
objective decision, we did not include the participant’s
data in the subsequent analyses. Two participants pro-
vided highly ambiguous statements in the open-text
boxes for both music conditions and the two partici-
pants were therefore excluded from the subsequent
analyses. Furthermore, one participant provided ambig-
uous information in the popular music condition and
a different participant in the classical music condition.
Thus, we had a total of 69 participants in each music
condition.
As a consequence of using the above mentioned pro-
cedure, we had a total of four possible criteria to deter-
mine whether participants fell for the repeated
recording illusion or not (see Appendix C for a decision
diagram depicting the decision procedure and criteria;
98 Manuel Anglada-Tort & Daniel Mu¨llensiefen

The tables in Appendix F and G show the information
used to make each individual decision per participant in
the two music conditions):
(1) When the information provided in the open-text
boxes specifically indicated any differences
between performances: In the popular music con-
dition, 37 out of 69 participants (53.62%) specif-
ically reported information indicating differences
between performances, such as ‘‘more upbeat than
the two others, a happier sounding performance’’
or ‘‘this piece sounds more aggressive than the
previous one. The tempo for me is faster.’’ In the
classical music condition, 42 out of 69 participants
(60.87%) specifically reported information indi-
cating differences between performances, such as
‘‘the mood in this piece seemed to escalate a lot
more naturally than in the other pieces’’ or ‘‘this
interpretation sounded a bit more hesitant. Again,
it was not as dramatic as the first performance, but
it was clearer than the second one.’’
(2) When the information in the open-text boxes
specifically indicated that the participant realized
that the recordings were the same: In the popular
music condition, 11 out 69 participants (15.94%)
specifically reported information indicating that
the recordings were the same (e.g., ‘‘I reckon this
is the same file repeated three times’’ or ‘‘this is
absolutely the same as the first two’’). In the clas-
sical music condition, 8 out 69 participants
(11.59%) specifically reported information indi-
cating that the recordings were the same (e.g.,
‘‘This sounds exactly like the two others’’ or ‘‘I
thought all 3 were the same’’).
(3) When the information provided in the open-text
boxes was not sufficient to make a clear and
objective decision but suggested that the partici-
pant was not aware that the recordings were iden-
tical: In these cases, in addition to the open-text
boxes, we took into consideration the scores from
the final comparative rating scales. If at least one
score from the final comparative ratings differed
by 10%from the midpoint of the scale (‘‘50’’), or
any two scores differed by 10%from each other,
we considered the participant as falling for the
illusion. Nineteen participants (27.54%) in the
popular music condition and 17 participants
(24.64%) in the classical music condition were
classified using this third criterion.
(4) When the information provided in the open-text
boxes was not sufficient to make a clear and
objective decision, but suggested that the
participant suspected that the performances were
the same: In these cases, in addition to the open-
text boxes, we took into consideration the scores
from the final comparative rating scales. If the
three scores from the final comparative ratings
did not differ more than 10%from the midpoint
of the scale (‘‘50’’), we considered the participant
as not falling for the illusion. Two participants
(2.90%) in the popular music condition and two
different participants (2.90%) in the classical
music condition were classified using this fourth
criterion.
Table 1 shows the number of participants who fell for
the repeated recording illusion. In the total sample of
participants, 52 out of 69 participants (75.36%) believed
that they had heard different musical performances in
at least one of the two music conditions. By contrast,
17 participants (24.64%) recognized that the perfor-
mance was the same in at least one of the two music
conditions. Only 6 out of 69 participants (8.7%) realized
that the recordings were identical in both music condi-
tions. When looking at the music conditions separately,
in the popular music condition 56 participants (81.16%)
fell for the illusion and 13 participants (18.84%) did not.
In the classical music condition, 59 participants
(85.51%)fellfortheillusionand10participants
(14.49%) did not. Additionally, in the nonprestige con-
dition (where the effect of explicit information was not
manipulated), 9 out of 12 participants (75%) were sus-
ceptible to the illusion. According to a X
2
test, there was
no significant association between the music conditions
(popular and classical piece) and the occurrence of the
repeated recording illusion, X
2
(1) ¼.47, p¼.49.
According to Fisher’s Exact test, there was no significant
association between the presence of prestige (i.e.,
prestige-suggestion and nonprestige group) and the
occurrence of the illusion (p¼.65).
Generally, participants rated the popular music piece as
more familiar (M¼72.16, SD ¼21.93 on 100-point
TABLE 1.
Numbers of Participants Falling for the Repeated
Recording Illusion
Did participants fall for
the repeated recording illusion? Yes %No %
Total 52 75.36 17 24.64
Popular music 56 81.16 13 18.84
Classical music 59 85.51 10 14.49
Prestige-suggestion group
Nonprestige group
43
9
80.70
75
14
3
24.56
25
Note. Participants were classified as NO if they identified the three
recordings as identical in at least one of the two music conditions.
The Repeated Recording Illusion 99

rating scale) than the classical piece (M¼13.73, SD ¼
21.10). This difference in familiarity was highly signifi-
cant as indicated by a paired samples t-test, t(68) ¼16.43,
p< .001.
INDIVIDUAL DIFFERENCE FACTORS
The analysis of individual difference factors was con-
ducted using a data classification method known as the
random forest (Breiman, 2001), in which the aim was to
examine whether individual differences contributed to
the repeated recording illusion. Random forest proce-
dures differ in a number of ways from other classifica-
tion methods in that they can handle large sets of
predictor variables and do not assume a linear relation-
ship between predictors (see Hastie, Tibshirani, & Fried-
man, 2009; see Pawley & Mu
¨llensiefen, 2012, for the use
of random forests in music psychology). We used the
conditional random forest based on permutation tests
as implemented in the R package ‘‘party’’ (Hothorn,
Buehlmann, Dudoit, Molinaro, & Van der Laan, 2006;
Hothorn, Hornik, & Zeileis, 2006; Strobl, Boulesteix,
Kneib, Agustin, & Zeileis, 2008; Strobl, Malley, & Tutz,
2009). The random forest model was run with a size of
5000 trees. We employed a measure of variable impor-
tance for each predictor variable, which is designed to
produce unbiased estimates of variable importance even
in situations where significant correlations between pre-
dictor variables exist and when the dependent variable
is very unequally distributed (Janitza, Strobl, & Boules-
teix, 2013).
As predictor variables, we used 6 demographic vari-
ables as well as musical variables that were collected
during the experimental session (age, gender, Gold-
MSI Musical Training and Active Engagement scores,
STOMP preference scores for Reflective & Complex,
Intense & Rebellious, Upbeat & Conventional, and
Energetic & Rhythmic). Data for 9 additional variables
were collected via the follow-up questionnaire measuring
the big five personality traits (Extraversion, Agreeable-
ness, Conscientiousness, Neuroticism, and Openness) as
well as suggestibility (Authority score, Consensus score,
Persuadability score, and Social Desirability score). Using
these 17 predictor variables we computed two different
models with two different binary dependent variables: (a)
a strict criterion model in which only those participants
who fell for the illusion in both music conditions were
considered as not falling for the illusion, and (b) a less
strict criterion model where we considered as not falling
for the illusion those participants who fell for the illusion
in at least one of the two music conditions. A variable
importance score was obtained for each predictor vari-
able, describing how predictive each variable was
compared to the others. We applied a ‘‘confidence inter-
val’’ criterion in order to select the top performing vari-
ables. Only the variables whose variable importance
scores were positive and greater than the absolute value
of the lowest negative variable importance score were
selected (Strobl et al., 2008; Strobl et al., 2009).
The two models (strict and less strict criterion) deliv-
ered very similar results, indicating that there were two
variable importance scores that met the above criterion
(neuroticism and openness). In both models, neuroti-
cism was the most important variable contributing to
the repeated recording illusion, followed by openness
(see Appendix D for graphs with the 17 variable impor-
tant scores in the two models). In the strict criterion
model, neuroticism was approximately 3.5 times more
important than openness. In this model, those partici-
pants falling for the illusion in the two music conditions
scored higher in neuroticism (M¼23. 41, SD ¼5.17)
and openness (M¼40.12, SD ¼5.14) than those parti-
cipants who did not fall for the illusion (M¼17.43,
SD ¼6.85 on the neuroticism factor; M¼35.28, SD ¼
7.02 on the openness factor). In the less strict criterion
model, neuroticism was approximately 3 times more
important than openness. In this model, those partici-
pants who fell for the illusion in at least one of the two
music conditions scored higher in neuroticism (M¼
23.14, SD ¼5.55) and openness (M¼40.12, SD ¼
5.42) than those participants who did not fall for the
illusion (M¼17.43, SD ¼6.85 on the neuroticism
factor; M¼35.28, SD ¼7.02 on the openness factor).
EXTRINSIC FACTORS: THE EFFECTS OF EXPLICIT INFORMATION
AND REPEATED EXPOSURE
The subsequent analyses included the sixty participants
of the main experimental group (i.e., where we manip-
ulated the effect of explicit information). In the popular
music condition, three participants were excluded from
the analyses and ten fell for the illusion. Therefore, in
the popular music condition we had a total of 47 parti-
cipants. In the classical music condition, three partici-
pants were excluded from the analyses and nine fell for
the illusion. Therefore, in the classical music condition
we had a total of 48 participants.
Participants’ ratings on the ten Likert rating scales were
aggregated into a single scale. First, the ratings of each
participant on each rating scale were transformed into
z-scores across the ratings of all six recordings (three in
the popular music condition and three in the classical).
Then, a principal component analysis (PCA) was con-
ducted on the z-transformed data of the ten rating scales.
The Kaiser-Meyer-Olkin (KMO) measure verified the
sampling adequacy for the analysis, KMO ¼.93
100 Manuel Anglada-Tort & Daniel Mu¨llensiefen

(‘‘marvellous’’ according to Hutcheson & Sofroniou,
1999). In addition, all KMO values for individual rating
scales were greater than .86, which is well above com-
monly accepted limit of .5 (Field, 2013). The scree plot of
the different factor solution was very clear and indicated
a solution with just one factor. Moreover, there was only
one PCA component with an eigenvalue >1 which
explained 64.56%of the variance. Thus, this one-factor
PCA solution was accepted and component scores for all
participants’ ratings were computed using the regression
method.
Because the two music recordings used in the popular
and classical music conditions differed substantially in
several aspects (i.e., music genre, familiarity, presence of
words/ vocalizations, duration of the excerpt, and qual-
ity of the recording), we ran two separate models, one
with the ratings obtained in the popular music condi-
tion and one with the ratings obtained in the classical
music condition (see Appendix E for a summary table of
both models). In addition, the ratings were standardized
separately for each music condition.
To test the hypothesis regarding the effects of explicit
information and repeated exposure we used the R
packages lme4 (Bates, Ma¨chler, Bolker, & Walker,
2015) and lmerTest (Kuznetsova, Brockhoff, & Chris-
tensen, 2016) to perform a linear mixed effects analysis
with the z-scores of the participants’ ratings as the
dependent variable. In the two models, explicit infor-
mation (low, medium, and high prestige of the text) and
repeated exposure (first, second, and third position)
were the fixed effect independent factors, whereas par-
ticipants were the random effect factor.
The linear mixed-effect model of the popular music
condition revealed that there were significant main
effects of explicit information (p< .001) and repeated
exposure (p< .001). Because the interaction between
explicit information and repeated exposure was not sig-
nificant we ran the model again only with the two main
factors. The effects of explicit information and repeated
exposure become visible in Figure 1. The effect of
explicit information shows that when the recording was
presented with a high prestige text the ratings were
significantly higher than when presented with low and
medium texts. The effect of repeated exposure of the
recording shows that when the recording was heard in
the second and third positions the ratings were signif-
icantly higher than when heard in the first position.
The linear mixed-effect model of the classical music
condition revealed that there was a significant main
effect of explicit information (p< .001). However, the
effect of repeated exposure and the interaction between
explicit information and repeated exposure were not
significant. Because the interaction between explicit
information and repeated exposure was not significant
we ran the model again only with the two main factors.
The effect of explicit information shows that when the
recording was presented with a high prestige text the
ratings were significantly higher than when presented
with low and medium texts (Figure 2).
The R
2
for the classical music model was .16 and
therefore lower than the R
2
of .28 of the popular music
model, indicating that the extrinsic factors explained
more of the variance in the more familiar popular music
condition.
EXPLORATORY ANALYSIS (REGRESSION TREE MODEL)
In order to capture higher order interactions between
extrinsic and individual difference factors and identify
conditions that lead to particularly low and high ratings,
FIGURE 1. Effects of explicit information and repeated exposure in the
popular music condition. Error bars represent the standard error.
FIGURE 2. Effects of explicit information and repeated exposure in the
classical music condition. Error bars represent the standard error.
The Repeated Recording Illusion 101

we computed a regression tree model based on permu-
tation tests as implemented in the R package ‘‘party’’
(Hothorn, Buehlmann et al., 2006; Hothorn, Hornik,
& Zeileis, 2006; Strobl et al., 2008, 2009). Statistical tree
models differ in a number of ways from linear regres-
sion models (see Hastie et al., 2009) in that they use
a built-in variable selection mechanism and therefore
can handle large sets of predictor variables. In addition,
tree models do not assume a linear relationship between
predictors and the dependent variable and they are very
useful for modelling higher-order interaction effects
between predictor variables automatically. For this
study we used a particular family of tree models called
conditional inference trees that combine the rigorous
theory of permutation statistics (Hothorn, Hornik, &
Zeileis, 2006) with the principle of recursive partition-
ing (Zeileis, Hothorn, & Hornik, 2008).
For the regression tree model, the z-transformed par-
ticipants’ ratings served as the dependent variable. In
addition to the two extrinsic factors (explicit informa-
tion and repeated exposure), we added the factor music
genre (popular and classical music) and six individual
difference variables (1. music training, 2. self-rated
familiarity with the music piece, 3. preference for the
STOMP meta-genre reflective & complex, 4. preference
for the STOMP meta-genre Intense & Rebellious, 5.
neuroticism, and 6. Openness), resulting in a total of
nine independent variables. Figure 3 shows the struc-
ture of the regression tree. The model makes use of only
3 of the nine independent variables and has an R
2
value
of .23. For each node of the tree, the pvalues indicating
the significance of the split based on the permutation
statistics are presented as well as a description of the two
subgroups of the split on the independent variable. For
the terminal nodes at the bottom of the graph, the dis-
tribution of the ratings on the standardized rating scale
are depicted as box-and-whisker plots.
The tree model can be interpreted by starting at the
top and following each branch down, to arrive at a ter-
minal node. A path to a terminal node describes the
interaction of experimental conditions that lead to a par-
ticular subset of ratings. To arrive at the subset with the
highest (i.e., most positive) average ratings, follow the
first ‘‘Explicit Information’’ node down the ‘‘High Pres-
tige’’ branch (left-hand side) and then descend to the
left at the ‘‘Repeated Exposure’’ node down the ‘‘2
nd
and
FIGURE 3. Regression tree model.
102 Manuel Anglada-Tort & Daniel Mu¨llensiefen

3
rd
Positions’’ branch. This branch can be interpreted as
follows: when participants listened to the music record-
ing presented with a high prestige text in the second and
third positions, the average ratings were around 1 and,
therefore, the highest compared to the other terminal
branches of the model. In contrast, the lowest ratings,
which were around -1, were given when the recording
was presented with low and medium prestige texts, in the
popular music condition, and when the recording was
heard for the first time. Overall, the regression tree model
confirms the effects of explicit information and repeated
exposure, but it also shows higher-level interactions
between the extrinsic factors and the two pieces of music.
None of the individual difference factors were significant
in the tree model. This indicates that after participants
had fallen for the illusion, individual difference factors
did not play an important role and musical judgments
were mainly influenced by the extrinsic factors.
Discussion
The primary aim of the present study was to construct
an experimental paradigm to enable the systematic
measurement of the repeated recording illusion. Parti-
cipants were misled to think that they had heard three
different performances of an original piece when in fact
they were exposed to the same repeated recording.
Each time, the recording was accompanied by a differ-
ent text suggesting a low, medium, or high prestige of
the performer. Most participants (75.36%) believed that
they had heard different musical performances. In con-
trast, seventeen participants (24.64%) recognized that
the performance was the same in at least one of the two
music conditions. Only six participants (8.7%) realized
that the recordings were identical in both music con-
ditions. Nearly three-quarters of the participants pro-
vided verbal comments indicating specific differences
between the performances (e.g., ‘‘this piece sounds
more aggressive than the previous one. The tempo for
me is faster’’) or that they were the same (e.g., ‘‘I reckon
this is the same file repeated three times’’). Thus, it can
be concluded that the majority of the participants fell
for the repeated recording illusion. This finding sug-
gests that musical judgments are sometimes not based
on perceptual features and musical cues but are influ-
enced by factors that do not depend on the music itself.
This is at least true when a mild deception is applied
and participants believe that they had heard different
performances.
It could be argued that the repeated recording illusion
occurs in part because participants are not familiar with
the original piece of music. We examined the illusion
using two different pieces that were significantly differ-
ent on familiarity, a highly familiar piece of popular
music (‘‘Jailhouse Rock’’ by Elvis Presley) and a highly
unfamiliar piece of classical music (Bruckner’s Sym-
phony No. 4). The repeated recording illusion occurred
similarly in the two music conditions. However, these
two recordings differed substantially in several other
aspects, including music genre, complexity, length of
the excerpt, presence of vocals, and quality of the
recording. Thus, these variables are confounded in this
experimental setup. Any interpretation of differences
between the two musical stimuli will have to take this
into account. Further studies should explore the
repeated recording illusion with a larger range of differ-
ent performances and recordings.
It is important to note that there is a main methodo-
logical restriction to be considered in the experimental
design used here: an implicit bias of authority figure. In
other words, the fact that participants were told they
would listen to ‘‘three different performances’’ by an
investigator in a lab situation may account, at least partly,
for the occurrence of the illusion. It would be interesting
for future research to investigate the repeated recording
illusion using an experimental paradigm without any
implicit bias of authority. This paradigm could consist
in presenting participants with pairs of different and
identical musical performances. Participants would be
instructed to rate how different are the two performances
using several rating scales. In the cases where the perfor-
mances were identical, participants’ ratings would indi-
cate to what extent people hear differences when listening
to the same repeated recording without relying on a judg-
ments bias excreted by a figure of authority.
The second aim of the study was to investigate pos-
sible individual difference factors that contribute to the
repeated recording illusion. The most important indi-
vidual difference factor related to the illusion was the
personality trait of neuroticism, which is in line with
previous research showing a positive (but low) link
between vulnerability to suggestion and neuroticism
(see Gudjonsson, 2003). This finding suggests that peo-
ple who tend to be anxious, pessimistic, shy, fearful,
vulnerable, and emotionally unstable are more likely
to fall for the repeated recording illusion. Although less
important, openness to experience also was a significant
factor related to the occurrence of the illusion, suggest-
ing that people who tend to be curious, imaginative,
artistic, excitable, and unconventional are more likely
to fall for the illusion. Importantly, none of the other
individual difference factors that were expected to con-
tribute to the illusion were significant, including music
training, suggestibility, and preferences for music style.
The Repeated Recording Illusion 103

We consider particularly interesting that different levels
of suggestibility (including bias to authority, consensus,
persuadability, and social desirability) were not related
with the occurrence of the illusion. Moreover, in our
sample of participants, highly trained musicians were
not any more or any less susceptible to the repeated
recording illusion than participants with low levels of
music training. Thus, it remains still open the question
of which are the main individual differences contribut-
ing to the repeated recording illusion. For instance,
what would occur when using participants with a greater
range of music training and expertise (e.g., top-level
professional musicians and music critics)? Would other
individual differences (e.g., intelligence, memory, per-
ceptual abilities) be able to explain why some people fall
for the illusion while others seem no be unaffected by it?
The third aim of the present research was to investi-
gate extrinsic factors responsible for differences in musi-
cal judgments when the acoustic input remains the
same. As predicted, we found that the effect of explicit
information contributed significantly to differences in
musical judgments. This effect was clear in the two
music conditions, where participants rated the same
music recording significantly better when presented with
a high prestige text than when presented with low and
medium prestige texts. This finding is consistent with
previous research on the effects of explicit information
upon aesthetic reactions to music (e.g., Kroger & Mar-
gulis, 2016; Margulis, 2010; Margulis et al., 2015; North
& Hargreaves, 2005). Using a similar paradigm, where
identical artworks were presented with different contex-
tual explicit information varying in prestige, Kirk et al.
(2009) found that prefrontal and orbitofrontal cortices
recruited by aesthetic judgments were significantly influ-
enced by the explicit information presented with the
same stimuli. We suggest that this neural system could
also be responsible for the modulation of aesthetic reac-
tions to music by explicit contextual information.
The effect of repeated exposure was only significant in
the more familiar popular music condition, but not in
the more unfamiliar classical music condition. This
finding supports partly previous research on the effects
of repeated exposure to music (North & Hargreaves,
2008, for a review). In one of the few studies using
musical performances as stimuli, Kroger and Margulis
(2016) found that evaluations of performances were
driven by a combination of repeated exposure and the
actual identity of the performer. Interestingly, in a sec-
ond experiment, Kroger and Margulis (2016) found that
the effect of explicit information was mitigated by the
influence of the actual performer and repeated expo-
sure, showing interplay between intrinsic and extrinsic
factors. In the present study, the two original pieces of
music differed in a number of important aspects. For
instance, the classical piece was a minute longer than
the popular piece, did not contain vocals, and was
highly unfamiliar to most of the participants. Further-
more, while the popular music piece was a live record-
ing from 1968 that had a notably worse recording
quality than ordinary studio recordings, the quality of
the classical music piece (recorded live in 1998) was
superior. Therefore, it may be possible that the effect
of repeated exposure did not affect participants in the
classical music condition because of the nature of the
music recording. Moreover, the explicit information
presented with the recordings might have had a different
impact on participants in the two music conditions.
Future studies will need to explore the strength of the
effect of repeated exposure across a larger range of dif-
ferent performances and recordings.
In an attempt to explore higher-order interactions
between the extrinsic and individual difference factors,
we used a regression tree model in which we identified
conditions that lead to particularly low and high ratings.
The highest ratings were given when the music record-
ing was presented with a high prestige text and heard in
the second and third positions. In contrast, the lowest
ratings were found when participants listened to the
popular music piece in the first position and presented
with low and medium prestige texts. Overall, the
regression tree model confirmed the effects of explicit
information and repeated exposure, but it also showed
higher-level interactions between the extrinsic factors
and the two pieces of music. None of the individual
difference factors used in the model (music training,
familiarity with the original piece, music preferences,
neuroticism, and openness) were significant in the
regression tree model. This finding suggests that after
participants had fallen for the illusion, individual dif-
ference factors did not play an important role and
musical judgments were mainly influenced by the
extrinsic factors.
The present study focussed on extrinsic factors in
order to examine differences in musical judgments
when the acoustic input remains the same. Nevertheless,
one could argue that the factors of explicit information
and repeated exposure might also be responsible, in
part, for the occurrence of the illusion. The results from
a nonprestige group, where the effect of explicit infor-
mation was not manipulated, indicated that 75%parti-
cipants were susceptible to the illusion. This finding
suggests that the effect of explicit information is not
essential for the occurrence of the illusion. By contrast,
we consider it likely that the effect of repeated exposure
104 Manuel Anglada-Tort & Daniel Mu¨llensiefen

contributes to the illusion. In an extensive investigation
of repetition in musical experience, Margulis (2014)
provides relevant insights to this matter. She stated that,
‘‘[a]t a minimum, a repeated element will sound differ-
ent from its initial presentation by virtue of coming later
and having been heard before’’ (p. 35). Although in this
quote Margulis refers to repetition within individual
pieces of music, we find it plausible that the same prin-
ciple should apply to the repeated recording illusion:
while the musical input remains the same, repeated
exposure modifies the listening experience, giving rise
to the feeling that the performances are different.
Two relevant questions arise from the results of this
study. Why are some individuals more susceptible to the
illusion than others? One way to approach this question
is the study of further individual difference factors (e.g.,
intelligence, memory, perceptual abilities) that may be
associated with the repeated recording illusion. The sec-
ond question refers to a more fundamental issue: did
participants in this study actually perceive differences
between the repetitions of the same recording? Or, alter-
natively, did they believe they heard differences because
they were misled to think so? We encourage the use of
neuroimaging techniques as one possible approach to
investigate whether the illusion is a perceptual phenom-
enon or rather a bias in a secondary and later stage of
cognitive processing and decision-making.
Taking a wider perspective, the research framework
developed by Tversky and Kahneman (Kahneman &
Tversky, 1984; Tversky & Kahneman, 1974; see Kahne-
man, 2011, for a review) could provide a theoretical
framework by which the results of the current study
could be interpreted. Although it does not involve music
and is mainly concerned with economic decision pro-
cesses, Tversky and Kahneman’s framework offers
insight into how to investigate traditional psychological
biases in musical judgments by using recent research on
human judgments and decision-making. However, this
framework has not yet been applied explicitly to the
study of evaluative judgment processes involving music.
The effect of explicit information may fall within
a broad heuristic principle, namely, the affect heuristic
(Kahneman & Frederick, 2002; Slovic, Finucane, Peters,
& MacGregor, 2002), which refers to the reliance on
good or bad feelings experienced in relation to a stimu-
lus. Thus, if the emotions associated with a stimulus are
positive, people will be more likely to judge character-
istics of the pertinent stimulus more positively, as found
in the present study when the music recording was
presented with a high prestige text. Similarly, the effect
of repeated exposure is one of several mechanisms
within the bias of perceptual fluency (see Kahneman,
2011, for a review), which has been widely shown to
influence human judgments and decision-making in
many areas (see Reber, Schwarz, & Winkielman, 2004,
for a review). Such findings suggest that perceptual flu-
ency gives rise to feelings of familiarity and a positive
affective response that results in an increase in prefer-
ence judgments. In the present study, this is evident only
when participants listened to the more familiar popular
music recording.
Our results suggest that at least in certain situations,
evaluations of music rely on judgment biases and heur-
istics that do not depend on the stimuli themselves,
which is in line with models of decision-making and
the research framework developed by Tversky and Kah-
neman. However, when applying Tversky and Kahne-
man’s framework to the study of evaluative and
judgment processes involving music, one should con-
sider the implications and difficulties of using music as
stimuli (e.g., familiarity, complexity, presence of vocals,
individual preferences to music, personality). This
approach wherein biases in musical judgments are
linked to comparable research in behavioral economics
could be used to investigate and better understand
musical judgments, preferences, and choice behavior.
This general approach, that could be termed the behav-
ioral economics of music, would attempt to create a solid
understating of the role that behavioral economics can
play in the study of musical judgments and preferences,
two fields that have been surprisingly unconnected in
the literature so far.
In summary, the findings of the present study show
that most participants believed that they had heard dif-
ferent musical performances when in fact they were
identical. This illusion occurred regardless of partici-
pants’ levels of suggestibility, music training, and pre-
ferences for music style. However, high levels on the
personality traits of neuroticism and openness made it
significantly more likely that an individual would fall for
the illusion. While the explicit information presented
with the music influenced participants’ evaluations of
music significantly, the effect of repeated exposure
affected participants’ ratings only in the more familiar
popular music recording. These findings support previ-
ous research showing that musical judgments are some-
times not based on musical cues and features but are
influenced by factors that do not depend on the music
itself. Beyond the findings and limitations of the present
research, the repeated recording illusion can constitute
a useful paradigm for investigating psychological biases
and individual differences in aesthetic and musical
judgments because the illusion allows for the study of
their effects while the music remains the same.
The Repeated Recording Illusion 105

Author Note
Manuel Anglada-Tort is now at Department of Audio
Communication, Technische Universita¨t Berlin, Berlin,
Germany.
Correspondence concerning this article should be
addressed to Manuel Anglada-Tort, Department of
Audio Communication, Technische Universita¨t Berlin,
Berlin, Germany. E-mail: m.angladatort@campus.tu-
berlin.de
References
BATE S,D.,MA
¨CHLER,M.,BOLKER,B.,&WALKER, S. (2015).
Fitting linear mixed-effects models using lme4. Journal of
Statistical Software, 67(1), 1-48. DOI: 10.18637/jss.v067.i01
BEHNE, K.-E. (1987). Urteile und Vorurteile: Die
Alltagsmusiktheorien jugendlicher Ho¨rer [Judgements and
prejudices: The musical everyday theories of adolescent
listeners]. In H. Motte-Haber (Ed.), Psychologische
Grundlagen des Musiklernens (pp. 221-272). Kassel, DE:
Ba¨renreiter.
BEHNE,K.-E.,&WO
¨LLNER, C. (2011). Seeing or hearing the
pianists? A synopsis of an early audiovisual perception
experiment and a replication. Musicae Scientiae,15,324-342.
DOI: 10.1177/1029864911410955
BREIMAN, L. (2001). Random forests. Machine Learning,45,5-32.
DOI: 10.1023/A:1010933404324
CASSIDY,J.W.,&SIMS,W.L.(1991).Effectsofspecialeducation
labels on peers’ and adults’ evaluations of a handicapped youth
choir. Journal of Research in Music Education,39,23-34.DOI:
10.2307/3344606
CAVI T T, M. E. (1997). Effects of expectations on evaluators’
judgments of music Performance. Texas Mu si c E d u c a t i o n
Research (pp. 7-10). Retrieved from https://www.tmea.org/
assets/pdf/research/Cav1997.pdf
CAVI T T, M. E. (2002). Differential expectation effects as factors in
evaluations and feedback of musical performance. Te x a s M u s i c
Education Research (pp. 2-6). Retrieved from https://www.
tmea.org/assets/pdf/research/Cav2002.pdf
DELIE
`GE, I. (1987). Grouping conditions in listening to music:
An approach to Lerdahl and Jackendoff’s grouping preference
rules. Music Perception, 4,325-60.
DOWLING, W. J. (1978). Scale and contour: Two components of
a theory of memory for melodies. Psychological Review, 85,
341-354. DOI: 10.1037/0033-295X.85.4.341
DOWLING,W.J.,&BARTLETT, J. C. (1981). The importance of
interval information in long-term memory for melodies.
Psychomusicology,1,30-49.
DUERKSEN, G. L. (1972). Some effects of expectation on evalua-
tion of recorded musical performance. Journal of Research in
Music Education,20,268-272.DOI:10.2307/3344093
ELLIOTT, C. A. (1995). Race and gender as factors in judgments
of musical performance. Bulletin of the Council for Research in
Music Education, 127, 50-56.
FIELD,A.(2013).Discovering statistics using IBM SPSS statistics.
London, UK: Sage.
GREASLEY,A.,&LAMONT, A. (2016). Musical preferences. In S.
Hallam, I. Cross, & M. Thaut (Eds.), Oxford handbook of music
psychology (2nd ed., pp. 263-281). Oxford, UK: Oxford
University Press.
GRIFFITHS, N. K. (2008). The effects of concert dress and physical
appearance on perceptions of female solo performers. Musicae
Scientiae,12,273-290.DOI:10.1177/102986490801200205
GUDJONSSON,G.H.(2003).The psychology of interrogations and
confessions: A handbook. West Sussex, UK: John Wiley & Sons.
HALPERN.A.R.,BARTLETT,J.,&DOWLING, W. (1995). Aging
and experience in the recognition of musical transpositions.
Psychology and Aging,10,325-342.DOI:10.1037/0882-
7974.10.3.325
HASTIE,T.,TIBSHIRANI,R.,&FRIEDMAN,J.(2009).Hierarchical
clustering. In T. Hastie, E. Tibshiran, & J. Friedman (Eds.), The
elements of statistical learning: Data mining, inference and
prediction (2nd ed., pp. 520-528). New York: Springer.
HOTHORN,T.,BUEHLMANN,P.,DUDOIT,S.,MOLINARO,A.,&
VAN DER LAAN,M.(2006).Survivalensembles.Biostatistics, 7,
355-373. DOI: 10.1093/biostatistics/kxj011
HOTHORN,T.,HORNIK,K.,&ZEILEIS, A. (2006). Unbiased
recursive partitioning: A conditional inference framework.
Journal of Computational and Graphical statistics,15,651-674.
DOI: 10.1198/106186006X133933
HUTCHESON,G.D.,&SOFRONIOU, N. (1999). The multivariate
social scientist. Introductory statistics using generalized linear
models.DOI:10.4135/9780857028075
JANITZA,S.,STROBL,C.,&BOULESTEIX,A.–L.(2013).AnAUC-
based permutation variable importance measure for random
forests. BMC Bioinformatics, 14,119.DOI:10.1186/1471-
2105-14-119
JOHN,O.P.,&SRIVASTAVA, S. (1999). The Big Five trait taxon-
omy: History, measurement, and theoretical perspectives.
Handbook of Personality: Theory and Research,2,102-138.
JUCHNIEWICZ, J. (2008). The influence of physical movement on
the perception of musical performance. Psychology of Music,
36,417-427
KAHNEMAN,D.(2011).Thinking, fast and slow.NewYork:Farrar,
Straus and Giroux.
KAHNEMAN,D.,&FREDERICK, S. (2002). Representativeness
revisited: Attribute substitution in intuitive judgment. In T.
Gilovich, D. Friffin, D. Kahneman (Eds.), Heuristics and biases:
The psychology of intuitive thought (pp. 49-81). New York:
Cambridge University Press.
106 Manuel Anglada-Tort & Daniel Mu¨llensiefen

KAHNEMAN,D.,&TVERSKY, A. (1984). Choices, values, and
frames. American Psychologist,39,341-349.DOI:10.1037/
0003-066X.39.4.341
KAPTEIN,M.,DERUYTER,B.,MARKOPOULOS,P.,&AARTS,E.
(2012). Adaptive persuasive systems: A study of tailored per-
suasive text messages to reduce snacking. ACM Transactions
on Interactive Intelligent Systems,2(2), 1-25. DOI: 10.1145/
2209310.2209313
KIRK,U.,SKOV,M.,HULME,O.,CHRISTENSEN,M.S.,&ZEKI,S.
(2009). Modulation of aesthetic value by semantic context: An
fMRI study. NeuroImage,44(3), 1125-1132. DOI: 10.1016/
j.neuroimage.2008.10.009
KROGER,C.,&MARGULIS, E. H. (2016). ‘‘But they told me it was
professional’’: Extrinsic factors in the evaluation of musical
performance. Psychology of Music,45,49-64.DOI:10.1177/
0305735616642543
KUZNETSOVA,A.,BROCKHOFF,P.B.,&CHRISTENSEN R. H. B.
(2016). lmerTest: Tests for random and fixed effects for linear
mixed effect models. RPackageVersion.DOI:http://CRAN.R-
project.org/package¼lmerTest
MARGULIS,E.H.(2010).Whenprogramnotesdon’thelp:Music
descriptions and enjoyment. Psychology of Music, 38,285-302.
MARGULIS,E.H.(2014).On repeat: How music plays the mind.
New York: Ocford University Press.
MARGULIS,E.H.,KISIDA,B.,&GREENE, J. P. (2015). A knowing
ear: The effect of explicit information on children’s experience
of a musical performance. Psychology of Music,43,596-605.
DOI: 10.1177/0305735613510343
MILGRAM, S. (1963). Behavioral study of obedience. Journal of
Abnormal Psychology,67,371-378.DOI:10.1037/h0040525
MU
¨LLENSIEFEN,D.,GINGRAS,B.,MUSIL,J.,&STEWART,L.
(2014). The musicality of non-musicians: An index for asses-
sing musical sophistication in the general population. PloS
ONE, 9,e89642.DOI:10.1371/journal.pone.0089642
NORTH,A.C.,&HARGREAVES, D. J. (2005). Brief report:
Labelling effects on the perceived deleterious consequences of
pop music listening. Journal of Adolescence,28,433-440.DOI:
10.1016/j.adolescence.2004.09.003
NORTH,A.,&HARGREAVES, D. (2008). The social and applied
psychology of music.NewYork:OxfordUniversityPress.
ORSMOND,G.I.,&MILLER,L.K.(1999).Cognitive,musical
and environmental correlates of early music instruction.
Psychology of Music,27, 18-37. DOI: 10.1177/03057356
99271003
PAWL E Y,A.,&MU
¨LLENSIEFEN,D.(2012).Thescienceofsinging
along: A quantitative field study on sing-along behavior in the
north of England. Music Perception, 30,129–146.DOI:
10.1525/mp.2012.30.2.129
PEARCE,M.T.(2015).Effectsonprocessesinvolvedinmusical
appreciation. In J. P. Huston, M. Nadal, F. Mora, L. Agnati, F.
Mora, & C. J. Cela-Conde (Eds.), Art, aesthetics and the brain
(pp. 319-338). Oxford, UK: Oxford University Press.
RADOCY, R . E. (1976). Effects of authority figure biases on
changing judgments of musical events. Journal of Research in
Music Education,24,119-128.DOI:10.2307/3345155
REBER,R.,SCHWARZ,N.,&WINKIELMAN,P.(2004).Processing
fluency and aesthetic pleasure: Is beauty in the perceiver’s
processing experience? Personality and Social Psychology
Review,8,364-382.DOI:10.1207/s15327957pspr0804_3
RENTFROW,P.J.,&GOSLING, S. D. (2003). The do re mi’s of
everyday life: The structure and personality correlates of music
preferences. Journal of Personality and Social Psychology,84,
1236-1256. DOI: 10.1037/0022-3514.84.6.1236
SILVEIRA,J.M.,&DIAZ, F. M. (2014). The effect of subtitles on
listeners’ perceptions of expressivity. Psychology of Music,42,
233-250.
SILVEY, B. A . (2009). The effects of band labels on evaluators’
judgments of musical performance. Update: Applications of
Research in Music Education,28(1), 47-52. DOI: 10.1177/
8755123309344111
SLOVIC,P.,FINUCANE,M.,PETERS,E.,&MACGREGOR,D.G.
(2002). Rational actors or rational fools: Implications of the
affect heuristic for behavioral economics. Journal of Socio-
Economics,31,329-342.DOI:10.1016/S1053-5357(02)00174-9
STO
¨BER, J. (2001). The Social Desirability Scale-17 (SDS-17):
Convergent validity, discriminant validity, and relationship
with age. European Journal of Psychological Assessment,17,
222-232. DOI: 10.1027//1015-5759.17.3.222
STROBL,C.,BOULESTEIX,A.-L.,KNEIB,T.,AUGUSTIN,T.,&
ZEILEIS, A. (2008). Conditional variable importance for ran-
dom forests. BMC Bioinformatics, 9, 307. DOI: 10.1186/1471-
2105-9-307
STROBL,C.,MALLEY,J.,&TUTZ, G. (2009). An introduction to
recursive partitioning: Rationale, application, and characteris-
tics of classification and regression trees, bagging, and random
forests. Psychological Methods,14,323–348.DOI:10.1037/
a0016973
TVERSKY,A.,&KAHNEMAN, D. (1974). Judgment under uncer-
tainty: Heuristics and biases. Science,185,1124-1131.DOI:
10.1126/science.185.4157.1124
UNAL,P.,TEMIZEL,T.T.,&EREN,P.E.(2014).Anexploratory
study on the outcomes of influence strategies in mobile
application recommendations. Proceedings of the Second
International Workshop on Behavior Change Support Systems
(pp. 27-40). Padova, Italy: BCSS.
VUOSKOSKI,J.K.,&EEROLA, T. (2013). Extramusical informa-
tion contributes to emotions induced by music. Psychology of
Music,43,262–274.DOI:10.1177/0305735613502373
ZAJONC, R. B. (1968). Attitudinal effects of mere exposure.
Journal of Personality and Social Psychology,9,1-27.DOI:
10.1037/h0025848
ZEILEIS,A.,HOTHORN,T.,&HORNIK , K. (2008). Model-based
recursive partitioning. Journal of Computational and Graphical
Statistics,17,492-514.DOI:10.1198/106186008X319331
The Repeated Recording Illusion 107

Appendix A
Prestige Texts (Low, Medium, and High) Used in the
Popular Music Condition (‘‘Jailhouse Rock’’ by Elvis
Presley)
Popular Music Condition: Low Prestige Text – Larry
Leigh
Larry Leigh was a humble truck driver who developed
an obsessive love for Elvis Presley from an early age. His
lack of musical training did not hold him back from
impersonating his beloved star on stage.
Although Leigh participated in the Georgia Elvis Festival
and the ETA Preliminary Competition, he never succeed
in his career as an Elvis imitator and his critics labelled
him as an amateur singer.
Leigh had some similarities to Elvis’ voice, however, due
to the lack of success he gave up his dream. A close
friend expressed that his unsuccessful career as an Elvis
impersonator made him spiral into a deep depression.
Information source: Impersonators in America (2001)
by Esther Newton
Popular Music Condition: Medium Prestige Text – Drew
Polsun
Drew Polsun was not just an owner of a music record
store, but also a lifelong lover and impersonator of Elvis.
With some musical training from an early age, he began
to perform as Elvis in his local area.
Putting his business on hold he began to compete more
seriously in Elvis impersonator contests. At the Ultimate
Elvis Tribute Artist Contest he placed 17th out of
roughly 40 impersonators.
Never able to reach the top, he turned back to his busi-
ness at the music store and focused on his family. Still to
this day, he continues to perform in his hometown shar-
ing his love for the King.
Information source: Impersonating Elvis (2009) by
Leslie Rubinkowski
Popular Music Condition: High Prestige Text – Shawn
Klush
Starting at the age of 6, Shawn Klush would sing and
dance like Elvis. He now works as a professional actor
and entertainer, and has a strong musical background,
from classical to jazz music, in singing and guitar.
Klush became very successful as one of the top Elvis
impersonators. He released 3 CDs, became the grand
champion at the $150,000 World Elvis Tribute Artist
Competition, and received the Heart of the Kind, World-
wide Ambassador of Elvis Award.
In 2007, he was named the World’s Greatest Elvis by 6.5
million international viewers on BBC1 Television in
the United Kingdom. Since then, Klush is considered
one of the world’s most professional Elvis Tribute
Artists.
Information source: www.shawnklush.com
Appendix B
Prestige Texts (Low, Medium, and High) used in the
Classical Music Condition (Bruckner Symphony No. 4)
Classical Music Condition: Low Prestige Text – Kurt
Schlichter
With an undergraduate degree from Royal Holloway,
University of London, Kurt Schlichter, continued his
education as a Masters student in conducting at the
Royal College of Music.
As a young student, Kurt found it quite challenging to
conduct a broad range of classical works in real-life
rehearsal and performance situations. But these experi-
ences allowed him to grow and become one of the top
students in his class.
Talking to Kurt about his experiences he says, ‘‘The
opportunities at the Royal College have been highly
varied and rewarding. For my final project I was able
to conduct one of my favourite Bruckner pieces which
allowed me to engage fully with the complex composi-
tional style of this genius composer’’.
Information source: www.rcm.ac.uk/conducting/story/
kurt
Classical Music Condition: Medium Prestige Text – Pablo
Gime
´nez
As the principal conductor of the Royal Seville Sym-
phony Orchestra, Pablo Gime
´nez, put Spain on the map
as a place for beautiful interpretations of classical music.
Based in Seville, his orchestra has played some very well
know interpretations of classical favourites.
Gimenez had become an up-and-coming conductor,
performing Carmen at the Liceu’s theatre of Barcelona
and international concert halls in China and Australia.
108 Manuel Anglada-Tort & Daniel Mu¨llensiefen

His album of Zarzuela has been released throughout
Europe, but did not receive any awards.
Even though his interpretations of works by Brahms
and Handel have been regarded as ‘‘beautiful orchestral
performances’’, his interpretations of Beethoven and
Wagner were labelled as ‘‘lifeless and inexpressive’’ by
some reviewers.
Information source: Guide to Spanish Conductors
(2010) by Jose Luis-Garcia
Classical Music Condition: High Prestige Text – Claudio
Abbado
Claudio Abbado was widely considered one of the great-
est conductors of the 20
th
century. He served as princi-
pal conductor of the Berlin Philharmonic and London
Symphony Orchestra, one of the best ensembles in the
world.
Claudio’s career couldn’t be more successful. He won
the International Competition for Composers and the
Grammy Award in the Best Small Ensemble Perfor-
mance. Additionally, the Orchestra Academy of the Ber-
lin Philharmonic established the Claudio Abbado
Composition Prize in his honour.
In 2004, Abbado conducted the Berlin Philharmonic to
performed Bruckner’s Symphony No. 4 in a series of
recorded live concerts. The resulting CD won Best
Orchestral Recording of the Year in Gramophone
awards.
Information source: Stories of the Great Contemporary
Conductors (2012) by Maurice Hinson
Appendix C
Decision Diagram of the Procedure used to Determine Whether Participants Fell for the Repeated Recording Illusion
The Repeated Recording Illusion 109

Appendix D
Variable Importance Scores for the 17 Variables
Variable Importance Scores for Predictor Variables in Random Forest Model using Strict Criterion (i.e., Participants
Falling for the Illusion in Both Music Conditions)
Variable Importance Scores for Predictor Variables in Random Forest Model using Less Strict Criterion (i.e.,
Participants Falling for the Illusion in One Music Condition)
110 Manuel Anglada-Tort & Daniel Mu¨llensiefen

Appendix E
Summary Tables of the Two Linear Mixed-Effects Mod-
els (Popular Music and Classical Music)
Popular Music Condition
Sum of
Squares df F p value
Explicit Information (EI) 7.89 2 7.79 < .001***
Repeated Exposure (RE) 17.42 2 17.20 < .001***
CI * RE 1.34 4 .66 .62
Classical Music Condition
Sum of
Squares df F p value
Explicit Information (EI) 12.61 2 10.66 < .001***
Repeated Exposure (RE) .23 2 .19 .82
CI * RE 3.96 4 1.67 .16
Appendix F
Criteria and Information Used to Determine Whether
Participants Fell for the Repeated Recording Illusion in
the Popular Music Condition
Legend:
Decision.
Yes - Participant is considered as falling for the illusion.
No - Participant is not considered as falling for the
illusion.
Why? - Justification of the decision made:
•A(when decision ¼yes): The information from
the open-text boxes indicates specifically any dif-
ferences between performances.
•B(when decision ¼yes): The information from the
open-text boxes is not sufficient but suggests that
the participant was not aware that the recordings
were identical. In addition, at least one score from
the final comparative rating scales differ by 10%
from the midpoint of the scale (‘50’), or any two
scores differ by 10%form each other.
•C(when decision ¼no): The information from the
open-text boxes indicates specifically that the par-
ticipant realized that the recordings were the
same.
•D(when decision ¼no): The information from the
open-text boxes is not sufficient but suggests that
the participant suspected that the performances
were the same. In addition, none of the three
scores from the final comparative rating scales
differ more than 10%from the midpoint of the
scale (‘50’).
•Eliminated: The information from the open-text
boxes and final comparative rating scales are not
enough and/ or too ambiguous to make a clear
objective decision based on the above criteria.
N"Information from open-text boxes
Comparative
ratings
(when needed) Decision Why?
1Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
55 75 35 YES B
2 ‘‘some parts sounded a bit too fast compared to the previous version’’ – – – YES A
3Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
25 75 40 YES B
4 ‘‘I have a feeling that it was the same song as before’’ – – – NO C
5Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
74 70 35 YES B
6 ‘‘I think that in this case, the music invites more to move. The voice is deeper and
maybe rougher’’/ ‘‘Yes I did. The voice was more similar to the original song and I
like it.’’ vs. ‘‘In this case apart that the voice was quite different to the original one.
The drums seemed to be quieter than in the original version.’’
– – – YES A
7 ‘‘the sound quality was much better and the tambourine was much more prominent’’/
‘‘I enjoyed this performance much more than the first as it was more upbeat’’
– – – YES A
8Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
61 42 85 YES B
9Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
81 35 66 YES B
(continued)
The Repeated Recording Illusion 111

Appendix F. (continued)
N"Information from open-text boxes
Comparative
ratings
(when needed) Decision Why?
10 ‘‘I thought this version sounded more like the original by Elvis sounds like’’/ ‘‘I enjoyed
this performance of the song more than the first one’’
– – – YES A
11 ‘‘To me, this interpretation was not as emotionally charged’’/ ‘‘I found Polsun’s voice to
sound quite strained and raspy and not as melodic and smooth as the original
version’’
– – – YES A
12 ‘‘the emotion put into the song seemed to have a personal quality and was sung with
more affection’’/ ‘‘whilst maybe technically better this version [ ...] seemed to be
expressed in lines with less emotion and more inclination to mimic Elvis’’
– – – YES A
13 ‘‘Would be money it’s all the same tune’’ – – – NO C
14 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
62 79 59 YES B
15 ‘‘perhaps the band was not as good as the in the last one i.e., the second’’ / ‘‘It was
helped by the good supporting band, who were better in quality to the band playing
in the first piece. He lack a little more in passion than the first one, so while it was
perhaps technically better, it could have had a little more emotion’’
– – – YES A
16 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
40 50 60 YES B
17 ‘‘I didn’t feel that the instrumentation was quite as tight in this clip, however, with the
bass being looser and the piano not as prominent’’
– – – YES A
18 ‘‘Still an enjoyable clip, this singer did his best to inject the passion and colour into
Jailhouse Rock as Elvis’’/ ‘‘it was easier to hear the different instruments within the
band, including piano and sax, and the band were very tight’’
– – – YES A
19 It cannot be determined whether the participant fell or not for the illusion, but it seems
that he/she suspected that the recordings were the same: ‘‘Was it the same as the last
performer?’’
50 50 50 NO D
20 ‘‘with this one (singer) I must say is even higher’’/ ‘‘I really enjoyed his interpretation
probably because his voice was more accurate to Elvis’s’’
– – – YES A
21 ‘‘The vocal quality of the interpretation wasn’t as good for this one’’/’’I didn’t enjoy this
performance as much as the others, mainly due to his vocal quality ‘‘/’’This version
had much more energy that the previous one and it felt that the performer was
having much more fun’’
– – – YES A
22 ‘‘I enjoyed it because it seemed more fun and playful in a way. Possibly I was also more
friendly minded towards this guy after learning about his tragic history’’
– – – YES A
23 ‘‘I have been able to rate easily each interpretation’’/ ‘‘The voice of the singer is very
poor, he can’t get a good imitation’’ vs. ‘‘The singer has made a wonderful
performance’’
– – – YES A
24 ‘‘Larry Leigh’s performance reminded me of Polsun’s. When I heard Klush’s
performance I realized that it was the same as the previous one’’
– – – NO C
25 Information is not enough and/or ambiguous ––– Eliminated
26 ‘‘I’m starting to suspect these are all the same version!!’’ – – – NO C
27 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
74 44 24 YES B
28 ‘‘It is a little difficult to judge which one I preferred the most since they are the exact
same clip’’
– – – NO C
29 ‘‘Although perhaps not as melodically varied, the singer’s tone and timbre was very
similar to the original, and really good to listen to’’/ ‘‘the rhythm section was fairly
average’’ vs. ‘‘The tempo felt almost exactly right (perhaps marginally too fast?), and
the rhythm section was really strong’’
– – – YES A
30 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
44 85 50 YES B
31 ‘‘I thought the little vocal impros he threw in were what made him slightly better’’ – – – YES A
32 ‘‘This performance of ‘Jailhouse’ is the closest to the original [ ...] Aurally, the
performance sounds more professional’’
– – – YES A
33 ‘‘it seemed to me that each performance was the same one performance, which thereby
made it difficult to ‘interpret’ each of the them’’
– – – NO C
(continued)
112 Manuel Anglada-Tort & Daniel Mu¨llensiefen

Appendix F. (continued)
N"Information from open-text boxes
Comparative
ratings
(when needed) Decision Why?
34 ‘‘ the vocals were not as professional as on the other two performances’’/ ‘‘A more
pationate performance. It was definitely more rock’n’roll than the previous
performance’’
– – – YES A
35 ‘‘seemed more in sync with the band and with the audience than the previous
performance’’/ ‘‘I liked this performance because he spent more time on his words
than the previous performance, and sounds more confident than the last’’
– – – YES A
36 ‘‘I reckon this is the same file repeated three times. The instrumentation is identical,
the vocal timbres are the same and even the crowd is the same. I’m fairly certain it’s
the same person’’.
– – – NO C
37 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical.
100 90 53 YES B
38 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical.
60 55 70 YES B
39 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical.
59 61 60 YES B
40 ‘‘I enjoyed the overall performance and thought the singer did a pretty decent job’’ vs.
‘‘I did not really enjoy the version. I did not like the vocals - sounded too forced and
tuning issues seemed to jar a little’’
– – – YES A
41 ‘‘Rhythm was better here, helped by the more relaxed tempo’’/’’This was my favourite
by far. The vocal performance wasn’t monotonous, like the other ones’’
– – – YES A
42 ‘‘More upbeat than the two others, a happier sounding performance’’/’’Liked it the best
of the three, the voice of the singer sounded ‘‘cleaner’’ than the other two’’
– – – YES A
43 It cannot be determined whether the participant fell or not for the illusion, but it seems
that he/she suspected that the recordings were the same: ‘‘The three versions sounded
incredibly similar to me’’
50 50 50 NO D
44 ‘‘I like this one a bit better than the two previous ones, even though again it was pretty
similar. I just liked the guys voice more’’
– – – YES A
45 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
79 83 71 YES B
46 Information is not enough and/or ambiguous ––– Eliminated
47 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
10 53 61 YES B
48 ‘‘Similar mood as with the previous two, but I didn’t like his interpretation that much’’ – – – YES A
49 ‘‘I thought it was a decent cover of the original song, but the singers voice although
fairly good, was the weakest element’’/ ‘‘I thought the performance was okay, again
the vocalist and the quality of the recording made me rate it slightly lower’’ vs. ‘‘I
thought the performance was really led and pushed forward by the vocalist which
gave it a good sense of movement and energy’’/ ‘‘I did enjoy the performance of the
song, I liked the quality of the singers voice as I thought it was strong, and fairly
accurate to Elvis’’
– – – YES A
50 ‘‘I believe that the songs were very similar and difficult to distinguish one from the
other. At some point I believed that I was listening at the same song all along.’’
– – – NO C
51 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
80 60 30 YES B
52 ‘‘I think this one was far more true to the original that the others’’/ ‘‘The three
interpretation were different but did not vary that much’’
– – – YES A
53 ‘‘I am not a big fan of interpretations unless they do something original, and neither of
these did. Despite that, I could tell the qualitative difference’’
– – – YES A
54 ‘‘I feel like he was shouting rather than singing, so this overall wasn’t enjoyable to me
musically’’/ ‘‘This was definitely my favorite because of his energy and vocal
dynamics’’
– – – YES A
55 Information is not enough and/or ambiguous ––– Eliminated
56 ‘‘The singer couldn’t quite control his voice as Elvis could, yet there was an emotion in it’’
vs. ‘‘His voice had the gravelly quality of Elvis’s. The instrumentation was more taught
too’’/ ‘‘altogether it was smoother which I don’t think was necessarily a good thing’’
– – – YES A
(continued)
The Repeated Recording Illusion 113

Appendix G
Criteria and Information Used to Determine Whether
Participants Fell for the Repeated Recording Illusion in
the Popular Music Condition
Legend:
Decision.
Yes - Participant is considered as falling for the illusion.
No - Participant is not considered as falling for the
illusion.
Why? - Justification of the decision made:
•A(when decision ¼yes): The information from
the open-text boxes indicates specifically any dif-
ferences between performances.
•B(when decision ¼yes): The information from the
open-text boxes is not sufficient but suggests that
the participant was not aware that the recordings
were identical. In addition, at least one score from
the final comparative rating scales differ by 10%
from the midpoint of the scale (‘50’), or any two
scores differ by 10%form each other.
•C(when decision ¼no): The information from the
open-text boxes indicates specifically that the par-
ticipant realized that the recordings were the same.
•D(when decision ¼no): The information from the
open-text boxes is not sufficient but suggests that
the participant suspected that the performances
were the same. In addition, none of the three
scores from the final comparative rating scales
differ more than 10%from the midpoint of the
scale (‘50’).
•Eliminated: The information from the open-text
boxes and final comparative rating scales are not
enough and/ or too ambiguous to make a clear
objective decision based on the above criteria.
Appendix F. (continued)
N"Information from open-text boxes
Comparative
ratings
(when needed) Decision Why?
57 ‘‘Closest of the three to the original ‘Elvis feeling’ in terms of voice.’’/ ‘‘It’s still a cover. It
was closer to the original than the one before’’
– – – YES A
58 ‘‘But, I think this one that the tempo is better than the last one’’/ ‘‘I think this one is the
best of the tree interpretations’’
– – – YES A
59 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
25 57 97 YES B
60 ‘‘For this piece I feel more lively and more energetic. The voice of the singer is quite
affective to me.’’ vs ‘‘For me the singer’s voice is too tender, and in some part I
cannot feel the lively spirit from the music’’
– – – YES A
61 ‘‘This version has been the best of the three songs that I listened’’/ ‘‘For me, the three
songs are a good interpretation of the Elvis original song’s’’
– – – YES A
62 ‘‘I don’t like the tone and rhytym that much’’/ ‘‘yes but the others were better’’ – – – YES A
63 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
57 55 65 YES B
64 ‘‘I liked this performance because the energy of the song was more consistent from the
outset. His voice sounded more naturally husky than shouty’’/ ‘‘The performance
felt a little bit flatter/less exciting’’
– – – YES A
65 ‘‘the overall mood was less fun than in the two previous recordings / music was less
vibrant and I didn’t receive the disco feeling as in the performance 2 / not catchy
enough’’
– – – YES A
66 ‘‘I did not exactly enjoy this performance as the singer, at least to my mind, tried too
much to sound like Elvis’’ vs. ‘‘I enjoyed this performance. The vocal quality of the
singer reminded me strongly of Elvis’’
– – – YES B
67 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
82 24 66 YES B
68 ‘‘this was the least accurate performance, but the emotional impact was really good’’ – – – YES A
69 ‘‘ this is absolutely the same as the first two’’ – – – NO C
70 ‘‘I like this one most. The singer interpretation was better’’/ ‘‘This is my favorite
interpretation. The singer had a lot of emotion and the tempo was perfect for me.’’
– – – YES A
71 ‘‘I honestly couldn’t distinguish any noticeable difference between this recording and
the last one - it sounded the same’’
– – – NO C
72 ‘‘Same performance as last time’’ – – – NO C
114 Manuel Anglada-Tort & Daniel Mu¨llensiefen

N"Information from open-text boxes
Final
comparative
rating Decision Why?
1Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
80 60 35 YES B
2 ‘‘I was the same with the two other ones’’/ ‘‘ It all was the same for me’’ – – – NO C
3Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
75 25 0 YES B
4 ‘‘I liked it, more unity than Pablo’s version (winds)’’/ ‘‘Liked the version, mostly the
strings (compared to the others)’’
– – – YES A
5 ‘‘I could only feel something more with Schlichter’s songs, I don’t know why but it
probably made more emotional impact than the others’’
– – – YES A
6 ‘‘It was much more emotional. The piece was cleaner and synchronized’’/ ‘‘I enjoyed
this performance of the song the most.’’/ ‘‘Again it did not give me the spark that the
second one gave. In this case you could notices the synchronization was not perfect.’’
– – – YES A
7 ‘‘I felt the mood of this piece was much more positive than the last’’/ ‘‘I thought this
performance was the most beautiful. It seemed softer and the sound quality was very
good’’
– – – YES A
8Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
72 20 94 YES B
9Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
25 42 97 YES B
10 ‘‘I enjoyed this performance, but not as much as the performance by Claudio Abbado’’/
‘‘For some reason, the tone and key at times sounded different (more major, less
minor)’’
– – – YES A
11 ‘‘This interpretation of the piece felt a bit more rushed to me than the other two. I felt
that this version didn’t have the same dramatic pauses as the other two’’
– – – YES A
12 ‘‘Initially I felt the intro to this piece had a lack of emotion - in comparison to the
previous piece - there seemed to be flatness to the work’’
– – – YES A
13 ‘‘Would be money that they are all the same recording’’ – – – NO C
14 ‘‘I did not find this song as good as the previous interpretation.’’ – – – YES A
15 ‘‘I preferred it to the first piece, as I felt that the emotional current of the piece moved
a bit quicker ‘‘/’’perhaps it was a little quicker in tempo, perhaps a little too rushed at
times’’
– – – YES A
16 ‘‘not as deep and compacted to the previous one’’ – – – YES A
17 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
30 70 80 YES B
18 ‘‘it perhaps didn’t reach the same levels of quiet nearer the beginning as other
interpretations did to contrast with the loud ending’’
– – – YES A
19 It cannot be determined whether the participant fell or not for the illusion, but it seems
that he/she suspected that the recordings were the same: ‘‘were they all the same? if not
then they were very similar’’
50 50 50 NO D
20 ‘‘With this interpretation of the song I get the feeling that the first one was too slow and
the second one too fast and this one the perfect tempo and pace. I liked it the most’’
– – – YES A
21 ‘‘The mood in this piece seemed escalate a lot more naturally than in the other pieces’’ – – – YES A
22 ‘‘Didn’t feel as strong as the last one’’/ ‘‘I could really find anything I disliked in this
one’’
– – – YES A
23 ‘‘Rating a orchestra has been much more difficult than rating a vocal song. Despite of
this, I have realize about the differences between each interpretations with quite
facility’’
– – – YES A
24 ‘‘I could see only very little, slight differences between the different interpretations’’ – – – YES A
25 ‘‘Exactly the same as the one prior. I can’t differentiate between the two’’/ ‘‘Exactly the
same as the previous two.’’
– – – NO C
26 ‘‘The instruments felt more like they were competing than cooperating in the crescendo
at the end, giving it a more frantic feel. I noticed the brass more than the timpani’’ vs.
‘‘This had the sense of building towards something big from early on’’
– – – YES A
27 ‘‘I enjoyed this one the best. The trumpets and brass at the end didn’t beat me in the
face but rather sounded more like a ray of sunlight through clouds on a rainy day’’
– – – YES A
(continued)
The Repeated Recording Illusion 115

Appendix G. (continued)
N"Information from open-text boxes
Final
comparative
rating Decision Why?
28 ‘‘Again, it’s hard to place an order of which clip I liked the most since it was the same
clip played three times’’
– – – NO C
29 ‘‘I enjoyed the performance of the song, and the interpretation was merely different in
it’s dynamics and style’’
– – – YES A
30 ‘‘I’ve enjoyed Abbado’s piece because it’s strong and with a lot of personality [ ...]
Anyway, I like the performance because even being a bit more chaotic than the Burke
performance it has more variety of instruments in each family’’
– – – YES A
31 ‘‘For me, this interpretation was slightly faster than the others, something which I
found that I enjoyed more than in the other interpretations. The build was faster and
more dramatic’’
– – – YES A
32 ‘‘I enjoyed the arrangement more so in this performance, the instruments were in
harmony and the piece seemed tighter’’/’’The performance was cathartic and
extremely captivating. I can tell it’s a younger interpretation of an older work, the
sound feels fresher’’
– – – YES A
33 Information is not enough and/or ambiguous ––– Eliminated
34 ‘‘This interpretation sounded a bit more hesitant. / Again, it was not as dramatic as the
first performance, but it was clearer than the second one. /Maybe this conductor is
a lot younger than the previous ones’’
– – – YES A
35 ‘‘I thought all 3 were the same’’ – – – NO C
36 ‘‘Some of the timing was slightly scrappier and the and the dynamics were less varied.
Brass section were very present, as were violins.’’ vs. ‘‘It felt like this performance was
a bit more mechanical and less emotional’’/ ‘‘I enjoyed it, but it lacked the fluidity of
the Berlin version.’’
– – – YES A
37 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
83 95 100 YES B
38 ‘‘I know I enjoyed the first one a lot more than the other two’’/ ‘‘The song is beautiful
but I can’t help but feel like this particular performance was not as emotive as it
could have been’’
– – – YES A
39 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
78 90 67 YES B
40 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
40 60 59 YES B
41 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
57 36 95 YES B
42 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
60 32 84 YES B
43 It cannot be determined whether the participant fell or not for the illusion, but it seems
that he/she suspected that the recordings were the same: ‘‘Again, I found each version
to be incredibly similar, so found it difficult to distinguish aspects’’
53 54 52 NO D
44 ‘‘This sounds exactly like the two others’’ – – – NO C
45 ‘‘It felt like more drama and more action, it also felt a bit fast’’/ ‘‘I don’t know why, but I
liked the middle on best’’
– – – YES A
46 Information is not enough and/or ambiguous ––– Eliminated
47 ‘‘I enjoyed it, although it wasn’t my favourite interpretation. I felt it was a little ‘darker’
in tone colour/timbre and felt that the dynamics and tempo could have been more
dramatic’’
– – – YES A
48 ‘‘stimulates strong and deep feelings, perhaps more spiritual, and contemplation’’/ ‘‘I
liked it slightly better than the previous one’’
– – – YES A
49 ‘‘The performance had some weak elements, the overall balance of the recording had
some inadequacies’’/ ‘‘The performance was ok’’ vs. ‘‘I thought the performance was
strong and there was a good tonal quality to the instruments and natural balance
between them’’/ ‘‘I liked the performance of the song, I think it was performed well
and it had an effective emotional impact’’
– – – YES A
50 ‘‘I still think that is the same song interpreted by the same person all along.’’ – – – NO C
(continued)
116 Manuel Anglada-Tort & Daniel Mu¨llensiefen

Appendix G. (continued)
N"Information from open-text boxes
Final
comparative
rating Decision Why?
51 ‘‘This interpretation felt much more agitated throughout, as a result of the faster
tempo’’/ ‘‘Though less ‘brassy’ (as compared to the Seville Orchestra), this lowered
the intensity a little bit.’’
– – – YES A
52 ‘‘I didn’t quite enjoy this performance as much as the last one. Where Gimenez version
was determined and executed with clear intentions of the mood it wanted to set,
swelling with emotional intensity, Schlichter’s execution lacked purpose’’/
‘‘Potentially the most professionally executed. However, I feel like it lacked emotion
compared to Gimenez’’
– – – YES A
53 ‘‘This had so many layers, I would say a great deal of details in the dynamic’’/ ‘‘It was
like this one had more details than the first one’’/ ‘‘It kind of felt like it was longer’’
– – – YES A
54 ‘‘The mood in this one was more anxious rather than emotionally empowering. I felt
that the increased volume and intensity made me nervous, which I didn’t like that
much’’ vs. ‘‘This was an emotional interpretation, which pure feeling behind the
entire piece’’
– – – YES A
55 Information is not enough and/or ambiguous ––– Eliminated
56 ‘‘It seemed more tame than the other versions’’/ ‘‘some were more emotional than
others’’/ ‘‘I’m not quite so sure what to look for, but some were more emotional than
others.’’
– – – YES A
57 ‘‘I can’t say what the difference is, but I liked this a little better than the other ones’’/ ‘‘ I
think it made me less nervous. There was less of the war’’
– – - YES A
58 ‘‘I think this one make me feel the key is not really clear’’/ ‘‘I think this final part of this
one is good ending’’
– – – YES A
59 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
28 0 49 YES B
60 ‘‘This piece sounds more aggressive than the previous one. The tempo for me is faster’’ – – – YES A
61 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
8 49 70 YES B
62 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
29 59 85 YES B
63 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
59 47 63 YES B
64 ‘‘I enjoyed this performance because it felt like a smoother performance than the last.
And the final note wasn’t as harsh so it felt like a more natural conclusion’’
– – – YES A
65 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
77 50 40 YES B
66 ‘‘I feel completely neutral about this performance. It was neither uplifting or
depressing’’ vs. ‘‘It was an enjoyable piece of music, though I can not explain why’’
– – – YES A
67 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
73 65 82 YES B
68 Information not sufficient but seems to indicate that the participant was not aware that
the recordings were identical
100 0 48 YES B
69 ‘‘I preferred this one much more than the last. I thought that the tone qualities of the
soprano voices were better balanced, which really made the brass sound more
effortless’’
– – – YES A
70 ‘‘I enjoyed but not as much as the others. I felt that there was no emotion on the
beginning of the song’’
– – – YES A
71 ‘‘Although I preferred performance 1, I enjoyed all of the performances to some extent
because I felt that, in many ways, each conductor made similar decisions, such as the
choice of tempo and dynamic range’’/’’this one felt a bit more’’
– – – YES A
72 ‘‘It is the exact same interpretation’’/ ‘‘Same as the last interpretation’’ – – – NO C
The Repeated Recording Illusion 117