Psychophysiology, 42 (2005), 465472. Blackwell Publishing Inc. Printed in the USA.

Copyright r 2005 Society for Psychophysiological Research

DOI: 10.1111/j.1469-8986.2005.00306.x

Absolute pitch and pupillary response: Effects of timbre

and key color

KATHRIN B. SCHLEMMER, FRANZISKA KULKE, LARS KUCHINKE,

and ELKE VAN DER MEER
Department of Psychology, Humboldt University at Berlin, Germany

Abstract
The pitch identication performance of absolute pitch possessors has previously been shown to depend on pitch range,
key color, and timbre of presented tones. In the present study, the dependence of pitch identication performance on
key color and timbre of musical tones was examined by analyzing hit rates, reaction times, and pupillary responses of
absolute pitch possessors (n 5 9) and nonpossessors (n 5 12) during a pitch identication task. Results revealed a
significant dependence of pitch identication hit rate but not reaction time on timbre and key color in both groups.
Among absolute pitch possessors, peak dilation of the pupil was significantly dependent on key color whereas the effect
of timbre was marginally significant. Peak dilation of the pupil differed significantly between absolute pitch possessors
and nonpossessors. The observed effects point to the importance of learning factors in the acquisition of absolute pitch.
Descriptors: Absolute pitch, Pupillary response, Pitch memory, Pitch learning

Absolute pitch is the ability to name or produce tonal pitches
without the use of an external reference such as the piano or
tuning fork. With an incidence between 0.01% and 0.10% in the
general population and around 5% among professional musi-
cians, this ability is comparatively rare. In contrast to relative
pitch, which is the ability to identify relations or intervals be-
tween pitches, absolute pitch is not required for professional
musicians, although the fact that famous composers like Bach,
Mozart, or Beethoven are assumed to have had absolute pitch
has led to an idealization of absolute pitch possessors as highly
gifted musicians. This study examines the inuence of key color
(i.e., whether a pitch corresponds to black or white piano keys)
and timbre (i.e., the instrumental sound in which a pitch is pre-
sented) of presented pitches on error rates, reaction times, and
pupillary responses of absolute pitch possessors and nonposses-
sors during pitch identication.
Characteristics of Pitch Identication
Relative pitch judgments are assumed to be based on working
memory (Baddeley & Hitch, 1974): The two tones to be com-
pared are maintained in the phonological loop until the decision
on the interval has been made. Then, the tones are no longer
rehearsed and, as a consequence, forgotten. On the other hand,
The authors thank Oliver Vitouch, Susanne Raisig, and two anon-
ymous reviewers for providing helpful comments on an earlier version of
this article.
Lars Kuchinke is now at Free University of Berlin.
Address reprint requests to: Kathrin Schlemmer, Department of
Psychology, Humboldt University at Berlin, Rudower Chaussee 18,
12489 Berlin, Germany. E-mail address: hahn@music-evaluation.de.
465
absolute pitch is dened as long-term memory for individual
pitches (Ward, 1999): The absolute pitch possessor is assumed to
map a heard pitch to an internal reference system for pitches to
nd the verbal label of the heard pitch. Nonpossessors are sup-
posed to rely on relational strategies (e.g., relating a heard pitch
to their own vocal range) in pitch identication tasks, usually
leading to performance around chance level (with each octave
containing 12 pitch categories or pitch classes, chance in pitch
identication is 1/12 or 8.3%).
Of all methods to measure absolute pitch (pitch identication,
pitch production, and memory tasks), pitch identication tasks
are most widely used. In pitch identication tasks, participants
hear a number of musical pitches and have to label them with the
corresponding pitch class, such as C or C#. Because pitch class
labels are repeated in each octave, a designation of octave, such
as C4 or C5, is required to dene the exact pitch height of a pitch.
Results of pitch identication studies (for a review, see Takeuchi
& Hulse, 1993) revealed a wide range of pitch identication rates
among absolute pitch possessors (as well as nonpossessors), de-
pending on the kind of stimuli, the selection of participants (self-
report or absolute pitch screening), and on the tolerance of
semitone (i.e., the smallest unit for pitch notation in Western
music, corresponding to adjacent piano keys) and octave errors.
The latter are typical errors of absolute pitch possessors, but not
of nonpossessors, which has led to the conclusion that absolute
pitch possessors rely on pitch class in their judgment, whereas
nonpossessors can only make imprecise pitch judgments on the
basis of overall pitch height (Ward, 1999).
A number of studies have examined the effect of stimulus-
specific inuences on pitch identication by absolute pitch
possessors. It has been found that speed and accuracy of pitch
466
identication depend on pitch range (Heyde, 1987; Miyazaki,
1989), timbre (Marvin & Brinkman, 2000; Miyazaki, 1989), and
key color (Marvin & Brinkman, 2000; Miyazaki, 1990; Takeuchi
& Hulse, 1991). The method of these studies is similar: A number
of pitches differing in the variable of interest, for example, key
color or timbre, are labeled by the absolute pitch possessor,
whose pitch labeling performance (error rates, reaction times) is
then compared for pitches of the different timbres or key colors.
It has been shown that tones of the middle frequency region are
identied better and faster than tones of the very high or very low
regions. Second, it has been shown that piano tones are identied
better and faster than synthesized complex tones or pure tones
(Miyazaki, 1989), and also faster than violin tones (Marvin &
Brinkman, 2000). Third, white key pitches have been shown to be
identied better and faster than black key pitches.
The effects of pitch range, timbre, and key color point to
the importance of learning factors in absolute pitch acquisition.
They can be explained as base rate effects, because pitches in the
middle frequency region, white key pitches, and pitches in
piano timbres occur more often in most Western music than
pitches of extreme frequency regions, black key pitches, and pure
tones. The interpretation of the key color effect as an effect of
base rate is supported by Simpson and Huron (1994) who
could relate Miyazakis (1990) reaction time data for the 12 pitch
classes to their expected frequency of occurrence, calculated
from a sample of Western baroque music. According to Simpson
and Huron, 45% of the observed variance in Miyazakis
data could be explained by the HickHyman law (Hick, 1952;
Hyman, 1953), stating that the greater the expectancy for a
given stimulus, the shorter the reaction time will be for that
stimulus.
While existing evidence supports the importance of base rate
learning, this study focuses on base rate effects in the individual
learning history. Instead of comparing articial and instrumental
timbres (like Miyazaki, 1989) we compared the pitch labeling
performance for pitches played in familiar versus less familiar
instrumental timbres. Additionally, we investigated absolute
pitch possessors as well as nonpossessors. Because we assume
that base rate learning affects both groups, we expected similar
effects of timbre and key color on the pitch labeling performance
of absolute pitch possessors and nonpossessors, with possessors
showing an overall higher performance level.
Pupillometrics
We investigated the amount of resource allocation needed during
the identication of musical pitches by analyzing changes in pupil
diameter as well as behavioral data (error rates, reaction times).
Although the pupils main function is the control of light entering
the eye (the functional path of the pupillary response; cf. Hoeks &
Levelt, 1993), numerous studies have shown that pupillary re-
sponses are a reliable and sensitive index of the extent of central
nervous system processing allocated to a task. This is called the
nonfunctional path of the pupillary response, in which the lateral
hypothalamus, the locus coeruleus, and the cerebral cortex are
assumed to play a crucial role (Hoeks & Ellenbroek, 1993). Ac-
cording to Granholm and Steinhauer (2004), the extent of pupil
dilation reects the amount of information processing. Increases
in pupil dilation in response to increased information processing
demands have been found in studies involving a variety of tasks,
such as language processing (Hyona, Tommola, & Alaja, 1995;
Just & Carpenter, 1993), perceptual tasks (Verney, Granholm, &
Dionisio, 2001), and memory tasks (Granholm, Asarnow, Sar-
K.B. Schlemmer et al.
kin, & Dykes, 1996; van der Meer, Friedrich, Nuthmann, Stelzel,
& Kuchinke, 2003). In a study on pitch discrimination, Kahne-
man and Beatty (1967) found a significant relation between dif-
culty of discrimination between two pitches and the magnitude
of pupil dilations.
Pupillary responses have also been linked to emotional
processing, although the evidence is somewhat mixed. For in-
stance, Mudd, Conway, and Schindler (1990), who compared
pupil responses and verbal preference judgments to six pieces of
music, found that ratings of liking were associated with pupil
dilation whereas disliking was associated with pupil constriction.
This is one of the few studies supporting the controversial hy-
pothesis of bidirectional pupil responses formulated by Hess
(1965). However, studies controlling the initial pupil constriction
following stimulus presentation found increases in pupil dilation
when both pleasant or unpleasant stimuli as opposed to neutral
stimuli were processed (Steinhauer, Boller, Zubin, & Pearlman,
1983; Libby, Lacey, & Lacey, 1973). These studies suggest that
both liking and disliking require the allocation of resources and
that the level of emotional involvement rather than the emotional
valence of stimuli is related to pupil dilation.
Pupillary responses were measured during pitch identication
to assess stimulus-specific differences in mental effort that
may not be able to be seen in error rate or reaction time. The
latter reect accuracy and speed of information processing rather
than mental effort needed in the task. The analysis of mental
effort is of particular interest for the investigation of key color
differences, because a higher error rate for black key pitches is
confounded by a response bias (i.e., a higher probability to guess
the name of a white key pitch when uncertain). Reaction time,
on the other hand, is inuenced by artifacts during answer
registration, either through the use of piano keys (with longer
ways to black keys; as in Miyazaki, 1989, 1990, Exp. 1) or a
decision between two labels for the same key (e.g., C# and D[)
for black key pitches only (as in Marvin & Brinkman, 2000).
Because differences in overall emotional involvement in
black and white key pitches are expected to be minimal, pupil
dilation could provide ne-grained information on differences in
mental effort related to behavioral effects of key color (and
correspondingly, timbre).
To examine differences in mental effort, we analyzed peak
dilation of the pupil (i.e., the maximal pupil diameter in the time
interval of interest), which according to Beatty (1982) is the most
relevant characteristic of the pupillary response. Because we ex-
pected higher mental effort to be required for the labeling of
black key pitches (compared with white key pitches) as well as
pitches in less familiar timbres (compared with familiar timbres),
we expected higher peak dilation of the pupil during the iden-
tication of black key pitches as well as pitches in less familiar
timbres. On the behavioral level, we expected a faster and more
accurate identication of white key pitches and pitches played in
familiar timbres compared with black key pitches and pitches
played in less familiar timbres.
The hypothesized stimulus-specific effects during pitch iden-
tication were expected to affect absolute pitch possessors as well
as nonpossessors. On the behavioral level we expected an overall
higher accuracy and shorter reaction time among absolute pitch
possessors compared to nonpossessors. On the psychophysio-
logical level, we expected higher peak dilation of the pupil among
absolute pitch nonpossessors compared to possessors, because
they are assumed to rely on costly relational strategies when
labeling pitches.
Absolute pitch and pupillary response
Method
Participants
Twenty-one participants took part in the experiment (9 men, 12
women, mean age: 28 years, SD: 5.8 years). Twelve participants
were professional musicians or music students and 9 participants
were amateur musicians. All participants have had musical
training on at least one instrument (mean duration of instru-
mental training: 14.6 years, SD: 6 years), 14 participants played
more than one instrument. The instruments played by the par-
ticipants included piano (all participants), organ (1 participant),
violin (8 participants), viola (2 participants), cello (3 partici-
pants), guitar (5 participants), recorder (5 participants), ute
(1 participant), clarinet (2 participants), saxophone (1 partici-
pant), bassoon (1 participant), French horn (1 participant), and
drums (2 participants). Nine participants reported having abso-
lute pitch. However, the nal classication of participants into
absolute pitch possessors and nonpossessors was based on their
pitch naming performance in the experiment. In accordance with
the subjective judgment of the participants, 9 were classied as
absolute pitch possessors (with more than 70% correct pitch
judgments) and 12 as nonpossessors.
Participants were recruited from music schools and choirs in
Berlin and received no payment for their participation. They
completed a questionnaire on momentary medical problems and
medication. None of the participants suffered from neurological
or psychiatric diseases, nor did they take any medication that
could inuence pupillary responses. They reported no oph-
thalmologic problems other than correctable vision. Informed
consent was obtained.
Stimuli
The stimuli consisted of 88 tones from the two octaves between
C3 and C5. The tuning standard was set to A4 5 440 Hz. Each
tone was presented for 2000 ms, including fade-in and fade-out
of 200 ms each. To test for effects of key color, 44 black key tones
and 44 white key tones were used. The frequency of occurrence of
all 12 pitch classes was balanced through the experiment between
six and nine times. To test for timbre effects, 10 different timbres
were used. Because most musicians (and all participants) are
familiar with the piano, we used 24 piano tones as familiar tim-
bre. Twenty-four triangular tones were used as less familiar tim-
bre. In these two timbres, all 24 tones of the two octaves appeared
once. Additionally, eight other instrumental timbres were chosen
from the three main groups of orchestral instruments, namely
string, woodwind, and brass instruments. For each of the three
groups the selection included instruments playing in a higher as
well as lower pitch range. Thus, we used violin, cello, ute, clar-
inet, bassoon, trumpet, trombone, and French horn timbres and
presented ve tones per timbre. The number of black and white
key pitches was balanced in each of these eight timbres. The
triangular tones were generated with the software CoolEdit
(Syntrillium) and all other tones were recorded from the per-
formance of professional musicians on digital audio tape (DAT)
and edited in duration and loudness with CoolEdit. Loudness
was normalized.
To classify familiar and less familiar timbres, each participant
was asked to name all instruments they had taken lessons in.
These instruments were then classied as familiar for this par-
ticipant. This procedure resulted in different numbers of familiar
and less familiar tones for each participant. It is justied, though,
by the fact, that for every participant, at least one very familiar
467
timbre was included, because not everybodys main instrument
was the piano.
To minimize the success of skilled musicians in using relative
pitch for naming successive tones, a strategy that is usually less
successful with large intervals, the tones were presented in a xed
order in which distances between two tones were never smaller
than 10 semitones. Tones of different timbres were randomized
so that two successive tones always differed in timbre.
Procedure
The experiment took place in a quiet, lighted room (background
luminance about 300 lux) and the participants were seated in a
comfortable chair, with a distance of 0.8 m to the computer
screen at eye level and 1 m to the loudspeakers. The participants
were familiarized with the experimental procedure by written
instructions. The experiment started with a xation cross ap-
pearing on the screen for 1000 ms. Then the tone was presented
from the loudspeakers. Participants were asked to react by mouse
click indicating identication as soon as they knew the name of
the respective tone. They were instructed to react as fast as pos-
sible, but only when they were quite sure about the note name.
The reaction time was the time between tone onset and mouse
click. After the participants mouse click, the cross remained on
the screen for a further 1000 ms because the pupillary response
has been shown to lag behind behavioral responses by 300500
ms (Beatty, 1982; Hoeks & Levelt, 1993). Then a smiley face
appeared on the screen indicating that participants were to an-
nounce the name of the tone and were allowed to blink. This
procedure was chosen to prevent light reactions and movement
artifacts during the experimental trials. Participants were asked
to extend the blinking pause as long as needed and to start the
next trial by mouse click.
Tone names were registered by the experimenter who gave no
feedback on the accuracy of responses. If participants did not react
within 5000 ms, the smiley face appeared 5000 (11000) ms after
tone onset and the respective tones were registered as not named.
Participants were tested individually. There were four practise
trials before the experiment started. The experiment lasted ap-
proximately 45 min.
Apparatus
The experiment was run by two IBM-compatible microcomput-
ers. One computer was used for stimulus presentation and the
other for the registration of pupillary data. The experiment was
written in Matlab (version 6.5), using the Psychophysics Toolbox
extensions (Brainard, 1997; Pelli, 1997). The pupillary data were
recorded with an iView system (SensoMotoric Instruments). An
infrared-sensitive video camera including an infrared light source
was xed on the participants head to record pupillary responses
from the right eye. A near-infrared light source (l 5 7001049
nm) illuminated the eye and produced a corneal reection. The
corneal reection was recorded by the video camera, and at
50 Hz (i.e., every 20 ms), the horizontal pupil diameter was au-
tomatically measured and stored on the computers hard disk
along with the 5 mV TTL signals marking the beginning of trials
and reaction times.
The xation cross and the smiley face were presented on a
computer screen that was 14 in. in size. The gray background had
a luminance of 39.65 cd/m2 and the two dark gray symbols had
approximately the same luminance near 30 cd/m2 (xation cross:
30.58 cd/m2, smiley face: 29.56 cd/m2). External reections were
eliminated as well as possible.
468
Analysis of Pupillary Responses
A computer algorithm was used to discard trials with major ar-
tifacts or excessive blinking and to correct trials with smaller
artifacts by linear interpolation. Trials with incorrect pitch labe-
ling responses as well as trials with no mouse click responses were
excluded from the averaging. Trials were also excluded when
reaction times exceeded mean reaction time (determined per
participant group) by two standard deviations. Altogether,
55.6% of trials were discarded due to major artifacts, excessive
blinking, or no or too slow responses. A further 9.7% of trials
were discarded due to incorrect pitch labeling responses, so there
remained 640 trials (34.7%) for further analyses. The loss of data
was systematically distributed across participant groups: Among
absolute pitch possessors, 564 trials (71.3%) were valid, whereas
20.6% of trials were discarded due to artifacts, blinking, or no or
too slow responses and 8.1% of trials were discarded due to
incorrect pitch labeling responses. Among nonpossessors, only
76 trials (7.2%) were valid, whereas 81.8% of trials were dis-
carded due to artifacts, blinking, or no or too slow responses and
11.0% of trials were discarded due to incorrect pitch labeling
responses.
A ve-point weighted average lter was then run over the
data. Baseline pupil size was dened per individual trial as the
average pupil diameter (in millimeters) recorded 200 ms before
tone onset. The overall average baseline pupil diameter was 4.03
mm (range 2.776.96 mm). To rule out an inuence of baseline
pupillary diameter on our results, baseline pupillary diameter
was averaged for each experimental condition and each partic-
ipant (cf. Table 2), and subjected to a 2  2 repeated-measures
ANOVA. Furthermore, to rule out a drift of baseline pupillary
diameter over the course of the experiment, the rst 10 and the
last 10 baselines of each participant were averaged and compared
with a repeated measures t test.
Peak dilation of the pupil was computed by individual trial of
each participant. Peak dilation was dened as the maximal base-
line-corrected pupil diameter in the interval between tone onset
and reaction time1500 ms, so pupillary responses lagging behind
the behavioral responses by up to 500 ms were included. Pupil-
lary responses were averaged separately for each participant and
each experimental condition. Exploratory analyses revealed a
large variability in the pupillary responses to tones in less familiar
timbres, which we attribute to the joint analysis of articial tri-
angular tones and tones in instrumental timbres with a more
complex overtone spectrum. To compare only pupillary respons-
es to tones with similar sound complexity, we omitted the pupil-
lary data of triangular tones from the analysis of timbre effects
and compared only familiar with less familiar instrumental tim-
bres. Therefore, the four conditions included the following num-
bers of valid trials: blackless familiar: 145 (absolute pitch
possessors: 131, nonpossessors: 14), blackfamiliar: 93 (posses-
sors: 86, nonpossessors: 7), whiteless familiar: 102 (possessors:
92, nonpossessors: 10), and whitefamiliar: 135 (possessors: 104,
nonpossessors: 31).
Results
Classication of Absolute Pitch Possessors and Nonpossessors
There were large individual differences in hit rate in the pitch
identication task among participants who judged themselves to
be absolute pitch possessors (73%100% correct identication)
as well as among those who judged themselves to be nonposses-
K.B. Schlemmer et al.
100
80
Hit Rate (in %)
60
absolute pitch absolute
pitch
nonpossessors
40 possessors
20
0
All Participants (n = 21)
Figure 1. Distribution of hit rates (in percent) among all participants
(n 5 21). Participants with a hit rate above 70% were classied as
absolute pitch possessors, the rest as nonpossessors.
sors (2%43% correct identication). Figure 1 displays hit rates
of all participants.
Because the largest gap in the hit rate distribution was be-
tween 43% and 73%, this was chosen as a cut-off point for the
classication of participants into absolute pitch possessors and
nonpossessors. With the chosen cut-off point, the classication
of participants by hit rate in the pitch identication task con-
formed with participants own judgment of their pitch labeling
ability. Although it could be argued that the distribution of hit
rates resembles an absolute pitch continuum (for a review of
the absolute pitch continuum discussion, see Vitouch, 2005),
there were significant differences in the hypothesized direction
between absolute pitch possessors and nonpossessors in hit rate,
F(1,19) 5 148.03, po.0001, Z2 5 .886, as well as reaction time,
F(1,19) 5 18.93, po.0001, Z2 5 .513, justifying this classication
(cf. Table 1). For both groups, the following rule was proven: The
faster the averaged reaction time of a participant, the higher his
individual hit rate, r 5 .79, po.0001. Furthermore, the char-
acteristic distribution of error sizes described by Ward (1999) was
found among our participants, too. Whereas absolute pitch pos-
sessors errors were almost exclusively semitone errors, nonpos-
sessors errors ranged from one to six semitones (because octave
designation was not required in our task, the maximal error was 12
octave 5 6 semitones). Figure 2 displays the distribution of error
sizes for absolute pitch possessors and nonpossessors.
Behavioral Results
Table 1 displays descriptive evidence including the means and
standard errors of hit rate and reaction time, comparing black
and white key pitches as well as familiar and less familiar timbres
for absolute pitch possessors as well as nonpossessors.
The significance of key color and timbre effects on the pitch
labeling performance of both participant groups was tested by
subjecting hit rate to a 2  2  2 repeated-measures ANOVA
with key color (black vs. white) and timbre (familiar vs. less
familiar) as within-participant variables and participant group
(absolute pitch possessors vs. nonpossessors) as the between-
participant variable. The analysis revealed a significant main ef-
fect of key color, F(1,19) 5 4.53, p 5 .047, Z2 5 .193, with a
higher hit rate for white compared to black key pitches. The main
effect of timbre was significant, F(1,19) 5 7.50, p 5 .013,
Z2 5 .283, with a higher hit rate for familiar compared to less
familiar timbres. There was a significant main effect of partic-
ipant group, F(1,19) 5 129.57, po.0001, Z2 5 .872, with a higher
hit rate among absolute pitch possessors compared to nonpos-
Absolute pitch and pupillary response 469

Table 1. Means and Standard Errors (SE) of Hit Rate and Reaction Time, Dependent on Key Color and Timbre of Pitches, for Absolute
Pitch Possessors (n 5 9) and Nonpossessors (n 5 12)

Key color
Black White

Less familiar timbre Familiar timbre Less familiar timbre Familiar timbre
Absolute pitch possessors
Hit rate
Mean (%) 81.87 91.54 87.64 88.27
SE (%) 6.67 3.88 3.49 4.63
Reaction time
Mean (ms) 1973 1711 1852 1651
SE (ms) 348 269 283 284
Absolute pitch nonpossessors
Hit rate
Mean (%) 14.21 12.88 15.73 32.94
SE (%) 3.40 4.48 4.13 8.56
Reaction time
Mean (ms) 2869 2522 3013 3327
SE (ms) 265 488 480 428

sessors. Interactions were all insignificant except for the triple
interaction key color  timbre  participant group, F(1,19) 5
5.57, p 5 .029, Z2 5 .227.
Further analyses of absolute pitch nonpossessors pitch iden-
tication performance compared their hit rate in each experi-
mental condition against chance level (8.3%). One-sample t tests
revealed that only hit rate in the easiest condition white
familiar exceeded chance performance significantly, t(11) 5 2.88,
p 5 .015. Because only reaction times of comparable trials, that
is, correctly labeled pitches, could be compared, chance per-
formance in three out of four conditions among absolute pitch
nonpossessors led to the methodological problem that the inu-
ences of key color and timbre could not be further distinguished
in the reaction time data of this participant group.
Therefore, reaction time of only absolute pitch possessors was
subjected to a 2  2 repeated-measures ANOVA with key color
(black vs. white) and timbre (familiar vs. less familiar) as within-
participant variables. The analysis of reaction time revealed an
only marginally significant main effect of timbre, F(1,8) 5 4.31,
p 5 .071, Z2 5 .350, whereas there was no significant main effect
of key color, F(1,8) 5 2.127, p 5 .183, Z2 5 .210, and no signif-
icant interaction.
diameter of the rst 10 trials and the last 10 trials of each par-
ticipant, t(8) 5 1.802, p 5 .109. Thus, there was neither evidence
for an inuence of baseline pupillary diameter on pupillary re-
sponses in the experimental conditions nor for a drift in baseline
pupillary diameter in the course of the experiment.
The significance of the key color and timbre effects on peak
dilation was tested by subjecting peak dilation to a 2  2 re-
peated-measures ANOVA with key color (black vs. white) and
timbre (familiar vs. less familiar) as within-participant variables.
The analysis revealed a significant main effect of key color,
F(1,8) 5 7.93, p 5 .023, Z2 5 .498, with greater peak dilation for
black compared to white key pitches. The main effect of timbre
was marginally significant, F(1,8) 5 4.08, p 5 .078, Z2 5 .338,
with less familiar timbre showing slightly larger pupil dilation
than familiar timbre. There was no significant interaction.
To analyze differences between absolute pitch possessors and
nonpossessors pupillary responses, peak dilation of the pupil
during nonpossessors only above-chance performance condition
(whitefamiliar) was compared. An independent-samples t test
revealed a significant difference, t(14) 5 2.175, p 5 .047. Figure 4
100
Relative Frequency (in %)

Pupillometric Results absolute pitch

Because only 7.2% of absolute pitch nonpossessors trials were 80 possessors
valid and because their pitch labeling performance exceeded absolute pitch
chance in only one experimental condition, effects of key color 60 nonpossessors
and timbre on pupil dilation were analyzed for absolute pitch
40
possessors only. Mean pupillary responses, averaged across all
possessors, are shown in Figure 3. Table 2 displays descriptive
20
evidence including the means and standard errors of baseline
pupillary diameter and peak dilation, comparing black and white
0
key pitches as well as familiar and less familiar timbres. 0 1 2 3 4 5 6 no
The effect of baseline pupillary diameter differences in the answer
four experimental conditions was tested by subjecting baseline Error Size (in semitones)
pupillary diameter to a 2  2 repeated-measures ANOVA with
Figure 2. Distribution of error sizes ( 5 distances between correct pitch
key color (black vs. white) and timbre (familiar vs. less familiar) class labels and incorrect pitch labeling answers, calculated in semitones)
as within-participant variables. There were no significant main among absolute pitch possessors and nonpossessors. For example, the
effects of key color, F(1,8) 5 1.507, p 5 .254, Z2 5 .159, or tim- answer C# to a presented A was counted as an error of four semitones,
bre, F(1,8) 5 1.096, p 5 .326, Z2 5 .120. Furthermore, there was because these two pitch classes are four semitones apart on the piano
no significant difference between the averaged baseline pupillary keyboard.
470 K.B. Schlemmer et al.

0.3 0.3
Key Color Timbre

Response (mm)

Response (mm)
Mean Pupillary

Mean Pupillary
0.2 0.2

0.1 0.1
black keys familiar
0.0 white keys 0.0 less familiar

0 500 1000 1500 2000 0 500 1000 1500 2000

Time After Tone Onset (ms) Time After Tone Onset (ms)

Figure 3. Averaged pupillary responses of absolute pitch possessors, comparing black and white key pitches (left) as well as familiar
and less familiar timbres (right). Pupil values were millimeter calibrated and baseline corrected. The time window chosen for
averaging lasted from tone onset to tone offset (2000 ms).

displays mean pupillary responses of absolute pitch possessors
and nonpossessors in the condition whitefamiliar. To show that
the group difference in pupil dilation was specific for correctly
labeled pitches, mean pupillary responses of both groups for
incorrect pitch labeling responses in all conditions are also
displayed.
Discussion
The experiment yielded four main results. First, on the behavi-
oral level, we obtained significant timbre and key color effects on
the pitch labeling performance of absolute pitch possessors and
nonpossessors. Second, on the behavioral level, we obtained no
significant timbre and key color effects in the reaction times of
absolute pitch possessors. Third, on the psychophysiological
level, we obtained a significant key color effect and a marginally
significant timbre effect on pupillary responses of absolute pitch
possessors. Fourth, on the psychophysiological level, we ob-
tained a significant difference between pupillary responses of
absolute pitch possessors and nonpossessors during the condi-
tion whitefamiliar.
The experiment was motivated by three questions. The rst
question was whether the key color effect observed by Marvin
and Brinkman (2000), Miyazaki (1990), and Takeuchi and Hulse
(1991) can be conrmed by psychophysiological data. Results
revealed a key color effect in hit rates but not reaction times of
absolute pitch possessors. That is, white key pitches were labeled
correctly more often but not faster than black key pitches. Com-
pared with the other reported studies, differences in both hit rate
and reaction time are rather small. Miyazaki (1990) and Marvin
and Brinkman (2000) obtained larger reaction time differences
around 200 ms whereas Takeuchi and Hulse (1991) used a dif-
ferent paradigm (a Stroop-like task in which participants had to
respond with same or different to an aurally presented pitch
in combination with a visually presented pitch name) so that their
reaction time data are not quite comparable with ours. The
smaller hit rate differences in the present study can be attributed
to the specific (and small) population tested here whereas the
smaller reaction time differences can be attributed partly to our
method of answer registration, which was identical for both key
colors and thus avoided effects of motor and other biases.
Interestingly, pupillary response data were more sensitive
than reaction time data in revealing a significant key color effect,
because peak dilation of the pupil was higher when participants
labeled black key pitches than when they labeled white key
pitches. This result suggests that the key color effect is really an
effect of differential mental effort during the labeling of black and
white key pitches, and not simply an effect of response bias or
experimental artifacts: Labeling a black key pitch is more de-
manding than labeling a white key pitch. On a theoretical level,
the higher demand in labeling black key pitches compared to
white key pitches suggests an easier retrieval of white key pitch
names from memory possibly because more frequent exposure to
white key pitches during the individual music learning experience
produces stronger associations of white key pitches with their
labels than of black key pitches with their respective labels.
The second question that motivated the present experiment
was whether a timbre effect could be shown when absolute pitch
possessors are asked to label pitches played in familiar versus less
familiar timbres. On the behavioral level, labeling pitches played
in less familiar timbres produced more errors and marginally
longer reaction times than labeling pitches in familiar timbres.
This conrms data by Marvin and Brinkman (2000) and Mi-

Table 2. Means and Standard Errors (SE) for Pupil Parameters, Dependent on Key Color and Timbre of Pitches, Absolute Pitch
Possessors Only (n 5 9)

Key color
Black White

Less familiar timbre Familiar timbre Less familiar timbre Familiar timbre
Baseline pupil diameter
Mean (mm) 4.385 4.353 4.367 4.440
SE (mm) 0.238 0.245 0.244 0.260
Peak dilation
Mean (mm) 0.383 0.354 0.358 0.296
SE (mm) 0.056 0.056 0.056 0.045
Absolute pitch and pupillary response
0.4
Responses (mm)
0.3
Mean Pupillary
0.2
0.1 Correct
Responses
0.0
White-Familiar
0.1
0 500 1000 1500
Time after Tone Onset (ms)
absolute pitch possessors
Figure 4. Averaged pupillary responses of absolute pitch possessors and nonpossessors for correctly labeled pitches in the condition
whitefamiliar (left) and for incorrect pitch labeling responses in all conditions (right). Pupil values were millimeter calibrated and
baseline corrected. The time window chosen for averaging lasted from tone onset to tone offset (2000 ms).
yazaki (1989), and extends their results to the individual famil-
iarity of timbres, because we did not test the effect of articial
versus instrumental timbres, but rather the effect of familiar ver-
sus less familiar instrumental timbres.
An effect of timbre was also suggested by pupillary data as
peak dilation of the pupil was marginally higher for pitches in less
familiar timbres than for pitches in familiar timbres. We attribute
this to a lack of power in analyzing a sample of only 9 partic-
ipants, but could not increase the number of participants due to
the low incidence of absolute pitch in the general population.
Upon debriefing, some participants mentioned great difculties
in avoiding the calculation of intervals from a correctly labeled
pitch (i.e., a trial in which they were quite sure about the respec-
tive pitch name), especially when a pitch played in less familiar
timbre was preceded by a pitch played in a very familiar timbre.
Thus, a possible explanation for the higher demand of labeling
pitches in less familiar timbres compared to familiar timbres is the
costly use of relational pitch labeling strategies, even if this was
perceived as a highly automatic process by the participants.
However, further research, possibly including a manipulation of
pitch labeling strategies, is needed to conrm this interpretation.
An alternative interpretation of both key color and timbre
effects on pupil dilation is that the effects are linked to differences
in emotional involvement in presented tones. We do not feel that
differences in emotional involvement can explain the key color
effect very well, because there is no reason to believe that such
differences exist except as a consequence of different mental ef-
fort associated with the labeling process. However, we cannot
exclude an inuence of differences in emotional involvement in
tones in different timbres on pupillary responses, although we do
not know the direction of such an inuence. Although familiarity
of timbres is certainly associated with positive emotionality, fa-
miliar timbres could still be less interesting than more novel, that
is, less familiar timbres. This issue cannot be resolved in the
current study because it requires a comparison between verbal
preference judgments and pupillary responses to pitches in dif-
ferent timbres, similar to the method used by Mudd et al. (1990)
for the comparison of entire musical pieces.
The third question motivating this study was whether stim-
ulus-specific effects could also be found among absolute pitch
nonpossessors, as suggested by Marvin and Brinkman (2000).
Hit rates above chance level only in the condition whitefamiliar
suggest an effect of both key color and timbre on the pitch labe-
ling performance of absolute pitch nonpossessors. Although the
471
0.3
Responses (mm)
0.2
Mean Pupillary
0.1
0.0
False Responses
0.1
2000 0 500 1000 1500 2000
Time after Tone Onset (ms)
absolute pitch nonpossessors
respective contribution of key color and timbre to this effect
cannot be differentiated due to the overall small number of cor-
rectly labeled pitches, a preliminary conclusion is that the factors
shown to inuence the pitch labeling performance of absolute
pitch possessors do also affect that of nonpossessors.
Higher peak dilation of the pupil among nonpossessors com-
pared to possessors in the condition whitefamiliar conrm the
hypothesis that nonpossessors have to put higher mental effort
into a pitch labeling task than possessors do. Again, an alter-
native interpretation involves emotional processes, for instance,
a higher anxiety among nonpossessors. Such an explanation is
circular, though, because an assumed higher anxiety would likely
be caused by a perceived higher task difculty of nonpossessors.
The lack of group differences in pupillary responses during in-
correctly labeled trials (cf. Figure 4) supports our interpretation
that the differences observed in valid trials are caused by differ-
ences in mental effort during pitch identication rather than by
differences in emotional reactions to a more or less solvable task.
On a theoretical level, effects of key color and timbre among
absolute pitch nonpossessors support the base rate theory (Take-
uchi & Hulse, 1991) for the explanation of key color effects,
because in most Western music absolute pitch possessors and
nonpossessors alike are more often exposed to white key pitches
than to black key pitches. Similarly, a personal base rate effect
can explain the timbre effect found in this experiment, because
pitch labels can more easily be retrieved from memory if the
pitches are presented in the timbre of an instrument which the
participant has played (and hence, heard very often). The reason
for this effect could be a larger representation of pitches in fa-
miliar timbre in the auditory cortex, as found by Pantev and his
coworkers (Pantev, Roberts, Schulz, Engelien, & Ross, 2001;
Pantev et al., 1998). Their participants listened to pitches in dif-
ferent instrumental timbres while cortical representation of
pitches was measured with magnetoencephalography (MEG).
Results showed that cortical representation is enlarged for pitch-
es in instrumental timbres compared to articial sine tones by
roughly 25% (Pantev et al., 1998), and that there is a specific
enlargement for the tones of ones own instrument compared to
another instrumental timbre (Pantev et al., 2001). Although the
task of these experiments was a passive one, that is, listening to
presented tones, it is interesting to note that the effect occurred in
both absolute pitch possessors and nonpossessors.
Taken together, Pantevs and our results suggest that frequent
exposure to specific tones enlarges the cortical representation for
472
these tones and facilitates their labeling for absolute pitch pos-
sessors as well as nonpossessors. The combination of behavioral
measures with the measurement of pupillary responses provides
meaningful information about cognitive processes underlying
pitch identication as well as pitch learning processes of both
absolute pitch possessors and nonpossessors. The observed dif-
REFERENCES
Baddeley, A. D., & Hitch, G. (1974). Working memory. In G. H. Bower
(Ed.), The psychology of learning and motivation (vol. 8, pp. 4789).
New York: Academic Press.
Beatty, J. (1982). Task-evoked pupillary responses, processing load, and
the structure of processing resources. Psychophysiological Bulletin, 91,
276292.
Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10,
433436.
Granholm, E., Asarnow, R. F., Sarkin, A. J., & Dykes, K. L. (1996).
Pupillary responses index cognitive resource limitations. Psychophys-
iology, 33, 457461.
Granholm, E., & Steinhauer, S. R. (2004). Introduction: Pupillometric
measures of cognitive and emotional processing. International Journal
of Psychophysiology, 52, 16.
Hess, E. H. (1965). Attitude and pupil size. Scientific American, 4, 4654.
Heyde, E.-M. (1987). Was ist absolutes Horen? Eine musikpsychologische
Untersuchung. Munchen: Profil.
Hick, W. E. (1952). On the rate of gain of information. Quarterly Journal
of Experimental Psychology, 4, 1126.
Hoeks, B., & Ellenbroek, B. A. (1993). A neural basis for a quantitative
pupillary model. Journal of Psychophysiology, 7, 315324.
Hoeks, B., & Levelt, W. J. (1993). Pupillary dilation as a measure of
attention: A quantitative system analysis. Behavior Research Meth-
ods, Instruments & Computers, 25, 1626.
Hyman, R. (1953). Stimulus information as a determinant of reaction
time. Journal of Experimental Psychology, 45, 423432.
Hyona, J., Tommola, J., & Alaja, A.-M. (1995). Pupil dilation as a
measure of processing load in simultaneous interpretation and other
language tasks. Quarterly Journal of Experimental Psychology, 48A,
598612.
Just, M. J., & Carpenter, P. A. (1993). The intensity of thought: Pupillo-
metric indices of sentence processing. Canadian Journal of Experi-
mental Psychology. Special Issue: Reading and language processing,
47, 310339.
Kahneman, D., & Beatty, J. (1967). Pupillary responses in a pitch-dis-
crimination task. Perception & Psychophysics, 2, 101105.
Libby, W. L. Jr., Lacey, B. C., & Lacey, J. I. (1973). Pupillary and
cardiac activity during visual attention. Psychophysiology, 10,
270294.
Marvin, E. W., & Brinkman, A. R. (2000). The effect of key color and
timbre on absolute pitch recognition in musical contexts. Music Per-
ception, 18, 111137.
K.B. Schlemmer et al.
ferences in pupil dilations conrm Beattys (1982) description of
pupillary responses as an index of both within-task (stimulus-
specific effects among absolute pitch possessors) and between-
individual (absolute pitch possessor vs. nonpossessor) variations
in mental effort and add to the validity of the pupillary response
as a measure of resource allocation.
Miyazaki, K. (1989). Absolute pitch identication: Effects of timbre and
pitch region. Music Perception, 7, 114.
Miyazaki, K. (1990). The speed of musical pitch identication by abso-
lute-pitch possessors. Music Perception, 8, 177188.
Mudd, S., Conway, C. G., & Schindler, D. E. (1990). The eye as music
critic: Pupil response and verbal preferences. Studia Psychologica, 32,
2330.
Pantev, C., Oostenveld, R., Engelien, A., Ross, B., Roberts, L. E., &
Hoke, M. (1998). Increased auditory cortical representation in mu-
sicians. Nature, 392, 811814.
Pantev, C., Roberts, L. E., Schulz, M., Engelien, A., & Ross, B. (2001).
Timbre-specific enhancement of auditory cortical representations in
musicians. NeuroReport, 12, 169174.
Pelli, D. G. (1997). The VideoToolbox software for visual psychophysics:
Transforming numbers into movies. Spatial Vision, 10, 437442.
Simpson, J., & Huron, D. (1994). Absolute pitch as a learned phenom-
enon: Evidence consistent with the Hick-Hyman Law. Music Per-
ception, 12, 267270.
Steinhauer, S. R., Boller, F., Zubin, J., & Pearlman, S. (1983). Pupillary
dilation to emotional visual stimuli revisited. Psychophysiology, 20,
472.
Takeuchi, A. H., & Hulse, S. H. (1991). Absolute pitch judgments of
black- and white-key pitches. Music Perception, 9, 2746.
Takeuchi, A. H., & Hulse, S. H. (1993). Absolute pitch. Psychological
Bulletin, 113, 345361.
van der Meer, E., Friedrich, M., Nuthmann, A., Stelzel, C., & Kuchinke,
L. (2003). Picture-word-matching: Flexibility in conceptual memory
and pupillary responses. Psychophysiology, 40, 904913.
Verney, S. P., Granholm, E., & Dionisio, D. P. (2001). Pupillary re-
sponses and processing resources on the visual backward masking
task. Psychophysiology, 38, 7683.
Vitouch, O. (2005). Absolutes Gehor. In Th. H. Stoffer & R. Oerter
(Eds.), Allgemeine Musikpsychologie (Enzyklopadie der Psychologie,
Vol. D/VII/1, pp. 717766). Gottingen: Hogrefe.
Ward, W. D. (1999). Absolute pitch. In D. Deutsch (Ed.), The psychology
of music (2nd ed., pp. 265298). San Diego, CA: Academic Press.
(Received October 28, 2004; Accepted March 24, 2005)