DEV (WILEJ) RIGHT BATCH

J. P. Lecanuet*
C. Graniere-Deferre
A.-Y. Jacquet Fetal Discrimination of Low-
Laboratoire Cognition et
Développement,
CNRS (UMR-8605), Université de
Pitched Musical Notes
Paris V, France

A. J. DeCasper
Department of Psychology
University of North Carolina at
Greensboro (UNCG),
Greensboro, NC 27410

Received 26 January 1999; accepted 8 July 1999

ABSTRACT: Cardiac responses of 36- to 39-week-old (GA) fetuses were tested with a no-
delay pulsed stimulation paradigm while exhibiting a low heart rate (HR) variability (the HR
pattern recorded when fetuses are in the 1f behavioral state). We examined whether fetuses
could discriminate between two low-pitched piano notes, D4 (F0 ⫽ 292 Hz/292–1800 Hz) and
C5 (F0 ⫽ 518 Hz/518–300 Hz). Seventy percent of all fetuses reacted to the onset of the first
note (D4 or C5) with the expected cardiac deceleration. After heart rate returned to baseline,
the note was changed (to C5 or D4, respectively). Ninety percent of the fetuses who reacted to
the note switch did it with another cardiac deceleration. Control fetuses, for whom the first
note did not change, displayed few cardiac decelerations. Thus, fetuses detected and responded
to the pulsed presentation of a note and its subsequent change regardless of which note was
presented first. Because perceived loudness (for adults) of the notes was controlled, it seems
that the note’s differences in F0 and frequency band were relevant for detecting the change.
Fetuses’ ability to discriminate between spectra that lay within the narrow range of voice F0
and F1 formants may play an important role in the earliest developmental stages of speech
perception. 䉷 2000 John Wiley & Sons, Inc. Dev Psychobiol 36: 29– 39, 2000

Keywords: fetus; auditory stimulation; musical notes; decelerative cardiac response; low heart
rate variability state

This study of fetal auditory perception is the latest in
a series from our laboratory that uses the autonomic
cardiac deceleration, an attentional response if not an
orienting response, as a measure of whether near-term
fetuses can detect relatively subtle changes in their au-
ditory environment. Here we looked for the cardiac
reaction when we changed the spectral properties of
low-pitched musical notes: D4 and C51 using a no-
delay pulsed paradigm.
Near term [over 35 – 36 weeks gestational age
(GA)]2 fetal heart rate responses to sound show the
same characteristics as those observed in newborns.
They are largely dependent on stimulus parameters
such as sound pressure level (SPL), frequency, band-

*Present address: Laboratoire Cognition et Développement, In-

stitut de Psychologie, 71 Avenue Edouard Vaillant, 92774 Boulogne 1 D4 and C5 refer to the piano keyboard notation, numbers cor-

Billancourt, Cedex, Paris, France respond to octave rank (from 0 to 8 starting on a C note.
Correspondence to: J.-P. Lecanuet 2 Gestational age.

short
䉷 2000 John Wiley & Sons, Inc. CCC 0012-1630/00/010029-11 standard
long
DEV (WILEJ)
30 Lecanuet et al.
width, and on subjects’ parameters such as behavioral
states (Granier-Deferre, Lecanuet, Cohen, & Busnel,
1985; Lecanuet, Granier-Deferre, Cohen, Le Houezec,
& Busnel, 1986; Schmidt, Boos, Gniers, Auer, &
Schulze, 1985). Acceleration responses are reliably
evoked with broadband stimuli at or above 105-dB
SPL (ref: 20 ␮Pa), and their frequency of occurrence
is significantly enhanced as sound pressure level in-
creases (for a review see Kisilevsky & Low, 1998;
Lecanuet & Schaal, 1996). With less intense pressure
level (80- to 90-dB SPL) stimulation, such as low or
medium frequency noises, music, and speech, there is
a shift from cardiac acceleration to deceleration. This
change corresponds (with a 15- to 20-dB lag) to the
change displayed by young infants and newborns
when the sound pressure level of the stimulation de-
creases (Berg & Berg, 1987; Eisenberg, 1976; Gra-
ham, Anthony, & Zeigler, 1983). For Graham and col-
leagues (Graham, 1984, Graham et al., 1983; Graham
& Clifton, 1966, Graham & Jackson, 1970), it reflects
a shift from startle or defense response mode (DR) to
orienting response mode (OR), according to Sokolov’s
(1963) adult model of information processing, trans-
ferred to the infant and the newborn. As for the new-
born, deceleration responses evoked in the “end of
gestation fetus” are not accompanied by movements
and have a smaller average amplitude than the accel-
eration responses (Lecanuet, Granier-Deferre, & Bus-
nel, 1988; Lecanuet et al., 1987).
This similarity between fetal and neonatal cardiac
responsiveness to auditory stimulation led us to at-
tempt investigations of fetal discrimination abilities
with a procedure derived from the one employed by
Clarkson and Berg (1983) to study newborn syllabic
discrimination of [A] and [i]: the “no-delay pulsed
stimulation paradigm.” This procedure, used in nu-
merous human infant and primate speech perception
studies, is sometimes called “the 20/20 paradigm.” It
is based on the idea that because HR response to a
long-lasting stimulus “usually returns to a steady base-
line while the stimulus is still present, it is possible,
by introducing a second stimulus immediately follow-
ing a sufficiently long-lasting first stimulus to evoke a
second orientation response to the change in stimula-
tion” (as cited in Brown, Morse, Leavitt, & Graham,
1976, p. 1). This procedure became a “no-delay” par-
adigm after Leavitt, Brown, Morse, and Graham
(1976) suppressed the intertrial because of the memory
load it imposed to subjects.
Using this procedure with 36- to 40-week-old (GA)
fetuses, we showed that decelerative cardiac responses
induced by onset of 16 pulsed presentations (at a rate
of 3,000 ms) of the paired syllables [bi] – [bA] (emitted
LEFT BATCH
at 92-dB SPL ⫾ 2 dB) were renewed after reversal of
the order of the syllables: [bA] – [bi]. This demon-
strated that near-term fetuses perceived the change in
the organization of the stimulus (Lecanuet et al.,
1987). We used the same paradigm to demonstrate that
same-age-range fetuses discriminated a male voice
(F0 ⫽ 80 – 100 Hz) and a female voice (F0 ⫽
165 – 200 Hz) uttering the same sentence; duration,
prosodic contour, and loudness for adults were
equated (Lecanuet, Granier-Deferre, Jacquet, & Bus-
nel, 1992; Lecanuet, Granier-Deferre, Jacquet, Cap-
poni, & Ledru, 1993). It should be emphasized that in
both studies cardiac deceleration occurred within the
first seconds of exposure to the initial stimulus and
recovery of this deceleration occurred within the first
seconds of a new stimulus, confirming that for fetuses
as well as for newborns only a short acoustic sample
is needed for the fetal auditory system to detect an
acoustically relevant change.
Since our 1987 experiment, fetal cardiac decelera-
tive reactiveness to different versions of the [A] and
[i] vowels (paired or not] has been confirmed at higher
sound pressure levels between (100 – 110 dB) (Zim-
mer et al., 1993) and down to 80-dB SPL (Groome,
Mooney, Holland, Bentz, & Atterbury, 1997). Shahi-
dullah and Hepper (1994) demonstrated that 35-week-
old fetuses could discriminate between [[bAbA] and
[bibi] by habituating and dishabituating startle re-
sponses to loud, 110-dB SPL stimuli.
The [bibA]/[bAbi] experiment had been designed at
first as a [bi]/[bA] study, similar to Clarkson and
Berg’s (1983) study and most discrimination studies,
the two syllables being emitted at the same sound pres-
sure level in air (92-dB SPL). However, due to differ-
ences in the structure of formants of the vowels: [i] F1
⫽ 240 Hz, F2 ⫽ 2160 Hz and [A]: F1 ⫽ 680 Hz, F2
⫽ 1200 Hz, [bi] sounded louder than [bA]. Because
we had no way to determine the role of loudness in
fetal perception and discrimination of acoustical stim-
uli, a first attempt at testing fetal discrimination abil-
ities was made with the paired syllables as a global
stimulus. Equating loudness (for adults) of male
(F0 ⫽ 83 – 100 Hz) and female (F0 ⫽ 165 – 200 Hz)
voices in the next experiment, which resulted in a 3-
dB SPL difference in air between these voices was
another attempt. Again, we could not be certain
whether the relevant dimension of difference was
stimulus intensity or fundamental frequency, fre-
quency band, and/or timbre. We do not know whether
the slight SPL difference between the voices was pre-
served in the amniotic fluid and/or whether fetuses’
perceived loudness functions are adultlike. We do
know that the intensity of the voices suffered little short
standard
long
DEV (WILEJ)
attenuated when transmitted in utero (Gagnon, Ben-
zaquen, & Hunse, 1992; Querleu, Renard, Boutteville,
& Crépin, 1989; Querleu, Renard, Versyp, Paris-Del-
rue, & Vervoort, 1988). We also know that, except
frequencies below 100 Hz, the voice stimuli were not
masked by the low-frequency components of maternal
and placental vascular noises (50- to 60-dB SPL) be-
cause they were presented at least 30- to 40-dB higher
than the background (Abrams, Gerhardt, & Peters,
1995; Armitage, Baldwin, & Vince, 1980; Gagnon et
al., 1992; Gerhardt, 1989; Peters, Abrams, Gerhardt,
Burchfield, & Wasserman, 1992; Querleu, et al., 1989;
Querleu, Renard, & Versyp, 1981; Vince, Armitage,
Baldwin, Toner, & Moore, 1982). Thus, in order to
further investigate the respective role of audio – acous-
tic parameters, we studied discrimination of two low-
pitched musical notes electronically generated in the
same scale. Similar overtone organization and emis-
sion at equal adult loudness should induce their dis-
crimination by adults on F0 ground and orient fetal
discrimination to be processed mainly on a similar
basis.
We tested fetuses during episodes of low heart rate
variability that could reliably be considered as heart
rate patterns (FHRPA), which is one of the character-
istics that define behavioral state 1F (usually consid-
ered as corresponding to quiet sleep state in the neo-
nate) (Nijhuis, Prechtl, Martin, & Bots, 1982). Cardiac
and motor reactions to sound are weaker in 1F than in
2F (active sleep), which is characterized by a high-
variability heart rate pattern and frequent movement-
induced heart rate accelerations. However, because fe-
tal movements are rare, testing in 1F offers the best
chance of detecting stimulus-induced, low-magnitude
heart rate changes and of avoiding the spontaneous
changes that could mislead our assessment of fetuses’
perceptual capacities.
METHODS
Subjects
The initial sample consisted of 138 healthy fetuses3
between 36 and 39 weeks GA. Their mothers, re-
cruited from the general obstetrical population of the
Baudelocque and the Port-Royal University Clinics in
Paris (France), were pregnant without complications
3 Fetuses were included on the basis of careful medical follow-
up systematically provided to mothers by the maternity hospital
during the whole pregnancy. Postnatal delivery and discharge ex-
amination were checked before inclusion of subjects’ data in the
sample.
RIGHT BATCH
Fetal Discrimination of Low-Pitched Musical Notes 31
and volunteered to participate in the study. Twenty-
two subjects were eliminated because at some time
during the experiment, their data signal was lost for
more than 2.5 s, they changed state, their heart rate
failed to recover within 65 s of the first stimulation
period, or their HR pattern reflected sucking move-
ments (van Woerden, van Geijn, Caron, van der Valk,
& Swartjes, 1988b). HR patterns that reflected mouth-
ing movements (van Woerden, van Geijn, Swartjes,
Caron, & Brons, 1988a) or breathing movements (Ni-
jhuis et al., 1982) did not lead to data rejection. The
final sample consisted of 116 fetuses. At birth, all fe-
tuses were healthy with weights appropriate for ges-
tational age.
Stimuli
In order to increase auditory conspicuousness, each
pulse was composed of two 500-ms emissions of the
same note interrupted by 150 ms of silence (total du-
ration ⫽ 1150 ms). D4, the low note (L) (F0 ⫽
292 Hz, overtones up to 1800 Hz) and C5, the high
note (H) (F0 ⫽ 518 Hz, overtones up to 3000 Hz) (see
Figure 1) were generated by a Yamaha synthesizer
(DX7T) set up in the no-scale piano mode to control
for a potential “brightness” effect4 that could contrib-
ute to the discrimination process. The notes were re-
corded in synchrony on separate channels of a D70
Revox stereo tape recorder so that it was possible to
switch notes at any time (via a BST switch cursor pre-
amplifier) without disrupting their temporal pattern.
The L amplified notes (Sony TA-AX330 amplifier)
were delivered at 93-dB SPL, a pressure level known
to evoke a high percentage of decelerative responses;
5- to 10-dB SPL higher increases the occurrence of
accelerative HR responses (Kisilevky, Muir, & Low,
1989; Lecanuet et al., 1986). Because there are — to
our knowledge — no equal loudness contours for pre-
term infants as young as our fetuses, the notes were
constructed to be equally loud to adult ears. Five nor-
mal-hearing adult subjects were asked to adjust the
loudness of the two stimuli in the same acoustic con-
ditions used during fetal testing. Sound pressure peak
level of the L note was first set at 93 ⫾ 2-dB SPL
(re: 20 ␮Pa) measured with an ACLAN audiometer
(SDH80) corresponding to 91 ⫾ 2-dB SPL (ref:
4
Brightness of a note generated by an instrument is related to
the spectral envelope of the note: the greater the number of high
frequency overtones the greater the brightness. Brightness in the
piano is not the same along the keyboard. In order to control for
this potential discrimination bias, we generated our stimuli using
the no-scale mode of the synthesizer. short
standard
long
DEV (WILEJ)
32 Lecanuet et al.
FIGURE 1 Frequency analyses of the two notes (Fast
Fourier Transform) show the bandwidth and pressure of
overtones.
20 ␮Pa) for the H note, at 20 cm from the loudspeaker
(AUDAX HR37).
Procedure
When mothers arrived for testing between 9 a.m. to
12:30 p.m., they were comfortably seated in a reclin-
ing armchair and briefed about the experiment. The
loudspeaker, supported by a movable C-shaped table,
was placed 20 cm above the maternal abdomen and
the probe of a Doppler cardiotocograph (Hewlett-
Packard Model 8030A) was positioned on the abdo-
men to pick up the best quality fetal heart rate signal.
Electric pulses, each of them corresponding to a
heart beat, were recorded of the cardiotocograph at the
lowest signal processing level of the device we could
LEFT BATCH
access and fed into a 386 SX PC. A lab-made software
package computed and saved beat-to-beat intervals in
milliseconds into separated files corresponding to test
periods. Fetal heart rate, in beats per minute (BPM),
computed by the cardiotocograph in beats per minute
(BPM) was continuously displayed in digital and an-
alog form, and was used only by the researchers to
monitor the fetuses’ cardiac activity during testing.
Two trained observers (A.-Y. J., J.-P. L.) scanned this
hard copy on-line for the low variability fetal heart rate
pattern. Once the fetus entered what both observers
defined as a stable low-variability period, earphones
(Koss Pro4/AAA) were placed on the mothers’ ears
that delivered guitar music at a high but still comfort-
able level (75- to 80-dB SPL) (Technics M215 tape
recorder) to mask the musical notes delivered to the
fetus. Masking was made complete by placing a thick
blanket over the loudspeaker and the lower part of the
mother’s body.
Half of the fetuses were scheduled randomly to first
encounter C5, the high (H) note and the others were
scheduled to encounter D4, the low (L) note. Half of
the fetuses in these groups were also randomly as-
signed to encounter a second stimulus (H or L) and
the other half no-change. In this way four groups were
formed: Group HL encountered the H note then the L
note; Group LH encountered the L note then H note;
for the control Groups HH and LL, the same note con-
tinued to be emitted. Each fetus was tested only once.
Each subject was randomly assigned to either one
of the two note-change groups: One group was first
stimulated with the high (H) note and then with the
low (L) note (HL) and the other with the L note fol-
lowed by the H note (LH). Others subjects were as-
signed to one of two control groups for which the emit-
ted note would continue unchanged: One group was
stimulated only with the H note (HH) and one only
with the L note (LL).
The three-part testing procedure began after the
low-variability heart rate pattern had been observed
for 5 min.
Prestimulation Period. The prestimulation period
was a silent period during which the low-variability
heart rate had to persist uninterrupted for 60 consec-
utive s before the onset of the stimulation (Period 1).
Any spontaneous cardiac acceleration or deceleration
greater than 4 BPM reset the 60-s clock.
First Stimulus Period. The first stimulus period com-
menced at the end of the 60-s prestimulation period.
During the first stimulus period the double H note or
double L note was presented once every 3 s. This short
pulsed stimulation persisted until the HR returned to standard
long
DEV (WILEJ)
a stable prestimulus rate and at least 45 but not more
than 65 s had elapsed. The initial occurrence of the
note was expected to elicit a heart rate deceleration
one could consider as an orienting response. The re-
turn was expected to occur as the pulsed stimulus be-
came a salient familiar feature of the general auditory
environment. If a stable heart rate was not recovered,
the test was cancelled. Cardiac reactions to the notes
were evaluated by comparing them to the reactions of
archival data from a control group of 34 similarly aged
fetuses who, in an earlier study (Lecanuet et al., 1992),
experienced the same 60-s prestimulation procedure
but never encountered a stimulus afterward. These
control subjects were not expected to show any sig-
nificant cardiac change at the end of the 60-s presti-
mulus period, the time when the experimental fetuses
were exposed to the notes.
Second Stimulus Period. The second stimulus period
began when heart rate had returned to a stable pattern.
For the HL and LH groups, the first stimulus was
switched to the second stimulus without disrupting the
temporal patterning of the stimuli and a key was tog-
gled on the computer’s keyboard to record the time of
stimulation change. For infants in the HH and LL
groups, only the key was toggled on the computer at
the time criteria for note changes were fulfilled.
Thirty-one fetuses receiving the H note and 31 re-
ceiving the L note reacted with a deceleration. Thus,
they were the only ones that could meet the deceler-
ation criteria again when the second stimulus was pre-
sented; these 62 fetuses were followed into the second
stimulus period. Of the 62 fetuses that oriented to the
first note, 14 were in the HL group, 20 in the LH
group, 17 in the HH control group, and 11 had been
assigned to the LL control group. A decelerative re-
sponse was expected of fetuses in the HL and LH
groups but not of fetuses in the HH and LL groups
because their pulsing stimulus did not change. The
second stimulus period lasted 1 min.
Data Treatment
Ultrasonographic detection of heart rate, based on
doppler analysis of heart contractions and dilations, is
less accurate than the EKG measure: Some beats may
not be detected or may be detected at the wrong time.
Thus, in order to restore in a biologically relevant way
values are stored in the files, every data file was proc-
essed with software running under the MS-DOS sys-
tem. (The design of which required interdisciplinary
cooperation.) Basically, this software defines as erro-
neous interbeat intervals with values that are 6% over
or under the average of the three previous ones. Re-
RIGHT BATCH
Fetal Discrimination of Low-Pitched Musical Notes 33
jection and substitution of such intervals by an esti-
mated value are based on several other factors (i.e.,
duration of the next interbeat intervals, phase shifts,
for the total file mean heart rate, etc.).
Interbeat intervals were then converted to weighted
heart rate (HR) measured in beats per minute per sec-
ond (BPM/s). To compute BPM/s, the prestimulus pe-
riod, the first stimulus period, and the second stimulus
period were each divided into 1-s bins from the mo-
ment they began. The number of R-wave events within
each second was counted and the proportion of an
R – R intervals within that second that did not end with
a beat was calculated. For example, a second that con-
tained 2.6 R-wave events was multiplied by 60 to pro-
duce a rate of 156 BPM/s.
Because the procedure used to define the nature and
amplitude of fetal responses requested a temporal con-
tinuity of heart rate data, the averaging process was
computed from the last beat of a period preceding on-
set or change of stimulation in two opposite directions:
backward through this period and forward through the
period immediately following. This ended with two
contiguous series of heart rate values in BPM/s, time
locked on the onset of each stimulation period. Thus,
the dependent variables for the prestimulus, first stim-
ulus, and second stimulus periods were strings of heart
rate in BPM/s.
Because of interfetus heart rate response differ-
ences in direction (acceleratory, deceleratory, bi-
phasic) and in slope, it was inappropriate to analyze
mean group differences. Thus, it was necessary to de-
sign a quantitatively based procedure that would de-
termine whether an individual fetus did or did not react
to a stimulus and to its change. We chose a measure
that was intentionally quite conservative in order to
increase its reliability and validity. Some known
sources of error in this kind of research include the act
of ultrasound recording of fetal cardiac activity itself,
the cardiotocograph’s built-in algorithm, our correc-
tive algorithm, and the conversion from interbeat in-
tervals to HR (BPM/s). Our measure also took into
account the ongoing level of heart rate variability,
which is known to modulate the amplitude of HR re-
activity in newborns (Porges, 1974) and in near-term
fetuses (Lecanuet et al., 1992). This was done using
the following procedure-based comparison of heart
rate variability ranges (see Figure 2, Panel c) that de-
cided whether the heat rate of a short period of time
(Period B) is showing a deceleratory change, an ac-
celeratory change, or no change with regards to the
immediately preceding time period of the same dura-
tion (Period A) (see Figure 2, Panel 2c).
The first quartile (Q1), the third quartile (Q3), and short
the interquartile range (IQR) were computed for the standard
long
DEV (WILEJ) LEFT BATCH

34 Lecanuet et al.

FIGURE 2 Upper Panel 2a shows an episode of low heart rate variability as displayed on the hard
copy printed by the cardiotocograph. Center Panel 2b gives an example of DC responses to the onset
of Note 1 and Note 2. Lower Panel 2c schematically describes, on a DC response, the procedure
used to characterize heart rate changes and measure their amplitude.

last 10 HR measures (BPM/second) of Period A and
for the first 10 HR measures of Period B. A deceler-
ation change (DC) was said to occur in Period B if:
(Q1 of the first 10 s of this period) ⬍ [(HR of the last
second of Period A) ⫺ (IQR of Period A)]. Recipro-
cally, an acceleration change (AC) was said to occur
if: (Q3 of Period B) ⬎ [(HR of the last second of
Period A) ⫹ (IQR of Period A)]. If both criteria were
fulfilled, a biphasic reaction occurred and the decision
was made in favor of the direction of the larger change,
either AC or DC. No change (NC) occurred when nei-
ther the AC or DC criterion was fulfilled. Period A
was the prestimulus period and B was the first stimulus
period when analyzing the effect of the first presen-
tation of a note; Period A was the first stimulus period
and B the second when studying the effect of a second
note or of the continuation of the first one. In addition, short
we computed the maximum cardiac decelerations standard
long
DEV (WILEJ)
(lowest HR of Period B ⫺ HR of last second of Period
A) and maximum cardiac accelerations (highest HR of
Period B ⫺ HR of last second of Period A).
RESULTS
Prestimulation Period
Heart rate characteristics during the last 10 s of the
prestimulation period for the final sample of 116 fe-
tuses are shown in the top panel of Table 1. Fifty-six
had been assigned to the H-note group and 60 to the
L-note group: Their means, medians, IQR of the last
10 measures, and the very last second of the presti-
mulation period did not differ. Therefore, the cardiac
activity of the H-note and the L-note groups was
equivalent before the first stimulus was presented.
First Stimulus Period
The proportion of fetuses in the H (n ⫽ 56) and L
(n ⫽ 60) conditions that showed a DC, an AC, or NC
was equivalent, ␹2(2) ⫽ 0.43, p ⬎ .05 (Table 1). H
and L groups seemed equally responsive to the notes
in that 73% of the fetuses exposed to the low note and
78% of those exposed to the high note displayed either
a DC or AC response. However, the two kinds of HR
changes were not equally likely: Thirty-one of the 44
reactions to the high note and 31 of the 44 reactions
to the low note were of the DC type, ␹2(1) ⫽
14.72, p ⬍ 001. Sixty-two of the 88 reactions (70%)
were of the DC type. Samples of HR decelerations are
shown in Figure 2, Panel b.
A nonstimulated control group of near-term fetuses
(n ⫽ 34) that had been tested with the same procedure
but had not been exposed to sound during the first
stimulation period displayed equally large proportions
of transient spontaneous ACs (15/34) and NCs (16/
34), but spontaneous DCs were rare (3/34; 9%) (ar-
chival data from Lecanuet et al., 1992). Here, 62/116
(53%) showed a DC response, 26/116 (22%) showed
an AC response and 28/116 (24%) showed a NC re-
sponse. The distribution of AC, DC, and NC is differ-
ent from those occurring when low-intensity musical
notes were presented ␹2(2) ⫽ 21.29, p ⬍ .005. A sig-
nificantly higher proportion of DC responses occurs
with the onset of the musical notes than during silence,
␹2(1) ⫽ 16,03, p ⬍ .001.
DC amplitudes ranged from ⫺1.5 BPM to ⫺9.9
BPM and AC amplitudes from 1.3 BPM to 8.9 BPM
(Table 1). Neither the DC mean scores, F(1, 60) ⫽
RIGHT BATCH
Fetal Discrimination of Low-Pitched Musical Notes 35
.011, p ⫽ .91, nor the AC mean scores, F(1, 24) ⫽
.007, p ⫽ .94, varied between the H and L notes.
Second Stimulus Period
Of the 62 who reacted with a HR DC, 14 had been
assigned to the HL group, 20 to the LH group, 17 to
the HH group, and 11 to the LL group (bottom panel
of Table 1). Here we assess whether HR activity of
fetuses in the HL and LH groups, who encountered a
stimulus change, differed from the HR activity of fe-
tuses in the HH and LL control groups, who did not
encounter a stimulus change.
Twenty-four fetuses reacted to the note change with
either an AC or DC response, and 22 of them (90%)
were of the DC type. In contrast, 17/28 control sub-
jects showed a spontaneous HR change at the point
when a note would have changed, but only 2/11 in
subgroup HH and 3/6 in subgroup LL were of the DC
type. The stimulus change condition produced a
disproportionately large number of DC reactions,
␹2(1) ⫽ 17.2, p ⬍ .001.
DC amplitude scores ranged from ⫺1.5 BPM to
⫺9.4 BPM and AC scores from 1.3 BPM to
7.5 BPM. AC changes occurred so infrequently for the
HL and LH conditions that their amplitude could not
be meaningfully compared to each other or to the AC
amplitudes of the control conditions (HH and LL). DC
mean scores of the LH stimulus-change groups did not
differ, F(1, 20) ⫽ .64, p ⫽ 0.43.
DISCUSSION
More than half of all fetuses reacted to the initial onset
of musical notes with a HR deceleration but only a
few nonstimulated control fetuses from an earlier
study did so (Lecanuet et al., 1992). Subsequently,
90% of the fetuses who had responded to the initial
onset of a musical note with a HR deceleration re-
sponded with another HR deceleration when the note
changed. In contrast, the no-change control subjects
who also had responded to the onset of first stimulus
with a DC reaction subsequently showed very few de-
celerations. This difference in the frequency of DC
reactions of fetuses in the experimental and control
conditions indicates that near-term fetuses can dis-
criminate between musical notes that differ primarily
in their constituent frequencies. The present results are
quite similar to those that we observed in the syllable
and male/female voice study. The procedures of the
three studies are also quite similar. Thus, we conclude
with some confidence that the cardiac decelerations short
seen at the onset of the first stimulus and again at the standard
long
DEV (WILEJ) LEFT BATCH

36 Lecanuet et al.

Table 1. Top Panel Shows the Mean, Standard Deviation (SD) of the Mean, Q1–Q3 and Last Sec HR of the Last 10
HR (BPM/S) of the Prestimulation Period for the H and L groups. Middle Panel Shows the Proportion of Cardiac
Decelerations, Accelerations, and No-Changes and Mean HR (SD) Amplitudes (in BPM/s) for the First 10 Sec of First
Stimulus Period for the H and L Groups. The Bottom Panel Shows the Proportion of Cardiac Decelerations,
Accelerations, and No-Changes and Mean HR (SD) Amplitudes (in BPM/s) for the First 10 Sec of the Second
Stimulus Period for the HL, LH, HH, and LL Groups
Prestimulus Period
GP M Median Q1–Q3 Last Sec
HL ⫹ HH 134.5 134.4 1.48 134.7
(7.9) (7.8) (0.73) (7.76)
LH ⫹ LL 136.0 135.9 1.45 136.1
(8.9) (8.9) (0.68) (8.8)
First Stimulus Period
GP DC AC NC
HL ⫹ HH (n ⫽ 56) %subjects 55.4 23.2 21.4
Xamp (SD) ⫺4.3(2.2) 3.6(2.4)
LH ⫹ LL (n ⫽ 60) %subjects 51.7 21.7 26.7
Xamp (SD) ⫺4.2(2.1) 3.7(1.3)
Second Stimulus Period
GP DC AC NC
HL (n ⫽ 14) %subjects 71.4 0 28.6
Xamp (SD) ⫺4.0(1.5)
LH (n ⫽ 20) %subjects 56.5 13.0 30.4
Xamp (SD) ⫺4.6(1.9) [3.9(2.1)]
HH (n ⫽ 17) %subjects 11.8 52.9 35.3
Xamp (SD) [⫺2.9(1.6)] 2.6(0.9)
LL (n ⫽ 11) %subjects 27.3 27.3 45.5
Xamp (SD) ⫺3.5(1.6) 1.1(0.3)
Note. Q1– Q2 ⫽ first quartile– second quartile. BPM/s ⫽ beats per minute per second.

onset of the second stimulus were not random heart
rate changes, but were autonomic cardiac responses.
These responses could be the cardiac component of an
orienting response, a point that is not fully established
by these data.
We have no firm explanation why 24% of the fe-
tuses showed no reaction at the onset of the first stim-
ulus. This number is not far from the one obtained in
the voice discrimination experiment: Twenty-three
percent of subjects exposed to the female voice and
18% of those exposed to the male voice did not re-
spond to these stimuli. Age of subjects, stimulus sound
pressure level, and data processing were similar in the
present experiment and in the voice one. Voices ut-
terances were slightly longer (1.8 s) than the note stim-
ulus (1.15 s). They also show more spectral changes
and their repetition rate was slower (3.5 s) than the
note pulsation rate (3.0 s). These differences could
“theoretically” account for a greater responsiveness to
the voices than to the notes, but the observed differ-
ences were not significant.
However, five possible factors may be relevant to
explain the no-change proportion. First, because our
fetuses ranged from 36 – 39 weeks of age, there may
be interfetus differences in auditory threshold devel-
opment (see Hepper & Shahidullah, 1994 on the on-
togenesis of fetal thresholds). Second, although testing
in State 1F is a methodologically sound practice, the
fact is that fetuses are less reactive in quiet sleep than
in active sleep. Third, State 1F is not a homogenous
state. The responsiveness of the autonomic nervous
system may vary within a 1F episode, which is rea-
sonable if one considers that near-term fetal quiet sleep
is the same as the newborn’s and that newborns’ quiet
sleep periods are not homogenous. EEG parameter
levels vary within quiet sleep and the most pronounced
EEG synchronization takes place at the beginning of
the quiet sleep phase (Salzarulo, Fagioli, Peirano, Bes,
& Schulz, 1991). Furthermore, not all quiet-sleep
phases contain slow wave periods (Bes, Schulz, & Sal-
zarulo, 1987; Schulz, Bes, & Salzarulo, 1989). Fourth,
the effective stimulus may have varied between fe- short
standard
long
DEV (WILEJ)
tuses. Because of the probability of the presence of
standing waves inside the amniotic fluid, perhaps the
ears of some fetuses were located in a minimal sound-
pressure zone during stimulation. (Lecanuet et al.,
1998, Nyman, Arulkumaran, Hsu, Ratnam, Till, &
Westgren, 1991; Richards Frentzen, Gerhardt,
Abrams, & McCann, 1992; Vince et al., 1982; Vince,
Billing, Baldwin, Toner, & Weller, 1985). Finally, the
signal/noise ratio might have been disadvantageous for
fetuses whose heads were near the highly vascularized
placenta (Querleu et al., 1988) or the maternal aorta
(Benzaquen, Gagnon, Hunse, & Foreman, 1990).
Fetal audition is not only constrained by transmis-
sion of external sounds (in particular for short wave
length sound components (frequencies ⬎ 500 Hz))
through various mismatched impedance media to the
amniotic fluid, but also, according to Gerhardt and col-
leagues (Gerhardt et al., 1992; Gerhardt et al., 1996),
by activation of the cochlea through bone conduction.
They found that fetal cochlear “acoustic isolation” in
the fetal sheep (defined as SPL differences required to
record the same amplitude cochlear microphonic po-
tential and outside the uterus) from an external white
noise emitted in air at 90-dB SPL was less than
20 – 30 dB up to 500 Hz, but was 40- to 45-dB SPL
from 500 Hz to 2000 Hz. In our studies, stimulation
was delivered between 90- and 95-dB SPL in air,
which should be — if sheep data can be transferred to
the human species — approximately 50-dB SPL above
fetal cochlear isolation for frequencies above 500 Hz.
For the present study, this gradual increase in “iso-
lation” affects more the C5 note, the highest overtones
of which range up to 3000 Hz (see Figure 1), than the
D4 note the 5th overtone of which peaks around
1700 Hz. This implies in utero note discrimination
cannot rely on the same acoustical cues in the adult
airborne situation. Because the defining low frequen-
cies of the notes suffered less attenuation and isolation
on their way to the fetal cochlea, they are good can-
didates to contribute to the discrimination of the notes.
Perceived loudness equalization of the notes for adults
controlled for this parameter of the situation. This lat-
est study together with our previous ones demonstrates
that fetuses, while in a low heart rate variability state
that quite certainly corresponds to a quiet-sleep epi-
sode, can detect differences between two low-pitched
complex sounds, as they do between pure tones within
the same range (250 – 500 Hz, Shahidullah & Hepper,
1994) with fundamental frequencies under 700 Hz.
These results strengthen the idea that this detection is
performed on F0 and low-frequency spectrum range
differences between sounds.
Postnatal preference studies (Fifer & Moon, 1989;
RIGHT BATCH
Fetal Discrimination of Low-Pitched Musical Notes 37
DeCasper & Fifer, 1980; DeCasper & Spence, 1986;
Moon, Cooper, & Fifer, 1993; Satt, 1984; Spence &
DeCasper, 1987) demonstrated that neonates can dis-
criminate between voices, speech sequences, lan-
guages, and songs to which they were regularly ex-
posed in the prenatal period. In addition, it is known
that the prosodic contours of speech and many of its
phonemic features are preserved in utero when voices
are emitted or spoken over 60-dB SPL (Griffiths,
Brown, Gerhardt, Abrams, & Morris, 1994; Querleu
et al., 1988; Smith Satt, Phelan, & Paul, 1990). Quer-
leu et al. (1988) and Benzaquen et al. (1990) found
that maternal vocalizations are around 10 dB higher
in utero than external male and female voices deliv-
ered at the same external SPL. Similarly, Richards et
al. (1992) observed that the maternal voice is amplified
by 5-dB SPL and that other external voices are atten-
uated only by 2 – 3 dB. The present article contributes
to this body of research and some of our previous data
(DeCasper, Lecanuet, Busnel, Granier-Deferre, &
Maugeais, 1994; Lecanuet et al., 1987; Lecanuet et al.,
1993) which clearly demonstrates that the fetal audi-
tory system can extract enough acoustic information
from sounds to perceive specific auditory patterns.
Specifically, this study confirms fetuses’ ability to dis-
criminate between spectra that lay within the range of
voice F0 and F1 formants, which may be important
during the earliest developmental stages of speech per-
ception. Highly analytical studies such as this one
might yield more definitive assessments of the bases
for fetal perception by studying acoustic features that
are not altered by the transmission through the mater-
nal tissues, such as the temporal pattern of complex
sounds.
NOTES
We express our gratitude to the mothers and to the babies
involved in this study. We also thank Pr. C. Sureau and Pr.
E. Papiernik, respectively, heads of the Baudelocque and the
Port-Royal Cliniques Universitaires, as well as Mrs. C. Con-
trepas and Mrs. M. Coulon, midwives, for welcoming our
group. We are also very grateful to M.-C. Busnel for her
help in the experiments and are especially indebted to S. Mac
Adams for devising the stimuli and to Y. Barbin and C.
Kervella for the HR data treatment software.
REFERENCES
Abrams, R. M., Gerhardt, K. J., & Peters, A. J. M. (1995).
Transmission of sound and vibration to the fetus. In J.-P. short
standard
long
DEV (WILEJ)
38 Lecanuet et al.
Lecanuet, W. P. Fifer, N. A. Krasnegor, & W. P. Smoth-
erman (Eds.), Fetal development: A psychobiological per-
spective (pp. 315– 330). Hillsdale, NJ: Erlbaum.
Armitage, S. E., Baldwin, B. A. & Vince, M. A. (1980). The
fetal sound environment of the sheep. Science, 208,
1173– 1174.
Benzaquen, S., Gagnon, R., Hunse, C., & Foreman, J.
(1990). The intrauterine sound environment of the human
fetus during labor. American Journal of Obstetrics and
Gynecology, 163, 484– 490.
Berg, W. K., & Berg K. M. (1987). Psychophysiological
development in infancy: State, sensory function, and at-
tention. In J. D. Osofsky (Ed.), Handbook of infant de-
velopment (pp. 238– 317). New York: Wiley.
Bes, F., Schulz, H., & Salzarulo, P. (1987). The temporal
interaction of slow wave sleep and paradoxical sleep in
infants. Sleep Research, 16, 168– 175.
Brown, J. W., Morse, P. A., Leavitt, L. A., & Graham, F.
K. (1976). Specific attentional effects reflected in the car-
diac orienting response. Bulletin of the Psychonomic So-
ciety, 7, 1– 4.
Clarkson, M. G., & Berg, W. K. (1983). Cardiac orienting
and vowel discrimination in newborns: Crucial stimulus
parameters. Child Development, 54, 162–171.
DeCasper, A. J., & Fifer, W. P. (1980). Of human bonding:
Newborns prefer their mother’s voice. Science, 208,
1174– 1176.
DeCasper, A. J., & Spence, M. J. (1986). Prenatal maternal
speech influences newborn’s perception of speech sounds.
Infant Behavior and Development, 9, 133–150.
DeCasper, A. J., Lecanuet, J.-P., Busnel, M. C., Granier-
Deferre, C., & Maugeais, R. (1994). Fetal reaction to re-
current maternal speech. Infant Behavior and Develop-
ment, 17, 159– 164.
Eisenberg, R. B. (Ed.) (1976). Auditory competences in
early life: The roots of communicative behavior (pp. 39–
153). Baltimore: University Park Press.
Fifer, W. P., & Moon, C. (1989). Psychobiology of newborn
auditory preferences. Seminars in Perinatology, 13, 430–
433.
Gagnon, R., Benzaquen, S., & Hunse, C. (1992). The fetal
sound environment during vibroacoustic stimulation in la-
bor: Effect on fetal heart rate response. Obstetrics and
Gynecology, 79, 950– 955.
Gerhardt, K. J. (1989). Characteristics of the fetal sheep
sound environment. Seminars in Perinatology, 13, 362–
370.
Gerhardt, K. J., Huang, X., Arrington, K. E., Meixner, K.,
Abrams, R. M. N., and Antonelli, P. (1996). Fetal sheep
in utero hear through bone conduction. American Journal
of Otolaryngology, 17, 374– 379.
Gerhardt, K. J., Otto, R., Abrams, R. M., Ale, J. J., Burch-
field, D. J., & Peters, A. J. M. (1992). Cochear micro-
phonics recorded from fetal and newborn sheep. Ameri-
can Journal of Otolaryngology, 13, 226–233.
Graham, F. K. (1984). An affair of the heart. In M. G. H.
Coles, J. R. Jennings, & J. Stern (Eds.), Psychophysiol-
ogy: A festschrift for John and Beatrice Lacey (pp. 171–
187). New York: Van Nostrand Rheingold.
LEFT BATCH
Graham, F. K., & Jackson, J. C. (1970). Arousal system and
infant heart rate responses. In L. P. Lipsitt & W. H. Reese
(Eds.), Advances in child development and behavior (Vol.
5, pp. 59–117). New York: Academic Press.
Graham, F. K., & Clifton, R. K. (1966.). Heart rate change
as a component of orienting response. Psychological Bul-
letin, 65, 305–320.
Graham, F. K., Anthony, B. J., & Zeigler, B. L. (1983). The
orienting response and developmental processes. In D.
Siddle (Ed.), Orienting and habituation: Perspectives in
human research (pp. 340–371). Sussex, UK: Wiley.
Granier-Deferre, C., Lecanuet, J.-P., Cohen, H., & Busnel,
M.-C. (1985). Feasability of prenatal hearing test. Acta-
Otolaryngology (Stockholm), 421 (Suppl.), 93–101.
Griffiths, S. J., Brown, W. S. Jr., Gerhardt, K. J., Abrams,
R. M., & Morris, R. J. (1994). The perception of speech
sounds recorded within the uterus of a pregnant sheep.
Journal of the Acoustical Society of America, 96, 2055–
2063.
Groome, L. J., Mooney, D. M., Holland, S. B., Bentz, L. S.,
& Atterbury, J. L. (1997). The heart rate deceleratory re-
sponse in low-risk human fetuses: Effect of stimulus in-
tensity on response topography. Developmental Psycho-
biology, 30, 106–113.
Hepper, P. G., & Shahidullah, B. S. (1994). Development of
fetal hearing. Archives of Diseases in Childhood, 71,
F81–F87.
Kisilevsky, B. S., & Low, J. A. (1998). Human fetal behav-
iour: 100 years of study. Developmental Review, 18,
1–29.
Kisilevsky, B. S., Muir, D. W., & Low, J. A. (1989). Human
fetal responses to sound as a function of stimulus inten-
sity. Obstetrics and Gynecology, 73, 971–976.
Leavitt, L. A., Brown, J. W., Morse, P. A., & Graham, E.
K. (1976). Cardiac orienting and auditory discrimination
in 6-week-old Infants. Developmental Psychology, 2,
514–523.
Lecanuet, J.-P., Granier-Deferre, C., Cohen, H., Le Houezec,
R., & Busnel, M.-C. (1986). Fetal responses to acoustic
stimulation depend on heart rate variability pattern, stim-
ulus intensity, and repetition. Early Human Development,
13, 269–283.
Lecanuet, J.-P., Granier-Deferre, C., DeCasper, A. J., Mau-
geais, R., Andrieu, A. J., & Busnel, M.-C. (1987). Per-
ception et discrimination foetales de stimuli langagiers,
mise en évidence à partir de la réactivité cardiaque, ré-
sultats préliminaires [Perception and discrimination of
language stimuli. Demonstration from the cardiac respon-
siveness. Preliminary results]. Compte-Rendus de
l’Académie des Sciences (Proceedings of the Academy of
Sciences), Paris, t. 3025, Série III, 161–164.
Lecanuet, J.-P., Granier-Deferre, C., & Busnel, M.-C.
(1988). Fetal cardiac and motor responses to octave-band
noises as a function of central frequency, intensity, and
heart rate variability. Early Human Development, 18, 81–
93.
Lecanuet, J.-P., Granier-Deferre, C., & Busnel, M.-C.
(1989). Perceptive and discriminative abilities of the near- short
term fetus. Seminars in Perinatology, 13, 421–442. standard
long
DEV (WILEJ)
Lecanuet, J.-P., Granier-Deferre, C., Jacquet, A.-Y., & Bus-
nel, M.-C. (1992). Decelerative cardiac responsiveness to
acoustical stimulation in the near-term foetus. Quarterly
Journal of Experimental Psychology, 44, 279–303.
Lecanuet, J.-P., Granier-Deferre, C., Jacquet, A.-Y., Cap-
poni, I., & Ledru, L. (1993). Prenatal discrimination of a
male and a female voice uttering the same sentence. Early
Development and Parenting, 2, 217– 222.
Lecanuet, J.-P., Gautheron, B., Locatelli, A., Schaal, B., Jac-
quet, A.-Y., & Busnel M.-C. (1998). What sounds reach
fetuses: Biological and nonbiological modeling of the
transmission of pure tones. Developmental Psychobiol-
ogy, 33, 203– 220.
Lecanuet, J.-P., & Schaal, B. (1996). Fetal sensory compe-
tencies. European Journal of Obstetrics and Gynecology
and Reproductive Biology, 68, 1– 23.
Moon, C., Cooper, R. P., & Fifer, W. P. (1993). Two-day-
olds prefer their native language. Infant Behavior and De-
velopment, 16, 495– 500.
Nijhuis, J. G., Prechtl, H. F. R., Martin, C. B., & Bots, R.
S. G. M. (1982). Are there behavioural states in the human
fetus? Early Human Development, 6, 177–195.
Nyman, M., Arulkumaran, S., Hsu, T. S., Ratnam, S. S., Till,
O., & Westgren, M. (1991). Vibroacoustic stimulation
and intrauterine sound pressure levels. Obstetrics and
Gynecology, 78, 803– 806.
Peters, A. J. M., Abrams, R. M., Gerhardt, K. J., Burchfield,
D. J., & Wasserman, D. E. (1992). Resonance of the preg-
nant sheep uterus. Journal of Low Frequency Noise and
Vibration, 11, 1– 6.
Porges, S. W. (1974). Heart rate indices of newborn atten-
tional responsiveness. Merrill-Palmer Quarterly, 20, 231–
253.
Querleu, D., Renard, X., & Versyp, F. (1981). Les percep-
tions auditives du foetus humain [Auditory perception of
the human fetus]. Medecine et Hygiéne, 39, 2101–2110.
Querleu, D., Renard, X., Versyp, T., Paris-Delrue, L., & Ver-
voort, P. (1988). Fetal hearing. European Journal of Ob-
stetric, Gynecology, and Reproduction Biology, 29, 191–
212.
Querleu, D., Renard, X., Boutteville, C., & Crépin, G.
(1989). Hearing by the human fetus? Seminars in Peri-
natology, 13, 430– 433.
Richards, D. S., Frentzen, K. J., Gerhardt, K. J., Abrams, R.
M., & McCann, M. E. (1992). Sound levels in the human
uterus. American Journal of Obstetrics and Gynecology,
166, 186– 190.
Salzarulo, P., Fagioli, I., Peirano, P., Bes, F., & Schulz, H.
(1991). Levels of EEG background activity and sleep
RIGHT BATCH
Fetal Discrimination of Low-Pitched Musical Notes 39
states in the first year of life. In J. Terzano, D. B. Halasz,
& H. V. Declerck (Eds.), Phasic events and dynamic or-
ganization of sleep (pp. 45–56). Raven Press.
Satt, B. J. (1984). On investigation into the acoustical in-
duction of intrauterine learning. PhD dissertation. Cali-
fornia School of Professional Psychology, Los Angeles.
Schmidt, W., Boos, R., Gniers, J., Auer, L., & Schulze, S.
(1985). Fetal behavioral states and controlled sound stim-
ulation. Early Human Development, 12, 145–153.
Schulz, H., Bes, E. & Salzarulo, P. (1989). Rhythmicity of
slow wave sleep: Developmental aspects. In A. Wauquier,
C. Dugovic, & M. Radulovacki (Eds.), Slow wave sleep:
Physiological, pathophysiological, and functional aspects
(pp. 49–60). New York: Raven Press.
Shahidullah, S., & Hepper, P. G. (1994). Frequency discrim-
ination by the fetus. Early Human Development, 36, 13–
26.
Smith, C., Satt, B., Phelan, J. P., & Paul, R. H. (1990). In-
trauterine sound levels: Intrapartum assessment with an
intrauterine microphone. American Journal of Perinatol-
ogy, 7, 312–315.
Sokolov, E. N. (1963). Perception and the conditionned re-
flex. New York: Macmillan.
Spence, M. J., & DeCasper, A. J. (1987). Prenatal experience
with low frequency maternal voice sounds influences neo-
natal perception of maternal voice samples. Infant Behav-
ior and Development, 10, 133–142.
Vince, M. A., Armitage, S. E., Baldwin, B. A., Toner, Y.,
& Moore, B. C. J. (1982) The sound environment of the
fetal sheep. Behaviour, 81, 296–315.
Vince, M. A., Billing, A. E., Baldwin, B. A., Toner, J. N.,
& Weller, C. (1985). Maternal vocalizations and other
sounds in the fetal lamb’s sound environment. Early Hu-
man Development, 11, 179–190.
van Woerden, E. E., van Geijn, H. P., Caron, F. J. M., van
der Valk, A. W. Swartjes, J. M., & Arts, N. F. T. H.
(1988a). Fetal mouth movements during behavioural
States 1F and 2F. European Journal of Obstetrics and
Gynecology and Reproductive Biology, 29, 97–103.
van Woerden, E. E., van Geijn, H. P., Swartjes, J. M., Caron,
F. J. M., Brons, J. T. J., & Arts, N. F. T. H. (1988b). Fetal
heart rhythms during behavioural State 1F. European
Journal of Obstetrics and Gynecology and Reproductive
Biology, 28, 29–38.
Zimmer, E. Z., Fifer, W. P., Kim, Y.-I., Rey, H. R., Chao,
C. R., & Myers, M. M. (1993). Response of the premature
fetus to stimulation by speech sounds. Early Human De-
velopment, 33, 207–215.
short
standard
long