See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/15823029 DeCaspar, A.J. & Fifer, W.P. Of human bonding: newborns prefer their mothers' voice. Science 208, 1174-1176 Article in Science · July 1980 DOI: 10.1126/science.7375928 · Source: PubMed CITATIONS READS 1,301 7,185 2 authors: Anthony J. Decasper William P Fifer University of North Carolina at Greensboro Columbia University 20 PUBLICATIONS 3,086 CITATIONS 173 PUBLICATIONS 5,254 CITATIONS SEE PROFILE Some of the authors of this publication are also working on these related projects: Safe Passage Study View project Safe Passage Study View project All content following this page was uploaded by Anthony J. Decasper on 03 September 2014. The user has requested enhancement of the downloaded file. SEE PROFILE Of Human Bonding: Newborns Prefer their Mothers' Voices Author(s): Anthony J. DeCasper and William P. Fifer Source: Science, New Series, Vol. 208, No. 4448 (Jun. 6, 1980), pp. 1174-1176 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/1683733 Accessed: 28/01/2009 14:55 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aaas. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org. American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science. http://www.jstor.org evidence for this in the sheep after day 100 of gestation (8, 9). A further question concerns the efficiency of the hearing mechanism within a totally fluid environment; the mammalian fetus is known to move in response to sound from outside the mother (10), and in the guinea pig, prenatal exposure to a specific sound changes the neonate's response to the sound (11). 50Implications for the human fetus are not clear because of postural, placental, and other anatomical differences. The main difference between results of re20 100 1000 20 100 1000 search with the goat (7) and ours can be Frequency(Hz) attributed to our use of a method of Fig. 2. Root mean squarevalues at peaks from spectrumanalysesfor differenttypes of internal greater physiological validity, namely, sound:0, gurgle;*, swallow; A, blowing;A, chewingcud; l, rushingnoise; *, quietewe; x, recording from inside the intact amnion intestinalnoise; +, eating hay; 0, drinking;*, eating nuts. in the fetus's normal fluid environment. We suggest that the auditory experience of the fetal mammal may be considerably it was reduced below and above this fre- sounds from outside the animal without more extensive, more varied, and, as in quency and at higher frequencies re- being able to hear the reported pulsa- birds, possibly of greater postnatal sigmained at about 20 dB up to the highest tions from inside. It is possible, and con- nificance than has been believed. recorded, 5 kHz. The amount of attenua- sistent with spectrum analyses, that SALLYE. ARMITAGE tion fluctuated, however: conversation these sounds occur at very low frequenB. A. BALDWIN at normal levels outside the animal could cies and, when attenuated, are below the MARGARET A. VINCE Institute of Animal Physiology, often, but not always, be understood human threshold for sound. when transmitted from inside. Raised Our results suggest that sounds avail- Cambridge CB2 4AT, England voices were almost always distinct. able to the sheep fetus, within its normal References Sounds generated within the ewe her- fluid environment, are varied and of 1. B. Tschanz, Z. Tierpsychol. 4, 1 (1968). self were picked up by the implanted hy- rather low frequency when they are gen2. J. Bench, J. Gen. Psychol. 113, 85 (1968). 3. J. C. Grimwade, D. W. Walker, C. Wood, Aust. drophone, amplified, and recorded by a erated by, or within, the mother. ExJ. Ment. Retard. 2, 63 (1970). tape recorder or fed directly to the spec- ternal sounds are attenuated by about 16 4. H. Murooka, Y. Koie, N. Suda, J. Gynecol. Obstet. Biol. Reprod. 5, 367 (1976). trum analyzer (as the amplifier had two to 37 dB, most attenuation occurring at 5. L. Salk, Trans. N. Y. Acad. Sci. 24, 753 (1962). outputs we were able to listen while ana- frequencies around 1 kHz. In the sheep, 6. D. Walker, J. Grimwade, C. Wood, Am. J. Obstet. Gynecol. 109, 91 (1971). lyzing the sounds). In the main, sounds external sounds of above 65 dB at the 7. R. J. Bench, J. H. Anderson, M. Hoare, J. heard were characteristic and identi- body wall should often penetrate to the Acoust. Soc. Am. 47, 1602 (1970). 8. C. G. Bernhard, I. H. Kaiser, G. M. Kolmodin, fiable: drinking, eating, swallowing, ru- uterus. Acta Physiol. Scand. 47, 333 (1959). The extent to which sound signals inmination, and sometimes heavy breath9. G. Gottlieb., in The Biopsychology of Development, E. Tobach, L. R. Aronson, E. Shaw, Eds. ing could be heard. Rumination, unex- side the amniotic sac are heard by the fe(Academic Press, New York, 1971), pp. 67-128. 10. E. W. Rubel, in Handbook of Sensory Physiolowas tus is rather A another pectedly, quiet. question currently being rushing gy, M. Jacobson, Ed. (Springer, Berlin, 1978), sound sometimes accompanied move- explored; in precocial mammals, the vol. 9, pp. 133-237. 11. M. A. Vince, Anim. Behav. 27, 908 (1979). ment by the ewe and irregular gurgles, auditory system is believed to become 13 November 1979; revised 11 March 1980 probably of digestive origin, occurred functional well before birth, and there is frequently. Periods of quiet were not unusual. Figure 2 shows root mean square values at peaks taken from spectrum analyses, for different types of internal Of Human Bonding: Newborns Prefer Their Mothers' Voices sound. These were of low frequency, Abstract. By sucking on a nonnutritive nipple in different ways, a newborn human tailing off above 500 Hz. Although we found attenuation of ex- could produce either its mother's voice or the voice of anotherfemale. Infants learned ternal sounds to be less than in other spe- how to produce the mother's voice and produced it more often than the other voice. cies (2, 3, 6), the loudness peaks and fre- The neonate's preference for the maternal voice suggests that the period shortly after quency of internal sounds were similar to birth may be important for initiating infant bonding to the mother. those recorded by others (3, 6). HowHuman responsiveness to sound be- ability of a newborn to discriminate its ever, average sound levels were lower than those previously reported (2, 3, 6, gins in the third trimester of life and by mother's voice from that of other females. However, evidence for differ7), especially as we often observed peri- birth reaches sophisticated levels (1), esods of quiet. In one particular our results pecially with respect to speech (2). Ear- ential sensitivity to or discrimination were at variance with those of other ly auditory competency probably sub- of the maternal voice is available only for workers: we were not able to hear serves a variety of developmental func- older infants for whom the bonding prosounds from the maternal cardiovascular tions such as language acquisition (1, 3) cess is well advanced (6). Therefore, the system. By holding a hydrophone firmly and mother-infant bonding (4, 5). role of maternal voice discrimination in against the ewe's skin in the brachial Mother-infant bonding would best be formation of the mother-infant bond is area we were able to pick up heart served by (and may even require) the unclear. If the newborn's sensitivities to 1174 0036-8075/80/0606-1174$00.50/0 Copyright ? 1980 AAAS SCIENCE, VOL. 208, 6 JUNE 1980 -2 I 0 I -1 I +1 I 5.8 4.4 4.3 4.5 +2 I +3 +4 I..I=.. corded during the next 5 minutes, but voices were never presented. This baseline period was used to determine the median interburstinterval (IBI) or time elapsingbetween the end of one burst of I A" suckingandthe beginningof the next (9). 0. A burst was defined as a series of indi- 0 0. vidual sucks separatedfrom one another with the than 2 seconds. less 4) Testing by voices began after the baseline had been established. 0 For five randomly selected infants, 2t t .2t sucking burst terminatingIBI's equal to Interburst intervals or greater than the baseline median (t) 2. Interburst interval per opportunity producedonly his or her mother'svoice Fig. when the maternal voice followed intervals (IBI rt), and bursts terminatinginter- less than the baseline median(solid line) and 0 3.7 3.4 I 1.0 3.8 3.3 C: 1-U 4.5 I I I I I I I I I vals less man ncemeuian prouuceu onmy Mother Other the voice of anotherinfant'smother(10). Fig. 1. For each subject, signed difference scores between the medianIBI's withoutvo- Thus, only one of the voices was precal feedback (baseline) and with differential sented, stereophonically, with the first vocal feedback(session 1). Differencesof the suck of a burstand remainedon untilthe four reversal sessions (*) are based on medi- burst ended, that is, until 2 seconds ans with differentialfeedback in sessions 1 and 2. Positive values indicate a preference elapsed without a suck. For the other for the maternalvoice and negative values a five infants, the conditions were repreferencefor the nonmaternalvoice. Filled versed. Testing lasted 20 minutes. bars indicatethat the mother'svoice followed A preference for the maternalvoice IBI's of less than the baseline median;open was indicated if the infant produced it bars indicatethat her voice followed intervals equal to or greaterthan the median. Median more often than the nonmaternalvoice. IBI's of the baseline (in seconds) are shown However, unequal frequencies not inopposite the bars. dicative of preference for the maternal intervalsequal to or greater than the median (dashed line). The IBI's are representedon the abscissa by the lower bound of interval classes equal to one-fifththe baseline median (t). voice per se could result either because short (or long) IBI's were easier to prospeech subserves bonding, discrimina- duce or because the acoustic qualitiesof tion of and preference for the maternal a particularvoice, such as pitch or invoice should be evident near birth. We tensity, renderedit a more effectiveform now report that a newborn infant young- of feedback. The effects of response reer than 3 days of age can not only dis- quirements and voice characteristics criminate its mother's voice but also will were controlled (i) by requiringhalf the work to produce her voice in preference infants to respond after short IBI's to to the voice of another female. produce the mother's voice and half to The subjects were ten Caucasian neo- respondafterlong ones and (ii) by having nates (five male and five female) (7). each maternal voice also serve as the Shortly after delivery we tape-recorded nonmaternalvoice for anotherinfant. Preferencefor the mother'svoice was the voices of mothers of infants selected for testing as they read Dr. Seuss's To shown by the increase in the proportion of IBI's capable of producingher voice; Think That I Saw It On Mulberry Street. Recordings were edited to provide 25 the medianIBI's shifted from their baseminutes of uninterrupted prose, and test- line values in a direction that produced ing of whether infants would differ- the maternal voice more than half the time. Eight of the ten medians were entially produce their mothers' voices shifted in a direction of the maternal began within 24 hours of recording. Sessions began by coaxing the infant to a voice (mean = 1.90 seconds, a 34 perstate of quiet alertness (8). The infant cent increase) (sign test, P = .02), one shiftedin the directionthat producedthe was then placed supine in its basinette, earphones were secured over its ears, nonmaternalvoice more often, and one and a nonnutritive nipple was placed in median did not change from its baseline its mouth. An assistant held the nipple value (Fig. 1). If these infants were working to gain loosely in place; she was unaware of the experimental condition of the individual access to their mother's voice, reversing infant and could neither hear the tapes the response requirementsshould result nor be seen by the infant. The nipple was in a reversalof their IBI's. Four infants, connected, by way of a pressure trans- two from each condition, who produced ducer, to the solid-state programming their mother's voice more often in sesand recording equipment. The infants sion 1 were able to complete a second were then allowed 2 minutes to adjust to session 24 hours later, in which the rethe situation. Sucking activity was re- sponse requirementswere reversed(11). The average magnitude of the difference 6 JUNE 1980 Differential feedback in session 2 began immediately after the 2-minute adjustment period. The criterion time remained equal to the baseline median of the first session. For all four infants, the median IBI's shifted toward the new cri- terionvalues and away fromthose which voice. previously producedthe nmaternal between the medians of the first and reversal sessions was 1.95 seconds. Apparentlythe infant learned to gain access to the mother's voice. Since specific temporalpropertiesof suckingwere required to produce the maternal voice, we sought evidence for the acquisition of temporally differentiated responding. Temporal discrimination within each condition was ascertained by constructing the functionfor IBI per opportunity: IBI's were collected into classes equal to one-fifth the baseline median, and the frequency of each class was divided by the total frequency of classes having equal and larger values (12). When IBI's less than the baseline median were required, the likelihood of terminatinginterburstintervalswas highest for classes less than the median (Fig. 2), whereas when longer intervals were required, the probability of terminating an IBI was maximalfor intervalsslightlylongerthan the median. Feedback from the maternal voice effectively differentiatedthe temporal character of responding that produced it: the probability of terminating IBI's was highest when termination resulted in the maternal voice. Repeating the experiment with 16 female neonates and a different discrimination procedure confirmed their preference for the maternal voice (13). The dis- criminativestimuliwere a 400-Hztone of 4 seconds duration(tone) and a 4-second 1175 period of silence (no tone). Each IBI contained an alternating sequence of tone-no-tone periods, and each stimulus was equally likely to begin a sequence. For eight infants, a sucking burst initiated during a tone period turned off the tone and produced the Dr. Seuss story read by the infant's mother, whereas sucking bursts during a no-tone period produced the nonmaternal voice. The elicited voice remained until the sucking burst ended, at which time the tone-notone alternation began anew. The discriminative stimuli were reversed for the other eight neonates. Testing with the voices began immediately after the 2minute adjustment period and lasted 20 minutes. Each maternal voice also served as a nonmaternal voice. During the first third of the testing session, the infants were as likely to suck during a stimulus period correlated with the maternal voice as during one correlated with the nonmaternal voice (Table 1). However, in the last third of the session the infants sucked during stimulus periods associated with their mother's voice approximately 24 percent more often than during those associated with the nonmaternal voice, a significant increase [F(l, 14) = 8.97, P < .01]. Thus, at the beginning of testing there was no indication of stimulus discrimination or voice preference. By the end of the 20minute session, feedback from the maternal voice produced clear evidence of an auditory discrimination; the probability of sucking during tone and no-tone periods was greater when sucking produced the maternal voice. The infants in these studies lived in a group nursery; their general care and night feedings were handled by a number of female nursery personnel. They were fed in their mothers' rooms by their mothers at 9:30 a.m. and at 1:30, 5:00, and 8:30 p.m. At most, they had 12 hours of postnatal contact with their mothers before testing. Similarly reared infants prefer the human voice to other acoustically complex stimuli (14). But, as our data show, newborns reared in group nurseries that allow minimal maternal contact can also discriminate between their mothers and other speakers and, moreover, will work to produce their mothers' voices in preference to those of other females. Thus, within the first 3 days of postnatal development, newborns prefer the human voice, discriminate between speakers, and demonstrate a preference for their mothers' voices with only limited maternal exposure. The neonate's capacity to rapidly acquire a stimulus discrimination that con1176 View publication stats Table 1. Mean (X) and standard deviation (S.D.) of the relative frequency of sucking duringa stimulusassociatedwith the maternal voice divided by the relative frequency of suckingduringa stimulusassociatedwith the nonmaternalvoice. A ratioof 1.0 indicatesno preference. Stimulus associated with maternal voice Tone No tone Combined First third Last third * S.D. 0.97 1.04 1.00 .33 .31 .32 1.26 1.22 1.24 .33 .19 .27 trols behavior (15) could provide the means by which limited postnatal experience with the mother results in preference for her voice. The early preference demonstrated here is possible because newborns have auditory competencies adequate for discriminating individual speakers: they are sensitive to rhythmicity (16), intonation (17), frequency variation (1, 13), and phonetic components of speech (18). Their general sensory competency may enable other maternal cues, such as her odor (19) and the manner in which she handles her infant (20), to serve as supporting bases for discrimination and vocal preference. Prenatal (intrauterine) auditory experience may also be a factor. Although the significance and nature of intrauterine auditory experience in humans is not known, perceptual preferences and proximity-seeking responses of some infrahuman infants are profoundly affected by auditory experience before birth (21). ANTHONY J. DECASPER WILLIAM P. FIFER Department of Psychology, University of North Carolina at Greensboro, Greensboro 27412 References and Notes 1. R. B. Eisenberg, Auditory Competence in Early Life: The Roots of Communicative Behavior (UniversityParkPress, Baltimore,1976.) 2. P. D. Eimas, in Infant Perception: From Sensation to Cognition, L. B. Cohen and P. Sala- patek, Eds. (AcademicPress, New York, 1975), vol. 2., p. 193. 3. B. Friedlander,Merrill-PalmerQ. 16, 7 (1970). 4. R. Bell, in The Effect of the Infant on Its Care- giver, M. Lewis and L. A. Rosenblum,Eds. (Wiley,New York, 1974),_p.1; T. B. Brazelton, E. Tronick,L. Abramson,H. Als, S. Wise,Ciba Found. Symp. 33, 137 (1975). 5. M. H. Klausand J. H. Kennel,MaternalInfant Bonding (Mosby, St. Louis, 1976); P. DeChateau, Birth Family J. 41, 10 (1977). 6. M. Milesand E. Melvish,Nature(London)252, 123 (1974);' J. Mehler, J. Bertoncini, M. Bauri6re,D. Jassik-Gershenfeld,Perception7, 491 (1978). 7. The infantswere randomlyselected from those meetingthe followingcriteria:(i) gestation,full term; (ii) delivery, uncomplicated;(iii) birth weight, between 2500 and 3850grams;and (iv) APGARscore, at 'least eight at I and 5 minutes after birth. If circumsized,males were not observed until at least 12 hours afterward.Informed writtenconsent was obtainedfrom the mother,and she was invitedto observethe testing procedure.Testing sessions beganbetween 2.5 and 3.5 hours after the 6 a.m. or 12 p.m. feeding.All infantswere bottle-fed. 8. P. H. Wolff,Psychol. Issues 5, 1 (1966).The infants were held in front of the experimenter's face, spoken to, and then presentedwith the nonnutritivenipple.Infantsfailingto fixatevisually on the experimenter'sface or to suckon the nipple were returnedto the nursery.Once begun, a session was terminatedonly if the infant cried or stopped sucking for two consecutive minutes. The intitial sessions of two infants were terminatedbecause they cried for 2 minutes. Their data are not reported. Thus, the results are based on 10 of 12 infants meetingthe behavioralcriteriafor enteringand remainingin the study. 9. With quiet and alert newborns, nonnutritive suckingtypicallyoccurs as burstsof individual sucks, each separatedby a second or so, while the bursts themselves are separatedby several seconds or more. Interburstintervalstend to be unimodallydistributedwith modal values differing amonginfants. [ K. Kaye, in Studies in Mother-Infant Interaction, H. R. Schaffer, Ed. (Academic Press, New York, 1977)]. A suck was said to occur when the negativepressure exerted on the nipple reached20 mm-Hg.This value is almost always exceeded duringnonnutritivesuckingby healthyinfants,butis virtually never produced by nonsucking mouth movement. 10. The tape reels revolved continuously,and one or the other of the voices was electronically switched to the earphones when the response thresholdwas met. Becausethe thresholdswere detectedelectronically,voice onset occurredat the moment the negative pressure reached 20 mm-Hg. 11. Two infantswere not tested a second time, because we could not gain access to the testing room, which served as an auxillarynurseryand as an isolationroom.The sessionsof two infants who cried were terminated.Two other infants were tested a second time, but in theirfirstsession one had shown no preferenceandthe other had shown only a slightpreferencefor the nonmaternalvoice. Their performancemay have been affectedby inconsistentfeedback.Because their peak sucking pressures were near the thresholdof the apparatus,very similarsucks would sometimesproducefeedbackand sometimes not, and sometimes feedback would be terminatedin the midstof a suckingburst.Consequently, second session performances of these two infants, which were much like their initialperformances,were uninterpretable. 12. D. Anger, J. Exp. Psychol. 52, 145 (1956). 13. Three other infants began testing with the voices, but their sessions were terminatedbecause they cried. Their data are not included. This studyis partof a doctoralthesis submitted by W.P.F. 14. E. Butterfieldand G. Siperstein,in OralSensation and Perception: The Mouth of the Infant, J. Bosma, Ed. (Thomas,Springfield,Ill., 1972). 15. E. R. Siquelandand L. P Lipsitt,J. Exp. Child. Psychol. 3, 356 (1966);R. E. Kron, in Recent Advances in Biological Psychiatry, J. Wortis, Ed. (Plenum,New York, 1967),p. 295. 16. W. S. Condonand L. W. Sander,Science 183, 99 (1974). 17. R. B. Eisenberg,D. B. Cousins, N. Rupp, J. Aud. Res. 7, 245 (1966); P. A. Morse, J. Exp. Child. Psychol. 14, 477 (1972). 18. E. C. Butterfieldand G. F. Cairns,in Language Perspectives: Acquisition, Retardation and In- tervention, R. L. Schiefelbusch and L. L. Lloyd, Eds. (UniversityParkPress, Baltimore, 1974),p. 75; A. J. DeCasper,E. C. Butterfield, G. F. Cairns,paperpresentedat the fourthbiennial conferenceon HumanDevelopment,Nashville, April 1976. 19. A. MacFarlane, Ciba Found. Symp. 33, 103 (1975). 20. P. Burns,L. W. Sander,G. Stechler,H. Julia,J. Am. Acad. Child Psychiatry 11, 427 (1972); E. B. Thoman,A. F. Korner,L. Bearon-Williams, Child Dev. 48, 563 (1977). 21. G. Gottlieb, Development of Species Identification in Birds: An Inquiry into the Prenatal Determinants of Perception (Univ. of Chicago Press, Chicago, 1971);E. H. Hess. Imprinting (Van Nostrand-Reinhold,New York, 1973). 22. Supportedby Research Councilgrant920. We thankthe infants,their mothers,andthe staffof Moses CaneHospital,wherethis workwas performed,andA. Carstensfor helpingconductthe research. 26 February1980 SCIENCE, VOL. 208
Purchase answer to see full attachment