and dissonance

8/13/2019 and dissonance

1/10


2/10


3/10

Consonance and Dissonance 485

quency separation between them that does notgive rise to dissonant effects. The critical bandis roughly a little lower than a minor third (3semitones). This means that, for pure sounds,any interval except a full tone (2 semitones) isnot dissonant. This generalization does nothold for more complex sounds because theymay have pairs of upper harmonics that fall inthe critical band. The theory implies that com-plex tones related by complex frequency ratiosare more likely to give rise to dissonant effectsnot because of the complexity of frequencyratios in itself but because of the greater num-ber of overlapping critical bands among adja-cent harmonics that characterizes intervalswith complex frequency ratios.Although Helmholtz and later physicistsbelieved that consonance judgments were thepsychological result of physical-acousticallaws operating on inborn properties of theauditory system (Plomp & Levelt, 1965; Ter-hardt, 1984), there is, at present, no proof ofthis biological preparedness. One could arguethat psychoacoustic laws, as posited by Helm-holtz and others, could reflect an acquired,rather than an inborn, form of auditory pro-cessing. Indeed, some investigators haveargued that consonance judgments areacquired competences based on exposure tothe music of a particular culture (Frances,1988; Lundin, 1985). This view is popularamong modern music theorists and composerswho adopt a sceptical stance toward an abso-lute notion of consonance (e.g. Boulez, 1971).This nurture view can be traced to Schdn-bergs declaration that the concept of conso-nance has no useful meaning. Not only didSchiinberg proclaim the emancipation of dis-sonance, but he also proscribed an elimina-tion of consonances (Schonberg, 1984).

Different strategies can be chosen to probethe origins of our perception of consonanceand dissonance. One strategy is to comparejudgments of subjects from cultures with dif-ferent musical systems. If consonance judg-ments are primarily a function of exposure weshould find variability in judgments across cul-tures. Unfortunately, evidence from such

cross-cultural studies is scant and ambiguous.Consonance judgements were invariant amongAmericans and Japanese (Butler & Daston,1968), but dissimilar among Canadians andIndians. Indians showed greater tolerancetoward dissonant intervals (Maher, 1976).Hence, the cross-cultural study of consonancepreferences has not yet proved to be informa-tive. Although additional evidence would be ofinterest, it is difficult, if not impossible, to findcultures that are totally isolated from Westernmusic.

A second strategy is to study the reactionsof animals to consonance and dissonance.Borchgrevink (1975) examined such conso-nance preferences in 34 albino rats. He con-nected a tape recorder to a test chambercontaining a loudspeaker and two pedals.When the rat pressed one pedal a consonantchord was heard, pressing the other pedal wasfollowed by a dissonant chord (unfortunately,the chords were not specified in the article).The association between the consonant anddissonant chords and the position of the pedalswas randomly distributed among animals.Each rat spent 15 min a day in the test chamberin a stipulated sequence for three weeks. Apreference for each animal was defined as thedifference between the number of presses onthe two pedals. The results show that the ratsdeveloped a consonance preference. After aninitial period of small and unreliable differ-ences, the number of presses on the conso-nance producing pedal increased, and was atthe end almost twice the number of presses onthe dissonance producing pedal (Borchgre-vink, 1975). Although a generalization fromrats to humans is speculative-and the priormusical experience of the rats is not known-this study provided support for a possible bio-logical basis for consonance preference.

In a more recent study European Starlings(Sturnus Vulgaris) were trained to peck at onekey when a consonant chord was presentedand at another key when a dissonant chord waspresented. The birds generalized the responseto a new pair of consonant and dissonantchords (Hulse, Bernard & Braaten, 1995). This


4/10

486 INFANT BEHAVIOR DEVELOPMENT Vol. 21, No. 3,1998

result suggests that consonance and disso-nance may be an important cue for auditorycommunication among songbirds.

The problems with conducting cross-cul-tural studies and the interpretation of animalwork motivate a search for other strategies.The study of young infants provides an oppor-tunity for the exploration of auditory predispo-sitions in humans. Although there is a body ofliterature on infants music perception, studiestypically focused on infants processing capa-bilities rather than preferences and/or affectiveresponses (Trehub & Trainor, 1993). More-over, these studies of infants interval percep-tion have examined responses to melodicintervals where tones are presented succes-sively, rather than harmonic intervals wheretones are presented simultaneously.

One of the exceptions is a study by Crow-der, Reznick and Rosenkrantz (1991) origi-nally designed to test 6 month-old infantsreactions to major and minor chords. Althoughthere were no differences to major and minormodes, the introduction of a dissonant chordseemed to provide reliable differences withprolonged visual fixation on the loudspeaker tothe consonant chords. Because the study wasnot designed to test infants reactions to conso-nance and dissonance, the number of infants inthe consonance/dissonance condition was verysmall (n = 9).

In a recent experiment by Schellenberg andTrehub (1996) nine month old infants wereable to detect changes from harmonic intervalswith simple frequency ratios to intervals withcomplex frequency ratios, but failed to detectchanges from complex to simple frequencyratios. The particular intervals used were thefourth and the fifth (= simple frequency ratios)as opposed to the tritone (= complex frequencyratio). The authors concluded that infants pos-sessed a greater facility to discriminate inter-vals with simple compared to intervals withcomplex frequency ratios. This result is inter-esting as it indicates that the distinctivenessand memorability of certain intervals mayunderlie a preference.

The present experiment was designed spe-cifically to test the hypothesis of an innatepreferential bias favoring consonance over dis-sonance. Unlike studies that used isolated pairsof intervals, infants were presented with actualmusic. Two different, unfamiliar melodieswere composed with a synthesizer, and a con-sonant and dissonant version was created foreach melody. Because our interest was in theinfants preference, and not in perceptual capa-bility, the dependent measures of visual fixa-tion of the music source, motor activity,vocalization, fretting and turning away fromthe music source were all coded as relevantindicators. It was expected that the infantswould be more attentive and show less fre-quent signs of distress to the consonant com-pared to the dissonant melodies.

METHOD

Part ic ipantsThirty-two full term, healthy 4 months old

infants (16 males, 16 females) participated inthe experiment (age range 16.7-21.1 weeks, M17.9 weeks). Nine additional infants weretested but excluded because they became fussy(7) or because of experimenter error (2). Theparents were middle-class Caucasian residentsof the Boston, Massachusetts metropolitanarea and were recruited through the birthrecords of local city halls.

Stimul iEach infant was presented with two different

melodies, each 3.5 set duration, in a consonantand a dissonant version for a total of 4 trials. Theconsonant and dissonant versions of the twomelodies were identical with regard to param-eters such as pitch, tempo, rhythm, timbre, con-tour. The dissonant version was composed inparallel minor seconds (interval size of 1 semi-tone), the consonant version in parallel majorand minor thirds (interval size of 4 and 3 semi-tones respectively), using only two synthetic


5/10


6/10

488 INFANT BEHAVIOR DEVELOPMENT Vol. 21, No. 3,1998

and was covered with an attractive pattern ofblack and white concentric circles. A videocamera recorded the infants behavior. Therewas a wall in front and to the left of the infant.A research assistant and the parent, whoremained in the room, sat behind the infant,outside of the infants visual field. Initially,there was a 35 set quiet baseline followed bythe 4 melodies with a quiet interval betweeneach melody of 8 sec. The order of the presen-tation of the melodies was counterbalanced(Latin Square design). The music was deliv-ered at a SPL of 60dB(A).

easuresThe videotape record was coded for (1) fix-

ation time directed at speaker (2) motor activ-ity (3) turning away from the speaker (4)vocalization and (5) fretting. The fixation time

variable was total fixation time during the 35set of each trial, the duration of the initial fix-ation following the onset of each trial, and theduration of fixation on speaker during theinterstimulus intervals. Motor activity wasdefined as the total duration of (a) flexion ofone or both arms and legs of at least 60degrees, (b) movement of one or both arms andlegs by at least 60 degrees to the right or leftand more than 2.5 cm in vertical direction.Turning away was defined as the frequency ofturning the head away from the pattern of con-centric circles by 90 degrees or more. Vocal-izations, non-distress vocal sounds, wasdefined as the frequency of vocalizations pertrial. There had to be a quiet period of at least3 set between two vocalizations in order tocode them as separate. Fretting, defined as dis-tressed vocal sounds, was coded in a similarmanner to vocalizations.

1312

Me 11an 10Tim

8e 7c

6

4 5 Melody A

Visual Fixation4

-Mel

Motor Activity

aelody A TAelody BCl Consonantm Dissonant I

FIGURE 3Left side: Mean total visual fixation time to music source during the consonant and dissonant versions oftwo melodies Right side: Mean time of motor movement during the consonant and dissonant versions ofthe two melodies Error bars represent standard errors


7/10

Consonance and Dissonance

Reliability of the coders was assessed byhaving two persons independently code thesevariables for 20 of the subjects. Pearsonproduct moment correlation coefficients were.94 for fixation time, .91 for motor activity, .96for turning away and .89 for vocalizations andfor fretting.

In addition, each parent filled out a ques-tionnaire about listening habits, musical back-ground and education, and their memories ofthe type of music the infants might have heardat home.

RESULTS

Visual Fixat ion and Motor Act iv i tyA repeated measures analysis of vari-

ance, with sex and order as the between-sub-ject variables and melody (A or B) andmode (consonant or dissonant) as therepeated measure, was implemented foreach dependent variable. A square root trans-formation was performed because the scoreswere not normally distributed. The repeatedmeasures ANOVA yielded a significantmain effect of musical mode on total tixa-tion time [F(1,24) = 11.50 p < 0.005], butalso on duration of initial fixation [F( 1,24)= 4.72 p < 0.051 and fixation time duringthe 8 set intervals [F(1,24) = 5.30p < 0.041. The infants looked significantlylonger at the speaker for the consonant, com-pared with the dissonant, version of eachmelody (see Figure 3).Infants also moved sigificantly more duringthe dissonant compared with the consonantversions of the melodies [F( 1,24) = 6.64p c 0.021 (see Figure 3). There was no signifi-cant effect for melody (A or B), sex, or orderof presentation. There was an unexpected, butsignificant melody x sex interaction[F(1,24) = 15.40 p c 0.0011. Girls were moreactive motorically during melody B, boys weremore active during melody A. Melody B had afaster tempo than melody A.

Vocal izat ions Frett ing and Avo idanceEight of 32 infants fretted or showed avoid-

ance during the dissonant versions, but neitherfretted nor avoided during the consonant ver-sions. No infant fretted or avoided only duringthe consonant versions but not during the dis-sonant versions. The probability of this resultoccurring by chance is p < 0.01 by the bino-mial theorem. Similarly, 7 infants vocalizedduring the consonant, but not the dissonantversions, while 1 infant vocalized during thedissonant but did not do so during the conso-nant versions [binomial test: p < 0.051.

Musical Backgro und VariablesIn order to examine possible effects of the

infants and parents musical experience wecreated a new variable by subtracting total fix-ation time to the dissonant versions from totalfixation time to the consonant versions: Thelarger this difference score, the longer theinfant fixated the concentric circles during theconsonant melodies. Similarly, we subtractedthe total motor score to the consonant melodiesfrom the total motor scores to the dissonantones: The larger the difference, the more activethe infant was during the dissonant conditions.Pearson product-moment correlations werecomputed between infants exposure to musicat home (average hours/week) and these differ-ence scores for fixation time and motor activ-ity: All correlations were nonsignificant(fixation: r = -.02; motor activity: r = -.lO).Similarly, there were no significant correla-tions between the infants fixation time ormotor behavior and either the parents musicalbackground or practice of a musical instrument(fixation: r = -.22; r = -.08 - motor activity:r = .19; r = -.06).

DISCUSSIONInfants looked significantly longer at thespeaker and were less motorically active whenhearing the consonant as compared to the dis-


8/10


9/10

Consonance and Dissonance

nance might be subject to a much stronger cul-tural impact.

Our primary objective in the present inves-tigation has been to examine the hypothesis ofan innate bias favoring consonance over disso-nance. We can offer no explanations, onlyspeculation, on the nature of the innate biasthat renders consonance perceptually moreattractive than dissonance. A number of inves-tigators contend that, intervals with matchingupper harmonics (= consonances) and inter-vals with non-matching upper harmonics (=dissonances) create distinct firing patterns inthe auditory neural network. This idea, put for-ward by Boomsliter and Creel (1961) andMoore (1989) (see also Schellenberg & Tre-hub, 1994), is that complex tones that arerelated by simple frequency ratios producesimilar firing patterns in the auditory networkbecause of the relatively frequent match ofupper partials (see Figure 1). The absence ofshared neural circuits presumed when there isprocessing of dissonances might be one reasonfor experiencing intervals with complex fre-quency ratios as dissonant.

Another explanation centers on the conceptof critical bandwith. In this view, the auditoryneural network has a limited ability to resolvedifferent tones (and its upper haromics) thatare too proximate in pitch. Tones that are veryproximate, but not identical, in frequency cannot be effectively processed by the basilarmembrane: hence the excitation patterns over-lap. These fluctuations result in a perception ofroughness or dissonance. Because the minorseconds used in the dissonant versions of ourmelodies are characterized by more complexfrequency ratios and more overlapping criticalbands than the thirds used for the consonantversions of the melodies, the two hypothesesare equally plausible.

These explanations are reminiscent of Gali-leos views for he wrote in 1638 (1963, p.100):

Agreeable consonances are pairs of toneswhich strike the ear with a certain regular-ity; this regularity consists in the fact thatthe pulses delivered by the two tones, in the

49

same interval of time, shall be commensu-rable in number, so as not to keep the ear

*drum in perpetual torment, bending in twodifferent directions in order to yield theever discordant impulses.

Acknowledgment: A portion of this workpreviously appeared in Nature (383, p. 29,1996). Preparation of this paper was supportedby a grant from the John D. and Catherine T.MacArthur Foundation Network on Psychopa-thology and Development and a fellowshipfrom the Swiss National Science Foundationawarded to the first author. We are grateful toCurt Stallman for his help with the creation ofthe stimuli, to Emily Hoffman, Steve Most andMeredith Rowe for helping us with testing andcoding, and to Carol Krumhansl and SandraTrehub for their helpful comments at differentstages of this research.

NOTES

Marcel R. Zentner is now at the Department ofPsychology at the University of Geneva.The pairs of tones can occur in succession orsimultaneously. The former type of interval iscalled sequential or melodic, the latter isreferred to as harmonic. Although consonanceand dissonance can refer to a sequence of singletones, they usually refer to harmonic intervals.It is only in this latter sense that we are con-cerned with consonance and dissonance here.The exclusive use of thirds would result in theviolation of tonality, introducing another typeof dissonance. Thus, in the few instances whereit was unavoidable the third was replaced by itsnatural complement, the sixth, which has avery similar consonance quality. Where thethird was replaced by the sixth in the consonantversion, in the dissonant version the naturalcomplement of the second, the seventh, wasused in order to match the interval size of theconsonant version (see Figure 2).

REFERENCES

Boomsliter, P. & Creel, W. (1961). The long patternhypothesis in harmony and hearing. Journal ofMusic Theory, 5,2-3 1.


10/10

and dissonance

Documents

Transcript of and dissonance