(Editor: Pamela Marek)
Created by changes in the air pressure, sound travels in waves that convey information about loudness and pitch. A basic review of the properties of sound waves will enhance understanding of the design and interpretation of research relevant to pitch and pitch memory. The pressure changes created by a pure tone, such as that produced by a tuning fork, are represented by a sine wave. As shown in Figure 1, the height (amplitude) of this sine wave relates to loudness, measured in decibels. The number of cycles the wave makes per second (frequency) relates to pitch, measured in Hertz. Frequencies audible to humans generally range from 20 to 20,000 Hz (Goldstein, 2002). The range of pitch on a piano is from approximately 27 to 4,186 Hz, with middle C at approximately 262 Hz.
In developing theories about the perception of pitch and organization of pitch memory, researchers have typically worked with pure tones. For example, Deutsch (1970) determined that short-term recognition of the pitch of pure tones was disrupted by six intervening tones, but not by six intervening spoken numbers, suggesting that immediate processing of musical pitch was in some way distinct from that of verbal information. However, Semal, Demany, Ueda, and Halle (1996) questioned the details of this distinction. Using complex tones and spoken numbers as stimuli, both with pitch systematically controlled, Semal et al. found that differences in pitch between the interfering sounds had a much greater effect on accuracy of recognition than did the type of interference (tone vs. speech), suggesting that only certain aspects of speech processing may be specialized. Subsequently, Moore, Keaton, and Watts (2005) reported that memory for pitch plays an important role in both pitch matching (being able to vocally duplicate the pitch of a tone) and pitch discrimination (being able make an accurate "same or different" judgment about two tones) task.
Laterality and hemispheric specialization (discussed in connection with the Dichotic Listening demonstration) are also relevant to pitch memory. For right-handed people, the left hemisphere is generally considered to be the center of speech processing. However, for left-handers the representation of speech is more diverse. Using a pitch memory task with pure tones, separated by six intervening tones, Deutsch (1979) found that people with a moderate preference for the left hand made fewer errors than did right-handed people. Considering this finding, Deutsch (1979) suggested that pitch information may be stored bilaterally for certain left-handed people, giving them an advantage in retrieving the information needed to make accurate same-different judgments.
The pitch memory task is patterned after Deutsch (1979). On each trial, a target tone and a test tone are presented with five distracter tones between them. You will be asked to judge whether the target and test tones are the same or different. There are 38 trials when the tones are the same and 38 when they are different. Performance will be measured as proportion correct on each of the trial types (i.e., same and different). Sample tones and sample tone sequences are presented prior to the start of the task.
The tones used in this demonstration were created as pure tones ranging from 220 Hz to 880 Hz (spanning two octaves). There are 25 tones in all, equally spaced across the interval. The target tone for each trial is selected randomly from the 12 even-numbered tones in the 25-tone series. Test tone selection is constrained to be either the same as the target tone or different by two semitones (a semitone, also called a half-step on a piano, is the distance between a key and the next highest or lowest key, white or black). Thus the same 12-tone set is used when choosing a target tone but the choice is not random. Distracter tones are selected from the 13 odd-numbered tones in the 25-tone series. Selection is random with the constraint that no tone can repeat as a distracter on any one trial. Thus five unique distracters are presented between the target and test tones. In order to distinguish target and test tones, there is a ½ sec delay between the test tone and the start of the 5-tone distracter sequence. The test tone appears ½ sec after the end of the 5-tone distracter sequence. When target and test tones are different, they can form an ascending or descending series with equal probability.
Data is downloadable in three formats (XML, Excel spreadsheet format, and comma delimited for statistical software packages). Figure 2 shows an excerpt from a sample Excel spreadsheet. The first nine columns provide classification data (participant ID number, gender, the class ID number, age, type of musical training, years of musical training, hand preference, completion date, and the amount of time the experiment took). The last two columns indicate the proportion of correct recognition when the target and test tones were similar and the proportion of correct recognition when the target and test tones were different. If the graphing function is used, the bars indicate the proportion of correct recognition for the tone when they are similar or different.
It may be of interest to examine the incorrect trial data. When participants were wrong, did they tend to be wrong by choosing "Same" when the tones were different or by choosing "Different" when they were the same? A chi-square analysis can compare these two proportions. Typically one finds an affirmative bias in same-different judgment tasks. This may reflect a characteristic of human decision making or it may reflect something about participants' discriminatory ability. The former explanation can be described as a tendency to say "same" whenever one is in doubt. The latter explanation is that different sounds tend to sound the same to those with less discriminating ears.
There are two between subject variables of interest.
The first between-subjects variable is differences in musical training. An independent samples t-test could be used to judge the significance of the difference in performance between people with and without musical training. (Be sure to look for differences on both the same and different trials, because you may find that musical training makes a difference on just one of the trial types). In larger samples, an attempt could be made to determine the functional relation between numbers of years of training and pitch memory. The first step toward this analysis would be a graph of the relation between years of training and proportion correct. Statistical analyses would employ trend analysis based either on an analysis of variance or a linear or curvilinear regression.
The second between subject variable is handedness. Again, an independent samples t-test could be used to judge the significance of the difference in performance between right-handed and left-handed people. Although a more discriminating measure of handedness is preferable to simply indicating whether people are right-handed or left-handed, a handedness effect may emerge in large data sets.
This demonstration deals with relative pitch, an ability to distinguish differences between pitches. In a paradigm based on the concept of preference for novelty, Plantinga and Trainor (2005) demonstrated that infants as young as 6 months old remembered melodies using relative pitch, the way in which most adults store musical information. In other developmental work, Keller and Cowan (1994) found evidence that that persistence of memory for pitch increases between the ages of 6 and 7 and adulthood.
Researchers have also probed memory for absolute pitch, the ability to label or produce specific tones. Findings suggest that development of absolute pitch requires early musical training (Deutsch, 2002). However, using excerpts from themes of popular TV shows, Schellenberg and Trehub (2003) demonstrated that people were able to distinguish melodies with pitches that were only slightly different from the original, revealing that, at the least, they were able to remember the specific pitch of the familiar theme.
Using advanced technology, researchers have also developed computer-based neural networks models to represent pitch memory (Franklin & Locke, 2005) and used have imaging techniques to determine gender related differences in brain areas used to process basic pitch (Gabb, Keenan & Schlaug, 2003). Regarding individual differences, researchers have acknowledged a better than average pitch memory among autistic children (Heaton, 2003), and identified a poorer than average pitch memory among children with developmental receptive language disorder (Lincoln, Dickstein, Courchesne, Elmasian, & Tallal, 1992). Thus, multiple strands of research have extended our understanding not only of memory for pitch, but for memory of timbre, relationships between pitches, and duration (Deutsch, 1999).
An interesting analysis to perform initially compares the observed proportion correct for each participant to the chance proportion of .5 to determine whether there is evidence that the individual has pitch memory. An appropriate test is a one-tail z-test that uses the statistic
||when the proportion is based on all 76 trials, and the statistic
||when the proportion is based on 38 trials. The general formula is
||where N is the number of trials, p is the probability of being correct when guessing, and q is the probability of being incorrect by guessing.
(Statistical Note: The denominator is the SD of the distribution for proportion correct based on N choices each of which has a probability of .5 of being correct. If one wants the same analysis performed on the mean percent correct for the class, the statistic is the same except that the first term in the denominator (the part under the radical) needs to be divided by the number of participants before the square root is taken.)
Deutsch, D. (1970). Tones and numbers: Specificity of interference in immediate memory.
Science, 168, 1604-1605.
Deutsch, D. (1979). Pitch memory: An advantage for the left-handed. Science, 199,
Deutsch, D. (1999). The processing of pitch combinations. In S. Deutsch, The psychology
of music (2nd ed.), pp. 349-411. San Diego, CA: Academic Press.
Deutsch, D. (2002). The puzzle of absolute pitch. Current Directions in Psychological
Science, 11, 200-204.
Franklin, J. A., & Locke, K. K. (2005). Recurrent neural networks for musical pitch memory
and classification. International Journal of Artificial Intelligence Tools,
Gabb, N., Keenan, J. P., & Schlaug (2003). The effects of gender on the neural substrates
of pitch memory. Journal of Cognitive Neuroscience, 15, 810-820.
Goldstein, B. (2002). Sensation and perception (6th ed.). Pacific Grove,
Heaton, P. (2003). Pitch memory, labeling, and disembedding in autism. Journal of Child
Psychology and Psychiatry, 44, 543-551.
Keller, T. A., & Cowan, N. (1994). Developmental increase in the duration of memory
for tone pitch. Developmental Psychology, 30, 855-863.
Lincoln, A. J., Dickstein, P., Courchesne, E., Elmasian, R., & Tallal, P. (1992). Auditory
processing abilities of non-retarded adolescents and young adults with developmental
language disorder and autism. Brain and Language, 43, 613-622.
Moore, R. E., Keaton, C., & Watts, C. (2005). Role of pitch memory in pitch matching and
pitch discrimination. The ASHA Leader, 10 (10), 4.
Plantinga, J., & Trainor, L. J. (2005). Memory for melody: Infants use a relative pitch code.
Cognition, 98, 1-11.
Schellenberg, E. G., & Trehub, S. E. (2003). Good pitch memory is widespread. Psychological
Science, 14, 262-266.
Semal, C., Demany, L., Ueda, K., & Halle, P. (1996). Speech versus non-speech in
pitch memory. Journal of the Acoustical Society of America, 100, 1132-1140.