Revisiting discrete versus continuous models of human behavior: The case of absolute pitch

Abstract

Many human behaviors are discussed in terms of discrete categories. Quantizing behavior in this fashion may provide important traction for understanding the complexities of human experience, but it also may bias understanding of phenomena and associated mechanisms. One example of this is absolute pitch (AP), which is often treated as a discrete trait that is either present or absent (i.e., with easily identifiable near-perfect “genuine” AP possessors and at-chance non-AP possessors) despite emerging evidence that pitch-labeling ability is not all-or-nothing. We used a large-scale online assessment to test the discrete model of AP, specifically by measuring how intermediate performers related to the typically defined “non-AP” and “genuine AP” populations. Consistent with prior research, individuals who performed at-chance (non-AP) reported beginning musical instruction much later than the near-perfect AP participants, and the highest performers were more likely to speak a tonal language than were the lowest performers (though this effect was not as statistically robust as one would expect from prior research). Critically, however, these developmental factors did not differentiate the near-perfect AP performers from the intermediate AP performers. Gaussian mixture modeling supported the existence of two performance distributions–the first distribution encompassed both the intermediate and near-perfect AP possessors, whereas the second distribution encompassed only the at-chance participants. Overall, these results provide support for conceptualizing intermediate levels of pitch-labeling ability along the same continuum as genuine AP-level pitch labeling ability—in other words, a continuous distribution of AP skill among all above-chance performers rather than discrete categories of ability. Expanding the inclusion criteria for AP makes it possible to test hypotheses about the mechanisms that underlie this ability and relate this ability to more general cognitive mechanisms involved in other abilities.

Individual differences in human frequency-following response predict pitch labeling ability

“…Notably, individual differences in early sensory encoding, as reflected by the FFR, are able to predict continuous variance in AP ability. Since the variation in pitch labelling ability has largely gone unexplained since researchers have argued that AP should be considered as a graded (rather than dichotomous) ability 20 this finding is novel. It has long remained an open scientific question why humans can place some types of stimulus characteristics into stable, barely changing categories (such as color) but less so others (such as pitch); understanding the relationship between individual differences in low-level sensory coding and in the higher-level cognitive ability to consistently categorize perceptual stimuli promises to shed light on broader theories of semantic memory, concepts, and categories 26 .…”

“…Absolute pitch possessors (N = 16) and musically matched subjects (N = 15) were recruited from the Chicagoland area. By including subjects that are expected to show a range of pitch perception ability, we hope that our sample is representative of the population distribution of absolute pitch ability described by Van Hedger et al 20 . Of the 31 remaining subjects, which included both males and females (16 females) with varying amounts of musical training, the average age was M = 21.6, SD = 3.01.…”

“…While AP has traditionally been construed as a dichotomous ability, in which subjects either have or do not have AP 17 , 18 , recent evidence has suggested that AP ability exists along a spectrum, where AP ability is best described as a continuously distributed variable 19 . While there is sizable variance in pitch labelling ability in the general population 20 , variables that predict continuous variation in absolute pitch perception ability are largely unknown and generally viewed as a consequence of cognitive factors rather than auditory ability 21 . The aim of the present study, then, is to investigate the extent to which individual differences in the FFR, reflecting low-level neural auditory encoding of sounds, predicts variation in pitch labelling ability, a higher-level cognitive process.…”

Use of explicit priming to phenotype absolute pitch ability

“…Our initial findings, however, argue against the classification of AP as a categorical ‘all-or-nothing’ trait, as shown by the lack of support for a two-group solution in the LPAs. The question of whether AP consists of more than two distinct phenotypes or is the endpoint of a continuum of ability has not been addressed here, and it should be highlighted that the notion of QAP as an extension of AP, rather than a distinct type of pitch-naming ability, has recently been supported by Van Hedger and colleagues [ 21 ] using data-driven methods. Of note, our distribution of ability appears consistent with either three categories (as per the LPA), or a continuum represented by a sigmoid function (see S2 Fig ).…”

Some like it sharp: Song familiarity influences musical preference for absolute tuning

“…Additionally, given the discussion of AP as a more distributed ability (e.g., Bermudez & Zatorre, 2009;Van Hedger et al, 2020), it is possible that the best approach to take in future TUNING JUDGMENTS investigations is to use performance-based assessments of AP and RP processing as an informative individual difference measure. Although we did not find any evidence that the administered musical measures were associated with individual differences in performance in the present experiments, these were all self-report measures and did not specifically probe relative and absolute pitch abilities.…”