Use of explicit priming to phenotype absolute pitch ability

Abstract

Musicians with absolute pitch (AP) can name the pitch of a musical note in isolation. Expression of this unusual ability is thought to be influenced by heritability, early music training and current practice. However, our understanding of factors shaping its expression is hampered by testing and scoring methods that treat AP as dichotomous. These fail to capture the observed variability in pitch-naming accuracy among reported AP possessors. The aim of this study was to trial a novel explicit priming paradigm to explore phenotypic variability of AP. Thirty-five musically experienced individuals (Mage = 29 years, range 18-68; 14 males) with varying AP ability completed a standard AP task and the explicit priming AP task. Results showed: 1) phenotypic variability of AP ability, including high-accuracy AP, heterogeneous intermediate performers, and chance-level performers; 2) intermediate performance profiles that were either reliant on or independent of relative pitch strategies, as identified by the priming task; and 3) the emergence of a bimodal distribution of AP performance when adopting scoring criteria that assign credit to semitone errors. These findings show the importance of methods in studying behavioural traits, and are a key step towards identifying AP phenotypes. Replication of our results in larger samples will further establish the usefulness of this priming paradigm in AP research.

Fig 1. Experimental paradigm for the explicit priming AP task (AP-P).

Fig 2. Decision tree for the explicit priming AP task (AP-P).

The first line shows the presented stimuli–a C prime, followed by target G. The second line shows the hypothetical options for responding, given differing chroma in a participant’s pitch template. The third line shows the outcome from each possible scenario. Note that chance performance corresponds to 1/12, or 8.33%.

Fig 3. Distribution of accuracy scores.

Distribution is shown across AP-S (squares) and AP-P (crosses), with groupings according to the three-group LPA solutions for each task (black = non-AP, dark grey = QAP, light grey = AP). The highlighted boxes refer to the three participants whose group allocation changed between AP-S and AP-P LPAs. For example, the participant in the right-most box scored 78% in AP-S and was allocated to the AP group, but only 64% in AP-P and was placed in the QAP group. Horizontal lines at 20% and 90% indicate commonly employed thresholds for AP (>90% accuracy) and non-AP (<20% accuracy).

Fig 4

Model fit statistics for 1–5 group solutions for AP-S showing (a) AP-S chroma accuracy and (b) AP-S chroma accuracy ST.

Fig 5

Model fit statistics for 1–5 group solutions for AP-P showing (a) AP-P chroma accuracy and (b) AP-P chroma accuracy ST.

Fig 6

Accuracy profiles for a typical (a) non-AP participant (Participant 24; AP-P task accuracy 7.58%) and (b) AP participant (Participant 30; AP-P task accuracy 97.73%). Target bars refer to the percentage of those trials for which each chroma was a target that was correctly identified by the participant. Prime bars refer to the percentage of those trials for which each chroma was a prime that preceded a correct target identification by the participant.

Fig 7. Accuracy profile for a QAP participant (participant 32) with significant target chroma, AP-P task accuracy 34.85%.

Marked target chroma were significantly less likely to be accurately identified than the statistical reference point of C, according to this participant’s logistic regression model. As the majority of “poor” target chroma corresponded to the black keys of the piano, this participant may be considered to have “white-note” AP (relatively good performance for white-key chroma). As prime chroma did not significantly predict the likelihood of a correct response, it is unlikely that this participant employed a reference chroma strategy for identifying pitches (**p < .01).

Fig 8. Accuracy profile for a participant (participant 27) with significant target and prime chroma, AP-P task accuracy 42.42%.

Marked target chroma were significantly less likely to be accurately identified than the statistical reference point of C, according to this participant’s logistic regression model. Marked prime chroma were significantly less likely to precede an accurately identified target than targets following C primes (*p < .05, **p < .01, ***p < .001).