Enhanced auditory disembedding in an interleaved melody recognition test is associated with absolute pitch ability

Abstract

Absolute pitch (AP) and autism have recently been associated with each other. Neurocognitive theories of autism could perhaps explain this co-occurrence. This study investigates whether AP musicians show an advantage in an interleaved melody recognition task (IMRT), an auditory version of an embedded figures test often investigated in autism with respect to the these theories. A total of N = 59 professional musicians (AP = 27) participated in the study. In each trial a probe melody was followed by an interleaved sequence. Participants had to indicate as to whether the probe melody was present in the interleaved sequence. Sensitivity index d' and response bias c were calculated according to signal detection theory. Additionally, a pitch adjustment test measuring fine-graded differences in absolute pitch proficiency, the Autism-Spectrum-Quotient and a visual embedded figures test were conducted. AP outperformed relative pitch (RP) possessors on the overall IMRT and the fully interleaved condition. AP proficiency, visual disembedding and musicality predicted 39.2% of variance in the IMRT. No correlations were found between IMRT and autistic traits. Results are in line with a detailed-oriented cognitive style and enhanced perceptional functioning of AP musicians similar to that observed in autism.

Figure 1

Graphical and musical notation of experimental material within the IMRT (interleaved melody recognition test). Each trial begins with a probe melody (A, blue) followed by an interleaved sequence with either the same probe melody (B, blue) or a modified melody (C, blue, here 2^nd and 4^th note modified compared to probe melody) interleaved with a distractor melody (red). Graphical notation of musical melodies in this example shows the pitch height of each note in ST relative to the A₄ (440 Hz). The distractor notes always encompassed the neighboring melody notes in terms of the pitch height.

Figure 2

Interleaved sequences of IMRT. Probe melody respectively modified probe melody (here probe melody, blue) and distractor melody (red) were interleaved and four separation conditions created. The two melodies were either fully interleaved and their mean frequency separated by 0 semitones (0 ST) or the distractor melody was shifted by 6, 12 or 24 ST downwards. Therefore melodies were sequentially separated further apart, increasing the possibility of hearing two musical streams.

Figure 3

Relation of IMRT and visual embedded figures test (a) and signal detection measures by group (b,c). (a) Overall performance on IMRT (d’) by mean time (in s) needed to solve an item on the visual embedded figures task (GEFT). Variables revealed a correlation of r = −0.407 (Pearson correlation; p < 0.001). Blue: IMRT version A, black: IMRT version B. (b,c) Results on sensitivity index d’ (b) and response bias c (c) by group (AP, absolute pitch; RP, relative pitch) regarding overall performance (“all”) and performance within separation conditions (0, 6, 12, 24 semitone (ST) separations) on IMRT. Higher values of d’ indicate better performance. Positive values of c indicate a tendency towards “no” responses, negative values a tendency towards “yes” responses. Bars represent standard errors. *p < 0.05. **p < 0.01. ***p < 0.001.

Figure 4

Influence of absolute pitch ability (Z_MAD), disembedding in vision (GEFT, time in s) and musicality (MSI) on perceptual sensitivity d’. Panel A corresponds to overall performance on IMRT, while panels B-E show the prediction of performance in different separation conditions (0, 6, 12, 24 semitone separation of probe and distractor melody). Color and shape scales correspond to disembedding in vision (GEFT, time in s) respectively MSI (score on questionnaire, higher values indicate greater musical sophistication). The regression line always takes the intercept and beta-weight of the simple linear regression of absolute pitch ability (Z_MAD, standardized to mean of RP group) on sensitivity index (d’-d24’, higher values indicate better performance). Ceiling effects in the 12-ST- and especially 24-ST-condition (panel E) are clearly visible (see Table 4 for statistical values).

Figure 5

Relationship between absolute pitch performance and autistic traits. Higher scores on the online pitch identification screening (PIS, maximum = 36) are associated with more autistic traits on the Autism-Spectrum-Quotient (AQ, maximum = 50). Size of the dots in the scatterplot corresponds to performance in the pitch adjustment test (PAT: Mean absolute deviation (MAD) from target tone in cent (100 cent = 1 semitone)) with smaller values associated with better AP performance. It is visible by eye that except one case, all participants with good performance on PIS also succeed on the PAT. Red line: cutoff for diagnosis of autism-spectrum diagnosis; green: boundary for critically/unusually high AQ; blue: general healthy population mean (Baron-Cohen et al., 2001). Black: regression line of PIS on AQ (r = 0.393, p = 0.003).