Retraction: “Role of perceptual integration in pitch discrimination at high frequencies” [JASA Express Lett. 2, 084402 (2022) DOI: 10.1121/10.0013429]

The authors are retracting the referenced publication (Mehta and Oxenham, 2022). The retraction was initiated by the second author and agreed to by the first author. This retraction is because discrepancies were discovered between the published results and the data as originally collected, which materially affect the conclusions of the paper. In particular, when the original data are analyzed, the difference between mean fundamental frequency difference limens (F0DLs) with harmonic and inharmonic components, as described in experiment 1, is no longer statistically significant (t15 ¼ –1.699, p ¼ 0.110, two-tailed test). In addition, the difference between the measured and predicted thresholds for harmonic tones under the early noise (peripheral) model is no longer significant, with no main effect of measure type (predicted vs obtained) and no interaction between measure type and harmonicity (harmonic vs inharmonic) (F1,15 ¼ 0.231, p ¼ 0.638, and F1,15 ¼ 1.161, p ¼ 0.298, respectively). For the late-noise (central) model, although the effect of measure type remains significant (F1,15 ¼ 23.5, p < 0.001), the interaction with harmonicity does not (F1,15 ¼ 1.468, p ¼ 0.244). These results do not support the conclusion that the late-noise model provides a better description of the harmonic than the inharmonic F0DLs. In fact, late-noise model predictions are significantly different from observed F0DLs with both harmonic (p ¼ 0.008) and inharmonic (p < 0.001) tones. A similar pattern of discrepancies was found in experiment 2, with the differences between the predicted and observed data for harmonic and inharmonic tones failing to support the original conclusion that harmonic and inharmonic tones are processed differently. In summary, the data as originally collected do not support the primary published conclusions. Instead, the F0DLs of high-frequency harmonic and inharmonic complex tones are not statistically significantly different from each other, providing no evidence that they are governed by different encoding mechanisms.