The influences of musical training and spectral centroid on perceptual interactions of pitch and timbre

Perceptual interactions of pitch and timbre have frequently been observed, and the nature of these interactions differs between musicians and nonmusicians. Yet, few researchers have investigated which aspects of timbre or musical training contribute to such interactions. Recently, Becker and Hall (2014) demonstrated that the spectral centroid contributed to pitch-timbre interactions in missing-F0 experiments, particularly for nonmusicians. The present experiment investigated whether the centroid also accounted for previously observed interactions between pitch and timbre (see Pitt, 1994) in a Garner speeded classification task designed to evaluate the perceptual independence of dimensions. There were two sets of synthetic stimuli involving orthogonal combinations of A4 and D#4 tones derived from violin and trumpet. Timbres in one set varied with respect to spectral envelope, amplitude envelope, and spectral centroid, whereas the other equated spectral centroids through slight manipulations of spectral slope. Tones with the same centroid were expected to reduce the magnitude of observed interference and redundancy gain effects. Contrary to hypotheses, such an effect was not observed, suggesting that the spectral centroid was not the aspect of timbre driving perceptual interactions in the current investigation. While musical training has been proposed to enhance the ability to distinguish pitch from timbre changes, the aspect of training that contributes to such enhancement remains unclear. This is complicated by most studies only considering total years of experience as a means of categorizing musicians versus nonmusicians, which directly impacts conclusions regarding statistical significance. The current investigation addressed these issues by introducing a musical training survey that measures a more diverse range of musical experiences (e.g., ensemble performance, recency/frequency of practice, level of coursework) in a more continuous manner (i.e., without fixed response options). This permitted statistically appropriate (regression) analyses of the relationship between years of training and perceptual independence of timbre and pitch, which was intended to identify relevant experiences for observed interactions. Increased amounts of musical training in general were associated with smaller interference effects with the adjusted stimulus set. Although specific experiences were not identified in the current investigation due to high correlations between musical predictor variables, such correlations raise the possibility that a single factor may be underlying the musical training items examined in this study. Collectively, these findings demonstrate that pitch and timbre perceptually interact regardless of level of musical training, although training can reduce the size of the interference effect in certain conditions. Additionally, rather than the spectral centroid being the attribute of timbre responsible for perceptual interactions, current results suggest that the spectral envelope may have a stronger influence