DRIFT AND TIMING VARIABILITY IN ISOCHRONOUS INTERVAL PRODUCTION WITH AND WITHOUT MUSIC IMAGERY

Guy Madison, Department of Psychology, Uppsala University, Box 1225, SE-751 42 UPPSALA, Sweden

Background. There has been some recent interest in the fact that isochronous serial interval production (ISIP) exhibits a substantial amount of drift - i.e. higher-order dependencies. However, these results are typically obtained in rather "unmusical" tapping experiments where singing or else subdividing the intervals is prohibited. The question is therefore how

valid drift is in the typical ISIP context; performing (to) music.

Aims. To show if and how the cognitive representation of music affects ISIP drift and dispersion. Specifically, one could hypothesize that music imagery would function as a subdividing temporal "glue", and (a) generally decrease drift and/or dispersion, (b) decrease the difference in drift between long and short inter response intervals (IRI) or, (c) impose

variability patterns that correspond to the music structure.

Method. Participants in a tapping experiment were later recruited to imagine listening to recordings of familiar songs (selected for being stable in tempo). They were asked to play along with specified multiples or subdivisions of the beat (corresponding to the IRIs in the tapping exp.: 0.5, 0.8, 1.1, and 1.5 s) and to maintain a stable tempo.

Results. The coefficient of variation (SD/M) and the local fluctuations in tempo were not affected by music imagery, whereas long-term (monotonous) drift was larger during imagery for 1.1 and 1.5 s IRI. Although autocorrelation functions were obscured by non-stationarity for IRIs above 0.8 s, there was a clear effect of imagery for 0.5 and 0.8 s: While the first two lags were typically moderately positive or negative ( .1) for the tapping data, the imagery data demonstrated periodicity for various higher lags.

Conclusions. A musical context does not seem to improve simple timing performance,

although it does affect the variability patterns. A comparison with subdivision in tapping experiments is the first step to explain these findings.