Auditory feedback affects the long-range correlation of isochronous serial interval production: support for a closed-loop or memory model of timing.

Long-range dependence is a characteristic property of successively produced time intervals, such as in un-paced or continuation tapping. We hypothesise in the present paper that serial dependence in such tasks could be related to a closed-loop regulation process, in which the current interval is determined by preceding ones. As a consequence, the quality of sensory feedback is likely to affect serial dependence. An experiment with human participants shows that diminished sensory information tends to increase the Hurst exponent for short inter-onset intervals and tends to decrease it for long intervals. A simulation shows that a simple auto-regressive model, whose order depends on the ratio between the inter-onset interval and an assumed temporal integration span, is able to account for most of our empirical results, including the duration specificity of long-range correlation.