Structuring temporal sequences: comparison of models and factors of complexity

Two stages for structuring tone sequences have been distinguished by Povel and Essens (1985). In the first, a mental clock segments a sequence into equal time units (clock model); in the second, intervals are specified in terms of subdivisions of these units. The present findings support the clock model in that it predicts human performance better than three other algorithmic models. Two further experiments in which clock and subdivision characteristics were varied did not support the hypothesized effect of the nature of the subdivisions on complexity. A model focusing on the variations in the beat-anchored envelopes of the tone clusters was proposed. Errors in reproduction suggest a dual-code representation comprising temporal and figural characteristics. The temporal part of the representation is based on the clock model but specifies, in addition, the metric of the level below the clock. The beat-tone-cluster envelope concept was proposed to specify the figural part.