Abstract: The purpose of the present two experiments was to study the predictive validityof three cognitive models in time estimation:the storage-size model(SS),the processingtime mode(PT),the change/ segmentation model(CD),but the experiments wereprimarily designed to test CS model . With the methods of both reproduction andparameter estimation of time, subjects were asked to prospectively (immediately ordelaying) estimate the duration between two red-colored squares, which were alsofilled with 12 digits appearing alternatively, At the same time, subjects were asked toreport 2 digits after a probing digit, which enabled subjects to perceive (or remember)the flow of digits in the predesigned chunking mode. the segmentation levels wereanticipatedly produced by different chunking size(2or 4 digits ). In experiment 1,external change was manipulated:diffirent duration lasting within two types of chunks,different intervals intervening two digits of the two types of chunks, In Experiment 2,internal change was manipulated, while external change was controlled: same intervalsintervening two digits, same duration within any digit, but subjects were asked to perceive these digits in diffirent chunking modes. The results showed that the higher the segmentation level in immediate response setting the longer the subjective duration estimation, but the subjetive duration was not influenced by the predetermined segmentation level in delayable response setting. Examining the way the subjective experience was processed indicated that in immediate response setting, the predesigned chunking of the list of digits was preserved, but in delayable response setting, such chunking diminished. These findings were discussed in terms of three models. The conclusion was that time estimation was best explained by the CS model: although the subjective duration in delayable prospective estimation was not influenced by the predesigned segmentation level, it was also influenced by the degree to which memorial representations of processed events segment experience. That is, in immediate estimation, the segmentation levels were determined by the predesigned chunked digits, while in remote estimation, the segmentation levels were determined by the discrete digits.

Key words: time estimation, change / segmentation model, reproduction time method, parameter estimation on method