The concept of attention is invoked to explain limitations on the amount of information that cognitive systems can process per unit time and differences in performance to identical input under different antecedent conditions. Although there are many approaches to explaining such observations, it seems appropriate to summarize attention as a process modulating the selection, sensitivity, and efficiency of a limited capacity for information processing (cf. Kinchla, 1992;Posner, Snyder, & Davidson, 1980). Such a concept implies a dynamic change, and indeed, many dynamic phenomena related to attention have been described. We investigated the ability of deterministic oscillator and stochastic timing models to account for the temporal allocation of attention, using both qualitative and quantitative approaches to chronometric data.A theoretical account of dynamic attention by Jones (1976) has been the focus of several recent studies (Barnes & Jones, 2000;Large & Jones, 1999;McAuley & Kidd, 1998;Olsen & Chun, 2001). According to dynamic attending theory (DAT;Jones, 1976;Jones & Boltz, 1989;Large & Jones, 1999), attention is guided through time by internal oscillators termed attending rhythms. Attending rhythms are coupled to the environment through perception. This coupling causes them to become correlated with the environment through a process of entrainment when the temporal structure of the environment is autocorrelated. Attention is then focused on an epoch with an intensity that depends on the phase relationship (i.e., the degree of correlation) between the environmental event stream and the attending rhythms. Specifically, the greater the degree of synchronization between the internal oscillator and the external sequence, the narrower and more intense the attentional pulse becomes.Much of the evidence for DAT comes from nearthreshold detection tasks for timing information. Jones and Yee (1997) found that temporal just-noticeable differences were smaller for intervals embedded in rhythmic sequences than for randomly timed sequences, even though the rhythmic sequences were more variable. Jones and Boltz (1989) found that discrimination between the duration of two melodies was influenced by whether or not the last note of one melody violated a temporal pattern established by the preceding notes. Large and Jones (1999) found that time change detection accuracy was better when stimuli occurred at unexpected times and that duration estimation accuracy was better when stimulus events occurred at expected times, where expected times were defined by DAT. Barnes and Jones (2000) found that duration judgments about a standard and a comparison interval were most accurate when the temporal context preceding the standard had a simple harmonic relationship to the standard interval. In all of these studies, temporal context altered sensitivity to temporal information in a direction consistent with a coupled oscillator mechanism.The use of near-threshold detection measures of timing has invited comparisons between DAT and temporal...