The divided time illusion (DTI) is the finding that brief intervals seem longer when they are subdivided by one or more additional events than when they are empty. Two previous multiexperiment studies using different synchronization-continuation tapping tasks demonstrated the DTI in musicians' perception and production of beat tempo (Repp, 2008; Repp & Bruttomesso, 2009): Subdivided seem slower than undivided beats. Each study, however, contained a single condition that yielded puzzling results inconsistent with the DTI; each had intermixed trials with duple and triple metrical subdivision of interbeat intervals (IBIs). Our attempt to replicate the deviant 2008 result failed, now yielding data consistent with the DTI. However, we did replicate the deviant 2009 result and then followed it up by assigning different forms of subdivision (duple, triple, quadruple) to separate trial blocks and varying IBI duration within each block. The results turned out to be quite systematic and consistent with the DTI. Moreover, they revealed that the DTI decreases as IBI duration decreases and vanishes when the subdivision event rate reaches 5- 6 Hz. We tentatively attribute the puzzling 2009 result to attenuation of the DTI by mental subdivision of undivided beats, due to memory requirements of the earlier task design.Keywords: divided time illusion, filled duration illusion, synchronization, tempo perception, rate limitsMany studies, some conducted as long ago as the 19th century, have demonstrated that brief temporal intervals are perceived to be longer when they contain one or more additional stimulus events than when they are empty (Adams, 1977; Buffardi, 1971; Craig, 1973; Goldfarb & Goldstone, 1963; Hall & Jastrow, 1886; Meumann, 1896; Ornstein, 1969; Thomas & Brown, 1974; Wearden, Norton, Martin, & Montford-Bebb, 2007). This divided time illusion (DTI; ten Hoopen, Miyauchi, & Nakajima, 2008) is one kind of filled duration illusion, another kind being the finding that continuous sounds are perceived as longer than equally long silent intervals (called the sustained sound illusion by Repp & Marcus, 2010). Most research on these effects used single intervals, though Grondin and collaborators (Gamache, Bisson, Hawke, Roussel, & Grondin, 2008; Grondin, 1992; Grondin, Guay, Lapointe, & Poulin, 2008; Grondin, Metthe, & Koren, 1994), who were primarily interested in timing variability, investigated cyclic reproduction of relatively long intervals from cyclic models containing various forms of metrical subdivision and found some evidence of the DTI for the shorter intervals used (1-1.2 s).In two recent studies, Repp (2008) and Repp and Bruttomesso (2009) similarly investigated whether the DTI is found in rhythmic contexts, where an isochronous sequence of presents the same interval repeatedly, with or without metrical subdivisions. These authors were specifically interested in the constant error occurring in musicians' judgment or reproduction of beat rates within a range characteristic of music (600-1,000 ms), the hypothesis being that subdivided would be perceived as being slower than undivided beats. (The term beats here refers simply to the sounds whose onsets delimit the successive intervals that are to be judged or reproduced.) Each of these previous studies yielded strong evidence in support of the DTI hypothesis in several experiments, but each contained one experiment or condition in which results were obtained that were inconsistent with the DTI hypothesis. The purpose of the present study was to try to replicate and possibly explain these puzzling exceptions.In Repp (2008), the deviant results occurred in Condition 4 of Experiment 2. In this condition, a synchronization-continuation paradigm was used where during the synchronization phase participants tapped with their right hand in synchrony with an isochronous auditory beat that was either undivided or divided by one or two left-hand taps, according to cues provided at the onset of the computer-produced beat sequence. …