Abstract
Numerosity estimates are proposed to critically depend on the integration of information across visual fixations via a serial accumulator mechanism. Accordingly, longer presentation durations engender more fixations, resulting in numerical quantity overestimation. Similarly, numerosity has been shown to affect duration estimates, i.e. larger numerosities result in duration overestimation; however, these studies 1) implicitly require saccadic suppression while estimating duration, and 2) explicitly require direct comparisons of duration and number, and so it is unknown if the time-numerosity interaction critically depends on saccades and/or comparisons. To explore these distinct findings, several studies were conducted with human participants (n=40) performing different visual tasks with no referents and dot arrays consisting of combinations of seven log-spaced durations (750–2250ms) and numerosities (10–90 dots), whereupon subjects classified stimuli as short(small) or long(large). In Study 1, separate groups of participants (20 ea.) were explicitly told to either maintain central fixation (eyes-fixed) or not (eyes-free) during a visual temporal bisection task. Contrary to previous reports, duration was linearly underestimated, rather than overestimated, with increasing numerosities, with no effect of duration on dot estimates. Crucially, results for both groups were identical, indicating the effect was not driven by fixation. In Study 2, a new group (n=20) performed a duration/dot bisection task in which participants were not told which feature to judge until after the stimulus had been extinguished. Again, only a linear underestimation of time was observed with increasing numerosity. In Study 3 (n=42), participants performed a duration/dot discrimination task, in which two quantities were explicitly compared. Consistent with previous reports, an overestimation of perceived duration was observed with increasing numerosities during the temporal task. However, perceived duration was near identical during the numerosity task. Overall, results from these studies emphasize the importance of considering task-dependent mechanisms involved in magnitude processing for visual tasks.