Subitizing: Direct apprehension or serial processing?

Tasks involving the judgment of numerosity in the subitizing range (1–5 elements) typically yield small but significant variations in performance as a function of the number of elements. Such a result is consistent with the existence of a serial component of numerosity processing in the subitizing range. In 1985, Sagi and Julesz reported the results of a subitizing task in which performance remained virtually constant as numerosity was varied. Three experiments are reported that explore this discrepancy. The results indicate that the discrepancy is not due to the nature of the stimuli, presentation mode, dependent measure, or level of practice used by Sagi and Julesz. All conditions showed variations in performance as a function of target numerosity. Our results are consistent with models that assume that there is a serial component to subitizing.