A parallel between single-letter and multiple-letter recall of tachistoscopic letter displays.

Abstract

Harcum (1965) has observed a close relationship between error functions for verbal serial learning and tachistoscopic pattern perception. This relationship suggests that common organizational and processing factors subsume both learning and perceptual tasks. A further parallel is here noted, between error curves arising from two different tachistoscopic tasks. Fig. 1 shows the distributions of errors from (a) a poststimulus sampling situation wherein rows of eight letters were exposed for 50 msec. and, at various delays before and after exposure, Ss were cued to report a single letter from the total display (Averbach & Sperling, 1961), and (b) an experiment conducted by the author where rows of eight letters were exposed for 100 msec. and Ss were instructed to report all letters shown. The remarkable similarity berween these curves may be taken as further support for the role of sequential trace-scanning factors, as opposed to sequential reporting factors, in determining the shape of the perceptual serial-position error curve. Contrary to the bow-shape curves reported by Harcum (1965), however, the present curves showed a dip in errors for letters near fixation. Unpublished work has shown that when lines of letters are exposed for 60 msec. or 120 msec., dips occur in the error curve; at 240 msec. a bowshape curve emerges. When lines of binary targets are exposed for 60 msec., 120 msec., or 240 msec., bow-shape error curves result. These differences may help reconcile Harcum's data with rhose reported here.