The role of visual working memory in temporary serial retention of verbal i
nformation was examined in four experiments on immediate serial recall of w
ords that varied in visual similarity and letters that varied in the visual
consistency between upper and lower case. Experiments 1 and 2 involved wor
ds that were either visually similar (e.g. fly, cry, dry; hem, new, few) or
were visually distinct (e.g. guy sigh, lie; mho, blue, eme). Experiments 3
and 4 involved serial recall of both letter and case from sequences of let
ters chosen such that the upper- and lower-case versions were visually simi
lar, for example Kk, Cc, Zz, Wm, or were visually dissimilar, for example D
d, Nh, Ru, Qq. Hence in the latter set, case information was encoded in ter
ms of both the shape and the size of the letters. With both words and lette
rs, the visually similar items resulted in poorer recall both with and with
out concurrent articulatory suppression. This visual similarity effect was
robust and was replicated across the four experiments. The effect was not r
estricted to any particular serial position and was particularly salient in
the recall of letter case. These data suggest the presence of a visual cod
e for retention of visually presented verbal sequences in addition to a pho
nological code, and they are consistent with the use of a visual temporary
memory, or visual "cache", in verbal serial recall tasks.