Finding the relevant attribute of visual or auditory stimuli.

The stimuli in a simple attribute-identification task were geometric designs or their verbal descriptions. Presentation was visual or auditory, in a constant order or a mixed order that varied the serial position of the relevant dimension (the dimension providing the relevant attribute) on each trial. Constant-auditory problems whose relevant dimensions were first or last were solved better, while mixed-auditory problems were solved least well, and visual problems were solved best overall with no serial-position effects. However, the results of an experiment in which the stimuli were noun pairs suggest that visual presentation is advantageous only if it produces a nonverbal code from which subjects can recover attributes lost from their descriptions during rehearsal.