Attention and Chunking in Visual Search among Letter Stimuli

Abstract

The experimental study of top-down influences upon visual search for a target letter in large letter arrays has been performed. The core question of this study was whether words embedded into random letter strings - and not perceptually segregated for an observer - can influence efficiency of the search for letters embedded either in words, or in random letter sequences between words. The experiment was based on an original modification of the classic selective attention test developed a century ago by H. Muensterberg. Participants performed visual search for a predefined letter of the Russian alphabet in letter matrices which included Russian words. In the first experimental condition, target letters always belonged to words, but the participants were not warned about the words, or about the arrangement of target letters in the matrix. In the second experimental condition, target letters never belonged to words. In the third (control) condition, there were no words embedded into letter matrices. The study revealed a dissociation between visual search efficiency and subjective representation of the search task. Although presentation conditions did not influence search rate, the participants significantly differed in their subjective experience of this influence. If target letters belonged to words, the words subjectively facilitated performance; whereas, if target letters did not belong to words, the words subjectively hampered the search. Moreover, if target letters were embedded in words, the participants noticed the words twice as often as in the opposite condition. We interpret this result as a dissociation between top-down processes in the visual system, and top-down influences upon visual search arising from chunking in visual information processing.