ИСТИНА

Muensterberg's selective attention test is a procedure of speeded visual search for words embedded in random letter strings (Burtt, 1917). Normally it is a slow effortful search, however, some words pop out, and it remains unclear whether words might somehow guide attention during the search, and to what extent unitization is possible without spatial segregation of a word. To address this issue, we studied letter search in large letter arrays containing words. In the initial experiment, we compared three conditions: 24 target letters in letter arrays were located always within words, always out of words, or there were no words in the array. Interestingly, the number of observers who noticed words during letter search differed: whereas 65.3% of participants reported noticing words when target letters belonged to them, only 37.5% reported noticing words when target letters were embedded in random letter sequences between them. Subjectively, words were perceived as helpful in the first condition and as distracting in the second condition. However, performance was the same in all three conditions, no matter whether the participants noticed words or not, and no matter whether they were naive or warned about the words. To understand this lack of influence from words embedded in letter arrays, we compared letter search and word search in the same arrays using both performance measures and eye-tracking. Letter search turned to be much faster than word search in the same arrays, with 75% of letters vs. 46% of words located within 1 minute. Surprisingly, there were no correlates of word detection in eye movements of the observers who noticed words when searching for letters. At the same time, there were significantly more fixations both within and out of words during word search as compared to letter search in the same arrays. In the further experiments, we found no transfer from letter search to the subsequent word search in the same arrays, and extremely scarce conscious and unconscious word processing, which we estimated using the Processes Dissociation Procedure by Jacoby (1991). Thus, our experiments clearly demonstrate the limitations of spontaneous unitization without spatial segregation of words, probably because word extraction requires the effortful letter-by-letter search for word borders. They also reveal the dissociation of subjective task representation and performance in the letter search task in large letter arrays containing words.