Abstract: PURPOSE: When viewing a rapid successive sequence of items, the second target (T2) was often missed if it presents 200-400ms after the first target (T1) onset—known as the atten- tional blink (AB)(Raymond,1992). A number of failed experiments that attempt to eliminate AB have led to the belief that it is impossible to get rid of the capacity limit associated with AB. However, Hoon Choia (2012) reported AB can be eliminated by short-term training with a color-salient T2. In this study, RSVP paradigm and categorical feature-controlled symbols(targets) were used to manipulate subjects' attention and they were asked to perform a categorization task. We wondered whether AB can be overcome without train- ing.
METHODS: Digits (2-9) or letters were randomly factitiously classified into three categories, such as furniture, electrical-equipment, and stationary. Letters or digits were used as dis- tractors. Subjects were asked to respond two classification questions (Which categories does T1 and T2 belong to?).
RESULTS: In Exp1, when T1 and T2 belonged to the same category, AB was absent. However, when they were in different categories, there was a strong AB. We changed the categories into self, friends and others (Exp2), increased items into six within each category (Exp3), or increased the number of categories into 6 (Exp4) while keeping other conditions. The results were extremely consistent with Exp1. No significant differences were observed among these results.
CONCLUSIONS: These experiments show a clear evidence that when T1 and T2 belongs to the same category, the AB disappears. It seems that if T1 and T2 are in different artificially defined categories, it will trigger AB. This may help to provide a new window for eliminate AB in selective attention.

Key words: Attentional Blink, Category, Selective attention