Abstract: The role of memory in visual search has become a controversial issue. Most theories of serial visual search support the view that visual search requires memory and a serial scanning mechanism, such as the processes of inhibition of return in serial search (Klein, 1988). However, Horowitz and Wolfe (1998) claimed that serial visual search worked without memory. They compared the search rates in both static and randomly dynamic conditions that prevented parallel accumulation of information of a particular letter and found no significant difference between the two conditions. In the present study, iconic memory was examined in dynamic conditions. Besides, the new object advantage was tested to explain the outcome of Wolfe’s study.
In experiment 1 a comparison was made between two conditions: In the restrained dynamic condition, the items moved to positions previously occupied by other items every 110ms. In the static condition, the display items remained in the same location from one frame to the next. In experiment 2 a comparison was made between a restrained dynamic condition and a random condition, in which the items moved to positions not occupied by any item in the last frame. In experiment 3, half of the items moved to new locations and half randomly moved to ones previously occupied. This experiment compared two conditions: the target moving to new locations and the target moving to old locations where items appeared in the last frame. Different groups of students participated in the experiment 1,2 and 3, 10, 14, and 12 students, respectively. In all experiments, the number of letters in the display (set-size) varied between 8, 12, and 16. The slope of the target-present reaction time × set-size function was used to measure the efficiency of display search. Repeated measures ANOVA and linear regression were used to analyze the data.
The main results of this study were as follows: first, the efficiency of search under static conditions was significantly higher than that under restrained dynamic conditions; secondly, the efficiency of search under random conditions was also significantly higher than that under restrained dynamic conditions; thirdly, the two conditions in experiment 3 showed no difference.
Our results showed that under both random and static conditions visual search required memory, which did not support the visual search model proposed by Horowitz and Wolfe (1998). According to our results, under static conditions, observers might use a static search process and check the items one by one. In addition, under random conditions observers first compare the locations of items in the present frame with the previous locations in the last frame, which were kept in their iconic memory, and then guide attention to focus on new locations and suppress old ones. The findings of our study would help to explain why Wolfe did not observe a difference between static and random conditions and to reveal the role of iconic memory in visual search under random conditions. Finally, our study also indicated that the new object advantage was not significant under random conditions. Generally speaking, iconic memory contributes to the performance of visual search under random conditions by effective guidance of attention.

Key words: visual search, iconic memory, new object advantage, search mode