Abstract: The biased competition model for attentional selection proposed by Desimone and Duncan (1995) provided a framework for researchers to understand the interaction between top-down working memory factors and bottom-up sensory factors in attentional control. This model suggested that the stimulus representation stored in visual working memory would automatically capture attention in a visual search. However, other studies found that the content of visual working memory played no role in attentional capture. In order to explore the influence of visual working memory content on top-down attentional control, we employed a visual search paradigm with a secondary memory task and recorded ERP’s.
Sixteen participants performed the task under long or short ISI conditions. The short ISI varied randomly from 400 to 600ms, while the long ISI varied randomly from 800 to 900ms. Each trial began with a black fixation cross displayed on a gray background for 500ms, followed by a memory item (abstract shape) for 1000 ms, and a blank interval (ISI). After the interval, a search array of four abstract symmetrical shapes containing an arrow each was presented for 2000ms, and participants had to use a button box to indicate the direction of the only horizontal arrow presented in these shapes. This was followed by a 200ms interval. Finally, a probe stimulus was displayed for 1500ms, and participants had to indicate whether this item was the same as the initial abstract shape presented on its own at the beginning of the trial. The next trail commenced after a 500ms intertrial interval. The search arrays were divided into valid (the memory item appears in the search array and contains the horizontal arrow), invalid (the memory item appears in the search array, but does not contain a horizontal arrow), and neutral (the memory item is absent from the search array).
The behavioral data demonstrated that the RTs were significantly shorter in the valid condition than those in the invalid condition regardless of ISI duration. The ERP data showed that the amplitudes of frontal P2 in the valid condition were significantly larger than those in the neutral condition. The amplitudes and latencies of occipital N1 and P1 did not show any significant differences between all visual working memory content conditions (valid, invalid and neutral). However, the main effect of ISI was reliable on the amplitudes of both P1 and N1. The amplitudes of P1 in the short ISI condition were larger than those in the long ISI condition, by the contrast, the amplitudes of N1 in the short ISI condition were smaller than those in the long ISI condition. That finding was considered to be related to the time course of visual working memory processing. Furthermore, the amplitudes of P300 in the valid condition were larger than those in the invalid condition when the ISI was short, and in contrast, the amplitudes of P300 in the valid condition were significantly smaller than those in the invalid condition when the ISI was long.
The overall findings suggest that the influence of visual working memory content on top-down attentional control was affected by the time course of visual working memory processing. In the valid condition where search targets matched the content of working memory, an automatic selection advantage was observed for RTs and ERPs.

Key words: visual working memory, information content, ISI, top-down attentional control, ERPs