Priming effects in attentional gating

JAMES F.JUOLA 0 PRATHYUSHA DUVURU 0 MATTHEW S. PETERSON 0 0 University of Kansas , Lawrence, Kansas Three experiments were conducted to determine the effects of amount of prior target information (Experiment 1) and semantic priming (Experiment 2) in an attentional gating task. The goal was to determine some causes of the processing deficits commonly observed in perceiving successive visual stimuli. Items in a rapid serial visual presentation (RSVP) stream are subject to processing deficits before they are processed to the level of recognition (early selection) and after they have been recognized (late selection). Deficits in the former case presumably are due to an early filter that prevents complete recognition and semantic analysis, whereas deficits in the latter case arise from interference or response competition, producing forgetting among a set of recognized items. The semantic-priming effects found between a cue and a target (Experiment 2) and between two successive targets (Experiment 3) indicate that top-down processes can increase the subjective availability of related items. The results are consistent with the idea that most processing deficits observed in search through an RSVP sequence are due to limited capacity in our ability to form episodic representations of all the items in the sequence. - correspond to traditional early- and late-selection theories viewed a single stream of black letters and were to identify of attention. The purpose ofthe present research is to gain a target letter appearing in white and then to detect the a better understanding of mechanisms that underlie post presence or absence of a T2 (probe), which was always target processing deficits in RSVP tasks. the black letter X. In a control condition, participants were Broadbent and Broadbent (1987) were among the first to ignore the target and merely detect the probe. The probe to report the interfering effect of identifying one target on was present on 50% of the trials, and the number of inter the processing of a later one. Their participants were to vening items between the target and the probe was varied identify two target words defined by either physical fea between zero and seven. Relative to the control condition, tures or category membership. Successive words were pre in which probe detection was uniformly high, participants sented at rates of 80 or 120 msec/item, and the number who were required to identify the target showed a signif of intervening words between targets was varied. They icant deficit in probe detection when it occurred between found that correct identification of the first target (T 1) 180 and 450 msec after the target. Even though both impaired identification of the second target (T2) ifit oc groups received the same sensory information, the probe curred within about 500 msec ofT 1, and that, within this detection deficit occurred only when the target was to be interval, correct T2 identifications increased steadily with identified, suggesting that the temporary impairment in increases in the temporal interval between targets. They visual processing was due mainly to attentional, rather than interpreted the results in terms of a two-stage, detect to sensory, limitations. This temporary deficit in visual then-identify model (Broadbent, 1977). In the first stage, processing resulting from attentionallimitations has been observers scan the series of items to detect the target termed the attentional blink (AB; Raymond et aI., 1992). defining feature, and once the target is detected, it enters There are different views about when during processing a limited-capacity second stage, at which it is identified. the AB occurs. One important difference is whether the They argued that it is the attention-demanding process of deficit occurs before or after target identification, which identification that is responsible for impaired processing corresponds to traditional ear1y- and late-selection theo of subsequent items. ries of attention. The inhibitory model proposed by Ray Using a slightly different procedure involving two mond et al. (1992) is similar to Weichselgartner and Sper streams of items, Reeves and Sperling (1986) observed ling's (1987) and explains the AB in terms of a gate that similar posttarget deficits. They asked participants to filters out posttarget items before they can be identified. monitor an RSVP stream presented to the left offixation In their model, if the item following the target (T+1 item) for the presence of a predefined cue letter. On detecting appears before target identification is complete, it leads the cue, participants were to shift their attention to a digit to interference with target processing. To prevent further stream presented to the right of fixation and report the interference, a suppressive mechanism blocks process earliest visible numeral. Typically, the numeral reported ing of subsequent items until the target is identified. The occurred about 300 msec after the cue, although there was finding that replacing the T+1 item with a blank field al considerable spread across positions. The average time most eliminated the AB was taken as support for this delay between cue onset and the onset of the reported model. However, the inhibitory model is incompatible digit was taken as a measure of attentional reaction time with results such as the finding that a reduction in target (ART), which is the time needed to identify the cue in the identification difficulty does not always influence the letter stream, shift attention to the numeral stream, and magnitude of the AB deficit (Shapiro, Raymond, & Ar then identify the first visible numeral. Weichselgartner nell,1994). and Sperling (1987) found deficits in reporting posttarget A second explanation for the AB is the interference items even when shifting attention to a different location model (Shapiro et aI., 1994), adapted from Duncan and was not required. In their study, a single stream of digits Humphreys's (1989) similarity theory. Following the ini was presented, and observers were to identify the first four tial representation of each item in the stream, items that items following a cue (an outline square around one of closely match internal templates for the target and the the digits). Most ofthe digits recalled occurred either be probe are selected for entry into a visual short-term mem tween 0 and 100 or between 300 and 400 msec after the ory (VSTM). By virtue of being temporally contiguous cue. They proposed the operation of two consecutive, to the target and probe, items that immediately follow partially overlapping attentional processes. One is a fast, them can also gain entry into VSTM. It is interference automatic process, triggered by cue detection, that is re among these items in VSTM that leads to failure in de sponsible for identification of the cue and the item fol tecting the probe. Since the target is usually the first item lowing it. The second is a slower, (...truncated)