We studied exogenous cuing caused by an uninformative abrupt onset during a time when subjects were under the influence of the attentional blink. In two experiments, we found a reduced impact of exogenous cuing during the blink time of the attentional blink. The results indicate that involuntary orienting caused by abrupt onsets is sensitive to manipulation of available attentional resources. Thus, onset capture requires attention.
Reflexive orienting, also known as exogenous orienting, has been distinguished from voluntary orienting for decades (e.g., Jonides, 1981; Posner, 1980). It is believed that voluntary orienting is under a person's volitional control and is attention demanding, whereas reflexive orienting is effortless and involuntary. This distinction can be seen when an event such as an abrupt onset summons a person's attention in the absence of voluntary control (Christ & Abrams, 2006; Jonides, 1981; Yantis & Jonides, 1984). In addition, a very rapid search rate for onset targets relative to nononset targets is consistent with the idea that onset capture is highly automatic or load insensitive (Yantis & Jonides, 1984; Yantis & Hillstrom, 1994).
The insensitivity of onset capture to a concurrent perceptual or attentional load was corroborated by a recent study on the attentional blink (AB). The attentional blink refers to an impairment in the detection or identification of a second target that follows within about 500 msec of an earlier target in the same location (Raymond, Shapiro, & Arnell, 1992). The impairment, or blink, lasts for a few hundred milliseconds. If onset capture is truly not attention demanding, abrupt onsets in the periphery should capture attention even if they occur during the blink time of the AB. Consistent with this prediction, Ghorashi, Di Lollo, and Klein (2007) reported an intact cuing effect elicited by abrupt onsets during the AB (see Joseph, Chun, & Nakayama, 1997, for an experiment with a similar approach). In Ghorashi et al.'s study, after the appearance of the first target in a central RSVP stream, a solid square was presented as a peripheral cue shortly before the appearance of a ring of 12 letters. Subjects were required to report the identity of the first target and the orientation of the lone T in the ring. At all target-probe lags (of 90, 270, or 630 msec), subjects consistently showed higher accuracy in orientation discrimination when the T appeared at cued locations (where the square had been presented) than when it appeared at uncued locations. The finding that the cuing effect induced by onsets survived the AB is consistent with the traditional idea that the involuntary orienting elicited by onsets is effortless, demanding only a negligible amount of attentional resource.
However, there is still debate regarding the extent to which attentional capture caused by onsets is truly automatic. A few recent studies have challenged the traditional opinion by showing that attentional capture by abrupt onsets can be interrupted by a concurrent monitoring task. For example, Boot, Brockmole, and Simons (2005) found that abrupt onsets failed to capture attention in visual search when subjects had to perform a concurrent auditory one-back task. Another recent study (Santangelo, Olivetti Belardinelli, & Spence, 2007) showed that both reflexive visual and auditory orienting were disrupted when subjects were instructed to attend to an RSVP or RSAP stream.
There is clearly a discrepancy between the results of Ghorashi et al. (2007) on one hand and Boot et al. (2005) and Santangelo et al. (2007) on the other. How can we reconcile the discrepancy? One possibility is that a concurrent dual task differs from an AB task. In particular, the dual task requires continuous engagement of attention to the primary task, whereas AB depletes attention only for a brief period of time. This difference could account for the intact cuing effect observed at longer target-probe lags in the Ghorashi et al. …