alleges to provide evidence for pure bottom-up capture, however, like the findings of Theeuwes and Burger there may be an alternative explanation for such evidence. In a visual search task Turatto and Galfano (2001) asked participants to search for a rotated T among rotated Ls and detect if the target was present or absent. Each letter was presented on one of several coloured discs arranged in a circle around fixation. A uniquely coloured disc (singleton) was also presented and RTs were recorded in relation to the distance of the target from the irrelevant singleton. They found that the singleton captured attention despite being task-irrelevant (RTs were longer to detect the target when it appeared closer to the irrelevant singleton, revealing that more inhibition had to be applied to the singleton, leaving fewer resources to detect the target) and concluded that participants were unable to prevent the automatic capture and that the search strategy could not override bottom-up salience. Turatto and Galfano state that they provide “direct behavioural proof that a task-irrelevant color singleton can elicit an automatic attentional capture in the absence of any set” (pp. 290). However, it is entirely possible that participants used a singleton detection search mode (Bacon & Egeth, 1994), in which case it would be no surprise that the irrelevant singleton captured attention. This argument is strengthened by their results which show that capture of attention by the irrelevant singleton reduced as the task went on (similar to the results of Theeuwes & Burger and possibly showing that participants modified their search strategy with experience to avoid automatic
capture), and that the singleton failed to capture attention when participants were told that it did not predict where the target would appear. This clearly shows that
automatic capture was reliant upon top-down knowledge about the task (despite the claims made by the authors).
The results from an experiment completed by Pratt, Sekuler, and McAuliffe (2001) are consistent with the above argument. Using Folk et al.’s 1992 spatial cuing paradigm they studied the ability of cues to capture attention when they did not match the target-defining features (and therefore did not match the top-down control
settings). When the experiment was blocked so that participants knew the target search feature on each trial, only the matching cues captured attention and elicited a cuing effect. When the trials were randomised so that participants did not know whether the target would be an onset or a colour singleton, all cues resulted in a cuing effect. Therefore when participants were unable to use top-down control settings to complete the task an irrelevant item which did not match the target feature captured attention, but when they could exert top-down control this automatic capture was suppressed. This illustrates that the level at which the control settings can be configured has a significant impact upon the capture of attention by task-irrelevant stimuli.
There is still no consensus over the relative influence of top-down and bottom- up control on selective attention, and some researchers are now resorting to additional forms of evidence to provide support for their respective viewpoints. One popular method has been to measure the event related potential (ERP) associated with activity during target detection (the N2pc). Using a visual search task Hickey, McDonald, and Theeuwes (2006) found that when participants had to search for a target in the
presence of a more salient distracter, the distracter elicited activity from the N2pc prior to the target, indicating that the most salient item will always capture attention regardless of the task demands. Conversely, using a spatial cuing paradigm, Eimer
and Kiss (2008) and Lien, Ruthruff, Goodin, and Remington (2008) found that a cue only elicited strong N2pc activity when it matched the target-defining features (e.g., an onset cue only evoked an N2pc effect when the target was also an onset). In both spatial cuing experiments the onset was the most salient item yet automatic capture was contingent upon the experimental condition, indicating that top-down control suppressed any bottom-up influence of stimulus salience.
It is important to note that of the studies reported in the literature those
claiming evidence for contingent capture often use a spatial cuing paradigm and those claiming to support stimulus-driven capture involve a visual search task. Theeuwes, Atchley, and Kramer (2000) claim that spatial cuing results in contingent capture (or rather the appearance of contingent capture) because disengagement of attention from a stimulus which matches the control settings takes longer than disengagement from a stimulus which does not match the control settings. This explanation is supported by evidence from Schreij, Owens, and Theeuwes (2008) which shows that presenting the cue at the same time as the target in a spatial cuing task eliminates the ‘contingent capture’ effect (because it eliminates the requirement to disengage from the cue), and all irrelevant singletons capture attention regardless of their task-relevance. The present author would argue that the task used by Theeuwes and colleagues (e.g., Schreij et al., 2008; Theeuwes, 1991b; Theeuwes & Burger, 1998) is equally
problematic for their conclusions. It is predicted that Theeuwes et al. consistently fall prey to underestimating the level at which the orienting system can be configured. For instance, they state that an irrelevant unique item captures attention because it is the most salient item in the display, yet there is evidence to show that this capture is contingent upon the top-down control setting because participants are actually searching for a unique item (using a singleton detection mode; e.g., Bacon & Egeth,
1994). In addition to this, even when Theeuwes et al. presume that the task can be completed using a specific feature search mode (and find that an irrelevant singleton which does not match the target-defining feature captures attention), participants may instead be allocating attention to a different feature of the target. For example, in the study completed by Theeuwes and Burger (1998) participants were asked to search for the letters ‘E’ or ‘R’ and even when the irrelevant singleton did not share the same colour as the target letter it still captured attention. Theeuwes and Burger assumed that participants used colour to search for the target but it may be reasonable to suggest that they are actually searching for ‘form’ because all items were grey and a colour search would not facilitate target detection. As the singleton was also always the letter ‘E’ or ‘R’ it would match the top-down control settings, therefore the
capture of attention was due to the task demands, not the salience of the stimulus. One aim of this thesis is to determine the levels at which the orienting system can be configured. This will help to clarify whether the design of the studies completed by Theeuwes et al. can really provide evidence for pure bottom-up capture.
1.2.5 Additional evidence for top-down control over stimulus-driven capture