Attentional capture by singletons is contingent on top-down control settings: Evidence from electrophysiological measures

Abstract

The present study examined whether the capture of spatial attention is driven by stimulus salience (e.g., object uniqueness) or by a match to current attentional control settings (contingent capture). We measured the N2pc effect, a component of the event-related brain potential thought to reflect lateralized attentional allocation. On every trial, a noninformative cue display containing a colour singleton box was followed by a target display of letters. Participants searched for a target letter in a specified colour (in Experiments 1–3) or within a specified shape (in Experiment 4) while ignoring other stimuli. The key manipulation was whether the singleton cue contained the target-defining feature (e.g., a specific colour). Experiment 1 revealed signs of attention capture—a cue validity effect and an N2pc effect—only for singleton cues that contained the target-defining feature. This pattern persisted even when we increased the salience of the singleton box (Experiments 2 and 3). Irrelevant colour singletons also failed to produce a significant N2pc effect when the target was defined based on shape rather than colour (Experiment 4). We conclude that attention capture is strongly contingent on top-down attentional control settings, not bottom-up stimulus salience.

Keywords:

• Attention capture
• Spatial attention
• Visual attention
• N2pc effect

Acknowledgements

This research was supported by funding from the Oregon State University Research Office to M-CL and Oregon State University Undergraduate Research, Innovation, Scholarship & Creativity to LC.

Notes

¹A coloured version of the event sequence is available at www.oregonstate.edu/~lienm/ACS.html

²A reasonable concern is that by eliminating trials with EEG artefacts (such as eye movements), we are actually eliminating the very trials on which attention was captured. Likewise, by eliminating participants with HEOG activity greater than 3µV in either direction, we are perhaps eliminating participants with the greatest amount of capture. To address this issue, we reanalysed the data without these exclusions. The results were indistinguishable from our original data with respect to the key dependent measures. Similar to the data reported in the main analyses, the cue validity effects for the irrelevant and competing singleton conditions were only 4 ms (ns) and −15 ms, F(1, 15) = 13.91, p<.01, respectively. Furthermore, the irrelevant singleton condition still showed no sign of an N2pc effect (the average effect was 0.153 µV), ts(15) ≤ 1.05, ps ≥ .3083. The competing singleton condition still showed a reverse N2pc effect (the average effect was 0.787 µV), ts(15) ≥ 3.24, ps<.01. Similarly, Experiment 2 showed no cue validity effect for both irrelevant and competing singleton conditions (the effect was only 3 ms, Fs < 1). Most importantly, the irrelevant and competing singleton conditions still showed no sign of a singleton-elicited N2pc effect (−0.036 µV and 0.098 µV, respectively; ps ≥ .6234). Thus, the absence of the capture by irrelevant singleton cues in our experiments was not due to the artefact rejection criteria we adopted.

³We wanted to use diamonds and squares as targets, since they are well-matched, differing only in orientation. This meant that we could no longer use squares in the fixation and cue displays. So we replaced the squares with circles, against which the diamond and square targets were roughly equally easy to discriminate.

⁴We define the N2pc effect with respect to the singleton cue location. As a consequence, when the singleton cue and target are in different hemifields, the polarity of the N2pc effect elicited by the target becomes positive. Thus, before averaging across the same and different hemifield conditions, we first reversed the polarity of the target-elicited N2pc effect for the different hemifield condition.