Preemptive control of attentional capture by colour: Evidence from trial-by-trial analyses and orderings of onsets of capture effects in reaction time distributions

Abstract

According to the preemptive-control hypothesis, participants can specify their control settings to attend to relevant target colours or to ignore the irrelevant distractor colours in advance of the displays. Two predictions of this hypothesis were tested. First, with the control settings being specified in advance, capture by a stimulus that better matches the settings was expected to temporally precede capture by a stimulus that matches the setting less well. Second, with the control settings being specified in advance, stronger capture by the better matching than by the less matching stimulus was predicted not to be a stimulus-driven consequence of the target colour in a preceding trial. Both predictions were shown to hold true under different conditions in three experiments.

Acknowledgments

Supported by Deutsche Forschungsgemeinschaft Grant AN 393/1–1 to Ulrich Ansorge, Holk Cruse, and Odmar Neumann. Thanks to Bradley Gibson, Christian Olivers, Roger Remington, Massimo Turatto, and three anonymous reviewers for insightful comments on previous drafts of the manuscript, and to Heike Hartwig-Jakobs for help with the final preparation of the manuscript.

Notes

¹ We are grateful to Chris Olivers for pointing out this alternative hypothesis to us.

² By averaging single-trial performance across participants and by focusing of the additional analyses on (shifts of) onsets of capture effects, information about (a) individual distributions and (b) distribution characteristics besides onsets, such as skewness or maxima, was lost. We think, however, that these shortcomings of our additional analyses were justified for at least three reasons. First, some of the lost information was contained in the more traditional mean analysis of our results presented above. For example, different maxima of the capture effects in set-matching relative to less set-matching conditions were reflected in a main effect of singleton type. Second, no advance predictions can be derived from the preemptive control hypothesis for parameters of the RT distributions besides onsets and maxima. This is true, for instance, for skewness. Finally, averaging across single trials provided a better temporal resolution for the analysis of the onsets of the capture effects than averaging across a range of individual trials, which is what is essentially done in vincentizing (cf. Ansorge et al., 2005).

³ In Experiment 1's matching condition, targets were white, whereas in the less set-matching condition, targets were coloured. To rule out that perceptual factors were responsible for different capture effects in matching and less matching conditions, we ran a brief control with 8 participants searching for coloured (red and green targets) in both matching and less set-matching conditions. Main results were essentially the same as before: There was a significant main effect of singleton type, F (1, 7) = 5.72, p < .05, and a significant interaction of Singleton Type × Distance, F (3, 21) = 6.01, p < .01. In set-matching and less set-matching singleton conditions, RT started from about the same level in condition p0 (set-matching: RT = 699 ms; less set-matching: RT = 725 ms), and a stronger capture effect of the set-matching singleton was apparent from its significantly stronger interference than that for the less set matching singleton for all distances beyond p0 (RT in set-matching conditions was 719 ms, 754 ms, and 761 ms from p1 to p3, respectively, and RT in less set-matching conditions, was 668 ms, 686 ms, and 676 ms from p1 to p3, respectively), all three ts(7) > 2.42, all ps < .05. Also, in a follow-up analysis, including the factor preceding target colour, neither a significant main effect of preceding target colour, nor a Singleton Type × Distance × Preceding target colour interaction was observed, both Fs < 1.00. Finally, a corresponding ANOVA of the arc-sine transformed error rates yielded neither a significant main effect, nor a significant interaction, all Fs < 1.20, all ps > .34. Thus, a speed–accuracy trade-off does not explain the RT results. Hence, results of Experiment 1 cannot be attributed to differences between conditions with white and coloured targets, a conclusion that is also corroborated by the present Experiment 3.

⁴ On the basis of the procedures used by Maljkovic and Nakayama (1994), one might speculate that what matters for a stronger singleton capture effect in a current trial N is the colour of the singleton in the preceding trial N – 1, and not the colour of the target in trial N – 1. However, this alternative interpretation of the results of Maljkovic and Nakayama does not explain the different results between the studies. In the present investigation, trial-by-trial singleton-colour repetitions were restricted to those conditions in which the singleton displayed in trial N had a different colour from that of the target in trial N – 1. Therefore, if it were true that trial-by-trial singleton colour repetitions account for the stronger capture effect of the better matching colour singleton, we would have found a significant influence of the variable preceding target colour, only with an inverted pattern of results as compared to the priming-by-pop-out hypothesis outlined in the Introduction.