On the generality of the contingent orienting hypothesis

Abstract

The contingent-orienting hypothesis states that attentional capture by a task-irrelevant stimulus is contingent on whether that stimulus shares a feature property that is critical to the task at hand [Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntary covert orienting is contingent on attentional control settings. Journal of Experimental Psychology: Human Perception and Performance, 18, 1030–1044]. Studies supporting this hypothesis have mostly used set size four displays throughout the experiment and thus constrict its ecological validity, since conclusions drawn from experiments using fixed set-size displays may not be generalized to other conditions with different set sizes. We used a spatial cueing paradigm in which a non-informative onset or color cue preceded an onset or a color target, and manipulated set size as a within- or between-subject factor. In four experiments, the original finding of Folk et al. (1992) was replicated only when a fixed set size (four) was used throughout. When both set-size four and eight were used in an experiment, stimulus-driven capture by onset in search of a color target was found even for set-size four displays. These results raise doubts as to the generality of the contingent-orienting hypothesis and help to delineate the boundary conditions on this hypothesis.

Introduction

It has long been recognized that attentional control involves both stimulus-driven and goal-directed components (James, 1890). Recently, the issue of whether and how a task-irrelevant stimulus receives attentional priority in a stimulus-driven fashion has been intensively studied under the subject of attentional capture (e.g., Gibson, 1996, Hillstrom and Yantis, 1994, Horstmann, 2002, Johnson et al., 2001, Rauschenberger and Yantis, 2001, Remington et al., 1992, Theeuwes, 1994, Turatto and Galfano, 2000, Turatto and Galfano, 2001, Yantis, 1993, Yantis and Egeth, 1999). On the one hand, some researchers argue for stimulus-driven attentional capture even when the stimulus does not match with the defining feature of the target (the stimulus-driven orienting hypothesis, e.g., Theeuwes, 1994, Yantis and Jonides, 1984). On the other hand, other researchers doubt whether purely stimulus-driven capture ever exists and postulate instead that attentional capture is contingent on the attentional control setting adopted by the participant to perform the task (the contingent-orienting hypothesis, Folk, Remington, & Johnston, 1992).

Studies that support the stimulus-driven orienting hypothesis have mostly used a visual search paradigm. Two indices are used as evidence for attentional capture. The first index is poorer performance when the target is presented along with a task-irrelevant distractor. For example, Theeuwes (1994) asked his participants to search for a target that was defined by a particular color, and found longer reaction times (RTs) when there was an onset distractor than when there was no onset distractor. Evidently, the appearance of the onset distractor affects the allocation of attention and thus is detrimental to the detection of the color target. The second index is the near-zero search slope when RT is plotted as a function of set size. For example, Yantis and Jonides (1984) asked their participants to search for a target letter. When the target happened to be an onset (which occurred only at chance level), the RT was relatively independent of the total number of items in the display, suggesting that the onset stimulus captures attention.

Studies that support the contingent orienting hypothesis, on the other hand, usually use a spatial cueing paradigm in which a task-irrelevant cue is presented before the target. RTs to a target appearing at a cued location are compared to those when the location is uncued. The cue is non-informative because it is no more likely that the target location would be the same as the cue than that it would be different from it. Thus, for the task at hand, the best strategy is to ignore the cue and respond to the target as soon as possible. If RT is shorter when the target appears at the cued location than when it appears at the un-cued one (i.e., the validity effect), attentional capture by the cue is inferred. The absence of this validity effect, conversely, indicates that the appearance of the cue does not affect the performance; that is, no attentional capture occurs.

Folk et al. (1992) used this spatial cueing paradigm and found that attentional capture is contingent on top-down attentional setting for static or dynamic discontinuity. For example, to find a red target letter, a red cue showed the validity effect, but an onset cue did not. In this case, the participant was supposedly set for static discontinuity elicited by the task of searching for a red target among white distractors since this is the property useful in locating the target, and a red cue was able to capture attention because it matches with the property that is critical for the task at hand. The onset cue, in contrast, could not capture attention because the cue property does not match with the attentional set elicited by the behavioral goal. They also tested the condition of a color or an onset cue that preceded an onset target, and the result was also consistent with the contingent orienting hypothesis; that is, only an onset cue (or a motion cue, Folk, Remington, & Wright, 1994) that signaled the dynamic discontinuity as in the target display captured attention. These findings are important because they suggest that involuntary orienting is in fact contingent on endogenous attentional control setting, and only within the same setting can stimulus salience play a role (Folk et al., 1994).

However, only displays with set-size four were used and they were used throughout the experiments in the series of studies supporting the contingent-orienting hypothesis (e.g., Folk and Remington, 1998, Folk and Remington, 1999, Folk et al., 1992, Folk et al., 1994). Hence, it remains unknown whether the results can be generalized to other conditions. There are two reasons to doubt the generalizability of this hypothesis. First, our visual world is full of various numbers of stimuli, and thus using a fixed (and rather small) set size throughout the experiments would lead to the problem of constricted ecological validity. Secondly, previous studies have used a larger set size (e.g., six items, Lamy & Egeth, 2003) or different set sizes within blocks in an experiment (e.g., 12 and 20 items, Theeuwes, 2004) and found evidence for stimulus-driven attentional capture. Nevertheless, the displays and the paradigms in these studies were different from those used by Folk and colleagues, which preclude a fair comparison if using fixed set size is a problem in the series of studies by Folk and colleagues.

We provide a direct comparison in the current study by using a similar spatial cueing paradigm as in Folk et al. (1992) to test whether the contingent-orienting hypothesis remains true by manipulating set size as a within- or between-subject factor. In Experiment 1, we modified the original design of Folk et al. (1992) by adding a set-size eight condition. In Experiment 2, we further added another distracting color along with the target color so that the color target was no longer a singleton in the target display and thus searching for a singleton was not possible. Experiments 3 and 4 more closely followed the displays and the designs in Folk et al. (1992) to illuminate the boundary conditions on the contingent-orienting hypothesis.