Using Rapid Serial Visual Presentation to Measure Set-Specific Capture, a Consequence of Distraction While Multitasking

Contingent attentional capture refers to a performance cost (slower reaction times and lower accuracy) that occurs when a participant erroneously directs attention to a distractor similar to their search goal. Indexing top-down orienting of attention, contingent attentional capture only occurs when a goal-relevant distractor is present (e.g., a green digit when searching for green letters), but not when a goal-irrelevant stimulus is present (e.g., a blue digit). Studies of contingent attentional capture have been integral to the understanding of top-down orienting and the limitations of information processing, namely, that once a stimulus captures attention, it is processed in a serial and effortful manner^1,2,3. Contingent attentional capture is most often measured using static displays that mimic a common visual search, such as searching for a red pepper in the produce section of a grocery store^3,4. In this example, an item sharing features with the target, such as a red apple, might capture attention, slowing down the search. Contingent attentional capture can be observed for color^3,5,6,7, shape⁸, motion⁹, time¹⁰, and semantic relevance^11,12. In addition to static displays, contingent attentional capture has been measured using dynamic displays that mimic situations such as searching for a landmark while driving along a road, or looking for a person in a quickly moving crowd^13,14.

More recently, researchers have investigated the consequences of attending to distractors when more than one search goal is active (such as searching for a red pepper and garlic at the same time^{7,8,15,16,17,18,19,20,21,22,23}.) In such situations, distraction costs can be especially devastating. While evidence is mixed as to whether multi-goal searches impair performance when distraction is not present, attentional capture from goal-related distractors can cause very large deficits in performance. In particular, we identified a new form of attentional capture called "set-specific capture," which occurs when multiple goals are concurrently maintained. In the case of set-specific capture, performance costs are especially large when a distractor resembling one target goal (e.g., an apple) grabs attention from the target item matching the other goal (e.g., the garlic)^7,20,21,22. See Figure 1 for an explanation of a typical finding, using this grocery example.

As in the case with contingent attentional capture, set-specific capture reveals that information is processed in a serial and effortful manner: when a distractor captures attention, attentional resources are drawn away from the target. In addition, set-specific capture shows that directing attention to the distractor's features leads to enhancement of the related goal within working memory. Thus, when more than one goal is concurrently maintained, this goal enhancement comes at the expense of any other current goals^7,21,22. Set-specific capture is a consequence of multitasking, akin to switch costs and mixing costs found in task-switching studies, but also distinct from these measures²⁴. It is important that future studies investigate this multitasking cost, both in order to understand the magnitude and nature of the impairment for practical reasons (e.g., safety-related situations involving dual-tasking), as well as to refine our understanding of the mechanics of visual search and how goals are maintained. For example, set-specific capture provides support for the idea that a single goal can be focused upon while a target or target-resembling distractor is attended, but that more goals are maintained in an accessory state during visual search^25,26,27.

The present method provides a robust way of measuring both contingent attentional capture and set-specific capture within a single paradigm. It uses a dynamic display, inspired by previous work on the attentional blink and contingent attentional capture with rapid serial visual presentations (RSVPs) of stimuli^{13,14,28,29,30}. This type of display yields much larger effects than do static display tasks, which usually rely on reaction time as a dependent measure, rather than accuracy^3,31,32. These larger effects allow researchers to use this paradigm to measure more sensitive manipulations of set-specific capture, such as the effect of practice²⁰.

In this task, participants search a heterogeneously colored, centrally-located RSVP for letters appearing in either of two "target" ink colors (e.g., green and orange; see Figure 2 for example stimulus colors). Any time a participant detects a target-colored letter appearing in the central display, they indicate whether the letter was from the first half of the alphabet ("press the 'J' key") or the second half of the alphabet ("press the 'K' key"). Meanwhile, participants ignore two RSVP displays consisting of mostly grey letters that appear on either side of the central display. Thus, at any given time, there are three letters on the screen at once – one centrally located and two peripheral. The letters change identity and color every 116 ms.

An experiment may consist of the following trial types: Target Alone, Distractor Alone, Non-Target Colored Distractor (NTC), Same Target Colored Distractor (STC), and Different Target Colored Distractor (DTC). In the Target Alone trial type, a target letter (e.g., a green C) appears in the central RSVP, without any color changes occurring in the peripheral RSVPs preceding it. In the Distractor Alone trial type, a target-colored item appears in one of the peripheral RSVP displays without a target item appearing afterward. The purpose of this trial type is to prevent participants from using a peripheral color change to predict an upcoming target, by including some trials in which a distractor did not predict a target. In the NTC, STC, and DTC trial types, a colored letter distractor appears in one of the peripheral displays before the target appears centrally, with a "lag" of 1 – 4 display frames (116 – 464 ms) between the appearance of the distractor and the target. For NTC trials, the distractor is not target-colored (e.g., a purple 'V'). In STC trials, the distractor (e.g., an orange 'B') is the same color as the following target (e.g., an orange 'T'). In DTC trials, the distractor (e.g., an orange 'C') is target-colored, but not the same color as the upcoming target (e.g., a green 'V'). See Figure 3 for a schematic of the task, including examples of each trial type. See Video 1 (video) for an example of the task. Viewed on loop, the example includes two targets. Video 2 (video) is the same video at a reduced speed for clarity.

Contingent attentional capture is indicated by the difference between NTC and STC performance, as a target-colored item captures attention only when it bears resemblance to one of the current goals (i.e., not on NTC trials, which usually yield the same accuracy level as Target Alone trials). Set-specific capture is indicated by the difference between STC and DTC performance. We have published several versions of this task, with slightly different configurations of trial types (i.e., with or without NTC and Distractor Alone trials; with just lags 1 and 3, with a variety of target colors, with three targets, etc.^7,20,21,22).

One notable feature of this method is that it uses a continuous display. Each trial includes the minimum components to represent that trial type, (e.g., a peripheral distractor, a target, and any letters that appeared in time between the distractor and target.) "Filler" stimuli connect one trial to the next seamlessly, and participants respond during this intertrial interval, whenever they detect a target. The interval lasts from 15 – 21 frames (1740 – 2436 ms), which is sufficient time to respond; most responses occur within 700 ms. An advantage of this method is that chance performance is near 0%; participants are not explicitly aware that a trial has ended if they miss a target item. This allows for three types of outcomes: 1) an identified letter, which will lead to a correct response, 2) a detected but not identified item (e.g., "I saw something green"), which will lead to a 50% chance of a correct response, and 3) an undetected / missed item, which leads to no response (coded as inaccurate). These three outcomes provide more information about the degree of stimulus processing than do tasks with a two alternative forced choice response, which cannot differentiate between detection-without-identification (i.e., a response error) and an outright miss (i.e., an omission error).

We describe the method here as we have used it in published work, in which participants search for colored letters. However, it can be modified for use with images³³ and potentially other stimuli, such as words³⁴. Moreover, distractors can appear as other colored items in the central display rather than just as colored letters appearing in the periphery (e.g., a target-colored digit in the central display)²¹. It is also likely that set-specific capture can be identified in static displays. The further development of the extensions of this method will allow researchers to investigate topics such as the effect of reward and motivation on distraction³⁵, or whether distraction costs are modulated by the number of concurrently maintained goals³³. Other applications could include measuring distraction costs in real-world contexts such as when completing a demanding visual search task (e.g., airport baggage screening or radiology screening)^36,37,38.