Abstract
Attention is often captured by irrelevant but salient changes in the environment, and usually results in slowed search speeds and increased errors during a typical visual search task. Nonetheless, a recent study conducted by Moher (2020, Psychological Science, 31[1], 31–42) found that the effect of a highly salient distractor on visual search depended on whether or not a target was also present in the display. While the distractor slowed search and increased errors for target-present trials, it speeded search for target-absent trials. Here, we aimed to replicate this finding and explore a potential boundary condition to the effect by manipulating the overall salience of the distractor. We did this by changing the size of the distractor to make it more or less salient. In Experiment, participants conducted a target-present and target-absent visual search task in the presence of a large, delayed-onset color distractor similar to that used in Moher’s Study. In Experiment 2, a distractor that was much smaller than that used in the original Moher study was utilized. Critically, when a large distractor was used, the original findings of Moher were largely replicated; large salient distractors speeded target-absent visual search and increased errors for target-present visual search. However, when a smaller distractor was used, the results differed. For target-absent trials, search speeds were slower when the distractor was present compared with when it was absent. Thus, it appears that a highly salient distractor might be needed to trigger a shift in visual search strategy, and subsequently, lower quitting thresholds.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
We use visual attention to select relevant information for further processing and ignore irrelevant information (Broadbent, 1982; Carrasco, 2011; Desimone & Duncan, 1995). For instance, when responding to emails, one might voluntarily focus their attention on each new email’s subject line while ignoring junk emails from advertising companies. However, it is often the case that salient events in our environment also capture attention, seemingly outside of volitional control. For example, a phone call might break a concentrated visual search of our emails (Carrasco, 2011; Corbetta & Shulman, 2002; Jonides, 1981; Petersen & Posner, 2012; Posner, 1980).
A singleton distractor task developed by Theeuwes (1991) is a seminal method used to measure attentional capture during visual search. In the task, the participant searches for a target shape amongst an array of other shapes. After locating the target shape, the participant then identifies the orientation of a line located inside the shape. On some trials, one of the nontarget items in the array is made highly salient by changing its color such that it is unique from other items in the array. It is hypothesized that this salient color distractor would automatically capture attention and impair target processing. Indeed, in Theeuwes (1991), search speeds were slowed for the primary task when a salient color distractor was present compared with when it was absent.
Despite there being differing theories as to the degree of automaticity of attentional capture (e.g., Bacon & Egeth, 1994; Folk et al., 1992), salient distractors usually impair processing (e.g., Geyer et al., 2008; Hickey et al., 2006; Lamy & Egeth, 2003; Müller et al., 2009; Sayim et al., 2010; Theeuwes, 1991, 1992, 2004; Theeuwes, Kramer, & Kingstone, 2004b). For example, Hickey et al. (2006) replicated the finding of Theeuwes, where salient color distractors slowed response times in the attention capture task described above. Furthermore, the authors found the salient distractor to elicit an earlier N2pc when the target and color distractor were presented on different sides of the search array, suggesting that it captured attention.
Nonetheless, Moher (2020) recently, quite insightfully and correctly, noted that most research exploring visual search and attention capture tends to use tasks in which a target is always present. However, in the “real world,” it is often not the case that a target is consistently present. For example, returning to our email scenario, there will often be times where there is not an email waiting to be found, but where a phone call still distracts us. Likewise, in medical diagnosis and baggage search, a target is often absent. To address this substantial gap in the literature, Moher tested the influence of salient distractors on search speeds and error rates in a target detection visual search task based on the work of Yantis and Egeth (1999). Participants searched for the presence versus absence of a vertical blue rectangle amongst tilted rectangles. The search arrays contained four or eight items, and the target was present on 50% (Experiments 1, 3) or 20% (Experiment 2) of trials. Of note, a much larger, red color distractor rectangle was also present on 50% of trials across the three experiments. This highly salient distractor replaced one of the tilted rectangles and had a delayed onset of 100 ms (see Fig. 1).
Example of distractor absent (above) and distractor present (below) search task used in Moher (2020). Note. The target is present and is the vertical line. The color distractor is the tilted red line. Participants responded “target-present” by pressing “m” and “target-absent” by pressing “z.” (Color figure online)
Similar to the existing literature exploring distractor processing, on target-present trials, the salient distractor slowed search speeds, as well as increased error rates. Critically, when the target was absent from the search array, the opposite pattern of results emerged. Salient distractors speeded search compared with when they were absent (the error rates were unaffected). This finding is surprising when taken at face value. After all, if something distracts us, we would expect it to slow our attainment of primary goals. Moher (2020) suggests that the best way to understand this somewhat counterintuitive data is to consider the effect of the salient distractor on quitting thresholds—the amount of evidence one requires before they terminate their search in a target detection task (Wolfe & Van Wert, 2010). For instance, during search tasks where the observer has to detect the presence or absence of a target (e.g., a tumor in medical image screening), they must decide when they will end a search if they have not yet located the target. Moher posits that a large salient color distractor may cause the quitting threshold to lower, resulting in observers ending their target search earlier. Furthermore, error rates for target-present trials would also increase (i.e., misses of the target).
The pattern of results observed in Moher’s study can also be understood in term of the competitive guided search model (Moran et al., 2013). In this model, searching for a target is an iterative process that includes a search termination mechanism based on conditional probabilities—the probability of quitting a search increases for each additional nontarget item selected (and then rejected). Initially, the salience map determines which target is most likely to be selected (which item is most distinctive from others in the array based on both top-down and bottom-up factors). Once an item is selected, the observer decides if that item is the target. If it is, this results in search termination; however, if it is not, then the nontarget is suppressed from selection and the probability of terminating the search increases. Finally, before selecting the next most salient item in the search array for inspection, the observer decides whether to terminate the search. Critically, the overall amount of activity in the salience map influences the probability of quitting a search such that it can accommodate both “pop-out” effects in parallel search and serial search. Thus, visual search on target-absent trials can be exhaustive (all items are selected) or based on changes in the quitting threshold after each search cycle (a search may be terminated early).
In Moher (2020), it is likely that the large, highly salient color distractor initially captured the participants’ attention and resulted in a nontarget decision. This decision would then cause inhibition of the salient distractor and continued search for the target. Nonetheless, given that search was serial, array items might have been selected multiple times. If the salient color distractor was reselected for processing in this cycle,1 it might cause the observer to more hastily terminate the search. In terms of the competitive guided search model, this would be akin to saying that the salient distractor’s reinspection increased the probability of search terminating because the observer would remember that they had already selected and rejected this highly salient item earlier on.
Finally, Moher (2020) also discusses his findings in the context of the subsequent search misses (SSM) literature (e.g., Cain et al., 2013), where participants are more likely to miss a second target when there are two targets present in a visual search display. Although Moher (2020) notes that the SSM literature is unable to account for the distractor-quitting threshold effect entirely, there are some similarities. For instance, one proposed account for SSMs is that processing the first target in a multitarget display lowers working memory resources available for searching for and processing a second target, resulting in missed targets (Cain & Mitroff, 2013). Similarly, it is possible that when an observer first processes a salient distractor during target detection visual search, it may consume available cognitive resources and result in an observer “missing” the target item in the array (Moher, 2020).
Aside from these potential explanations of the data, it is important to note that Moher found the magnitude of the distractor effect to vary with overall target prevalence. Specifically, when target prevalence was lowered from 50% in Experiment 1 to 20% in Experiment 2, on target-absent trials, the difference in reaction times for trials with or without distractors was smaller. Thus, the effect of salient distractors on quitting thresholds may vary across differing visual search contexts. Indeed, Moher (2020) himself notes that “there may be limits to the robustness of salient distractor effects” (p. 40).
As such, the current study aimed to replicate and expand upon Moher’s original finding by exploring how the overall salience of the color distractor influences search speeds and error rates during target-absent and target-present visual search. Specifically, in Moher (2020), the salient distractor varied on three dimensions compared with the other array items (color, size, and onset), making it highly dissimilar to the other items in the array. Here, we wanted to see whether a distractor with lower salience that differed on only two dimensions to other array items (color and onset) would similarly affect search strategies and quitting thresholds. Indeed, it is possible that overall salience could influence the degree of attentional capture and thus have a different impact on quitting thresholds. Furthermore, the degree of capture from a lower salience distractor during visual search may depend on the size or scope of the focus of attention, which is determined by the perceptual load of a search array. Specifically, a high load task is thought to result in serial visual search, subsequently narrowing attentional focus and lowering capture from distracting items (Theeuwes, Kramer, & Belopolsky, 2004a). Therefore, if attentional capture is more likely to occur when distractors fall within the focus of attention, a lower salience distractor may not influence quitting thresholds compared with a higher salience distractor.
In Experiment 1, we used a salient distractor that was dissimilar to the other items in the array on three dimensions (onset, color and size; thus, conceptually identical to that used in Moher, 2020). In Experiment 2, the salient distractor was only different from other array items on two dimensions (color and onset). If the distractor’s overall salience (manipulated here using size changes) does not matter for the quitting threshold effect, the pattern of results obtained in Experiments 1 and 2 will be identical. That is, in both experiments, distractors should reduce search speeds for target-absent trials and increase error rates for target-present trials. However, if salience does matter, the results of Experiments 1 and 2 might differ.
Experiment 1
In Experiment 1, we aimed to replicate Moher (2020). On each trial, participants completed a visual search for a vertical blue target line in an array of tilted lines. However, on 50% of the trials, a delayed onset color distractor that was far larger than the other objects in the search array was presented. If salient distractors do lower quitting thresholds in target-detection visual search, we should see a similar pattern of results to Moher (2020). That is, the salient distractor should increase error rates on target-present trials and decrease reaction times on target-absent trials.
Method
Participants
For Experiment 1, we aimed to collect enough data to have approximately 200 complete data sets for analyses. Participants were recruited via Mechanical Turk, provided informed consent, and received $2.50 for their participation. Although 366 complete data sets were collected for the current study, demographic screening revealed that there were four likely cases where a single individual had completed the computer task twice (eight data sets total). As such, these cases were removed from further analyses. The mean age of the remaining participants was 35.22 years (SD = 9.93 years). Seventy-five reported being female, 111 reported being male, and 12 did not report their gender. One hundred and seventy-nine reported being right-handed, 11 reported being left-handed, and three reported being ambidextrous. Five participants did not report their handedness, and five did not report their visual status. Two participants reported having vision problems, and the remaining participants reported having normal or corrected-to-normal vision.
Stimuli and procedure
Participants completed a demographic survey hosted on LimeSurvey, followed by a computer task created using the PsychoPy builder interface (Peirce et al., 2019; hosted at www.pavlovia.org). As each participant completed the study using their personal computer, here, we report the general stimulus specifications used (Peirce & MacAskill, 2018). All stimuli were presented on a white background.
Similar to Fig. 1, participants searched for the presence or absence of a blue vertical rectangle in an array of titled blue rectangles presented on a white background. The blue rectangles subtended 40 pixels by 8 pixels in size, and the distractors were randomly tilted 30 degrees to the left or right off of vertical. The target was presented on 50% of trials, and participants had to indicate on each trial if it was present or absent by pressing “z” and “m” on the keyboard, respectively. The search array remained on the screen until the participant responded.
Two different-sized search arrays were used—one with four items and one with eight items (50% of trials each). For the four-item display, items randomly appeared within a grid of 400 pixels × 400 pixels where grid lines were spaced 50 pixels apart. For the eight-item display, items randomly appeared within a grid of 500 pixels × 500 pixels. No two items could appear in the same location, and no two items overlapped one another.
Critically, the search was completed in the presence or absence of a salient red color distractor that was much larger than the other array items, had a delayed onset and was present on 50% of trials (80 pixels × 16 pixels in size). Specifically, on trials where the color distractor was present, it replaced one of the titled blue rectangles and was presented 100 ms after the search array’s onset. A delayed onset was used to maximize salience (Moher, 2020).
Overall, participants completed 218 trials. There were 10 practice trials where participants received corrective feedback, and then two blocks of 104 experimental trials where no feedback was given. All conditions were equally likely to occur. Participants were offered a break at the end of each block of trials, and the experiment took 20 minutes to complete.
Results
Error rates
Firstly, error rates across our experimental conditions were examined. Participants’ data were excluded from analyses if they made search errors on more than 40% of trials or more than 10% of responses were errors in any one condition. Applying these criteria left a final sample of 198 participants. Next, we ran a 2 × 2 × 2 repeated-measures analysis of variance (ANOVA), with the factors search array size, distractor presence, and target presence. There was a main effect of array size, F(1, 197) = 91.73, p < .001, ηp2 = .32, a main effect of target presence, F(1, 197) = 304.67, p < .001, ηp2 = .61, and a main effect of distractor presence, F(1, 197) = 8.63, p = .004, ηp2 = .04. The two-way interaction between array size and distractor presence was significant, F(1, 197) = 4.67, p = .032, ηp2 = .02, as was the two-way interaction between array size and target-presence, F(1, 197) = 55.69, p < .001, ηp2 = .22. The interaction between distractor presence and target presence was also significant, F(1, 197) = 22.07, p < .001, ηp2 = .10. Finally, the three-way interaction between array size, distractor-presence and target-presence was significant, F(1, 197) = 5.72, p = .018, ηp2 = .03. Descriptive data are shown in Table 1 and Fig. 2.
To further explore the interaction of interest between target presence and distractor presence, two separate ANOVAs on error rates for target-present and absent trials were conducted. For target absent trials, there was no main effect of search array size on error rates, F(1, 197) = 2.82, p = .133, ηp2 = .01, however, there was a main effect of distractor presence, F(1, 197) = 9.53, p = .002, ηp2 = .05, where participants made more errors on distractor absent trials (M = 3.81%, SE = 0.70%) compared with distractor present trials (M = 2.39%, SE = 0.39%). The interaction between distractor presence and search array size was nonsignificant, F(1, 197) = 0.11, p = .738, ηp2 < .01.
For target-present trials, there was a main effect of array size, F(1, 197) = 92.59, p < .001, ηp2 = .32, a main effect of distractor presence, F(1, 197) = 22.87, p < .001, ηp2 = .10, and an interaction between array size and distractor presence, F(1, 197) = 5.76, p = .017, ηp2 = .03. For array size four, error rates were lower for distractor absent trials (M = 10.86%, SE = 0.71%) compared with distractor-present trials (M = 15.25%, SE = 0.95%), t(197) = 5.46, p < .001. For array size eight, error rates were lower for distractor absent trials (M = 17.85%, SE = 0.94%) compared with distractor present trials (M = 19.74%, SE = 1.04%), t(197) = 2.16, p = .032.
Finally, we ran a Bayesian paired-samples t-test using JASP to test the effect of distractor presence on error rates for target-present trials, collapsed across search array sizes (JASP Team, 2020). The null hypothesis was that there was no effect of distractor presence on error rates, and the alternative hypothesis was that error rates would be higher for distractor-present compared with distractor-absent trials. The default prior in JASP was used. In contrast to Experiment 1, the analysis revealed strong evidence in favor of the alternative, BF10 >10. Thus, it appears that distractors increased error rates on target-present trials.
Reaction time data
Next, we examined the reaction time (RT) data for correct responses. At the individual level, RTs were trimmed if they were faster than 200ms or slower than 10 s. On average, this meant that 9.89% (SD = 8.24%) of trials were excluded per participant. Next, we ran a 2 × 2 × 2 repeated-measures ANOVA exploring the effects of array size, distractor presence, and target presence on the data. There was a main effect of array size, F(1, 197) = 771.01, p < .001, ηp2 = .80, a main effect of target-presence, F(1, 197) = 292.24, p < .001, ηp2 = .60, and a main effect of distractor presence, F(1, 197) = 35.70, p < .001, ηp2 =.15. The two-way interaction between array size and target presence was significant, F(1, 197) = 249.37, p < .001, ηp2 = .56, as was the interaction between target-presence and distractor-presence, F(1, 197) = 44.90, p < .001, np2 = .19. In contrast, the two-way interaction between array size and distractor-presence was nonsignificant, F(1, 197) = 0.04, p = .842, ηp2 < .01. Finally, the three-way interaction between array size, target-presence, and distractor-presence was nonsignificant, F(1, 197) = 0.08, p = .785, ηp2 < .01.
To further explore these findings, we ran two separate ANOVAs testing the effect of array size and distractor presence on RTs for target-absent trials and target-present trials. For target-absent trials, there was a main effect of array size on RT, F(1, 197) = 640.56, p < .001, ηp2 = .77, where RTs were faster for array size four (M = 1,377 ms, SE = 41 ms) compared with array size eight (M = 2,143 ms, SE = 60 ms). There was also a main effect of distractor presence, F(1, 197) = 88.18, p < .001, ηp2 = .31, where participants were slower for distractor absent (M = 1,811 ms, SE = 50 ms) compared with distractor present trials (M = 1,709 ms, SE = 48 ms), but the interaction between distractor presence and search array size was nonsignificant, F(1, 197) = 0.10, p = .751, ηp2 < .01.
For target present trials, a different pattern of results emerged. Firstly, there was a main effect of array size on RT, F(1, 197) = 624.71, p < .001, ηp2 = .76, where RTs were faster for array size four (M = 1,168 ms, SE = 33 ms), compared with array size eight (M = 1,551 ms, SE = 40 ms). However, there was no main effect of distractor presence, F(1, 197) = 0.29, p = .591, ηp2 < .01, nor an interaction between distractor presence and array size, F(1, 197) < 0.01, p = .974, ηp2 < .01. As shown in Table 2 and Fig. 3, unexpectedly, RTs for distractor present and absent trials were almost identical.
Finally, we conducted two Bayesian paired-samples t-tests using JASP (JASP Team, 2020). Firstly, we tested the effect of distractor-presence on RTs for target-absent trials, collapsed across search array sizes. The null hypothesis was that there was no effect of distractor presence on RTs, and the alternative hypothesis was that RTs would be faster for distractor-present compared with distractor-absent trials. The default prior in JASP was used. In contrast to Experiment 1, the analysis revealed strong evidence in favor of the alternative, BF10 >10. Thus, it appears that distractors speeded visual search on target-absent trials. Secondly, we tested the effect of distractor-presence on RTs for target-present trials collapsed across search array sizes (JASP Team, 2020). The null hypothesis was that there was no effect of distractor presence on RTs and the alternative hypothesis was that RTs would be slower for distractor-present compared with distractor-absent trials. The default prior in JASP was used. This analysis favoured the null, BF10 = .13.
Discussion
Experiment 1 aimed to conduct a close replication of Moher (2020) to explore the effect of distractor salience on quitting thresholds. Using a large, highly salient color distractor, we replicated Moher’s key finding of salient distractors speeding visual search on target-absent trials and increasing error rates on target-present trials. Together, these results suggest that highly salient distractors can trigger a search strategy shift, subsequently lowering quitting thresholds.
One unexpected aspect of the data is also worth noting; we did not obtain the typical finding of distractors slowing RTs during target-present visual search (as also found by Moher, 2020). Specifically, we expected to see a difference in attentional capture magnitude of around 60 ms. However, RTs for distractor present versus distractor absent trials were almost identical when a target was also present in the search array. Why did we not find an effect of distractor presence on target present visual search trials? Perhaps participants did not prioritize speed during this visual search task. While the instructions to participants were to respond as quickly and accurately as possible, the participants included in our final sample might have been trying very hard to ensure they responded accurately. Given that MTurk workers are sometimes “bonused” for performing a task well, this may have meant that participants were extra careful in ensuring correct responses (Paolacci & Chandler, 2014). Nonetheless, this seems unlikely for two reasons. First, if participants prioritized accuracy, we would still expect them to be somewhat slower for more challenging distractor-present trials. Second, returning to our error rate data, we can see that distractors increased error rates for target-present visual search.
Alternatively, it is possible that throughout the experiment, participants became somewhat familiar with the distractor and then were able to use this familiarity to effectively ignore it. For instance, Vatterott and Vecera (2012) suggest that although a salient distractor may initially capture attention during visual search, an individual may be able to suppress distractor effects with experience through repeated exposure to that distractor. Therefore, it is possible that in our experiment, the salient color distractor may have initially been very distracting, but later on in the experiment, effectively ignored. In turn, this experience may minimize the overall likelihood of observing slowing on target-present distractor trials when data were analyzed across the entire experiment. Nonetheless, given the small number of trials per participant in the current study (208), it is unfeasible to explore blocking effects reliably using the data from this experiment. However, such an exploration may be of interest to future researchers. In any case, what we can conclude is that distractors influenced processing, just not in the way we expected.
Experiment 2
Having replicated Moher’s main finding in Experiment 1 of increased error rates in target present trials and decreased RTs in target-absent trials, in Experiment 2, we wanted to test the limits of the effect by manipulating the overall salience of the distractor. In both Moher (2020) and Experiment 1 of the current study, the distractor varied on three dimensions (color, delayed onset, and size). Thus, to lower salience, we used a distractor that varied on only two dimensions (color and delayed onset). That is, the distractor was the same size as other items in the array. If the salience of the distractor matters, we might expect to see a different pattern of results in this experiment compared with prior work.
Method
Participants
In determining the experiment’s appropriate sample size, we used Anderson et al.’s (2017) sample size planner. We chose a power of .8 and assurance of .5 and found that to observe an effect similar in magnitude to that seen in Experiment 2 of Moher (2020), we would need a sample of 84 participants. Eighty-five participants completed the experiment and were recruited via Mechanical Turk. Their mean age was 31.27 years (SD = 9.85 years), and all but two had normal or corrected-to-normal vision. Forty-six were male, and 39 were female. Seventy-eight reported being right-handed, three reported being left-handed, and two reported being ambidextrous. All participants provided informed consent and were compensated $3 for their participation.
Stimuli and procedure
Participants completed a demographic survey hosted on Qualtrics, followed by a computer task created using the PsychoPy builder interface (Peirce et al., 2019; hosted at www.pavlovia.org). Similar to Experiment 1, as each participant completed the study using their personal computer, here, we report the general stimulus specifications used (Peirce & MacAskill, 2018).
As shown in Fig. 4, participants searched for the presence or absence of a black vertical line in an array of tilted lines on a white background. Furthermore, the search was done in the presence or absence of a salient red color distractor with a delayed onset that was the same size as the other display items. The search array contained four or eight items, and the location of each of the search items was randomly determined on a trial-by-trial basis. This randomization was achieved by sampling 1 of 81 possible locations without replacement on each trial. These locations were determined by drawing an imaginary 9 × 9 grid that had a length and width of 80% of the participant’s browsers total height (whether participants viewed stimuli in full screen or not was uncontrolled). Targets were drawn at locations where the grid lines intersected. The length of all lines in the array was 5% of the participant’s browser’s total height, and the distractor lines were tilted 30 degrees to the left or right off of vertical. When the salient color distractor was absent, all of the diagonal lines in the search array were black. However, when the salient color distractor was present, one of the tilted black lines was replaced with a tilted red line. This line was the same size as the other lines, but had a delayed onset of 100 ms. The search array remained on the screen until the participant responded. When the target was present, participants were required to press “z” on the keyboard. When the target was absent, participants pressed “m.”
Example of distractor absent (above) and distractor present (below) search task used in Experiment 2. Note. The target is the vertical line, and the salient color distractor is the tilted red line. Participants responded ‘target-present’ by pressing “z” and ‘target-absent’ by pressing “m.” (Color figure online)
In the computer task, participants first completed 16 practice trials that contained all possible trial types. During the practice, participants received corrective feedback. Following this, participants completed two experimental blocks containing 128 trials each. Across the experiment, the target was present on 50% of trials. Furthermore, the singleton distractor was present on 50% of trials. Finally, participants completed an equal number of search trials with an array size of four items and eight items. Overall, the study took around 20 minutes to complete, and participants were offered rest breaks at the end of each block.
Results
Error rates
Similar to Experiment 1, participants’ data were excluded from analyses if they made search errors on more than 40% of trials or more than 10% of responses were errors in any one condition. One participant was excluded for this reason. Next, we ran a 2 × 2 × 2 repeated- measures ANOVA exploring the effect of array size (four versus eight items), distractor presence (present versus absent), and target-presence (present versus absent) on error rates. There was a main effect of array size, F(1, 83) = 44.27, p < .001, ηp2 = .35, a main effect of target presence, F(1, 83) = 103.67, p < .001, ηp2 = .56, and an interaction effect between array size and target presence, F(1, 83) = 42.25, p < .001, ηp2 = .34. Interestingly, there was no main effect of distractor presence, F(1, 83) = 0.01, p =.917, ηp2 < .01, and the interaction between array size and distractor presence was nonsignificant, F(1, 83) = 1.51, p = .222, ηp2 = .02. The interaction between target presence and distractor presence was also nonsignificant, F(1, 83) = 0.34, p = .559, ηp2 < .01. Finally, the three-way interaction between array size, target presence and distractor presence was nonsignificant, F(1, 83) = 0.08, p = .779, ηp2 < .01. For descriptive data, refer to Fig. 5 and Table 3.
In contrast to Moher’s findings, our analyses suggested that the color distractor did not increase error rates on target present trials. Therefore, to further explore this, we ran an additional Bayesian paired-samples t-test exploring the effect of distractor presence on error rates for target-present trials, collapsed across search array sizes. The null hypothesis was that there was no effect of distractor presence on error rates, and the alternative hypothesis was that error rates would be higher for distractor present compared with distractor absent trials. The default prior in JASP was used. There was strong evidence in favor of the null, BF10 = .10.
Reaction times
Having examined the error rate data, we next turned to the RT data for trials where participants responded correctly. Firstly, RTs were screened at the individual level for responses that were faster than 200 ms, or slower than 10 seconds (Moher, 2020). On average, 5.94% (SD = 6.76%) of trials were excluded from further analysis. We then conducted a 2 (array size) × 2 (distractor presence) × 2 (target presence) repeated-measures ANOVA on RT. There was a main effect of array size, F(1, 83) = 322.62, p < .001, ηp2 = .80, a main effect of target presence, F(1, 83) = 154.01, p < .001, ηp2 = .65, and an interaction between array size and target presence, F(1, 83) = 169.51, p < .001, ηp2 = .67. There was also a main effect of distractor presence, F(1, 83) = 7.77, p = .007, ηp2 = .09, where RTs were slower for distractor-present trials compared with distractor-absent trials. However, there was no significant interaction between distractor presence and target presence, F(1, 83) = 1.53, p = .220, ηp2 = .02, nor between array size and distractor presence, F(1, 83) = 0.71, p = .402, ηp2 = .01. Finally, the three-way interaction between array size, distractor presence and target presence was nonsignificant, F(1, 83) = 0.05, p = .827, ηp2 < .01. For descriptive data, refer to Fig. 6 and Table 4.
As we did not replicate Moher’s finding of salient color distractors speeding target-absent visual search, we ran an additional Bayesian paired-samples t-test using to test to explore the effect of distractor-presence on RTs for target-absent trials, collapsed across search array sizes. The null hypothesis was that there was no effect of distractor presence on RTs and the alternative hypothesis was that RTs would be faster for distractor present compared with distractor absent trials. The default prior in JASP was used. The analysis revealed very strong evidence in favor of the null, BF10 = .03. Thus, it appears that distractors did not speed visual search on target absent trials.
Discussion
In Experiment 2, we aimed to extend upon the findings of Moher (2020) by lowering the overall salience of a singleton distractor during a target detection visual search task and exploring quitting thresholds. Specifically, unlike the Moher experiment, where salient distractors varied to other items on three dimensions (color, size, onset), the distractor in Experiment 2 varied on only two dimensions (color, onset). That is, the color distractor was smaller in size to that used in Moher’s study (as well as in Experiment 1 of the current study).
Importantly, using a lower-salience distractor, we observed a different pattern of results. Firstly, whereas Moher (2020) found that compared with distractor absent trials, distractor present trials speeded target absent visual search and increased error rates on target present visual search, here, we found that our salient distractor slowed search, regardless of whether a target was present or absent in the visual search array. Furthermore, error rates in our study were unaffected. This finding is broadly consistent with past research where salient singleton distractors slowed search speeds and had little to no error rate effects (Theeuwes, 1991, 1992).
Secondly, and consistent with our salience manipulation, the magnitude of the distractor effect was much smaller in Experiment 2 compared with Moher (2020). Indeed, the magnitude of the attentional capture effect in Experiment 1 in Moher (2020) for target present trials was approximately 62 ms. In contrast, the current experiment’s average capture effect across both target absent and present trials was only 19 ms. As such, these findings suggest that the effect of salient distractors on quitting thresholds depends upon the overall salience level of the distractor itself. That is, while a lower salience distractor may partially capture attention and slow search speeds, highly salient distractors might shift search strategy and lower quitting thresholds in target absent/present visual search. Such findings are consistent with the small literature exploring attentional capture during inefficient visual search, which shows small distractor interference effects when attention is narrowly distributed during a serial search task (Theeuwes, Kramer, & Belopolsky, 2004a).
General discussion
The current study aimed to test some of the conditions under which salient distractors alter quitting thresholds during target-present and target-absent visual search. To do so, we manipulated salience by varying the size of a delayed onset color distractor presented during a visual search task. In Experiment 1, we used a highly salient distractor that differed from other items in the search array on three dimensions (color, size, onset). In contrast, in Experiment 2, we used a salient distractor that varied on only two dimensions (color, onset). Overall, the highly salient distractor used in Experiment 1 triggered a shift in the participants’ search strategy. Specifically, the highly salient distractors speeded target-absent visual search and increased error rates during target-present visual search (i.e., misses of the target). This finding closely replicates Moher’s (2020) original study. In contrast, the lower-salience distractors used in Experiment 2 slowed the visual search for both target-absent and target-present trials, while not affecting error rates.
Taken together, the current study’s findings suggest that salient distractors alter quitting thresholds only when the distractor itself is very different from other items in the search array (e.g., varying on three rather than two feature dimensions). For example, imagine searching for a USB flash drive in a cluttered draw that also contained a large set of shiny house keys. Although the house keys are likely to capture attention initially, they are also likely to be quickly rejected as the target, as they are very different from the USB flash drive. In contrast, imagine searching for the USB flash drive in a cluttered draw that contained a small but shiny padlock key. Like the house keys, the shiny padlock key might also capture attention. However, because the padlock key is more similar in size to the USB flash drive (i.e., lower salience), it might be more difficult to reject, resulting in slower search speeds overall. In the current study, the larger red delayed onset distractor in Experiment 1 might have been much easier to reject than the distractor used in Experiment 2, causing the different pattern of results obtained across both experiments.
Alternatively, the results of the current study could be due to differences in item reinspection and rejection rates in serial visual search. Indeed, across both Experiment 1 and 2 of Moher (2020), and the two experiments of the current study, participants completed a challenging serial visual search task that presumably required eye movements to various items in the array (in Experiment 3 of Moher, larger stimuli were used that resulted in shallower search slopes and likely minimal eye movements). During these serial searches, a participant might have initially fixated on the salient color distractor and then disengaged from it, suppressing that location from further reinspection (Moran et al., 2013). However, it is also possible that the participant might eventually reinspect the original distractor item (given that the suppression of the distractor might decrease as search time increases). Returning to reinspect a highly salient distractor (such as that used in Experiment 1) could hasten a decision to quit search earlier than a distractor lower in salience (such as that used in Experiment 2). As such, future work may wish to explore the use of covert versus overt attentional search strategies in more detail. Although Moher’s Experiment 3 used larger stimuli in an attempt to minimize eye movements, it would be interesting to see if exactly how eye movements in the context of serial search contribute to the quitting threshold effect.
Finally, the singleton distractor’s size, rather than the overall salience of the distractor, might have a differential effect on quitting thresholds. In particular, this could be due to changes in the color distractor’s relative proximity to other items in the search array. Specifically, with a larger distractor (Experiment 1), more items in the search array are likely to appear close by it compared with a smaller distractor (Experiment 2). When the distractor captures one’s attention, the participant might be less likely to attend to any other items in the nearby vicinity, resulting in a speedier search for larger distractor items. As such, future research may wish to disentangle these possibilities by testing whether salient singleton distractors that differ in onset, shape, and color rather than onset, size, and color result in similar findings to those observed in the current study.
Change history
31 October 2023
A Correction to this paper has been published: https://doi.org/10.3758/s13414-023-02811-4
References
Anderson, S. F., Kelley, K., & Maxwell, S. E. (2017). Sample-size planning for more accurate statistical power: A method adjusting sample effect sizes for publication bias and uncertainty. Psychological Science, 28(11), 1547–1562. https://doi.org/10.1177/0956797617723724
Bacon, W. F., & Egeth, H. E. (1994). Overriding stimulus-driven attentional capture. Perception & Psychophysics, 55(5), 485–496. https://doi.org/10.3758/BF03205306
Blanca, M. J., Alarcón, R., Arnau, J., Bono, R., & Bendayan, R. (2017). Nonnormal data: Is ANOVA still a valid option? Psicothema, 29(4), 552–557.
Broadbent, D. E. (1982). Task combination and selective intake of information. Acta Psychologica, 50(3), 253–290. https://doi.org/10.1016/0001-6918(82)90043-9
Cain, M. S., Adamo, S. H., & Mitroff, S. R. (2013). A taxonomy of errors in multiple-target visual search. Visual Cognition, 21(7), 899–921. https://doi.org/10.1080/13506285.2013.843627
Cain, M. S., & Mitroff, S. R. (2013). Memory for found targets interferes with subsequent performance in multipletarget visual search. Journal of Experimental Psychology: Human Perception and Performance, 39(5), 1398–1408. https://doi.org/10.1037/a0030726
Carrasco, M. (2011). Visual attention: The past 25 years. Vision Research, 51(13), 1484–1525. https://doi.org/10.1016/j.visres.2011.04.012
Corbetta, M., & Shulman, G. L. (2002). Control of goal-directed and stimulus-driven attention in the brain. Nature Reviews Neuroscience, 3(3), 201–215. https://doi.org/10.1038/nrn755
Cousineau, D. (2005). Confidence intervals in within-subject designs: A simpler solution to Loftus and Masson’s method. Tutorials in Quantitative Methods for Psychology, 1(1), 42–45
Desimone, R., & Duncan, J. (1995). Neural mechanisms of selective visual attention. Annual Review of Neuroscience, 18(1), 193–222. https://doi.org/10.1146/annurev.ne.18.030195.001205
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(4), 1030–1044. https://doi.org/10.1037/0096-1523.18.4.1030
Geyer, T., Müller, H. J., & Krummenacher, J. (2008). Expectancies modulate attentional capture by salient color singletons. Vision Research, 48(11), 1315–1326. https://doi.org/10.1016/j.visres.2008.02.006
Hickey, C., McDonald, J. J., & Theeuwes, J. (2006). Electrophysiological evidence of the capture of visual attention. Journal of Cognitive Neuroscience, 18(4), 604–613. https://doi.org/10.1162/jocn.2006.18.4.604
JASP Team. (2020). JASP (Version 0.14.1) [Computer software]. https://jasp-stats.org
Jonides, J. (1981). Voluntary versus automatic control over the mind’s eye’s movement. In Long, J. B. & Baddeley, A. D., eds., Attention and Performance IX. Hillsdale, New Jersey: Lawrence Erlbaum Associates. 187–203
Lamy, D., & Egeth, H. E. (2003). Attentional capture in singleton-detection and feature-search modes. Journal of Experimental Psychology: Human Perception and Performance, 29(5), 1003–1020. https://doi.org/10.1037/0096-1523.29.5.1003
Moher, J. (2020). Distracting Objects Induce Early Quitting in Visual Search. Psychological Science, 31(1), 31–42. https://doi.org/10.1177/0956797619886809
Moran, R., Zehetleitner, M., Müller, H. J., & Usher, M. (2013). Competitive guided search: Meeting the challenge of benchmark RT distributions. Journal of Vision, 13(8), 24–24. https://doi.org/10.1167/13.8.24
Müller, H. J., Geyer, T., Zehetleitner, M., & Krummenacher, J. (2009). Attentional capture by salient color singleton distractors is modulated by top-down dimensional set. Journal of Experimental Psychology: Human Perception and Performance, 35(1), 1–16. https://doi.org/10.1037/0096-1523.35.1.1
Paolacci, G., & Chandler, J. (2014). Inside the Turk: Understanding Mechanical Turk as a participant pool. Current Directions in Psychological Science, 23(3), 184–188. https://doi.org/10.1177/0963721414531598
Peirce, J. W., & MacAskill, M. R. (2018). Building experiments in PsychoPy. SAGE.
Peirce, J., Gray, J. R., Simpson, S., MacAskill, M., Höchenberger, R., Sogo, H., Kastman, E., & Lindeløv, J. K. (2019). PsychoPy2: Experiments in behavior made easy. Behavior Research Methods, 51(1), 195–203. https://doi.org/10.3758/s13428-018-01193-y
Petersen, S. E., & Posner, M. I. (2012). The attention system of the human brain: 20 years after. Annual Review of Neuroscience, 35, 73-89. https://doi.org/10.1146/annurev-neuro-062111-150525
Posner, M. I. (1980). Orienting of Attention. Quarterly Journal of Experimental Psychology, 32(1), 3–25. https://doi.org/10.1080/00335558008248231
Sayim, B., Grubert, A., Herzog, M. H., & Krummenacher, J. (2010). Display probability modulates attentional capture by onset distractors. Journal of Vision, 10(3), 10–10. https://doi.org/10.1167/10.3.10
Schmider, E., Ziegler, M., Danay, E., Beyer, L., & Bühner, M. (2010). Is it really robust? Reinvestigating the robustness of ANOVA against violations of the normal distribution assumption. Methodology: European Journal of Research Methods for the Behavioral and Social Sciences, 6(4), 147–151. https://doi.org/10.1027/1614-2241/a000016
Theeuwes, J. (1991). Cross-dimensional perceptual selectivity. Perception & Psychophysics, 50(2), 184–193. https://doi.org/10.3758/BF03212219
Theeuwes, J. (1992). Perceptual selectivity for color and form. Perception & Psychophysics, 51(6), 599–606. https://doi.org/10.3758/BF03211656
Theeuwes, J. (2004). Top-down search strategies cannot override attentional capture. Psychonomic Bulletin & Review, 11(1), 65–70. https://doi.org/10.3758/BF03206462
Theeuwes, J., Kramer, A. F., & Belopolsky, A. V. (2004a). Attentional set interacts with perceptual load in visual search. Psychonomic Bulletin & Review, 11(4), 697-702. https://doi.org/10.3758/BF03196622
Theeuwes, J., Kramer, A. F., & Kingstone, A. (2004b). Attentional capture modulates perceptual sensitivity. Psychonomic Bulletin & Review, 11(3), 551–554. https://doi.org/10.3758/BF03196609
Vatterott, D. B., & Vecera, S. P. (2012). Experience-dependent attentional tuning of distractor rejection. Psychonomic Bulletin & Review, 19(5), 871–878. https://doi.org/10.3758/s13423-012-0280-4
Wolfe, J. M., & Van Wert, M. J. (2010). Varying target prevalence reveals two dissociable decision criteria in visual search. Current Biology, 20(2), 121–124. https://doi.org/10.1016/j.cub.2009.11.066
Yantis, S., & Egeth, H. E. (1999). On the distinction between visual salience and stimulus-driven attentional capture. Journal of Experimental Psychology: Human Perception and Performance, 25(3), 661–676. https://doi.org/10.1037/0096-1523.25.3.661
Acknowledgments
This work was supported by an NSERC Grant (2016-06359) awarded to Jay Pratt. We would like to thank Jeffery Moher and Hermann Müller for their valuable assistance and feedback during this project.
Author Note
1. We would like to thank Hermann Müller for this helpful suggestion. Although the CGS assumes that distractors are suppressed with 100% accuracy, it is certainly possible that one might not be completely able to suppress a distractor in a difficult search with highly salient items.
2. Here, we report demographics for the sample of participants who completed our demographic survey and the computer task (hosted on www.pavlovia.org). However, we do note that in both Experiments 1 and 2, some individuals completed only the demographic survey, and their data is not reported here. Further, although we aimed to have a sample of 84 participants for Experiment 2, data for one extra participant was erroneously collected (N = 85). The first 84 data sets were analyzed.
3. The normality assumption was violated across all experimental conditions in the current study for both error rate and RT measures. Nonetheless, ANOVA is thought to be robust to departures from this assumption (e.g., Blanca et al., 2017; Schmider et al., 2010).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Disclosure statement
The authors declare no conflicts of interest associated with the research and that the research meets ethical guidelines. Data associated with the current study is found at the following link: (https://osf.io/usp7r/?view_only=cd3115a3cfbb4c5a90d043ac5d721286). Study materials can be made available upon request. None of the experiments was preregistered.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The original online version of this article was revised: There was an error in Figure 2 in the originally published manuscript. The error has been corrected in the online version and the updated Figure is included here
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Lawrence, R.K., Pratt, J. Salience matters: Distractors may, or may not, speed target-absent searches. Atten Percept Psychophys 84, 89–100 (2022). https://doi.org/10.3758/s13414-021-02406-x
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.3758/s13414-021-02406-x