Intertrial priming of popout search on visual prior entry

Abstract

When the features of a visual target and of nontargets are repeated between successive search displays, responding to a subsequent target is faster than when the features of the target and the nontargets switch between trials. This intertrial priming effect can influence perceptual processes, postperceptual processes (e.g., episodic retrieval), or both. Previous studies have shown that repeating irrelevant visual features that do not define the target or the response can influence postperceptual processes. The present study examined whether intertrial priming by irrelevant features also influences perceptual processes. Subjects completed a temporal order judgment task that appeared within a popout visual search display containing a color singleton among nonsingletons, all of which served as placeholders for two probes. Intertrial priming by the placeholder colors shifted the psychometric function. Specifically, the probes appearing at the color singleton in the switch condition needed to appear earlier than the probes at the color singleton in the repeat condition to be perceived simultaneous with the probes on a nonsingleton. This suggests there was an influence of intertrial priming by the irrelevant colors on visual prior entry; hence, repeating irrelevant features between trials can influence perpetual processes.

Introduction

In visual search tasks, by repeating object properties between successive displays, visual search for upcoming targets can be facilitated (e.g., Chun & Jiang, 1998; Chun & Nakayama, 2000; Kristjánsson & Campana, 2010; Kristjánsson, Saevarsson, & Driver, 2013; Lamy, Yashar & Ruderman, 2010; Maljkovic & Nakayama, 1994, 1996, 2000; Meeter & Olivers, 2006; Müller, Heller, & Ziegler, 1995; Wolfe, Butcher, Lee, & Hyle, 2003). For example, Maljkovic and Nakayama (1994) found that during visual search for a feature singleton target (a red diamond among green diamonds or a green diamond among red diamonds), responding was faster when the target and distractor colors repeated between trials than when the target and distractor colors switched between trials. This priming of popout (PoP) or intertrial priming effect has been observed with color (Goolsby & Suzuki, 2001), orientation (Hillstrom, 2000), shape (Lamy, Carmel, Egeth, & Leber, 2006), and spatial position (Maljkovic & Nakayama, 1996). Hence, priming's effect on visual search is ubiquitous and has been influential to our understanding of visual processing. One issue is whether repeating task-relevant (target-defining) visual features and task-irrelevant visual features affect perceptual processing or later stages of processing such as responding or episodic retrieval. This study used a temporal order judgment (TOJ) task to examine whether repeating irrelevant features between displays influences perceptual processing. 

Maljkovic and Nakayama (1994, 1996, 2000) found that only features that were relevant to locating a target influenced visual search on subsequent trials. For example, when the target-defining dimension was color, as in Maljkovic and Nakayama (1996), repeating the target's color (relevant feature) influenced responding, but repeating the target's shape (irrelevant feature) did not. They also found that repeating the target's color, shape, and spatial position produced an additive effect on performance, suggesting those three properties (color, shape, position) were represented separately in memory. Because repeating those properties affected performance independently, Maljkovic and Nakayama (1994, 1996, 2000) concluded that intertrial priming facilitated attentional deployment to objects that shared features with previous targets. Thus, because attended objects are perceived before unattended objects a la Titchener's Law of Prior Entry (Shore & Spence, 2005; Spence & Parise, 2010; Titchener, 1908), intertrial priming likely affected perceptual processing. 

Huang, Holcombe, and Pashler (2004; see also Huang & Pashler, 2005; Thomson & Milliken, 2011, 2012, 2013) challenged this conclusion by demonstrating that repeating a relevant, target-defining feature interacted with repeating an irrelevant, nonspatial feature. Huang et al. (2004) varied visual targets along three dimensions (size, color, orientation) with one being the target-defining feature (size in Experiment 1), one being irrelevant (color), and one being the response feature (orientation). They found that repeating the relevant feature (size) interacted with repeating the response feature (orientation) and interacted with repeating the irrelevant feature (color). Huang et al. (2004; Huang & Pashler, 2005) concluded that because the effect of repeating the relevant feature interacted with repeating an irrelevant feature and interacted with repeating the response feature, intertrial priming affected postperceptual processes. That is, priming affected a later stage of processing, during which information from independent memory stores combined to influence performance (Neill, 1997). 

However, postperceptual explanations of intertrial priming have been challenged. For example, Becker (2008a, 2008b) observed how repeating a relevant, target-defining feature sped target fixation latencies and sped saccades away from fixation, which are physiological measures unrelated to postperceptual response processes. Becker (2008a) also observed a greater number of nontarget fixations when the target and nontarget features switched between trials, suggesting the previous target feature was prioritized on the following trial. Additionally, Lamy et al. (2010; Yashar & Lamy, 2011) found that the interaction between target feature repetition and response feature repetition, which was observed by Huang et al. (2004), required time to manifest. In Lamy et al.'s (2010) study, a color change to the display objects occurred 100 ms to 400 ms after the display's onset. For example, a display could initially contain a red target and green nontargets, but after 100 to 400 ms the target changed to blue and the nontargets changed to yellow. Target color repetition sped responding at all color-change delays, but repeating the response feature interacted with repeating the target color only when the colors on the current trial were presented for at least 200 ms. Because response feature repetition required time to influence performance, but target color repetition did not, Lamy et al. (2010) proposed a dual-stage account of intertrial priming, which states that priming influences both perceptual processes and postperceptual response processes. Indeed, in their review of the literature, Kristjánsson and Campana (2010) concluded no single process accounted for all intertrial priming effects on visual search. 

One issue in these studies is that the repeated feature was always target-defining; hence, it was always task-relevant. It may be that intertrial priming by relevant features affects both perceptual processes and postperceptual processes, but it remains to be seen whether priming by irrelevant features, as observed by Huang et al. (2004), affects only postperceptual processes or also affects perceptual processes. Another issue is that previous studies used indirect measures to assess priming's influence on perceptual prioritization (e.g., accuracy on masked displays, eye movement). A more direct measure of perceptual prioritization would be a TOJ task, in which observers judge the order of the appearance of two asynchronous probes, and examine whether intertrial priming influenced TOJs (Shore & Spence, 2005; Spence & Parise, 2010). 

For example, Theeuwes and Van der Burg (2013) used a TOJ task to examine whether a centrally-located prime influenced visual prior entry—the perceived order of appearance—of two peripheral probes. A colored circle prime (red or green circle) appeared at fixation and was followed by one red probe and one green probe, one to the left and one to the right of fixation, and subjects judged which probe (left or right) appeared first. Theeuwes and Van der Burg (2013) observed a shift in the psychometric function between the primed probe and unprimed probe conditions. Specifically, the probe that matched the prime's color (primed probe) was perceived to appear before the probe that did not match the prime's color (unprimed probe), suggesting an influence of feature priming on visual prior entry. 

To examine whether repeating irrelevant visual features between successive search displays influences perceptual prioritization, in the present study subjects viewed displays that contained four colored squares (green and red), one of which was a color singleton. Unlike other intertrial priming studies in which a repeated feature was relevant to locating the target, the squares served as placeholders for where two probes appeared as part of a TOJ task. Two probes were presented on two placeholders and subjects indicated which appeared first. Hence, the placeholder colors were task-irrelevant, because they did not define the target or its position and did not require a response. Between trials the placeholder colors repeated or switched and the point of subjective equality (PSE)—time needed to perceive both probes simultaneously—was calculated when the colors repeated and when the colors switched. If repeating the irrelevant colors influenced visual prior entry, the PSE should be shifted between the switch and repeat conditions. Specifically, for subjects to perceive the two probes simultaneously, probes on the unprimed color singleton (switch condition) would need to appear earlier than probes appearing on the primed color singleton (repeat condition). 

Methods

Subjects

Ten undergraduates from the University of Scranton participated (six females; one left handed; age 18 to 21 years, M = 19.40, SD = 0.97) and received credit toward a requirement in a psychology course. All reported normal or corrected-to-normal vision. The study was approved by the Psychology Departmental Review Board and was conducted according to guidelines established by the American Psychological Association and US Department of Health and Human Services regarding research with humans. Subjects signed a consent form before beginning the study. 

Apparatus

Each subject's color vision was screened using the Ishihara colorblindness test, which all subjects passed. The experiment was programmed and presented using E-Prime software (v2.0.10242, Psychology Software Tools, Pittsburgh, PA) on a Dell 755 computer (Dell Computers, Round Rock, TX) equipped with a Pentium Core 2 Duo processor (1.96 GB RAM, 2.33 GHz). Subjects sat 60 cm away from a Dell E178Fpv monitor (Dell Computers) (1024 × 768) that was adjusted so the center of the screen was at each subject's eye level. 

Stimuli

Displays (Figure 1) were presented on a gray background (20.98 cd/m²; RGB: 125, 125, 125; CIE_xy: 0.313, 0.329). A black fixation cross (0.165 cd/m²; RGB: 0, 0, 0) measuring 0.38° × 0.38° was presented throughout each trial. All displays except the fixation and mask displays contained red (20.44 cd/m²; RGB: 255, 0, 0; CIE_xy: 0.634, 0.323) and green (20.62 cd/m²; RGB: 10, 177, 31; CIE_xy: 0.287, 0.514) squares (1.62° × 1.62°) that served as placeholders for where two probes appeared in the TOJ task. The squares appeared 45°, 135°, 225°, and 315° around fixation and the distance between fixation and the center of each square was 4.58°. One square was a color singleton and the other three appeared in a different color. The probes were black Os (0.38° diameter) in Courier font, centered in two of the placeholders. One probe appeared to the left of fixation and the other to the right, but both probes appeared above or below fixation. The mask display consisted of two rows of three Xs (0.38° × 0.48°) where each placeholder appeared. 

Figure 1

 

Example trial sequence (top, Trial N-1) used in the TOJ task and examples of a repeat trial and switch trial (bottom, Trial N). Assignment of trials to the repeat and switch conditions was based on whether the placeholder colors in the current trial repeated or switched, respectively, from the preceding trial. Displays are not to scale.

Procedures

Subjects were individually tested in a soundproof room under normal illumination and were informed the study was on perception of order. Subjects were told that on each trial four squares would be presented as placeholders for two black probes, but the squares would not help determine which probe appeared first. They were informed one probe would appear to the right and the other to the left and their task was to decide which probe appeared first. 

Each trial began with a fixation display for 300 ms followed by a display containing the placeholders for 300 ms. The color of the singleton and nonsingletons was chosen randomly on each trial, but the number of trials with a red singleton and a green singleton was equal. The first probe was presented in a randomly chosen square for 17 ms, 34 ms, 68 ms, or 136 ms before the second probe was presented on the opposite side of fixation and on the same vertical level as the first probe. Both probes appeared together for 72 ms before being masked until the subject responded. Subjects pressed the “z” key for the left probe first and the “/” key for the right probe. Subjects completed a practice block of 64 trials followed by eight blocks of 128 trials each. 

The singleton color, singleton location, first probe location, and stimulus onset asynchrony (SOA) between probes were chosen randomly on each trial. Only trials on which a probe appeared at the color singleton placeholder were analyzed as trials when both probes appeared in nonsingleton placeholders would not address priming; thus, only 50% of a subject's responses were analyzed. Each trial was classified by whether the placeholder colors on the current trial (Trial N) were repeated or switched from the preceding trial (Trial N – 1) and SOA (−136 ms, −68 ms, −34 ms, −17 ms, 17 ms, 34 ms, 68 ms, 136 ms). Negative values indicate the probe on the color singleton placeholder (singleton probe) appeared first and positive SOAs indicate the probe on the nonsingleton placeholder (nonsingleton probe) appeared first. To avoid location cuing, trials were not included if the location of the singleton repeated between trials (<1% of remaining trials). 

Results

One subject's data were not included due to overall high accuracy and the inability to fit a psychometric function to the data. Data were analyzed separately for the repeat and switch conditions and are presented in Figure 2, which depicts the proportion of the times the singleton probe was judged as appearing first as a function of SOA between the singleton probe and the nonsingleton probe. To examine whether priming by the placeholder colors influenced TOJs, the PSE was estimated for the repeat and switch conditions for each subject. This was done by fitting the following two-parameter logistic function to each subject's data, which minimized the root mean square error (RMSE) using Microsoft Excel's Solver toolkit (Van der Burg et al., 2008; Theeuwes & Van der Burg, 2013):  

 

 

Figure 2

 

Mean proportion of singleton probe first responses as a function of SOA between the two probes for the repeat and switch conditions, averaged over subjects.

SOA is the delay between the probes (−136 ms, −68 ms, −34 ms, −17 ms, 17 ms, 34 ms, 68 ms, 136 ms), and PSE and b (slope) were estimated. A difference in PSEs between the switch and repeat conditions would indicate a priming effect. Importantly, a larger positive PSE in the switch condition would suggest the singleton probe needed to be presented earlier in the switch condition than repeat condition for both probes to be perceived simultaneously. As can be seen in Figure 2, the psychometric function in the switch condition was shifted to the right of the repeat condition. 

Separate one-sample t tests revealed the PSE in the switch condition (M = 23.31 ms, 95% CI = [19.60, 31.86]) was significantly greater than 0 ms, t(8) = 4.285, SE = 5.440, p = 0.003 (two-tailed), d = 1.428; whereas the PSE in the repeat condition, 9.08 ms [0.52, 17.63], was not, t(8) = 2.044, SE = 4.440, p = 0.075 (two-tailed), d = 0.681. Importantly, a paired-sample t test revealed the PSE in the switch condition was greater than in the repeat condition, t(8) = 3.837, SE = 11.127, p = 0.005 (two-tailed), d = 0.955. The shift in PSEs between the repeat and switch condition indicates that repeating the irrelevant placeholder colors primed the TOJs for the probes. 

The slope in the repeat condition (b = −0.020 [−0.024, −0.016]) and switch condition (b = −0.019 [−0.021, −0.015]) were not significantly different, t(8) = 0.678, SE = 0.005, p = 0.517 (two-tailed), d = 0.084, but both slopes differed from 0 [repeat: t(8) = −4.105, SE = 0.005, p = 0.003 (two-tailed), d = 1.368; switch: t(8) = −4.062, SE = 0.005, p = 0.004 (two-tailed), d = 1.354]. 

Discussion

This study demonstrated that probes appearing on an unprimed color singleton (switch condition) needed to be presented earlier than probes appearing on a primed color singleton (repeat condition) to be perceived simultaneously with a probe appearing on a nonsingleton. This suggests that repeating the placeholder colors between trials influenced perceptual prioritization of the placeholders on the following trial, leading to a prior entry effect for probes on the primed color singleton. Importantly, the repeated singleton color was irrelevant, because it did not define the target, did not indicate where the first probe appeared, and did not require a response. Previous studies showed that repetition of irrelevant features affected later, postperceptual processes (Huang et al., 2004; Huang & Pashler, 2005). Other studies showed that repeating the relevant, target-defining features sped perceptual processes (e.g., Becker, 2008a, 2008b; Bichot & Schall, 2002), and sped both perceptual processes and post-perceptual response processes (e.g., Lamy et al., 2010; Yashar & Lamy, 2011). The results of the present study add to the literature by showing how repetition of irrelevant features can indeed facilitate perceptual processing. 

The results of this study are similar to those of Theeuwes and Van der Burg (2013), who observed that a centrally-located prime induced a prior entry effect for probes that matched the prime's color. Both studies demonstrated that priming influenced perceptual prioritization, but the results of the present study extend Theeuwes and Van der Burg (2013). In their study, the prime served as a go/no-go signal to complete the TOJ task so the prime's color was task-relevant, which made the repeated color of one probe relevant. In contrast, in the present study the placeholder colors were task-irrelevant so the repeated singleton color was irrelevant. Thus, the present study showed that repeating irrelevant features can produce a prior entry effect. 

This study is also somewhat similar to studies showing that properties of backgrounds can influence attentional deployment. For example, West, Pratt, and Peterson (2013) presented asynchronous probes on a background that consisted of a perceptually convex area and a concave area. They observed a prior entry effect for probes appearing on the convex area, suggesting attention was biased to select the perceptually “nearer” area. Similarly, Lester, Hecht, and Vecera (2009) examined the influence of figure versus ground on attention by presenting asynchronous probes on a background that consisted of a figure (an object) and ground (background). Lester et al. (2009) observed a prior entry effect for probes appearing on the figure, suggesting figures are perceptually prioritized. 

The results of the present study are somewhat inconsistent with claims that active processing of objects is necessary for intertrial priming (Fecteau, 2007; Kristjánsson et al., 2013). Kristjánsson et al. (2013) cued observers to respond to a search display (actively view) or not (passively view), and observed priming only when the preceding display was actively processed. If active processing is a necessary requirement for intertrial priming, priming by the placeholders should not have occurred in the present study, because they did not have to be searched and did not require a response. There are at least three reasons for the difference in results between the present study and Kristjánsson et al. (2013). First, observers in Kristjánsson et al. (2013) may have suppressed the display objects in the passive viewing condition since they were told to ignore the display, which would not have occurred in the present study as subjects were not instructed to do anything with the placeholders. Second, in the present study, active viewing of the placeholders may have occurred, since they indicated where the probes would appear. Lastly, in the present study priming was assessed by a difference in PSEs between the repeat and switch conditions, whereas priming was assessed by a difference in RTs between the repeat and switch conditions in Kristjánsson et al. (2013). It may be that TOJ tasks are more sensitive to performance differences resulting from intertrial priming when displays are passively viewed. 

In short, the present study demonstrated that repeating irrelevant visual features can influence perceptual processing and lead to a prior entry effect for probes appearing at a location coinciding with a repeated color. Thus, intertrial priming by relevant features as well as irrelevant features can speed perceptual processing and postperceptual processing. 

Acknowledgments

Commercial relationships: none. 

Corresponding author: Bryan R. Burnham. 

Email: [e-mail]; [e-mail]. 

Address: Department of Psychology, University of Scranton, Scranton, PA, USA.