Dissociating Orienting Biases From Integration Effects With Eye Movements

Abstract

Despite decades of research, the conditions under which shifts of attention to prior target locations are facilitated or inhibited remain unknown. This ambiguity is a product of the popular feature discrimination task, in which attentional bias is commonly inferred from the efficiency by which a stimulus feature is discriminated after its location has been repeated or changed. Problematically, these tasks lead to integration effects; effects of target-location repetition appear to depend entirely on whether the target feature or response also repeats, allowing for several possible inferences about orienting bias. To parcel out integration effects and orienting biases, we designed the present experiments to require localized eye movements and manual discrimination responses to serially presented targets with randomly repeating locations. Eye movements revealed consistent biases away from prior target locations. Manual discrimination responses revealed integration effects. These data collectively revealed inhibited reorienting and integration effects, which resolve the ambiguity and reconcile episodic integration and attentional orienting accounts.

Keywords

visual attention episodic memory selective attention eye movements implicit memory

Under what conditions is a return shift of attention to a prior target location facilitated or inhibited? A major reason why this fundamental question remains unanswered is that the reaction time (RT) data typically taken as evidence of orienting bias in leading paradigms are affected strongly by perceptual processing demands. Thus, it is often extremely difficult to unambiguously distinguish attentional-orienting biases from later processes involved with determining target identity (Posner & Petersen, 1990) and the response to it (Hommel, 1998).

Imagine the simplest cases. In the common target–target detection and localization tasks, participants simply detect or localize two serially presented target stimuli by pressing a button. RTs are slower when target stimuli randomly repeat instead of switch locations, an effect that lasts for several seconds (e.g., Maylor & Hockey, 1985; Tanaka & Shimojo, 1996, 2000; Taylor & Donnelly, 2002; Welsh & Pratt, 2006). This is called inhibition of return. As implied by the name, inhibition of return is widely thought of as a bias against shifting attention to previously attended regions (Klein, 2000; Posner, Rafal, Choate, & Vaughan, 1985), which promotes efficient exploration of the visual environment (Wang & Klein, 2010).

The patterns become more complex in these same target–target paradigms when target identities are discriminated instead. In these cases, if the second target response repeats the first, RTs are faster, or facilitated, when target location repeats. In contrast, if the second target response is different than the first, RTs are slower, or inhibited, when target location repeats (e.g., Hilchey, Rajsic, Huffman, & Pratt, 2017a, 2017b; Hommel, 2005; Notebaert & Soetens, 2003; Rajsic, Bi, & Wilson, 2014; Terry, Valdes, & Neill, 1994). These opposing facilitatory and inhibitory effects are roughly equal, making it unclear whether shifts of attention are biased in favor of (Tanaka & Shimojo, 1996) or against (Taylor & Donnelly, 2002) prior target locations.

Indeed, implicit-episodic-retrieval frameworks (Wilson, Castel, & Pratt, 2006), such as the theory of event coding (TEC; Hommel, Musseler, Aschersleben, & Prinz, 2001; Hommel, Proctor, & Vu, 2004), provide plausible alternative accounts for the facilitatory and inhibitory patterns. These frameworks suggest that a target location and the response to the target bind together to form a common representation, called an event file, that is stored in implicit episodic memory. When a second target appears, a conflict occurs when the codes that are retrieved only partially match the current event, which delays responding (Hommel, 2004). Results in line with these frameworks are often called integration or binding effects, in principle because costs are incurred from integrating the target into the representation of the old event. Applied to popular feature discrimination tasks, the TEC asserts that responses are faster when both the prior target location and the prior response either repeat or switch relative to when only the prior target location or prior response repeats (i.e., integration effects or costs). This is precisely the pattern that occurs in these feature discrimination tasks. Whereas orienting biases cannot be ruled out by such findings, they are rendered entirely hypothetical because the results can be accounted for in full by the TEC (Hilchey et al., 2017a, 2017b).

As it stands, there is no unambiguous evidence of orienting biases when feature discrimination responses are made to targets. This fact is surprising not only because this research area is quite mature but also because orienting is often used to obtain a meaningful, actionable percept. In this study, we resolved this long-standing ambiguity by requiring both eye movement localization and manual feature discrimination responses to targets. Our rationale for this approach and its success hinged on several key notions: (a) Integration effects are a type of “prepared reflex” (Hommel, 2005, p. 1080) brought on by feature discrimination (Hilchey et al., 2017a, 2017b); (b) integration effects are confined mainly to the responses involved with making discrimination judgments, similar in principle to distractor–response integration effects (Moeller, Hommel, & Frings, 2015); and (c) integration effects can completely overshadow orienting biases (e.g., Klein, 2004).

With this approach, eye movements are functionally divorced from the arbitrary visuomotor transformations needed for the discrimination responses and are used only to orient. In principle, this approach makes the eye movements independent of the integration effects that obscure orienting biases. Consequently, eye movements should provide a relatively pure index of orienting bias, unlikely to be contaminated by integration effects. On the other hand, manual RTs should reveal roughly symmetrical benefits and costs of target-location repetition for manual response repeats and switches, respectively (i.e., integration effects). That is, on the basis of our previous work (Hilchey et al., 2017a, 2017b), we expected no overall manual RT benefit or cost from the repetition of target location. Thus, hypothetically, manual RTs should reveal integration effects with minimal evidence of orienting biases, whereas the saccadic RTs should reveal orienting effects with minimal evidence of integration effects.

Experiment 1

Eye movements sample target locations, and, when fixated, target identities (“×” or “+”) are discriminated with manual responses. After the eye movement and manual response, the eye is summoned back to its original position and remains there until the next target appears randomly at either the same or different location as the first. Effects of target-location repetition are computed in the standard way, as the mean RT difference between target-location repeats and switches. In some conditions, to help ensure that eye movements could reflect only orienting biases, we made target identities very difficult to see until fixated. In other conditions, we allowed targets to be easily discriminable in the visual periphery while still requiring eye movements to them for a trial to proceed.

Method

Participants

We recruited undergraduate students from the University of Toronto. The target sample size was 20, but 24 participated for course credit. Four of the first 20 participants failed to comply consistently with the eye movement instructions. These participants made in excess of five eye movements on more than 30% of all trials (32%, 39%, 42%, and 47%), a pass/fail criterion that was established prior to testing. Historically, a target sample size of 20 is more than adequate for detecting integration effects in manual RT data (e.g., Hilchey et al., 2017a, 2017b) and orientation biases in eye movement RT data, presuming inhibited reorienting (e.g., Hilchey, Klein, & Ivanoff, 2012; Hilchey, Klein, & Satel, 2014).

Apparatus and stimuli

Eye movements were monitored by an EyeLink 1000 Desktop Mount eye tracker (SR Research, Ottawa, Ontario, Canada) with a temporal resolution of 1000 Hz and spatial resolution of 0.01° of visual angle. The velocity and acceleration thresholds for eye movements were set to 35.0°/s and 9500.0°/s². Stimuli were displayed on an 18-in. Dell P992 CRT monitor (Dell Computer, Round Rock, TX) with a refresh rate of 85 Hz and 1,024 × 768 pixels resolution. The stimulus display was connected to a 2.8 GHz dual-core processor, and the eye tracker was connected to a 2.93 GHz dual-core processor. Participants’ head position was stabilized by a chin and head rest 57 cm from the monitor. Standard 9-point calibration and validation procedures were used until the average measurement error on gaze position was within half a degree of visual angle. Responses were made by pressing the space bar (thumb), “Y” key (right index finger), and “B” key (left index finger) on a standard QWERTY keyboard.

All stimuli were displayed against the black background (RGB: 0, 0, 0; luminance: 3 cd/m²) of the CRT monitor. Placeholders were gray outline boxes (RGB: 128, 128, 128; luminance: 31 cd/m²) subtending 2.0° × 2.0° of visual angle. One placeholder appeared at the center of the monitor and was flanked on the left and right by a placeholder 7.5° away, center to center. A white fixation cross (RGB: 255, 255, 255; luminance: 84 cd/m²) subtended 0.3° × 0.3° and appeared in the center placeholder. In the visible-target condition, target stimuli were white X (×) and plus (+) signs subtending 1.5°. In the invisible-target condition, target stimuli were small white outline circles (width = 2 pixels; radius = 0.24°) that circumscribed ×s and +s. Targets appeared in the center of placeholders. The main difference between conditions was that the × was very difficult to discriminate¹ from the + unless the participant made an eye movement to it in the invisible-target condition, though the identity was not strictly gaze contingent. The cue back (i.e., the stimulus that guided the eyes back to fixation) was a filled white circle (radius = 0.25°) in the middle of the center placeholder.

Procedure

See Figure 1 for an example trial sequence. Each trial began with the presentation of three placeholders and the fixation cross. To begin the trial, the participant performed a drift correction by staring at the fixation cross and pressing the space bar. A successful drift correction was signaled by a tone, and 0.5 s later, a target appeared randomly in the left or right placeholder. The participant made an eye movement to the target and, on arriving at the target, discriminated its identity by pressing “Y” or “B” for + or × targets, respectively. When the eye was within 3° of visual angle of target center and a key press was recorded, the target vanished. After this step, the participant returned his or her gaze to center, which was highlighted by the transformation of the fixation cross into a cue back. When this return eye movement was within 3° of visual angle of center, the cue back transformed into a fixation cross. The participant refrained from making eye movements until the appearance of the next target, which occurred 1 s later. This target appeared randomly in the left or right placeholder and was randomly either the same as or different from the first. The participant made an eye movement to this target and discriminated its form with a key press. When the eye was within 3° of visual angle of a target and a key press was recorded, all stimuli vanished from the screen, ending the trial. If a key-press error occurred at any point during the trial, an error message appeared at the end of the trial with the stimulus/key-press mappings. If more than five eye movements were detected on any given trial, that trial was spontaneously aborted, all stimuli disappeared, and the message, “You have made too many eye movements” was displayed. The participant acknowledged these messages by pressing the space bar. Between every trial, there was a 1-s interval during which no stimuli were on screen.

Fig. 1.

Example trial sequence from Experiments 1 and 2. The target was always a + (shown here) or an ×; participants indicated which they saw by pressing the “Y” or “B” key, respectively. Their expected gaze locations as they looked toward the target are indicated here by a pair of eyes. In this case, the target location switches, with the participant’s eyes orienting to a new location, but the manual response (pressing “Y”) repeats from the first target to the next.

Prior to the experiment, participants were informed correctly that the locations and identities of successive targets were uncorrelated. They were further instructed that a maximum of five eye movements was permitted on each trial, that an eye movement was required to each stimulus, that a blink counted as one eye movement, and that only three eye movements were needed on each trial. They were further instructed that they could take breaks as necessary between trials, performing the drift correction at their leisure. Each participant first watched the experimenter complete about a dozen trials successfully. Then the experimenter watched the participant practice for 20 trials and answered any questions. Each participant then independently completed two consecutive blocks (80 experimental trials each) in each target-visibility condition (320 total trials). Either two visible- or invisible-target–target blocks were given first, and the order was counterbalanced across participants.

Results

We excluded 17.5% of the trials because more than five eye movements occurred. Seven trials were excluded for unreasonably long (> 5 s) target–target onset asynchronies. Of the remaining data, 8.0% of the total trials were removed because of key-press errors to either the first target (2.6%), the second target (4.7%), or both (0.7%); 3.7% were excluded because gaze was not within 3° of center fixation when a target appeared; 4.5% were excluded because the first eye movement following a target did not land within 3° of it; and 1.6% of trials were lost because a key press was made to a target before an eye movement.

Saccadic reaction times (SRTs) were computed as the time between the onset of a target and the initiation of the first eye movement to it. Trials (0.4%) with impossibly fast target saccades (< 80 ms) were excluded. We then computed z scores for SRTs for each participant for each condition (visible target and invisible target) to detect outliers (z scores > 3) to the targets. We excluded 1.2% and 1.4% of trials as outliers for the visible- and invisible-target conditions, respectively. Manual reaction times (MRTs) were computed as the temporal difference between the onset of a stimulus and the manual response. We computed z scores for MRTs for each participant and excluded 1.2% of trials as outliers (z scores > 3) in each condition.

SRTs

Mean SRTs to the first target were 171 ms and 187 ms for invisible and visible targets, respectively. Mean SRTs to the second target were analyzed with a 2 (target-location repetition: repeat or switch) × 2 (target-form repetition: repeat or switch) × 2 (condition: visible target or invisible target) repeated measures analysis of variance (ANOVA; see Fig. 2). Critically, there was an effect of target-location repetition, F(1, 19) = 63.88, p < .001, η_p² = .771, with slower mean SRTs for target-location repeats (196 ms) than for target-location switches (175 ms). There was also an effect of condition, F(1, 19) = 9.655, p = .006, η_p² = .337, with faster mean SRTs for invisible (172 ms) than visible (199 ms) targets. There was no effect of target-form repetition, F(1, 19) = 1.451, p = .243, η_p² = .071.

Fig. 2.

Mean saccadic reaction time (SRT; top row) and mean SRT difference between target-location repeats and switches (repeat – switch; bottom row) in Experiment 1. In both rows, results are shown as a function of target-form repetition and target-location repetition, separately for visible-target and invisible-target trials. For mean SRT differences, numbers greater than 0 indicate slower responding to the prior target location (inhibition), and numbers lower than 0 indicate faster responding to the prior target location (facilitation). All error bars show half Fisher’s least-significant differences computed from the mean square error term of the three-way interaction. Nonoverlapping error bars indicate significant simple effects.

Target-location repetition interacted with condition, F(1, 19) = 6.399, p = .020, η_p² = .252. Although repeating the target location led to slower SRTs in both conditions, this slowing was greater for visible—M = 27 ms, 95% confidence interval (CI) = [18, 36]—than for invisible targets—M = 14 ms, 95% CI = [7, 20] (see Fig. 2). Finally, there was an interaction between target-location repetition and target-form repetition, F(1, 19) = 4.540, p = .046, η_p² = .193. The main effect of slower SRTs when repeating the target location was greater when the target form switched (M = 24 ms, 95% CI = [18, 30]) than when it repeated (M = 17 ms, 95% CI = [10, 24]; see Fig. 2). No remaining interactions were reliable (F < 1). Ultimately, the data revealed inhibited reorienting, despite some weakening of the effect by less visible targets and target-form repetition.

MRTs

Mean MRTs to the first target were 877 ms and 672 ms for the invisible and visible targets, respectively. Mean MRTs to the second target were analyzed with a 2 (target-location repetition: repeat or switch) × 2 (target-form repetition: repeat or switch) × 2 (condition: invisible target or visible target) repeated measures ANOVA (see Fig. 3). There was no effect of target-location repetition, F(1, 19) = 0.459, p = .506, η_p² = .024, and, thus, no unambiguous evidence of orienting bias. Target-form repetition was significant, F(1, 19) = 26.01, p < .001, η_p² = .578, with faster mean MRTs when the target form repeated (708 ms) instead of switched (732 ms). The effect of condition was significant, F(1, 19) = 137.7, p < .001, η_p² = .879, with faster mean MRTs for visible (626 ms) than invisible (813 ms) targets.

Fig. 3.

Mean manual reaction time (MRT; top row) and mean MRT difference between target-location repeats and switches (repeat – switch; bottom row) in Experiment 1. In both rows, results are shown as a function of target-form repetition and target-location repetition, separately for visible-target and invisible-target trials. For mean MRT differences, numbers greater than 0 indicate slower responding to the prior target location (inhibition), and numbers lower than 0 indicate faster responding to the prior target location (facilitation). All error bars show half Fisher’s least-significant differences computed from the mean square error term of the three-way interaction. Nonoverlapping error bars indicate significant simple effects.

Critically, there was an interaction between target-location repetition and target-form repetition, F(1, 19) = 46.38, p < .001, η_p² = .709; this finding reveals that fully repeating or switching the prior target location and form led to the fastest MRTs (see Fig. 3), which are the standard integration effects. No other interactions were reliable (Fs < 1), and none of the MRT effects were undermined by speed/accuracy trade-offs.

Discussion

The SRT data decisively demonstrate that attentional orienting is biased against a prior target location, without much evidence of integration, despite some weakening of this orienting bias by less visible targets and target-form repetition. In contrast, the MRT data decisively demonstrate integration effects, with faster responding when both target location and form repeated or switched, without much evidence of orienting bias.

Experiment 2

Although there was compelling evidence for integration in the MRTs and inhibited reorienting in the SRTs of Experiment 1, the data loss, in combination with a couple of small and unanticipated effects in the SRT data, compelled us to firm up and extend these findings. The main difference in Experiment 2 was that targets also randomly varied on a response-irrelevant dimension (color: red or green). In the MRT data, we expected this to reveal an additional integration effect, with the fastest MRTs occurring when the response-relevant form and response-irrelevant color both repeated or switched (e.g., Hommel & Colzato, 2004). As for the eye movements, although there is little indication that integration effects should factor into the orienting biases, we wanted to give a salient dimension, such as color (Huffman, Al-Aidroos, & Pratt, 2017), an opportunity to do so, while we also reevaluated whether the relatively small effects of target visibility and form repetition on the orienting bias were reliable.

Method

Participants

We again recruited undergraduate students from the University of Toronto. The target sample size of 20 was chosen to match that of Experiment 1. However, 25 students participated for course credit because 5 of the first 20 participants failed to meet the inclusion criterion, making in excess of five eye movements on over 30% of all trials (31%, 33%, 34%, 42%, and 45%).

Apparatus and stimuli

The apparatus and stimuli were the same as those used in Experiment 1, except the × and + targets were randomly red (RGB: 225, 0, 0; luminance: 30 cd/m²) or green (RGB: 0, 125, 0; luminance: 30 cd/m²).

Procedure

The procedure was the same as in Experiment 1 with two exceptions. In the second experiment, the participants were explicitly instructed to ignore the target color because it was irrelevant and uncorrelated with target locations and forms. In addition, the trial count for each block was increased to 128 (256 across both blocks) to accommodate target-color repetition as a factor.

Results

We excluded 15.9% of their trials because more than five eye movements occurred. Three trials were excluded for unreasonably long (> 5 s) target–target onset asynchronies. Of the remaining data, 6.8% of the total trials were removed because of key-press errors to either the first target (1.9%), second target (4.4%), or both (0.5%); 3.2% were excluded because gaze was not within 3° of center when a target appeared; 3.9% were excluded because the first eye movement following a target did not land within 3° of it; and 1.7% of trials were lost because a key press was made to a target before an eye movement.

Trials (0.4%) with impossibly fast target saccades (< 80 ms) were excluded. We then computed z scores for SRTs for each participant for each condition (visible target and invisible target) to detect outliers (z scores > 3) to the targets, and we excluded 1.4% and 1.3% of trials as outliers for the visible-target and invisible-target conditions, respectively. Then, z scores for MRTs were computed for each participant. We excluded 1.1% and 1.0% of trials as outliers (z scores > 3) in the visible-target and invisible-target conditions, respectively.

SRTs

Mean SRTs to the first target were 159 ms and 172 ms for the invisible and visible targets, respectively. Mean SRTs to the second target were analyzed with a 2 (target-location repetition: repeat or switch) × 2 (target-form repetition: repeat or switch) × 2 (target-color repetition: repeat or switch) × 2 (condition: visible target or invisible target) repeated measures ANOVA (see Fig. 4). As we previously found, there was an effect of target-location repetition, F(1, 19) = 18.01, p < .001, η_p² = .487, with slower mean SRTs for target-location repeats (175 ms) than for target-location switches (157 ms). There was a marginal effect of condition, F(1, 19) = 3.807, p = .066, η_p² = .167, with faster mean SRTs for invisible (158 ms) than visible (173 ms) targets. Neither the effect of target-color repetition nor target-form repetition was significant (both Fs < 1).

Fig. 4.

Mean saccadic reaction time (SRT; top rows) and mean SRT difference between target-location repeats and switches (repeat – switch; bottom rows) in Experiment 2 for color repeats and color switches. In all rows, results are shown as a function of target-form repetition and target-location repetition, separately for visible-target and invisible-target trials. For mean SRT differences, numbers greater than 0 indicate slower responding to the prior target location (inhibition), and numbers lower than 0 indicate faster responding to the prior target location (facilitation). All error bars show half Fisher’s least-significant differences computed from the mean square error term of the four-way interaction. Nonoverlapping error bars indicate significant simple effects.

There were two marginally significant interactions. One of these concerned the relationship between target-location repetition and target-form repetition, F(1, 19) = 3.51, p = .077, η_p² = .156. Essentially, this finding trended in the opposite direction of what was observed in Experiment 1, such that the main effect of slower SRTs when repeating the target location was greater when the target form repeated (M = 20 ms, 95% CI = [10, 29]) instead of switched (M = 16 ms, 95% CI = 8, 25]; see Fig. 4). This inconsistency between experimental results, coupled with the relatively small effect sizes, suggests that this interaction may be spurious. Further, the three-way interaction among target-form repetition, target-color repetition, and condition was marginally significant, F(1, 19) = 4.278, p = .0525, η_p² = .184. In the invisible-target condition only, SRTs tended to be about 4 ms faster when both the prior color and form repeated or switched relative to when one feature remained the same and the other switched. No other interactions looked even remotely plausible (all ps > .10). The important point is that the data again revealed inhibited reorienting across all conditions.

MRTs

Mean MRTs to the first target were 803 ms and 617 ms for invisible and visible targets, respectively. Mean MRTs to the second target were analyzed with a 2 (target-location repetition: repeat or switch) × 2 (target-form repetition: repeat or switch) × 2 (target-color repetition: repeat or switch) × 2 (condition: visible target or invisible target) repeated measures ANOVA (see Fig. 5). There were main effects of target-form repetition, F(1, 19) = 7.051, p = .016, η_p² = .271; target-color repetition, F(1, 19) = 6.777, p = .018, η_p² = .263; and condition, F(1, 19) = 157.6, p < .001, η_p² = .892. Mean MRTs were faster when target form repeated (657 ms) instead of switched (669 ms). Mean MRTs were also faster when target color repeated (660 ms) instead of switched (666 ms). Mean MRTs were faster for visible (574 ms) than invisible (753 ms) targets. There was a marginal effect of target-location repetition, F(1, 19) = 3.884, p = .064, η_p² = .170, with generally slower mean MRTs when target location repeated (667 ms) instead of switched (659 ms).

Fig. 5.

Mean manual reaction time (MRT; top rows) and mean MRT difference between target-location repeats and switches (repeat – switch; bottom rows) in Experiment 2 for color repeats and color switches. In all rows, results are shown as a function of target-form repetition and target-location repetition, separately for visible-target and invisible-target trials. For mean MRT differences, numbers greater than 0 indicate slower responding to the prior target location (inhibition), and numbers lower than 0 indicate faster responding to the prior target location (facilitation). All error bars show half Fisher’s least-significant differences computed from the mean square error term of the four-way interaction. Nonoverlapping error bars indicate significant simple effects.

There were three reliable interactions. Target-location repetition interacted with target-form repetition, F(1, 19) = 41.96, p < .001, η_p² = .688, which, as before, revealed that MRTs were relatively fast when both target location and form repeated or switched compared with when only target location or form repeated (see Fig. 5). Target-form repetition interacted with target-color repetition, F(1, 19) = 47.69, p < .001, η_p² = .715, which revealed that MRTs were relatively fast when both the prior form and color repeated or switched compared with when only the color or form repeated. There was also a three-way interaction among target-location repetition, target-color repetition, and target-form repetition, F(1, 19) = 8.816, p = .008, η_p² = .317. Generally stated, the relative RT advantages for full binary (e.g., location and form, or color and form) repetitions and switches were strongest when a third dimension (e.g., color or location, respectively) also repeated. These are standard integration effects.

Finally, there was one marginal interaction between target-color repetition and condition, F(1, 19) = 4.231, p = .054, η_p² = .182, such that color repetition led to generally faster mean RTs in the invisible-target condition (12 ms) than in the visible-target condition (1 ms), but if present, this effect was very weak. No other relationships were significant (all ps > .10), and no MRT effects were undermined by speed/accuracy trade-offs.

Discussion

The SRT data again revealed impressively stable orienting biases against the prior target location, largely unaltered by any other higher order interactions. Any effect of repeating a target feature on this orienting bias was weak, was unreliable, and in no way undermined inhibited reorienting. In contrast, the MRT data showed standard integration effects between form and color and between form and location, which were amplified by repetition of a third dimension (see also, e.g., Hommel & Colzato, 2004). There was only weak to nonexistent evidence of orienting bias in the MRTs.

General Discussion

These data unambiguously establish inhibited reorienting in feature discrimination tasks, while also establishing that integration effects, as revealed by manual discrimination responses, can completely obscure this inhibition. The data also establish an important boundary on integration effects by revealing that they are confined mainly to responses involved in actualizing perceptual discrimination judgments. Our findings largely resolve ambiguity between episodic-retrieval and attentional-orienting frameworks in accounting for data from feature discrimination tasks by dissociating them within the same experiments, thereby validating both accounts.

Historically, because responding is slower to a prior target location only if a different response is made, there has been little unequivocal support for orienting biases in target discrimination tasks (Taylor & Donnelly, 2002; Terry et al., 1994). This has led to justifiably stronger emphases on the role of episodic retrieval and updating processes in accounting for the MRT data (e.g., Hommel et al., 2004; Lupianez, 2010). Simply put, the MRTs are particularly sensitive to processes related to making the perceptual judgment but much less so to orienting bias. As shown here, a straightforward behavioral solution for distinguishing between enduring attentional-orienting biases and integration effects is to decouple orienting from the processes of enacting a perceptual judgment. Variations on this general idea could be implemented in many attention paradigms for which there is ambiguity about the status of attentional-orienting biases and episodic retrieval processes (e.g., see D’Angelo, Thomson, Tipper, & Milliken, 2016; Frings, Schneider, & Fox, 2015, for reviews of spatial negative priming) or attentional-orienting biases and later selection effects (see Lamy & Kristjánsson, 2013, for a review of intertrial priming).

In accounting for effects of target-location repetition in the SRT data here, the only reliable processes are related to inhibition of return. Stimulus-elicited saccade priming or execution causes this effect, and the effect goes on to bias later orienting toward novel regions (Hilchey et al., 2014; Posner et al., 1985). For the MRT data, it is necessary to invoke processes related to implicit episodic retrieval and updating (Hommel, 2005, 2007). These processes are low-level heuristics that facilitate the reenactment of prior perceptual judgments when aspects (e.g., location) of an earlier target repeat and that facilitate novel responses when aspects of an earlier target switch. To account for the absence of inhibited reorienting in the MRT data, we have identified some possibilities. One is that the tendency to repeat the prior response when target location repeats is stronger than the tendency to alternate the prior response when target location switches. When inhibited reorienting is added to this tendency, it would result in the appearance of roughly symmetrical integration effects. A second possibility is that the response repetition and alternation tendencies for target-location repeats and switches are roughly equal, in which case another process would be necessary to offset inhibited reorienting. This third hypothetical process could be related to the effect of facilitatory target-location repetitions often reported in intertrial priming studies that require focused attention (i.e., the need to narrow in on and select the particular identity of a stimulus, once found; Yashar & Lamy, 2010). Stated simply, it may be easier to reengage focal attention to or extract information from a prior target location after reorienting to it has occurred.

Ultimately, our findings clearly dissociate orienting biases from integration effects by functionally divorcing oculomotor responding from manual feature judgments. Simply, there are unambiguous orienting biases against previously attended locations in feature discrimination tasks when integration effects are factored out. These data reveal that inhibited reorienting is a ubiquitous side effect of prior oculomotor orienting, which can be completely overshadowed by integration effects. At the same time, the methods used in the present experiments provide a powerful and relatively intuitive tool for dissociating attentional-orienting biases from later effects related to episodic retrieval, updating, and selection.

Footnotes

Action Editor

Edward S. Awh served as action editor for this article.

Author Contributions

M. D. Hilchey, J. Rajsic, and J. Pratt conceived the original experimental designs, and all authors provided constructive feedback on them. M. D. Hilchey conducted the experiments and analyzed the data. All authors provided critical revisions to the first draft and every version thereafter.

Declaration of Conflicting Interests

The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Notes

References

D’Angelo

M. C.

Thomson

D. R.

Tipper

S. P.

Milliken

(2016). Negative priming 1985 to 2015: A measure of inhibition, the emergence of alternative accounts, and the multiple process challenge. The Quarterly Journal of Experimental Psychology, 69, 1890–1909.

Frings

Schneider

K. K.

Fox

(2015). The negative priming paradigm: An update and implications for selective attention. Psychonomic Bulletin & Review, 22, 1577–1597.

Hilchey

M. D.

Klein

R. M.

Ivanoff

(2012). Perceptual and motor inhibition of return: Components or flavors? Attention, Perception, & Psychophysics, 74, 1416–1429.

Hilchey

M. D.

Klein

R. M.

Satel

(2014). Returning to “inhibition of return” by dissociating long-term oculomotor IOR from short-term sensory adaptation and other nonoculomotor “inhibitory” cueing effects. Journal of Experimental Psychology: Human Perception and Performance, 40, 1603–1616.

Hilchey

M. D.

Rajsic

Huffman

Pratt

(2017a). Intervening response events between identification targets do not always turn repetition benefits into repetition costs. Attention, Perception, & Psychophysics, 79, 807–819.

Hilchey

M. D.

Rajsic

Huffman

Pratt

(2017b). Response-mediated spatial priming despite perfectly valid target location cues and intervening response events. Visual Cognition. Advance online publication. doi:10.1080/13506285.2017.1349230

Hommel

(1998). Event files: Evidence for automatic integration of stimulus-response episodes. Visual Cognition, 5, 183–216.

Hommel

(2004). Event files: Feature binding in and across perception and action. Trends in Cognitive Sciences, 8, 494–500.

Hommel

(2005). How much attention does an event file need? Journal of Experimental Psychology: Human Perception and Performance, 31, 1067–1082.

10.

Hommel

(2007). Feature integration across perception and action: Event files affect response choice. Psychological Research, 71, 42–63.

11.

Hommel

Colzato

(2004). Visual attention and the temporal dynamics of feature integration. Visual Cognition, 11, 483–521.

12.

Hommel

Musseler

Aschersleben

Prinz

(2001). The theory of event coding (TEC): A framework for perception and action planning. Behavioral & Brain Sciences, 24, 849–937.

13.

Hommel

Proctor

R. W.

K. P. L.

(2004). A feature-integration account of sequential effects in the Simon task. Psychological Research, 68, 1–17.

14.

Huffman

Al-Aidroos

Pratt

(2017). Salience drives non-spatial feature repetition effects in cueing tasks. Attention, Perception, & Psychophysics, 79, 212–222.

15.

Klein

R. M.

(2000). Inhibition of return. Trends in Cognitive Sciences, 4, 138–147.

16.

Klein

R. M.

(2004). Orienting and inhibition of return. In Gazzaniga

M. S.

(Ed.), The cognitive neurosciences (3rd ed., pp. 545–559). Cambridge, MA: MIT Press.

17.

Lamy

D. F.

Kristjánsson

. (2013). Is goal-directed attentional guidance just intertrial priming? A review. Journal of Vision, 13(3), Article 14. doi:10.1167/13.3.14

18.

Lupianez

(2010). Inhibition of return. In Nobre

A. C.

Coull

J. T.

(Eds.), Attention and time (pp. 17–34). Oxford, England: Oxford University Press.

19.

Maylor

E. A.

Hockey

(1985). Inhibitory component of externally controlled covert orienting in visual space. Journal of Experimental Psychology: Human Perception and Performance, 11, 777–787.

20.

Moeller

Hommel

Frings

(2015). From hands to feet: Abstract response representations in distractor–response bindings. Acta Psychologica, 159, 69–75.

21.

Notebaert

Soetens

(2003). The influence of irrelevant stimulus changes on stimulus and response repetition effects. Acta Psychologica, 112, 143–156.

22.

Posner

M. I.

Petersen

S. E.

(1990). The attention system of the human brain. Annual Review of Neuroscience, 13, 25–42.

23.

Posner

M. I.

Rafal

R. D.

Choate

Vaughan

(1985). Inhibition of return: Neural basis and function. Cognitive Neuropsychology, 2, 211–228.

24.

Rajsic

Wilson

D. E.

(2014). Long-term facilitation of return: A response-retrieval effect. Psychonomic Bulletin & Review, 21, 418–424.

25.

Tanaka

Shimojo

(1996). Location vs feature: Reaction time reveals dissociation between two visual functions. Vision Research, 36, 2125–2140.

26.

Tanaka

Shimojo

(2000). Repetition priming reveals sustained facilitation and transient inhibition in reaction time. Journal of Experimental Psychology: Human Perception and Performance, 26, 1421–1435.

27.

Taylor

T. L.

Donnelly

M. P.

(2002). Inhibition of return for target discriminations: The effect of repeating discriminated and irrelevant stimulus dimensions. Attention, Perception, & Psychophysics, 64, 292–317.

28.

Terry

K. M.

Valdes

L. A.

Neill

W. T.

(1994). Does “inhibition of return” occur in discrimination tasks? Attention, Perception, & Psychophysics, 55, 279–286.

29.

Wang

Klein

R. M.

(2010). Searching for inhibition of return in visual search: A review. Vision Research, 50, 220–228.

30.

Welsh

T. N.

Pratt

(2006). Inhibition of return in cue–target and target–target tasks. Experimental Brain Research, 174, 167–175.

31.

Wilson

D. E.

Castel

A. D.

Pratt

(2006). Long-term inhibition of return for spatial locations: Evidence for a memory retrieval account. The Quarterly Journal of Experimental Psychology, 59, 2135–2147.

32.

Yashar

Lamy

(2010). Intertrial repetition affects perception: The role of focused attention. Journal of Vision, 10(14), Article 3. doi:10.1167/10.14.3