The attentional boost effect overcomes dual-task interference in choice-response tasks

Abstract

Dual-task interference often arises when people respond to an incoming stimulus according to an arbitrary rule, such as choosing between the gas pedal and the brake when driving. Severe interference from response selection yields a brief “Psychological Refractory Period,” during which a concurrent task is put on hold. Here, we show that response selection in one task does not always hamper the processing of a secondary task. Responding to a target may paradoxically enhance the processing of secondary tasks, even when the target requires complex response selection. In three experiments, participants encoded pictures of common objects to memory while simultaneously monitoring a rapid serial visual presentation (RSVP) of characters or colours. Some of the RSVP stimuli were targets, requiring participants to press one of the two buttons to report their identity; others were distractors that participants ignored. Despite the increased response selection demands on target trials, pictures encoded with the RSVP targets were better remembered than those encoded with the RSVP distractors. Contrary to previous reports and predictions from dual-task interference, the attentional boost from target detection overcomes increased interference from response selection.

Keywords

Attentional boost effect dual-task processing response selection target detection

Introduction

Daily activities such as driving often require us to monitor the visual environment for important “target” signals, such as stoplights. For years, research on attention has focused on cognitive processes that lead to the detection of the target (Wolfe, 2021). Yet, increasing evidence has shown that finding a target does more than terminating a task. Detecting and responding to a target facilitates the processing of concurrently presented background stimuli, resulting in a paradoxical dual-task enhancement. For example, in Swallow and Jiang (2010), participants encoded a stream of scenes into memory while simultaneously monitoring a rapid serial visual presentation (RSVP) of coloured squares, pressing a button for white squares and withholding responses to black squares. Although detecting a target often exerts greater attentional demands than rejecting a distractor (as shown in the “two-target cost,” Duncan, 1980, and the attentional blink, Raymond et al., 1992), scenes encoded concurrently with the RSVP targets were better remembered than scenes encoded with the RSVP distractors. Known as the “attentional boost effect” (ABE), this finding exemplifies the processing enhancement that results from target detection. Analogous effects have been reported in tests of perceptual learning, decision-making, visual working memory, source memory, and incidental memory (Lin et al., 2010; Makovski et al., 2011; Schonberg et al., 2014; Seitz & Watanabe, 2003; Yebra et al., 2019).

To date, few studies on the ABE have examined the nature of the RSVP task that resulted in this boost. Nearly all of the published studies have used a simple RSVP task that requires a Go/No-Go response. For example, participants may press the spacebar for a red square and make no response to a green square (Spataro et al., 2013). Using a detection task with low perceptual and response demands, studies have minimised competition between the RSVP and the background tasks. Consequently, the ABE sometimes manifests as an “absolute boost”—memory for images encoded with RSVP targets is better than memory for images encoded on their own (Lin et al., 2010; Prull, 2019; Spataro et al., 2013; Swallow & Jiang, 2014a). The ABE may also be a “relative boost” when the RSVP task causes sufficient interference, resulting in the target-paired images being better remembered than distractor-paired images, but not better than images encoded in a single-task baseline (Spataro et al., 2013; Swallow & Jiang, 2010).

Theories of the ABE, such as the dual-task interaction model (Swallow et al., 2022; Swallow & Jiang, 2013), have emphasised the need to consider both dual-task interference and target-induced enhancement. According to this theory, two concurrent tasks—RSVP and scene encoding—compete for processing resources. This competition occurs alongside an enhancement associated with target detection. Scene memory is typically worse in the dual-task condition than in a single-task condition. The degree of dual-task interference on scene encoding depends on the difficulty of the RSVP task—the more demanding the RSVP task is, the greater the dual-task interference. By incorporating both dual-task interference and target-induced boost, this model successfully accounts for a wide range of behavioural findings, including the relative boost frequently observed in the tests of background image memory.

Although the dual-task interaction model has received strong empirical support, the theory is underdeveloped with regard to how the RSVP task interacts with the target-induced boost. One possibility is that dual-task competition and target-induced boost exert independent effects. When the RSVP task becomes more demanding, memory for the background images declines relative to a single-task baseline, but memory for target-paired images may still receive a boost relative to distractor-paired images. Alternatively, an increase in task demand may affect how targets and distractors are processed, leading to an interaction between the RSVP task and the target-induced boost. This could happen, for instance, when the targets and distractors are so similar that reliable classification becomes difficult. As reviewed next, empirical studies on how task demands affect the ABE have revealed conflicting results.

On one hand, the ABE appears to be independent from increased perceptual demands. In Swallow and Jiang (2014b), participants searched for a target character defined by a combination of colour and shape (e.g., a red “2” among red “z”s and green “2”s) and made a Go response to the target. Despite the difficulty of finding the RSVP target, the experiment produced a robust ABE. In another experiment, participants searched for a target that could be either a red “2” or a green “2” among distractors that were red “z”s. One of the targets—the red “2”—was visually similar to the distractors, increasing the perceptual demand. The other target—the green “2”—was dissimilar to the distractors. Although different in perceptual load, the two types of targets produced comparable attentional boost for concurrently presented objects. Similar results were observed in Toh and Lee (2022), who asked participants to search for an RSVP target among simultaneously presented distractors, making a Go response on target-present trials. The target differed from the distractors in either a simple feature (colour) or a combination of features (line segments). Dual-task interference was greater in the conjunction search task than in the feature search task, as manifested by worse memory for background objects in the conjunction search condition. However, the attentional boost—better memory for objects paired with target-present than target-absent trials—was comparable between the two conditions. Thus, increasing the perceptual demand of the RSVP task impairs the encoding of background images, but the relative boost from target detection is preserved.

Increasing the response demand, however, appears to be detrimental to the attentional boost. In Experiment 5 of Swallow and Jiang (2010), participants encoded background scenes into memory while monitoring a stream of coloured squares. Most of the time the square was black, requiring no response. Sometimes, the square was red or green, requiring a response. Swallow and Jiang (2010) manipulated the response demand for the red or green squares. One group of participants pressed the spacebar for both the red and the green squares and made no response to the black squares (detection group). Another group of participants pressed the “r” key for red and the “g” key for green and made no response to the black squares (choice-response group). The ABE was found in the detection group: scenes paired with the RSVP targets (red or green squares) were better remembered than those paired with the RSVP distractors (black squares). However, this effect was eliminated in the group that performed the choice-response task.

Leclercq and Seitz (2012) provided further evidence that response demands may be detrimental to the ABE. In that study, participants encoded scenes into memory while monitoring a stream of black squares for an occasional arrow. The participant’s task was to press the left or the right arrow key to report the arrow’s direction and withhold responses to black squares. The same set of scenes was repeatedly used in the experiment, but a probe scene was presented every 16 trials for participants to decide whether it was in the preceding 16 presentations. Scenes presented with the arrow target were better remembered than scenes presented with the black squares, but the effect was modest, reaching statistical significance only when the target-paired scene was compared with the scenes preceding the target scene; there was no memory advantage relative to the scenes presented after the target scene. In another experiment, participants made a detection response to a white square presented among black squares. This experiment produced a large ABE that was significant relative to both the pre- and post- target scenes. Leclercq and Seitz (2012) attributed the reduced ABE in the arrow task to the arrow’s tendency to automatically draw spatial attention to the cued side of the screen. However, it is also possible that the two-choice response requirement for the arrow disrupted the ABE.

Not all choice-response tasks were disruptive. The ABE is preserved as long as the choice response is delayed. In a study on task-irrelevant perceptual learning, participants viewed a stream of black and white letters displayed against a moving field of dots. At the end of the stream, they reported the two white letters they saw in the stream. Participants acquired perceptual learning of the target-paired background motion, even though the letter task required identification and a choice response (Seitz & Watanabe, 2003). Other studies that delayed choice responses to the end of the stream likewise found target-induced enhancement of background images (Lin et al., 2010; Nishina et al., 2007). By delaying the response, these studies reduced response-related interference at the time when the background images are encoded.

What accounts for the different effects of perceptual demands and response demands on the ABE? The dual-task interaction model provided one explanation by considering the locus of the ABE. According to Swallow and Jiang (2014b), the attentional boost occurs after the successful classification of an RSVP stimulus as a target. Because perceptual demands are exerted before target classification, their effects are independent of the subsequent target-induced enhancement. In contrast, response selection occurs at or immediately after target classification, making it more likely that an increase in response demands would interact with the attentional boost. This explanation also fits with the larger dual-task interference literature, which has highlighted the engagement of a central processing bottleneck for the response selection (Pashler, 1994). Interference from response selection is often severe enough to cause a “Psychological Refractory Period,” a brief moment during which a secondary task must wait for the central bottleneck (for an alternative view, see Meyer & Kieras, 1997).

To date, the idea that response demands are uniquely detrimental to the ABE has gone unchallenged. Yet, both theoretical and empirical considerations raise doubts about its validity. At the conceptual level, it is unclear how response selection could have interacted with the attentional boost. Theories of the ABE have attributed the attentional boost to a transient temporal orienting response, which occurs when participants detect an RSVP target (Mulligan & Spataro, 2015; Swallow & Jiang, 2010, 2013). Target detection leads to the transient activation of neurons in the Locus Coeruleus (LC) and the release of norepinephrine (NE; Aston-Jones & Cohen, 2005; Swallow & Jiang, 2010, 2013; Yebra et al., 2019). Whether this transient activation interacts with response selection depends, in part, on the timing of response selection and the temporal orienting response. An overlap in processing time may result in an interaction between response selection and the ABE. However, if response selection happens after the transient temporal orienting, then response demands are unlikely to undo the effect of temporal orienting.

Theoretical considerations aside, empirical studies using the choice-response task are sparse and inconsistent. The ABE was eliminated in Swallow and Jiang (2010) but attenuated in Leclercq and Seitz (2012). The interpretation of the findings was also mixed. Whereas Swallow and Jiang (2010) emphasised the disruption from response selection, Leclercq and Seitz (2012) attributed the reduced ABE to perceptual interference from spatial orienting triggered by the arrow target. In addition, sample sizes in these previous studies were not large—16 participants in Swallow and Jiang’s (2010) choice-response task and 20 participants in Experiment 1a of Leclercq and Seitz (2012)—raising questions about their sensitivity in detecting an ABE.¹ Finally, neither study included a single-task baseline to help gauge the degree of interference from the RSVP task. Given the theoretical importance of understanding the role of response interference in the ABE, it is important to re-examine how choice-response tasks affect the ABE.

We conducted three experiments to examine the impact of choice-response tasks on the ABE. Participants encoded a series of objects into memory while simultaneously monitoring an RSVP stream of symbols (Experiments 1 and 3) or colours (Experiment 2). The RSVP stream contained targets that participants responded to, distractors that they withheld responses from, and blank (no-stimulus) baseline trials. In Experiment 1, participants made a choice response to identify the target (e.g., was the target a digit 1 or 2?). In Experiment 2, one group of the participants made a choice response to identify the target (e.g., was the square red or green?), whereas a second group made a detection response (e.g., was this a target colour?). Experiment 3 was a replication of Experiment 1 using a detection task. In all experiments, we tested participants’ memories for the background objects, asking them to identify both the basic-level category and specific exemplar (Sisk & Lee, 2022). By including the no-stimulus baseline trials, we could assess the degree of interference from the RSVP task. The sample sizes used in the study—48 in each condition—were about two to three times as large as those used in previous studies, providing greater statistical power. With this improved design, we investigated whether making a choice response to a target eliminates the ABE or whether the effect is preserved even under increased response demands.

Experiment 1

This experiment aimed to provide a conceptual replication of previous choice-response RSVP tasks while improving several features of the experimental design. Participants in this experiment viewed photographs of common objects presented at a pace of 1.2 s/object while monitoring an RSVP stream of characters (i.e., digits or symbols). The onset of the character was synchronised with the onset of the objects. The RSVP stimuli could be the digits 1 or 2, the symbols < or >, or nothing, with equal probability. Half of the participants responded to the digits by pressing either the “1” or the “2” key. The other half of the participants responded to the symbols by pressing either the “<” or the “>” key. They made no response on the other trials, including blank trials that served as a single-task baseline. After encoding 72 photographs into memory, participants completed a recognition memory test using a four-alternative forced-choice (4AFC) task (Figure 1). The four choices included an old object, a foil from the same basic-level category as the old object, and two exemplars from a new category not seen before. For example, suppose an object participant previously encoded was a green apple, the recognition task would include the green apple, a red apple, and two toothbrushes (the participant had not seen any toothbrushes).

Our study design improved upon previous choice-response RSVP tasks in several ways. First, we counterbalanced the designation of the RSVP targets and distractors. Targets for half of the participants (e.g., digits 1 or 2) were distractors for the other half of the participants. This differed from previous choice-response studies that always used the same stimuli as targets (e.g., red or green) and distractors (e.g., black). Together with a full counterbalancing of photographs paired with the targets, distractors, and blank trials, we eliminated systematic effects of stimulus memorability on recognition memory (Bainbridge, 2019). Second, in our design, the RSVP targets, RSVP distractors, and blank trials occurred with equal frequency. This differed from previous choice-response studies that presented occasional RSVP targets among frequent distractors. By equating the frequencies of the RSVP targets and distractors, our design bypassed extraneous effects associated with occasional targets, such as the distinctiveness effect, the memory isolation effect, and the oddball effect (Swallow & Jiang, 2010). Third, the inclusion of blank trials in the RSVP task provided a single-task baseline to gauge dual-task interference. Compared with the blank trials, the RSVP distractor trials required perceptual identification of the character, whereas the RSVP target trials required both perceptual identification and response selection and execution. Thus, the attentional demand from the RSVP task increased from blank trials, to RSVP distractor trials, to RSVP target trials. If dual-task interference is the only effect, we would expect the best memory for scenes presented on their own and the worst memory for scenes encoded with the RSVP targets. However, if target detection provides an additional enhancement, even in a choice-response task, then memory may be better for target-paired objects than distractor-paired objects.

Method

Participants

Participants were recruited from Prolific.co, an online crowdsourcing site for behavioural research, which we had previously used in ABE studies (Sisk & Lee, 2022; Toh & Lee, 2022). All participants were healthy adults between 18 and 45 years of age, fluent in English, and had normal visual acuity and normal colour vision.

Sample size determination

Previous studies using objects as background stimuli and a Go/No-Go RSVP task observed an effect size of approximately 0.70 in Cohen’s d (Toh & Lee, 2022). The effect size ranged from 0.76 to 1.26 in Cohen’s d when the perceptual demand of the RSVP task was high (Swallow & Jiang, 2014b). Assuming the smallest effect size, a sample size of 29 provides a power of .95 in detecting the effect in a two-tailed test with an alpha of .05. Because choice-response tasks may produce a smaller effect size and because full counterbalancing of stimuli required us to test in multiples of 24, we set 48 as the targeted sample size.

There were 20 females and 28 males in Experiment 1. Their mean age was 28.9 years (SD = 8.0). One additional participant received a duplicated participant number, resulting in their data being excluded from the analysis.² The study was approved by the University of Minnesota Institutional Review Board. Participants provided informed consent through Qualtrics, which then directed them to the experiment on Pavlovia.org. At the completion of the experiment, participants were redirected back to Prolific.co to receive payment.

Equipment

The experiment was coded using PsychoPy (Peirce et al., 2019), which read counterbalanced condition files pre-generated in MATLAB (www.mathworks.com). The experiment was converted into JavaScript for testing on Pavlovia.org. Testing was conducted on the participants’ own desktop or laptop computers.

Materials

We used photographs of common objects from Tim Brady’s database (https://bradylab.ucsd.edu/stimuli.html). The stimulus set contained 144 distinct categories of objects, each with two exemplars. Each object subtended 256 × 256 pixels. The dual-task encoding phase presented 72 categorically distinct objects. The memory recognition phase presented these 72 old objects, along with 72 foils from the same basic-level category, and 72 pairs of objects from new categories. Across the 48 participants, we fully counterbalanced old images, within-category foils, and between-category foils. The 72 objects for encoding were further divided into three groups, to be paired with the RSVP targets, the RSVP distractors, or blank. The assignments of the objects to these three conditions were fully counterbalanced across participants. We also counterbalanced images associated with the two possible types of targets (digits 1 or 2 and symbols < or >). Across the entire sample of 48 participants, there were no systematic differences in how images were assigned across conditions.

The RSVP task presented participants with the digit 1 or 2 or the symbol < or >, printed in black with a font size of 24, displayed inside a small white square (20 × 20 pixels).

Procedure

Participants completed 48 trials of practice, during which they were presented with 1, 2, <, or >, randomly with equal frequency, at the centre of the display. Half of the participants responded to the digits by pressing the corresponding keys and withheld responses to the symbols. The other half of the participants responded to the symbols by pressing the corresponding keys and withheld responses to the digits. The stimuli were presented at a fixed pace of 1.2 s/item, including a 500 ms stimulus presentation and a 700 ms blank period. Each incorrect response triggered a 1-s accuracy feedback that reminded participants how they should respond.

Next, participants completed the dual-task encoding phase for 144 trials. On each trial, an object was presented in the background for 1,000 ms, followed by a 200 ms blank period. Participants were asked to encode the objects for a later memory test and to monitor concurrently presented character. At the centre of the object was either a digit (1 or 2), a symbol (< or >), or nothing, with equal frequency. The character, when present onset at the same time as the object and was erased after 500 ms, leaving the object by itself for an additional 500 ms. The presentation pace was 1.2 s/item. Half of the participants monitored the digits and pressed the “1” key for 1 and the “2” key for 2. The other half of the participants monitored the symbols and pressed the “<” key for < and the “>” key for > . They made no response on blank trials without any character or on trials with a distractor character. Each incorrect response triggered a 1-s accuracy feedback that reminded participants how they should respond. The encoding phase contained 72 objects from different basic-level categories. After all of the objects were presented once, they were presented again in a different random order. The category of RSVP stimuli (digits, symbols, or nothing) remained the same for the two times that a specific object was encountered.

Finally, in the recognition test, participants completed 72 trials, each involving four options: an object they saw earlier, a foil from the same basic-level category as the old object, and two exemplars from a new category not presented before. Each image subtended 150 × 150 pixels. The four options were aligned horizontally in a random order. The numbers 1, 2, 3, and 4 were displayed below the objects from left to right. Participants were asked to press the number below the old object. Accuracy feedback (a green “Correct!” or a red “Incorrect!” for 300 ms plus a 300 ms blank) was provided after each response.

Data availability

De-identified data in an aggregated format and videos verifying script accuracy are available at the Open Science Framework (https://osf.io/fzg3d/).

Results

Encoding phase RSVP task

Mean accuracy on the RSVP task was 98.3% (SE = 0.2%). Across the three types of RSVP stimuli, participants made a correct choice response on 95.8% (SE = 0.7%) of the target trials and correctly withheld responses on 99.3% (SE = 0.2%) of the distractor trials and 99.96% (SE = 0.04%) of the blank trials. Paired-samples t-tests (critical alpha: .0167 adjusted for multiple comparisons) showed that accuracy was lower on target trials than on both the distractor trials, t(47) = 5.072, p < .001, and the blank trials, t(47) = 5.99, p < .001, reflecting the higher task demand on target trials. In addition, accuracy was lower on distractor trials than on blank trials, t(47) = 3.92, p < .001. Similar results were observed using nonparametric tests, which took into consideration the ceiling effect in some participants’ responses. Wilcoxon signed-rank test produced significant differences (ps < .001) in all pairwise comparisons.

Object memory

As shown in Figure 2 (left), the probability of choosing the old object varied across conditions. An analysis of variance (ANOVA) on encoding RSVP status (target, distractor, or blank) revealed a significant main effect, F(2, 94) = 6.77, p = .002, $η_{p}^{2} = . 126$ . We ran three planned contrasts using paired-samples t-tests, adjusting the critical alpha to .0167 for multiple comparisons. This analysis showed that target-paired objects were better remembered than distractor-paired objects, showing an ABE, t(47) = 2.77, p = .008, Cohen’s d = 0.399. In addition, distractor-paired objects were more poorly remembered than objects presented on their own, showing a dual-task cost, t(47) = 4.03, p < .001, Cohen’s d = 0.581. The ABE appeared to have overcome the dual-task interference from response selection, resulting in a lack of impairment for target-paired objects relative to the blank baseline, t(47) = 0.39, p = .701, Cohen’s d = 0.055. Bayesian analysis for null results showed a Bayes Factor of 8.233, providing evidence in favour of the null hypothesis (Dienes, 2014; Rouder & Morey, 2011). Thus, target detection produced a “relative boost,” with better memory for target-paired objects than distractor-paired objects, alongside dual-task interference.

In addition to overall accuracy, our design allowed us to separately examine visual category and exemplar memories. Following Sisk and Lee (2022), we defined category memory as correctly identifying the basic-level category of the old object. This was computed as the proportion of trials in which participants chose either the old object or the within-category foil, out of all trials. In contrast, exemplar memory was defined as correctly identifying the old exemplar given that a correct category had been chosen. This was computed as the proportion of trials in which participants chose the old object out of the trials in which either the old object or the within-category foil was chosen. As shown in Figure 2 (right), the results were qualitatively similar to the overall accuracy, though with weaker statistical significance. In category memory, the main effect of RSVP status missed significance, F(2, 94) = 2.21, p = .116, $η_{p}^{2} = . 045$ . Paired t-tests did not reach the critical alpha of .0167 for any of the comparisons, t(47) = 1.74, p = .088, Cohen’s d = 0.51 for target versus distractor, t(47) = 1.88, p = .067, Cohen’s d = 0.55 for distractor versus baseline. In exemplar memory, the main effect of RSVP status was significant, F(2, 94) = 5.87, p = .004, $η_{p}^{2} = . 111$ . Although the ABE narrowly missed statistical significance, t(47) = 2.362, p = .022, Cohen’s d = 0.69, the advantage for the baseline relative to the distractor condition was significant, t(47) = 4.093, p < .001, Cohen’s d = 1.19.

Discussion

Experiment 1 required participants to make a choice response to the RSVP targets and withhold responses to the RSVP distractors or blank trials. The task demand was greater on RSVP target trials relative to RSVP distractor trials, leading to worse RSVP task performance on target trials. Both conditions required perceptual discrimination, incurring a dual-task cost not present in the blank baseline. Using objects as background images for memory, we observed an attentional boost for target-paired objects relative to distractor-paired objects. This effect was significant when examining overall memory accuracy, a reflection of memory for both the basic-level category and the specific object exemplar. When these two components were separated, the results were qualitatively similar, but the ABE missed statistical significance. Our data also showed that the ABE occurred alongside dual-task interference. Distractor-paired objects were more poorly remembered than objects presented by themselves. This interference was expected given the additional task demands on distractor trials: participants had to identify the character in addition to remembering the background objects. The dual-task demand increased further on target trials which required not just perceptual identification but also response selection. The lack of a memory deficit on target trials, relative to the blank baseline, suggests that the ABE had overcome the cost of response selection.

Despite its moderate size, Experiment 1 demonstrates that choice responses do not abolish the ABE. This finding differs from Swallow and Jiang (2010), who found no ABE in a choice-response task involving red or green targets. Leclercq and Seitz (2012), however, observed a diminished attentional boost in a modified paradigm using arrows as the RSVP targets. In Experiment 1, the symbols < and > were visually similar to arrows, but the digits 1 and 2 were not. To test whether our findings differed between these two types of RSVP targets, we ran an ANOVA using target type (symbols or digits) as a between-subject factor and RSVP status (target, distractor, or blank) as a within-subject factor. This analysis found just a significant main effect of RSVP status, F(2, 92) = 6.645, p = .002, $η_{p}^{2} = . 126 .$ The main effect of target type was not significant, F(1, 46) = 0.443, p = .509, and neither was the interaction, F(2, 92) = 0.160, p = .852. Thus, the ABE in Experiment 1 was not restricted to targets that were visually similar to arrows.

Why did the ABE survive a choice-response task in Experiment 1, but not in Swallow and Jiang (2010)? One possibility is that the stimuli used in Experiment 1 could be more easily mapped onto responses than the colours used in Swallow and Jiang (2010). The digits “1” and “2” map onto a mental number line that runs from the left to the right (Dehaene et al., 1993), making it straightforward to map the stimulus to the response keys. Likewise, the symbols “<” and “>” imply left and right directionality, which can be easily mapped onto the response keys. In contrast, the red and green squares used in Swallow and Jiang (2010) do not have intrinsic spatial properties, making their correspondence to the “r” and “g” keys arbitrary. Because response selection is easier in ideomotor compatible tasks (Lien et al., 2002), the preservation of the ABE in Experiment 1 may be limited to tasks with intuitive response mappings. In the next experiment, we attempted to replicate Experiment 1 using colours in the RSVP task.

Experiment 2

The purpose of Experiment 2 was to test the generality of Experiment 1’s findings. To this end, we used an RSVP task that involved a more arbitrary response mapping. Participants viewed an RSVP stream of coloured squares that could be red, green, black, or white, along with blank trials. Half of the participants responded to the red and green squares by pressing the “r” or the “g” key; the other half of the participants responded to the black and white squares by pressing the “b” or the “w” key. They made no responses to the distractor colours or blank trials. The response mapping (e.g., “r” for red) relied on a learned association between colours and their English spelling. Nonetheless, unlike the digits and symbols used in Experiment 1, the colours do not readily map onto spatial locations, making the response mapping of Experiment 2 more arbitrary than that of Experiment 1. If choice responses involving more arbitrary mapping eliminate the ABE, then unlike Experiment 1, Experiment 2 should not produce a memory advance for objects paired with the RSVP targets relative to those paired with the RSVP distractors.

To further gauge the degree of dual-task interference, Experiment 2 included a second group of participants that made a Go/No-Go response. Similar to the choice-response group, this group of participants withheld responses to RSVP distractor and blank trials. The two groups differed in how they responded to the targets. The detection group pressed the spacebar for either of the two targets. The choice-response group pressed the corresponding keys to report the target’s colour (e.g., “r” for red and “g” for green). Of interest is whether the ABE differed qualitatively between the detection and choice-response tasks.

Method

Participants

Overall, 96 new participants from Prolific.co completed Experiment 2. The 48 participants in the choice-response group included 18 females and 30 males with a mean age of 25.5 years (SD = 6.7). The 48 participants in the detection group included 19 females and 29 males with a mean age of 29.1 years (SD = 7). Data from three additional participants were excluded because they received a duplicated participant number (see Note 2).

Equipment and materials

The experiment used the same object stimuli as in Experiment 1. The RSVP task used coloured squares (20 × 20 pixels) in one of the four colours: red, green, white, and black, enclosed inside a yellow border.

Design and procedure

The choice-response group was tested using a design similar to that of Experiment 1, except that the digits and symbols were replaced by coloured squares. Half of the participants were assigned red and green as targets, the other half white and black. As in Experiment 1, they made a choice response (e.g., “r” for red and “g” for green) to the targets and withheld responses on distractor and blank trials. The detection group was tested using the same stimuli, except that participants pressed the spacebar for either targets (e.g., spacebar for either red or green) and withheld responses on distractor and blank trials. Similar to Experiment 1, we counterbalanced the colours assigned as targets and distractors, and the background objects paired with the RSVP targets, distractors, and blank.

Following the encoding phase, participants completed the same 4AFC recognition memory test as that used in Experiment 1.

Results

Choice-response group

Encoding phase RSVP task

Mean accuracy on the two-choice RSVP task was 97.4% (SE = 0.3%). Participants correctly responded to the target colours 93.2% of the time (SE = 0.7%), which was significantly lower than their correct rejection rates of 99.3% (SE = 0.3%) on blank trials, t(47) = 8.06, p < .001, Cohen’s d = 1.163, and 99.6% (SE = 0.1%) on distractor trials, t(47) = 9.34, p < .001, Cohen’s d = 1.347. Correct rejection rate was comparable between blank trials and distractor trials, t(47) = 1.013, p = .316, Cohen’s d = 0.146, Bayes Factor = 5.381 in favour of the null hypothesis. The nonparametric Wilcoxon signed-rank test confirmed the significantly worse accuracy in the target condition, relative to both the baseline (p < .001) and the distractor (p < .001) conditions. The baseline and distractor conditions did not differ from each other (p = .252). Thus, there was an increase in task demand on RSVP target trials relative to RSVP distractor and blank trials.

Object memory

Accuracy in selecting the old objects varied as a function of encoding condition (Figure 3, Left), F(2, 94) = 14.277, p < .001, $η_{p}^{2} = . 233$ . Replicating Experiment 1, memory was better for objects paired with the RSVP targets than those paired with the RSVP distractors, t(47) = 3.536, p < .001, Cohen’s d = 0.510. Dual-task interference manifested as worse memory for objects paired with the RSVP distractors than for objects encoded on their own, t(47) = 5.968, p < .001, Cohen’s d = 0.861. The attentional boost on RSVP target trials partially overcame response-selection demands, producing comparable memory between the target and blank conditions, t(47) = 1.574, p = .122, Cohen’s d = 0.227, Bayes Factor = 2.705, showing weak evidence for the null hypothesis.

The pattern of results reported above held when we separated memory for the objects’ basic-level category and for the specific exemplar. These more detailed analyses can be found in the online Supplementary Material A.

Detection group

Encoding phase RSVP task

Mean accuracy in the detection task was 97.6% (SE = 0.3%), with errors occurring on target and distractor trials. Participants correctly responded to targets 97.6% (SE = 0.3%) of the time. They withheld responses on 98.9% of the blank trials (SE = 0.4%) and 96.1% of the distractor trials (SE = 0.5%). Paired-sample t-tests showed that the correct rejection rate was significantly lower on distractor trials than on blank trials, t(47) = 5.143, p < .001, Cohen’s d = 0.742. This difference was confirmed in the nonparametric Wilcoxon signed-rank test (p < .001), showing dual-task interference.

The pattern of accuracy in the detection task differed from that in the two-choice task. An ANOVA on RSVP status (target, distractor, and blank) and task (two-choice vs detection) revealed a significant interaction, F(2, 188) = 50.658, p < .001, $η_{p}^{2} = . 350$ . In the two-choice task, errors concentrated on target trials, with virtually no false responses on distractor and blank trials. In the detection task, errors included misses on target trials and false alarms on distractor trials. This latter pattern may reflect a tendency for participants to adopt an automatic response mode in detection tasks (Hawkins et al., 2019). In this mode, any stimulus presentation would trigger an automatic response, resulting in false alarms. Choice tasks are highly unlikely to trigger such a response, given the need to choose the response corresponding to the target colour.

Object memory

Memory accuracy varied as a function of encoding condition (Figure 3, right), F(2, 94) = 29.815, p < .001, $η_{p}^{2} = . 388$ . Target-paired objects were better remembered than distractor-paired objects, t(47) = 6.198, p < .001, Cohen’s d = 0.895. Objects encoded on their own were better remembered than distractor-paired objects, t(47) = 7.510, p < .001, Cohen’s d = 1.084, but not different from target-paired objects, t(47) = 0.039, p = .485, Cohen’s d = 0.006, Bayes factor = 8.851 in favour of the null hypothesis. This pattern of results held when we examined category memory and exemplar memory separately (see the online Supplementary Material A). Thus, target detection resulted in a pattern of “relative boost.”

To directly compare results across the two groups, we conducted an ANOVA using task (choice response vs detection) as a between-subject factor and square status (target, distractor, or blank) as a within-subject factor. The main effect of task was not significant, F(1, 94) = 2.554, p = .113, $η_{p}^{2} = . 026$ . The main effect of square status was significant, showing an ABE, F(2, 188) = 40.630, p < .001, $η_{p}^{2} = . 302$ , but this did not interact with task, F(2, 188) = 1.933, p = .148, $η_{p}^{2} = . 020$ . Thus, although the ABE was numerically greater in the detection task (M = 14.7%, SE = 2.4%) than in the choice-response task (M = 8.4%, SE = 2.4%) and had a larger effect size (Cohen’s d = 0.895 in the detection task and 0.510 in the choice-response task), the discrepancy was not large enough to reach statistical significance.

The ANOVA reported above included blank trials, which were not expected to differ between the choice-response and detection tasks. We conducted an additional ANOVA test that excluded the blank trials. This analysis used task (choice-response vs detection) as a between-subject factor and square status (target vs distractor) as a within-subject factor. It produced a significant main effect of square status, F(1, 94) = 47.293, p < .001, $η_{p}^{2} = . 335$ , showing an ABE. The main effect of task was not significant, F(1, 94) = 2.871, p = .093, $η_{p}^{2} = . 030$ . The interaction between square status and task was marginally significant, F(1, 940 = 3.465, p = .066, $η_{p}^{2} = . 036$ . Thus, although the ABE was associated with a larger effect size in the detection task than in the choice-response task, evidence for a task interaction was weak.

Discussion

Using colours as stimuli in the RSVP task, Experiment 2 replicated the findings of Experiment 1: objects paired with the RSVP targets were better remembered than objects paired with the RSVP distractors. Despite an increase in attentional demand for response selection and execution, target-paired objects received better memory than distractor-paired objects. Objects encoded on their own were better remembered than distractor-paired objects, demonstrating dual-task interference. Target detection and response overcame this interference.

Experiment 2 also showed that the ABE was numerically smaller when participants performed the choice-response task rather than the detection task. However, the interaction between task and square status only reached marginal statistical significance. In other words, the difference between the detection and choice-responses tasks was not qualitative. Both tasks yielded a “relative boost”—better memory in the target condition than the distractor condition, but comparable memory between the target condition and the single-task baseline. In both cases, the ABE manifested as enhanced memory for the basic-level object category and the specific exemplar. These similarities provide strong evidence for the generality of the ABE. Even substantial task demands, such as the need to make a choice response, failed to eliminate the processing advantage associated with the detection and response to the targets.

Experiment 3

The first two experiments showed that choice-response tasks can produce an ABE. In addition, Experiment 2 suggested that the ABE may be weaker in choice-response tasks than in detection tasks. The interaction between square status (target vs distractor) and task, however, only reached marginal statistical significance. The goal of Experiment 3, therefore, was to provide an additional test for the task modulation of the ABE. To this end, we repeated Experiment 1 but changed the RSVP task from a choice-response to a detection task. Participants pressed the spacebar to the targets (1 or 2 for half of the participants, < or > for the other half) and made no response on distractor or blank trials. If choice responses interfere with the attentional boost, then the ABE should be larger in Experiment 3 than in Experiment 1.

Method

Participants

Overall, 43 new participants from Prolific.co completed Experiment 3. There were 14 females and 34 males with a mean age of 30.3 years (SD = 6.0). Five additional participants completed the experiment, but their data were excluded because they received a duplicated participant number (see Note 2). One other participant was excluded due to low accuracy in the RSVP task that fell below the 90% accuracy cutoff.

Material, procedure, and design

This experiment used the same stimuli, procedure, and design as those of Experiment 1. The only difference was the RSVP task participants performed in the dual-task encoding phase. Half of the participants were asked to press the spacebar for the digits and make no responses to the symbols; the other half of the participants did the reverse.

Results

Table 1 presents the condition means of Experiment 3. For ease of comparison, we also included results from Experiment 1.

Table 1.

Results from Experiment 3 (detection task) and Experiment 1 (choice-response task). SE of the mean is in the parenthesis.

Square status	Experiment 3 (detection)			Experiment 1 (choice response)
Square status	Distractor	Target	Blank	Distractor	Target	Blank
RSVP accuracy	97.6% (0.4%)	99.2% (0.2%)	99.9% (0.1%)	99.3% (0.2%)	95.7% (0.7%)	99.96% (0.04%)
RSVP RT (ms)	N/A	494 (7)	N/A	N/A	561 (9)	N/A
Object memory	67.9% (2.8%)	73.0% (3.2%)	75.3% (2.6%)	71.4% (2.3%)	76.4% (2.3%)	77.1% (2.2%)

RSVP: rapid serial visual presentation. RT: Response Time.

Encoding phase RSVP task

Mean accuracy on the RSVP task was 98.9% (SE = 0.1%). Participants were highly accurate on target trials (M = 99.2%, SE = 0.2%). They correctly withheld responses on nearly all of the blank trials (M = 99.9%, SE = 0.1%), but made some false alarms on distractor trials (M = 97.6%, SE = 0.4%). Paired-samples t-tests comparing the false response rates showed a significant difference between the distractor and blank trials, t(47) = 5.072, p < .001, Cohen’s d = 0.732, a result confirmed by the nonparametric Wilcoxon signed-rank test (p < .001).

As in Experiment 2, the error pattern in the RSVP task differed between the choice-response and the detection tasks. In the choice-response task of Experiment 1, errors were concentrated on target trials, with virtually no false responses on distractor or blank trials. In the detection task of Experiment 3, errors reflected primarily false responses on distractor trials, with few mistakes on target or blank trials. An ANOVA using task as a between-subject factor and square status (target, distractor, and blank) as a within-subject factor revealed a significant interaction, F(2, 188) = 31.829, p < .001, $η_{p}^{2} = . 253$ .

Object memory

Accuracy for choosing the old object varied across conditions. An ANOVA on RSVP status (target, distractor, or blank) revealed a significant main effect, F(2, 94) = 6.398, p = .002, $η_{p}^{2} = . 120$ . Planned contrasts showed that target-paired objects were better remembered than distractor-paired objects, t(47) = 2.39, p = .021 (though this missed the Bonferroni-corrected alpha of .0167), Cohen’s d = 0.345. In addition, distractor-paired objects were more poorly remembered than objects presented on their own, t(47) = 3.926, p < .001, Cohen’s d = 0.567. Target-paired objects produced comparable accuracy as the blank baseline, t(47) = 1.003, p = .321, Cohen’s d = 0.145. The magnitude of the ABE in Experiment 3 (detection task) was smaller than that in Experiment 2’s detection group, F(1, 94) = 8.945, p = .004, $η_{p}^{2} = . 087$ . The difference in tasks—symbol/shape discrimination in Experiment 3 and colour discrimination in Experiment 2—may have contributed to the difference in the magnitude of the ABE, though additional research is needed.

Similar results were observed when we analysed category and exemplar memories separately, though the statistical significance was weaker (Supplementary Material B).

Task modulation

To examine whether the detection task yielded a greater ABE than the choice-response task, we conducted an ANOVA using task (Experiment 1’s choice-response task vs Experiment 3’s detection task) as a between-subject factor and encoding condition (target vs distractor character) as a within-subject factor. The use of the same symbol/shape discrimination task in these two experiments facilitated the direct comparison between the two. As in Experiment 2, this analysis excluded the blank trials, which were not expected to differ between the choice-response and detection tasks. This analysis revealed just a significant main effect of encoding condition, F(1, 94) = 13.022, p < .001, $η_{p}^{2} = . 122$ , showing an ABE. The main effect of task was not significant, F(1, 94) = 0.988, p = .323, $η_{p}^{2} = . 010$ , neither was the interaction between encoding condition and task, F(1, 94) = 0.004, p = .951, $η_{p}^{2} < . 001$ . The ABE was comparable in size between Experiment 1’s choice-response task (Cohen’s d = 0.399; 4.9% accuracy gain) and Experiment 3’s detection task (Cohen’s d = 0.342; 5.1% accuracy gain). Using the ABE (target minus distractor accuracy) as the dependent measure, a Bayesian analysis showed a Bayes factor of 6.366 (null vs alternative hypothesis) for the effect of task, in favour of the null hypothesis.

The lack of task modulation on the ABE raised the question of whether the task demand was higher in the choice-response task than in the detection task. To this end, we compared the RSVP performance between Experiments 1 and 3. Compared with the detection task, the choice-response task was associated with significantly lower accuracy, t(94) = 2.009, p = .047, Cohen’s d = 0.410, and slower Response Time (RT), t(94) = 5.819, p < .001, Cohen’s d = 1.188. This finding confirmed that the choice-response task was more demanding than the detection task.

Discussion

Experiment 3 used the same stimuli and procedure as in Experiment 1 but changed the RSVP task from a choice-response to a detection task. Participants in Experiment 3 responded to the RSVP targets more quickly and accurately than those in Experiment 1, verifying the reduced task demand. In addition, response errors reflected primarily a failure to withhold responses on distractor trials in Experiment 3, and this pattern differed from that of Experiment 1, where errors constituted primarily of an incorrect choice response on target trials.

If reduced response demands always yield a larger ABE, as was apparent in previous studies (Leclercq & Seitz, 2012; Swallow & Jiang, 2010) and in our Experiment 2, then Experiment 3 should produce a larger ABE than Experiment 1. The data did not support this prediction. The ABE was comparable in effect size between Experiments 1 and 3. Thus, although increased response demands may, under some conditions, interfere with the ABE, this is not always the case.

General discussion

Detecting a target in a stream of nontargets is often associated with increased attentional demands, leading to reduced attention to concurrent or subsequent stimuli (Duncan, 1980; Raymond et al., 1992). Despite this increase in attentional demand, detecting and responding to an RSVP target enhances the processing of background images, resulting in better memory for target-paired images than distractor-paired images. For over a decade, studies on this paradoxical ABE have been largely confined to one type of RSVP tasks—a detection task in which participants make a Go response to targets and a No-Go response to distractors. When the targets required more complex responses, such as pressing one key for a red target and another key for a green target, the ABE was much weaker or absent (Leclercq & Seitz, 2012; Swallow & Jiang, 2010). Theoretical explanations for such diminution have focused on either the perceptual component of the RSVP stimuli, such as the automatic cueing of spatial attention by arrow targets, or the response demands of the task (Swallow & Jiang, 2010). The use of infrequent targets in these studies, together with relatively small sample sizes, has clouded both the empirical results and their explanations. Using a sample size about three times as large as past studies and an improved experimental design, this study showed that the ABE was robust in choice-response tasks. By providing greater clarity on the empirical results, the finding challenged the idea that response demands are particularly detrimental to the ABE.

In Experiment 1, participants monitored an RSVP stream of characters while encoding background objects to memory. The target characters—either digits or symbols—required participants to make a choice response corresponding to the target’s identity. Distractor characters required no response. In both cases, participants had to identify the character, but the target trials required additional response selection and execution. Despite the increased task demands, target trials yielded better object memory than distractor trials. The ABE was moderate, with an observed effect size of 0.399 in Cohen’s d. Previous ABE studies had a typical sample size of 16. These studies were underpowered to detect a moderate effect size found in Experiment 1—the estimated power of detecting an effect size of 0.399 with a sample of 16 was only .32. The larger sample size was an important reason why we were able to find an ABE in Experiment 1.

Experiment 2 provided a replication and extension of the ABE in choice-response tasks. The colours used in the RSVP task had no intrinsic spatial properties, making the stimulus-to-response mapping more arbitrary. With these stimuli, participants could not rely on the mental number line or the implied directionality of the symbols to respond. Mean RT on target trials was slower in Experiment 2 (M = 589 ms, SE = 10 ms) than in Experiment 1 (M = 561 ms, SE = 9 ms), t(94) = 1.976, p = .051, Cohen’s d = 0.403, indicating that the response mapping was more difficult in Experiment 2. As in Experiment 1, errors in the RSVP task occurred predominantly on target trials, indicating greater task demands on target than distractor trials. Nonetheless, objects encoded with the RSVP targets were better remembered than objects encoded with the RSVP distractors. The effect size—Cohen’s d of 0.510—was again modest, but it showed once again that with a larger sample size, it was possible to observe an ABE in choice-response tasks.

The preservation of the ABE in choice-response tasks suggests that although response demands can increase dual-task interference, they do not fundamentally change the nature of attentional boost. This does not mean that the response is not important. Previous studies have found that the ABE occurs when an RSVP stimulus is classified as a target (Sisk & Jiang, 2020; Swallow & Jiang, 2014b). Distractors that are perceptually similar to the target do not trigger an ABE (Swallow & Jiang, 2014b). This finding suggests that the ABE occurs relatively late in the processing stage of the RSVP stimuli. Furthermore, when the RSVP stimuli require Go/No-Go responses, stimuli associated with the Go response receive a memory enhancement, even when these stimuli are nontargets (Toh & Lee, 2022). These findings converge on the conclusion that the ABE happens to RSVP stimuli requiring a behavioural update, such as a manual response, a covert count, or a delayed response. The current study does not challenge the importance of response to the ABE. What it shows, though, is that the manner of response to the RSVP targets is not critical. The ABE was found both when participants made a simple Go response to the targets and when they made choice responses to them.

How do we account for the lack of a clear ABE in previous choice-response tasks? In addition to a larger sample size, our study design incorporated features that increased our chance of finding clear results. Swallow and Jiang (2010) embedded occasional targets among nontargets in the RSVP stream, with the target occurring less than one-seventh of the time. Memory was then assessed not only for the occasional targets but also for distractors at various temporal positions relative to the target. This design limited the number of data points per temporal position. Likewise, in Leclercq and Seitz (2012), the arrow target appeared once every 16 presentations. In addition, because a single probe trial was tested after every 16 presentations, the test was subjected to serial position effects: images presented at the end of the series were likely remembered better owing to recency effects and reduced conceptual masking (Intraub, 1984; Potter, 1993; Potter et al., 2002). For a significant ABE to emerge, the target-induced boost would need to overcome not only dual-task interference but also serial position effects. In our study, the target, distractor, and blank trials occurred equally often and were randomly intermixed. This not only eliminated systematic serial position effects but also provided more data points per condition. The careful counterbalancing of images associated with different conditions also eliminated differences in the image memorability (Bainbridge, 2019), a factor that could have added noise to the data.

Other procedural differences across studies may have also contributed to a difference in results, though their impact requires further investigation. For example, whereas images were presented twice in the current study, they were presented 10 times in Swallow and Jiang (2010). This difference is unlikely to have affected the results, given that the ABE has been observed across a wide range of image repetitions, including one-exposure designs (Broitman & Swallow, 2020; Sisk & Lee, 2022). Another procedural difference that could have influenced the results was the presentation pace. The earlier studies presented the images at a pace of 500 ms/item (Leclercq & Seitz, 2012; Swallow & Jiang, 2010), which was faster than the presentation pace of 1,200 ms/item used in this study. Although the ABE has been found across a wide range of presentation pace (Mulligan & Spataro, 2015; Swallow et al., 2012), choice-response tasks may be particularly sensitive to presentation pace. In Experiments 2 and 3, the mean choice RT exceeded 500 ms, suggesting that the previous studies may have given participants insufficient time to respond. Time pressure, and the ensuing stress, may have interacted with the function of the LC–NE system (Yebra et al., 2019). By giving participants an adequate amount of time to respond, the current study may have removed the time pressure that interfered with the ABE.

The inclusion of the blank baseline condition helped clarify the nature of the ABE in our study. In all experiments, objects encoded with the RSVP targets were better remembered than those encoded with the RSVP distractors, but not better than the blank baseline. This pattern of results was described as a “relative boost.” It is a common finding in studies of the ABE (Spataro et al., 2013; Swallow & Jiang, 2010). Another pattern of results, an “absolute boost,” refers to better memory for target-paired images relative to both distractor-paired images and the single-task baseline. This pattern was observed when implicit, rather than explicit, memory test was used (Spataro et al., 2013). Tests of explicit memory had sometimes shown an absolute boost when the image encoding was very brief (e.g., 400 ms/item, rather than the typical 1 s/item; Mulligan & Spataro, 2015; see also Prull, 2019). Notably, an absolute boost was reported in Swallow and Jiang (2014a), who randomly intermixed RSVP target, distractor, and blank trials. Recognition memory was better for target-paired images than images encoded on their own. This pattern of results was replicated in subsequent studies and extended to tests of older adults (Bechi Gabrielli et al., 2021; Rossi-Arnaud et al., 2018). Given that the current study used the three-condition (target/distractor/blank) design, why did we find a relative boost rather than an absolute boost?

One reason why the ABE was relative in our study was that the RSVP task that we used produced greater dual-task interference than in previous studies. The previous three-condition studies have, for the most part, used simple perceptual stimuli, with participants making a Go response to (say) blue squares and making no response to (say) red squares and blank trials. Our RSVP task included four types of stimuli—four different characters in Experiments 1 and 3 and four colours in Experiment 2. The increased perceptual demand was reflected in the response time. The choice-response tasks had RTs exceeding 550 ms. Even the detection task produced a mean RT of around 490 ms, longer than the 396 ms RT reported in Swallow and Jiang’s (2014a) detection task. A second reason why our study incurred greater dual-task interference was the presentation of the RSVP stimuli on top of the background images. The coloured squares were less visible, especially if the background had similar colours. In addition, the RSVP stimuli occluded the centre part of the objects for 500 ms, interfering with object encoding. In contrast, Swallow and Jiang (2014a) presented the RSVP stimuli on the two sides of the background image, reducing perceptual interference between the two tasks. Thus, increased perceptual interference between the RSVP stimuli and the background objects in our study likely accounts for why the ABE was a relative boost.

Although our data are consistent with the idea that the target-induced ABE overcame interference from response selection, they also raised the question about whether our results reflected distractor-induced inhibition rather than target-induced boost. After all, in all experiments, distractor-paired objects were remembered more poorly than both the target-paired objects and the blank baseline. In fact, some studies have proposed that withholding responses led to the inhibition of concurrently presented stimuli (Chiu & Egner, 2015a, 2015b). However, it is unlikely that response inhibition could explain our results. First, participants withheld responses not only on distractor trials but also on blank trials. Given that all three trial types were randomly intermixed, participants could not have predicted which stimuli they would see. If they had initiated an anticipatory response, they would have likely done so on all trials. Second, in the choice-response tasks, false alarms to distractor or blank trials were nearly nonexistent. If participants had planned to make a response and then cancelled it, this should have produced more false alarms. This pattern can be contrasted with the detection group of Experiment 2 (and those in Experiment 3), who falsely responded to distractors on 3%–4% of the trials. Notably, participants in the detection task rarely committed false alarms on blank trials, suggesting that preparatory responses were rarely made on blank trials. Thus, whereas response inhibition may have been a factor in the detection group, it is unlikely significant in the choice-response task. Third, the response inhibition account does not explain why, despite the much greater task demands, memory was not impaired on target trials relative to the blank baseline. Thus, while it is possible that response inhibition may have contributed to the ABE in detection tasks, inhibition is unlikely to explain the ABE in the choice-response task.

By showing that the ABE survives increased demands in response selection, the current study clarifies the relationship between target-induced boost and dual-task interference. Like the previous version of the dual-task interaction model, we propose that two concurrent tasks—the RSVP task and image encoding—compete for limited processing resources. Thus, as the attentional demands for the RSVP task increase, memory for the concurrently presented images declines. However, the temporal orienting to the RSVP target produces an attentional boost even when the RSVP task is challenging. The ABE is preserved both with high perceptual load (Swallow & Jiang, 2014b; Toh & Lee, 2022) and with high response demands. In this updated view, response demands do not exert a uniquely detrimental effect on the ABE. Instead, the target-induced boost occurs with both high perceptual load and high response demands.

Our study does not rule out the possibility that under some circumstances, the nature of the RSVP task may interact with the attentional boost. Many studies have identified transient activation of neurons in LC as the neurophysiological basis of the ABE. Neurons in the LC not only fire transiently (i.e., phasic response) but they also maintain a baseline level of activity (i.e., tonic response). The tonic level of activity can range from low to high, reflecting, in part, the arousal level of the observer, which in turn, affects the effectiveness of the neuron’s phasic response (Aston-Jones & Cohen, 2005; Aston-Jones & Waterhouse, 2016). Characterised by the inverted-U function of the Yerkes–Dodson law, phasic responses are optimal when the tonic activity is neither too low nor too high. These neurophysiological considerations suggest that RSVP tasks could interact with the ABE if the tasks produce very high or very low levels of arousal. Future studies that use a broad range of RSVP tasks are needed to enrich our understanding of how concurrent tasks interact, both competitively as in dual-task interference, and synergistically as in the ABE.

Supplemental Material

sj-docx-1-qjp-10.1177_17470218231156375 – Supplemental material for The attentional boost effect overcomes dual-task interference in choice-response tasks

Supplemental material, sj-docx-1-qjp-10.1177_17470218231156375 for The attentional boost effect overcomes dual-task interference in choice-response tasks by Vanessa G Lee in Quarterly Journal of Experimental Psychology

Footnotes

Acknowledgements

The author thanks Linden Lee for the discussion and edits to the article and Gavin Oliver for the help with Experiment 3.

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Engdahl Research Fund. The funder was not involved in the design of the study or its execution and interpretation.

ORCID iD

Vanessa G Lee

Data accessibility statement

The data and materials from this experiment are publicly available at the Open Science Framework website:

Supplemental material

The supplementary material is available at .

Notes

References

Aston-Jones

Cohen

J. D.

(2005). An integrative theory of locus coeruleus-norepinephrine function: Adaptive gain and optimal performance. Annual Review of Neuroscience, 28, 403–450. https://doi.org/10.1146/annurev.neuro.28.061604.135709

Aston-Jones

Waterhouse

(2016). Locus coeruleus: From global projection system to adaptive regulation of behavior. Brain Research, 1645, 75–78. https://doi.org/10.1016/j.brainres.2016.03.001

Bainbridge

W. A.

(2019). Memorability: How what we see influences what we remember. In Federmeier

K. D.

Beck

D. M.

(Eds.), Psychology of learning and motivation (Vol. 70, pp. 1–27). Academic Press. https://doi.org/10.1016/bs.plm.2019.02.001

Bechi Gabrielli

Rossi-Arnaud

Spataro

Doricchi

Costanzi

Santirocchi

Angeletti

Sani

Cestari

. (2021). The attentional boost effect in young and adult euthymic bipolar patients and healthy controls. Journal of Personalized Medicine, 11(3), 185. https://doi.org/10.3390/jpm11030185

Broitman

A. W.

Swallow

K. M.

(2020). The effects of encoding instruction and opportunity on the recollection of behaviourally relevant events. The Quarterly Journal of Experimental Psychology, 73(5), 711–725. https://doi.org/10.1177/1747021819893676

Chiu

Y.-C.

Egner

(2015a). Inhibition-induced forgetting results from resource competition between response inhibition and memory encoding processes. Journal of Neuroscience, 35(34), 11936–11945. https://doi.org/10.1523/JNEUROSCI.0519-15.2015

Chiu

Y.-C.

Egner

(2015b). Inhibition-induced forgetting: When more control leads to less memory. Psychological Science, 26(1), 27–38. https://doi.org/10.1177/0956797614553945

Dehaene

Bossini

Giraux

(1993). The mental representation of parity and number magnitude. Journal of Experimental Psychology: General, 122(3), 371–396. https://doi.org/10.1037/0096-3445.122.3.371

Dienes

(2014). Using Bayes to get the most out of non-significant results. Frontiers in Psychology, 5, Article 781. https://www.frontiersin.org/article/10.3389/fpsyg.2014.00781

10.

Duncan

(1980). The locus of interference in the perception of simultaneous stimuli. Psychological Review, 87(3), 272–300. https://doi.org/10.1037/0033-295X.87.3.272

11.

Hawkins

G. E.

Mittner

Forstmann

B. U.

Heathcote

(2019). Modeling distracted performance. Cognitive Psychology, 112, 48–80. https://doi.org/10.1016/j.cogpsych.2019.05.002

12.

Intraub

(1984). Conceptual masking: The effects of subsequent visual events on memory for pictures. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10(1), 115–125. https://doi.org/10.1037/0278-7393.10.1.115

13.

Leclercq

Seitz

A. R.

(2012). The impact of orienting attention in fast task-irrelevant perceptual learning. Attention, Perception, & Psychophysics, 74(4), 648–660. https://doi.org/10.3758/s13414-012-0270-7

14.

Lien

M.-C.

Proctor

R. W.

Allen

P. A.

(2002). Ideomotor compatibility in the psychological refractory period effect: 29 years of oversimplification. Journal of Experimental Psychology: Human Perception and Performance, 28(2), 396–409. https://doi.org/10.1037/0096-1523.28.2.396

15.

Lin

J. Y.

Pype

A. D.

Murray

S. O.

Boynton

G. M.

(2010). Enhanced memory for scenes presented at behaviorally relevant points in time. PLOS Biology, 8(3), Article e1000337. https://doi.org/10.1371/journal.pbio.1000337

16.

Makovski

Swallow

K. M.

Jiang

Y. V.

(2011). Attending to unrelated targets boosts short-term memory for color arrays. Neuropsychologia, 49(6), 1498–1505. https://doi.org/10.1016/j.neuropsychologia.2010.11.029

17.

Meyer

D. E.

Kieras

D. E.

(1997). A computational theory of executive cognitive processes and multiple-task performance: Part I. Basic mechanisms. Psychological Review, 104(1), 3–65. https://doi.org/10.1037/0033-295X.104.1.3

18.

Mulligan

N. W.

Spataro

(2015). Divided attention can enhance early-phase memory encoding: The attentional boost effect and study trial duration. Journal of Experimental Psychology. Learning, Memory, and Cognition, 41(4), 1223–1228. https://doi.org/10.1037/xlm0000055

19.

Nishina

Seitz

A. R.

Kawato

Watanabe

(2007). Effect of spatial distance to the task stimulus on task-irrelevant perceptual learning of static Gabors. Journal of Vision, 7(13), 2. https://doi.org/10.1167/7.13.2

20.

Pashler

(1994). Divided attention can enhance early-phase memory encoding: The attentional boost effect and study trial duration. Psychological Bulletin, 116(2), 220–244. https://doi.org/10.1037/0033-2909.116.2.220

21.

Peirce

Gray

J. R.

Simpson

MacAskill

Höchenberger

Sogo

Kastman

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

22.

Potter

M. C.

(1993). Very short-term conceptual memory. Memory & Cognition, 21(2), 156–161. https://doi.org/10.3758/BF03202727

23.

Potter

M. C.

Staub

Rado

O’Connor

D. H.

(2002). Recognition memory for briefly presented pictures: The time course of rapid forgetting. Journal of Experimental Psychology: Human Perception and Performance, 28(5), 1163–1175. https://doi.org/10.1037/0096-1523.28.5.1163

24.

Prull

M. W.

(2019). The attentional boost effect for words in young and older adults. Psychology and Aging, 34(3), 405–417. https://doi.org/10.1037/pag0000337

25.

Raymond

J. E.

Shapiro

K. L.

Arnell

K. M.

(1992). Temporary suppression of visual processing in an RSVP task: An attentional blink? Journal of Experimental Psychology: Human Perception and Performance, 18(3), 849–860. https://doi.org/10.1037/0096-1523.18.3.849

26.

Rossi-Arnaud

Spataro

Costanzi

Saraulli

Cestari

(2018). Divided attention enhances the recognition of emotional stimuli: Evidence from the attentional boost effect. Memory, 26(1), 42–52. https://doi.org/10.1080/09658211.2017.1319489

27.

Rouder

J. N.

Morey

R. D.

(2011). A Bayes factor meta-analysis of Bem’s ESP claim. Psychonomic Bulletin & Review, 18(4), 682–689. https://doi.org/10.3758/s13423-011-0088-7

28.

Schonberg

Bakkour

Hover

A. M.

Mumford

J. A.

Nagar

Perez

Poldrack

R. A.

(2014). Changing value through cued approach: An automatic mechanism of behavior change. Nature Neuroscience, 17(4), 625–630. https://doi.org/10.1038/nn.3673

29.

Seitz

A. R.

Watanabe

(2003). Is subliminal learning really passive? Nature, 422(6927), 36–36. https://doi.org/10.1038/422036a

30.

Sisk

C. A.

Jiang

Y. V.

(2020). The yellow light: Predictability enhances background processing during behaviorally relevant events. Journal of Experimental Psychology: Learning, Memory, and Cognition, 46(9), 1645–1658. https://doi.org/10.1037/xlm0000838

31.

Sisk

C. A.

Lee

V. G.

(2022). Concurrent target detection is associated with better memory for object exemplars. Psychonomic Bulletin & Review, 29(1), 159–168. https://doi.org/10.3758/s13423-021-01983-0

32.

Spataro

Mulligan

N. W.

Rossi-Arnaud

(2013). Divided attention can enhance memory encoding: The attentional boost effect in implicit memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 39(4), 1223–1231. https://doi.org/10.1037/a0030907

33.

Swallow

K. M.

Broitman

A. W.

Riley

Turker

H. B.

(2022). Grounding the attentional boost effect in events and the efficient brain. Frontiers in Psychology, 13, Article 892416. https://www.frontiersin.org/articles/10.3389/fpsyg.2022.892416

34.

Swallow

K. M.

Jiang

Y. V.

(2010). The attentional boost effect: Transient increases in attention to one task enhance performance in a second task. Cognition, 115(1), 118–132. https://doi.org/10.1016/j.cognition.2009.12.003

35.

Swallow

K. M.

Jiang

Y. V.

(2013). Attentional load and attentional boost: A review of data and theory. Frontiers in Psychology, 4, Article 274. https://doi.org/10.3389/fpsyg.2013.00274

36.

Swallow

K. M.

Jiang

Y. V.

(2014a). The attentional boost effect really is a boost: Evidence from a new baseline. Attention, Perception, & Psychophysics, 76(5), 1298–1307. https://doi.org/10.3758/s13414-014-0677-4

37.

Swallow

K. M.

Jiang

Y. V.

(2014b). Perceptual load and attentional boost: A study of their interaction. Journal of Experimental Psychology: Human Perception and Performance, 40(3), 1034–1045. https://doi.org/10.1037/a0035312

38.

Swallow

K. M.

Makovski

Jiang

Y. V.

(2012). Selection of events in time enhances activity throughout early visual cortex. Journal of Neurophysiology, 108(12), 3239–3252. https://doi.org/10.1152/jn.00472.2012

39.

Toh

Y. N.

Lee

V. G.

(2022). Response, rather than target detection, triggers the attentional boost effect in visual search. Journal of Experimental Psychology: Human Perception and Performance, 48(1), 77–93. https://doi.org/10.1037/xhp0000977

40.

Wolfe

J. M.

(2021). Guided search 6.0: An updated model of visual search. Psychonomic Bulletin & Review, 28(4), 1060–1092. https://doi.org/10.3758/s13423-020-01859-9

41.

Yebra

Galarza-Vallejo

Soto-Leon

Gonzalez-Rosa

J. J.

de Berker

A. O.

Bestmann

Oliviero

Kroes

M. C. W.

Strange

B. A.

(2019). Action boosts episodic memory encoding in humans via engagement of a noradrenergic system. Nature Communications, 10(1), Article 3534. https://doi.org/10.1038/s41467-019-11358-8

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB