Faculty Mentor: Brennan Payne (Psychology, University of Utah)

Abstract

We explored how user-controlled cues impact distractor-induced quitting in visual search. Participants completed visual searches in four conditions: addition cue, removal cue, automatic, and baseline. In some trials, the cues highlighted the target, while in others, they highlighted a non-target. In the addition cue condition, participants pressed a key to trigger a salient cue before responding. Conversely, the removal cue condition required participants to remove the cue with a keypress before responding. The automatic condition presented cues with the display, whereas the baseline condition included no cues. We replicated past findings showing evidence for early quitting behavior. Contrary to our expectations, user-controlled cues did not reduce early quitting behavior and might have worsened it.

Background

Our research investigates how top-down control mechanisms can mitigate the early quitting effect in visual search tasks. The early quitting effect refers to the tendency of individuals to prematurely abandon a task when encountering distractions. In visual search, this effect can manifest when participants give up finding their target due to the presence of misleading or distracting cues. For example, Moher (2020) aimed to investigate whether task-irrelevant cues can produce an early quitting effect in three different experiments. In these experiments, on each trial, a salient distractor could or could not be present, and a target could or could not be present. These trial types were randomly intermixed independently so that each participant received an approximately equal number of trials in the different relevant combinations. They found that in target-present trials, participants had longer RTs when a salient distractor was present compared to when it was absent. On the other hand, target-absent trial RTs were shorter with salient distractors present. In addition, salient distractors increased errors in target-present trials. These findings suggest that salient distractors makeparticipants reduce their quitting threshold in visual search. In a follow-up study, Moher et al. (under review) explored whether task-relevant salient cues can produce a similar early quitting effect on trials where the cues fail to highlight the target. They found that salient cues can indeed trigger early quitting on trials when those cues highlight a non-target.

Our study investigates the impact of user-controlled cues on performance, advancing understanding of their influence on visual search efficiency and task engagement. This research is particularly relevant in fields like radiology, where accurate visual searches are crucial for diagnosing medical conditions from images. By examining how different cue types affect performance, our findings could inform the development of enhanced imaging systems, training programs and computer aided cues across a variety of clinical imaging types. These advancements may empower radiologists to better detect subtle abnormalities, thereby improving diagnostic accuracy and patient outcomes in clinical practice.

Methods

Participants were recruited online to complete a visual search task, randomly assigned to four between-subjects experimental conditions. They searched for a rotated “T” among rotated “L”s, all displayed on a gray 1/f noise on a black background. In target-present trials, one of the items was a rotated “T”, while in target-absent trials, all items were rotated “L”s. The study employed a design with four conditions- Addition Cue, Removal Cue, Automatic, and Baseline. Addition Cue Condition: Participants pressed the “N” key at some point during each trial to make a red circle cue appear, which sometimes highlighted the target “T”. Removal Cue Condition: Participants pressed the “N” key to make a red circle cue disappear, which also sometimes highlighted the target “T”. Automatic Condition: Red circle cues appeared simultaneously with the rest of the display on each trial. Baseline Condition: Red circle cues did not appear with the display. Participants responded solely on the presence of the target T among distractor L’s. Each trial began with images displaying rotated T’s and distractor L’s. Participants were instructed to respond as quickly and accurately as possible by pressing the appropriate keys on the keyboard. The experiment included a practice phase to familiarize participants with the task, followed by several blocks of trials with breaks in between.

Study Predictions

First, we hypothesized that we would replicate the early quitting effect such that accuracy would be poorer and reaction times (RTs) faster in the automatic condition (relative to the baseline condition) when the salient cue highlighted a distractor. The critical question concerned whether the user-controlled conditions (addition and removal cue conditions) would result in reducedearly quitting behavior. Specifically, we hypothesized that accuracy rates on trials where the salient cue was irrelevant would be higher in the Addition and Removal Cue conditions due to enhanced engagement. Similarly, we expected longer RTs on the Addition and Removal Cue conditions, relative to the Automatic condition when the salient cue highlighted a distractor, indicating sustained engagement with the task.

Result and Discussion

Our findings robustly replicated Moher et al. (under review), with greater evidence of early quitting in all cueing conditions compared to baseline, and with participants showing higher rates of prematurely abandoning the task when they had control over cue timing. In terms of RTs, when no target was present, reaction times (RTs) were as follows: Addition condition:2.5s, Removal condition:2.6s, Baseline condition had an RT of 4.1s, and Automatic condition: 2.8s. Notably, all three cue conditions (Addition, Removal, and Automatic) showed shorter RTs compared to the baseline no cue condition. The analysis of RTs across conditions highlights that while participants generally responded faster to relevant cues, user control RTs were shorter than the baseline condition which fitted that user control exacerbated early quitting. Our results also revealed unexpected findings contrary to our hypothesis. Participants did not demonstrate improved performance when given control over cue appearances. Specifically, we found that overall accuracy did not significantly differ across conditions ( Addition cue:82%, Removal cue:82%, Baseline:77%, Automatic:84%). Accuracy rates on trials where the salient cue was irrelevant were lower in conditions with user-controlled cues compared to the baseline and automatic conditions (off-target accuracy: Addition Cue 43%, Removal Cue 42%, Baseline 70%, Automatic 48%). These outcomes suggest that while participants may have engaged more with the task when given control over cues, it did not translate to improved task performance or reduced quitting behavior- if anything, early quitting was exacerbated when participants controlled the cue. Further research is needed to explore the underlying mechanisms behind these unexpected results for optimizing cue usage in visual search context.

References

Moher, J. (2020). Distracting objects induce early quitting in visual search. Psychological

Moher, J et al. Cue relevance drives early quitting in visual search. Psychological.