Handling visual distraction: Statistical learning of spatial and feature-based distractor regularities
Salient distractors have the potential to attract attention and, thus, interfere with the search for a target. Prior work has shown that distractor interference is more pronounced when the features of the target being searched for are less predictable and when there is an overlap between target and...
Saved in:
| Main Author: | |
|---|---|
| Contributors: | |
| Format: | Doctoral Thesis |
| Language: | English |
| Published: |
Philipps-Universität Marburg
2025
|
| Subjects: | |
| Online Access: | PDF Full Text |
| Tags: |
No Tags, Be the first to tag this record!
|
| Summary: | Salient distractors have the potential to attract attention and, thus, interfere with the search for a target. Prior work has shown that distractor interference is more pronounced when the features of the target being searched for are less predictable and when there is an overlap between target and distractor features. For example, distractor interference is larger when observers search for a shape singleton among similarly shaped nontargets (e.g., the target is a circle among diamonds or vice versa) than when the target shape is fixed (e.g., the target is always a circle), as only the latter allows observers to set up a precise feature-specific target template. Similarly, distractor interference is more pronounced when observers adopt a search mode that focuses on searching for singletons (the so-called singleton-detection mode) compared to when they are required to search for a unique target shape among heterogeneously shaped nontargets (e.g., a circle target among diamonds, triangles, and squares), thereby adopting feature-search mode to find the target. Interestingly, however, it has been demonstrated that distractor interference can be reduced when the distractor exhibits statistical regularities, such as appearing more often at one location or in one color, which enables either distractor location learning or distractor feature learning, respectively. However, it remains unclear how these findings connect and, thus, how target and nontarget features, as well as the overlap between target and distractor features, influence distractor location learning. In addition, the neural mechanisms of distractor feature learning are not yet well understood. In three studies, the present dissertation addresses these research gaps.
Studies I and II focused on distractor location learning, whereas Study III focused on distractor feature learning. All three studies used adapted variants of the additional singleton task (Theeuwes, 1991, 2018). In the additional singleton task, participants are asked to search for a shape target among circularly arranged nontargets and to report the orientation of the line inside the target while ignoring a salient color distractor (see Figure 1, page 9). In Study I and II, the distractor was more likely to appear at a specific location, while in Study III, the distractor appeared more often in a specific color. It was assumed that participants learn these regularities to reduce distractor interference. Additionally, it was expected that with more distractor interference, distractor location learning would be more pronounced, as it becomes more necessary for efficient target selection.
Study I examined whether the precision of the target template modulates distractor location learning. The shape of the target was either fixed or varied between participants, with only the former allowing the setting up of a precise target template. Results showed that participants were less susceptible to distractor interference with a more precise target template and relied less on distractor location learning. Also, it was observed that differences between groups developed late in the experiment, suggesting that experience with the task is required to enhance distractor handling. In short, the results demonstrate that when the target is highly predictable, there is less distractor interference and, thereby, less distractor location learning.
Study II investigated whether the search mode of the observer and the similarity of target and distractor influence distractor interference and, thereby, distractor location learning. The search mode was manipulated by varying the similarity of the nontargets, allowing participants to adopt only singleton-detection mode in the mixed-feature task but feature-search mode (Bacon & Egeth, 1994) in the fixed-feature task. The similarity of target and distractor was manipulated in two fixed-feature tasks by using either a chromatic (colored) target or an achromatic (gray) target, as it was assumed that the latter overlaps less with a color distractor. Results showed that distractor interference was more pronounced when participants could only rely on singleton-detection mode to find the target. Also, it was observed that participants used distractor location learning in order to reduce distractor interference. Conversely, when participants were required to adopt feature-search mode, the results differed depending on whether the target was chromatic or achromatic. With an achromatic target, distractor interference was observed in the first of two sessions. In contrast, with a chromatic target, the presence of the distractor resulted in a search benefit, presumably showing up-weighting of the target on the featural level (e.g., up-weighting of the color green). Furthermore, the results indicated that distractor location learning was only used with a chromatic target, suggesting that with an achromatic target, participants likely employed a different strategy to reduce interference, such as down-weighing the irrelevant color dimension (Liesefeld & Müller, 2019). Hence, distractor interference and distractor location learning are modulated by the search mode of the observer and the similarity of the target and distractor, indicating that observers can flexibly adjust their strategies to handle distraction based on the search context.
Study III examined distractor feature learning at the neural level by measuring EEG. Also, this study investigated whether distractor feature learning influences later cognitive processes, such as visual working memory (VWM) performance, using a change-detection task (Luck & Vogel, 1997, 2013). In the change-detection task, participants were asked to memorize the color and location of stimuli and, after a retention interval, to report whether the color of one stimulus had changed. Behavioral data showed that participants learned the distractor feature regularities to reduce interference from the distractor. Furthermore, at the neural level, it was observed that less suppression was required (smaller PD) and that attentional allocation was more efficient (larger target N2pc) when the distractor appeared in the more likely color. Also, the role of intertrial priming in distractor feature learning was examined (see also Golan & Lamy, 2022). Intertrial priming refers to the observation that previous encounters with target or distractor features enhance search performance when the same features are presented in subsequent trials (Maljkovic & Nakayama, 1994, 1996). The results showed that observers did not only benefit from distractor feature learning to handle distraction but also from intertrial priming. Distractor feature learning did not influence VWM performance, indicating that learned distractor feature inhibition is not maintained in VWM when the task context changes. Thus, Study III shows that the visual system can learn distractor feature regularities to reduce the impact of a salient distractor while also using intertrial priming to enhance attentional selection. The learned distractor features seem to be not further maintained in VWM, highlighting the brain’s flexibility to adapt to changes in the environment.
To summarize, the present dissertation demonstrates that spatial (Study I and II) and feature-based regularities (Study III) of a distractor can be learned to reduce distractor interference. Distractor location learning is more required when the target features are less predictable (Study I and II). Also, distractor feature learning enhances attentional selection, while observers also benefit from intertrial priming processes (Study III). In addition to the learning of statistical regularities, observers can apply other strategies to handle distraction, depending on the overlap between target and distractor features (Study II). These findings highlight that visual search settings can be variably adapted, demonstrating the remarkable flexibility of the visual search system, which allows humans to handle distraction efficiently. |
|---|---|
| DOI: | 10.17192/z2025.0213 |