What are the benefits of directed attention within verbal working memory?

Abstract

Information that is particularly relevant for upcoming behaviour can be prioritised within working memory, by directing attention to it. Receiving focused attention during retention is assumed to be associated with specific benefits, such as increased memory performance and reduced vulnerability to perceptual distractions. This has been demonstrated in visuospatial working memory. Given the domain-general nature of the focus of attention, these benefits should extend to verbal working memory as well. This was tested in the current study. In particular, we examined and compared the effects of cue-based and reward-based prioritisation in verbal working memory across a series of five preregistered experiments. These experiments varied in their memory materials, set size, interference, and memory task. Our results collectively revealed several key findings. First, both cue-based and reward-based prioritisation led to a clear and consistent memory boost for prioritised information in verbal working memory. Second, the memory boost induced by cue-based prioritisation was mostly comparable to that induced by reward-based prioritisation. Third, memory for verbal information did not drastically suffer when exposed to perceptual interference. And finally, the effect of perceptual interference on verbal information was not drastically influenced by whether the information was prioritised or not. Overall, this series of experiments contributes to understanding the consequences of directed attention in verbal working memory and highlights similarities and differences from findings in visuospatial working memory.

Keywords: Verbal working memory; attention; distraction; focus of attention; memory boost; vulnerability.

Figure 1.

Schematic illustration of a trial in Experiment 1. Participants memorised five short words, shown in five different frames. These five frames were also shown empty, both before and after memory item presentation (i.e., pre-encoding and post-encoding). On No-prioritisation trials, none of the frames was bolded. On Pre-prioritisation trials, one of the five frames was bolded before item presentation. On Post-prioritisation trials, one of the five frames was bolded after item presentation (as illustrated here). On all trials, after a brief retention interval, memory was tested through cued recall. During the brief retention interval, either the screen remained blank or a suffix item was shown centrally on screen.

Figure 2.

Mean recall performance in Experiment 1, as a function of Prioritisation mode (Cue-based prioritisation or Reward-based prioritisation), Distraction (No suffix vs. Suffix), and Prioritisation Status. For Cue-based prioritisation (Panel A), Prioritisation Status includes Uncued or Cued, with the Cued condition further distinguished by whether prioritisation occurred before or after encoding. For Reward-based prioritisation (Panel B), Prioritisation Status includes Low-reward, High-reward, or Equal-reward, with Low-reward and High-reward conditions further distinguished by whether prioritisation occurred before or after encoding. Error bars represent standard error of the mean.

Figure 3.

Mean vulnerability scores in Experiments 1–3 (mean recall performance in No-suffix condition minus mean recall performance in Suffix condition), as a function of Prioritisation mode (Cue-based prioritisation on the left, Reward-based prioritisation on the right), and Prioritisation Status (Unprioritized vs. Prioritised, corresponding to Uncued vs. Cued items, respectively, in the Cue group, and corresponding to Low-reward vs. High-reward items, respectively, in the Reward group). Grey points are showing individual data, and pink lines represent the means per condition. The dotted line represents a vulnerability score of 0; scores above this line indicate the presence of a disruptive effect of distraction on memory performance, whereas scores on or below this line indicate the absence of a disruptive effect of distraction on memory performance.

Figure 4.

Schematic illustration of a trial in Experiment 2. Participants memorised five short words, shown in five different frames. These five frames were also shown empty after memory item presentation (i.e., post-encoding). On No-prioritisation trials, none of the frames was bolded; on Post-prioritisation trials, one of the five frames was bolded after item presentation (as illustrated here). On all trials, after a brief retention interval, memory was tested through cued recall. During the brief retention interval, either the screen remained blank or a suffix item was shown in all five locations.

Figure 5.

Mean recall performance in Experiment 2, as a function of Prioritisation mode (Cue-based prioritisation or Reward-based prioritisation), Distraction (No suffix vs. Suffix), and Prioritisation Status. For Cue-based prioritisation (Panel A), Prioritisation Status includes Uncued or Cued. For Reward-based prioritisation (Panel B), Prioritisation Status includes Low-reward, High-reward, or Equal-reward. Error bars represent standard error of the mean.

Figure 6.

Schematic illustration of a trial in Experiment 3. Participants memorised four short words, shown in four different colours and in four different frames. These four frames were also shown empty after memory item presentation (i.e., post-encoding). On No-prioritisation trials, none of the frames was bolded, and on Post-prioritisation trials, one of the four frames was bolded after item presentation (as illustrated here). On all trials, after a brief retention interval, memory was tested through cued recall. During the brief retention interval, either the screen remained blank or a suffix item was shown in all four locations.

Figure 7.

Mean recall performance in Experiment 3, as a function of Prioritisation mode (Cue-based prioritisation or Reward-based prioritisation), Distraction (No suffix vs. Suffix), and Prioritisation Status. For Cue-based prioritisation (Panel A), Prioritisation Status includes Uncued or Cued. For Reward-based prioritisation (Panel B), Prioritisation Status includes Low-reward, High-reward, or Equal-reward. Error bars represent standard error of the mean.

Figure 8.

Schematic illustration of a trial in Experiment 4. Participants memorised seven short words, shown in seven different frames. These seven frames were also shown empty, after memory item presentation (i.e., post-encoding). On No-prioritisation trials, none of the frames was bolded; on Post-prioritisation trials, one of the seven frames was bolded after item presentation (as illustrated here). On all trials, after a brief retention interval that remained blank, memory was tested through cued recall.

Figure 9.

Mean recall performance in Experiment 4, as a function of Prioritisation Mode (Cue-based or Reward-based prioritisation) and Prioritisation Status. For Cue-based prioritisation (Panel A), Prioritisation Status includes Uncued or Cued. For Reward-based prioritisation (Panel B), Prioritisation Status includes Low-reward, High-reward, or Equal-reward. Error bars represent standard error of the mean.

Figure 10.

Schematic illustration of a trial in Experiment 5. Participants memorised five shape words, shown in five different frames. These five frames were also shown empty after memory item presentation (i.e., post-encoding). On No-prioritisation trials, none of the frames was bolded, on Post-prioritisation trials, one of the five frames was bolded after item presentation (as illustrated here). On all trials, after a brief retention interval that remained blank, memory was tested through cued recall. The corresponding shape had to be drawn using paper and pencil.

Figure 11.

Mean recall performance in Experiment 5, as a function of Prioritisation Mode (Cue-based or Reward-based prioritisation) and Prioritisation Status. For Cue-based prioritisation (Panel A), Prioritisation Status includes Uncued or Cued. For Reward-based prioritisation (Panel B), Prioritisation Status includes Low-reward, High-reward, or Equal-reward. Error bars represent standard error of the mean.