Figure 1
Design of the Visual Search Task. In separate blocks, neutral, negative, or positive cues were presented with randomized color selection per trial. Search trials constituted 80% of the task, requiring participants to locate gap-oriented up or down Landolt-C targets. Additionally, 20% of trials were probe trials, prompting reporting of the preceding cue’s color on a color-wheel.
Figure 2
Measures of the Visual working memory quality: An example for the negative cue (blue) condition regarding Figure 1. If working memory is linked to an automatic attentional guidance towards memory-matching items, we would anticipate that measures of working memory quality (Zhang & Luck, 2008) would show evidence of lower quality representations for negative templates than positive templates. This would appear as a larger standard deviation and/or higher guess rate for negative templates than positive templates.
Figure 3
Performance on search trials. Error bars represent between-subject standard error from the mean RTs. ***p < 0.001.
Figure 4
Performance on probe trials. (A) Distributions of recall errors with the fits of Zhang and Luck’s (2008) standard mixture model across the three cue conditions. (B) Absolute error deviation of reported value. (C) Standard Deviation estimates from the mixture model. (D) Guess rate estimates from the mixture model. Error bars represent between-subject standard error from the mean. ***p < 0.001.
Figure 5
Correlation between memory precision and benefits of negative and positive cues. (A) Individuals with higher VWM precision (lower s.d.) showed larger negative cue benefits. (B) No relationship between VWM precision and positive cue benefits. *p < 0.05.
Figure 6
t-tests compared to a zero baseline of the RT benefits of negative and positive cues during periods of good and bad visual working memory. No benefits of negative cues during periods of lower VWM quality. Error bars represent between-subject standard errors from the mean RT benefits. *p < 0.05, ***p < 0.001.