Focusing on attention: the effects of working memory capacity and load on selective attention

Abstract

Background: Working memory (WM) is imperative for effective selective attention. Distractibility is greater under conditions of high (vs. low) concurrent working memory load (WML), and in individuals with low (vs. high) working memory capacity (WMC). In the current experiments, we recorded the flanker task performance of individuals with high and low WMC during low and high WML, to investigate the combined effect of WML and WMC on selective attention.

Methodology/principal findings: In Experiment 1, distractibility from a distractor at a fixed distance from the target was greater when either WML was high or WMC was low, but surprisingly smaller when both WML was high and WMC low. Thus we observed an inverted-U relationship between reductions in WM resources and distractibility. In Experiment 2, we mapped the distribution of spatial attention as a function of WMC and WML, by recording distractibility across several target-to-distractor distances. The pattern of distractor effects across the target-to-distractor distances demonstrated that the distribution of the attentional window becomes dispersed as WM resources are limited. The attentional window was more spread out under high compared to low WML, and for low compared to high WMC individuals, and even more so when the two factors co-occurred (i.e., under high WML in low WMC individuals). The inverted-U pattern of distractibility effects in Experiment 1, replicated in Experiment 2, can thus be explained by differences in the spread of the attentional window as a function of WM resource availability.

Conclusions/significance: The current findings show that limitations in WM resources, due to either WML or individual differences in WMC, affect the spatial distribution of attention. The difference in attentional constraining between high and low WMC individuals demonstrated in the current experiments helps characterise the nature of previously established associations between WMC and controlled attention.

Figure 1

Experiment 1 trial sequence and displays. A) Sample trial sequence for a congruent (left) and incongruent (right) trial, under high load and low WML conditions respectively. B) Sample of the letter identification display in Experiment 1. C) Sample of the letter identification display in Experiment 2. Dashes represent possible target and distractor positions and were not displayed in the experiment.

Figure 2

Experiment 1 RT congruency effects graph. Mean RTs congruency effect under low and high WML conditions, for the low and high WMC groups in Experiment 1. Error bars are standard errors.

Figure 3. The proposed modulation of the Mexican-hat distribution as a function of WMC and WML.

The schematic representation illustrates the proposed dispersion of the Mexican-hat profile as a function of cognitive limitations and also explains the inverted-U pattern of congruency effects recorded at TD 4 in Experiment 1 & 2.

Figure 4

Experiment 2 RT congruency effects graph. Mean congruency as a function of target-to-distractor distance in A) High and B) Low WMC groups under High and Low WML. Note that the typical Mexican-hat profile is evident in High WMC under Low Load, with a relatively strong congruency effect at distance d1 (first attention zone, a1), followed by weaker congruency at distance d2 (first suppression zone, s1), stronger congruency at distance d3 (second attention zone, a2), and finally weaker congruency at distance d4 (peripheral suppression or unattended zone, s2).