Dopamine D2 agonist affects visuospatial working memory distractor interference depending on individual differences in baseline working memory span

Abstract

The interplay of dopaminergic striatal D1-D2 circuits is thought to support working memory (WM) by selectively filtering information that is to be remembered versus information to be ignored. To test this theory, we conducted an experiment in which healthy participants performed a visuospatial working memory (VSWM) task after ingesting the D2-receptor agonist cabergoline (or placebo), in a randomized, double-blinded, crossover design. Results showed greater interference from distractors under cabergoline, particularly for individuals with higher baseline dopamine (indicated by WM span). These findings support computational theories of striatal D1-D2 function during WM encoding and distractor-filtering, and provide new evidence for interactive cortico-striatal systems that support VSWM capacity and their dependence on WM span.

Keywords: Agonist; Basal ganglia; Cabergoline; Capacity; D2; Dopamine; Individual differences; Prefrontal cortex; Striatum; Visuospatial working memory; Working memory.

Fig. 1

a Trial events and temporal parameters of the visuospatial working memory (VSWM) task. Participants responded whether the square currently occupied by the probe (?) had been occupied by a red dot (in the preceding visual array). Participants were instructed to ignore yellow dots. Visual arrays showed either 3 targets (Condition 3), 3 targets plus two distractors (Condition 3+2; depicted in Fig.), or 5 targets (Condition 5). See main text for additional details. b Hypothetical inverted-U function relating baseline DA to distractor-filtering ability. It was predicted that cabergoline would push individuals with higher working memory (WM) span out of an optimal position on the curve, resulting in a greater VSWM capacity decrement for these individuals due to the presence of distractors, compared to placebo. In contrast, cabergoline would push individuals with lower WM span into the optimal position, resulting in a smaller VSWM capacity decrement for these individuals due to the presence of distractors, compared to placebo

Fig. 2

Results for visuospatial working memory (VSWM) performance. a K for low (left) and high (right) Ospan groups across Task × Drug conditions (placebo = blue, cabergoline = red). 3, 3 + 2, and 5 refer to task conditions (respectively, 3 targets, 3 targets plus 2 distractors, 5 targets). Error bars represent one standard error above and below the mean. The Condition × Drug × Ospan interaction was significant in ANCOVA, p = .02 (normalized Ospan as continuous covariate). Pairwise tests were not significant, p > .05. b KΔ, the VSWM capacity decrement due to the presence of distractors, under placebo (blue) or cabergoline (red), separately for low and high Ospan groups. The hypothetical inverted-U function relating baseline DA to distractor-filtering ability is superimposed. Error bars represent one standard error above and below the mean. The Drug × Ospan interaction was significant in ANCOVA, p < .01 (z-score normalized Ospan as continuous covariate). High/low Ospan groups formed by median-split; (Mdn = 68; high span N = 13, low span N = 12). (Color figure online)

Fig. 3

Relationships between Ospan and KΔ, the visuospatial working memory (VSWM) capacity decrement due to the presence of distractors, under placebo (a), and cabergoline (b). Ospan was not significantly correlated with KΔ under placebo, r(23) = .357, p = .080. Ospan was significantly negatively correlated with KΔ cabergoline, r(23) = −.502, p = .011. These correlations were significantly different from each other, Z = −2.657, p < .05, 95% CI for the difference (−1.608, −.243) (Meng et al., 1992). c Relationship between Ospan and KΔDD, the drug difference (cabergoline minus placebo) for the VSWM capacity decrement (KΔ) due to the presence of distractors. Scores below the zero point (negative KΔDD) indicate the participant was more impaired at filtering distractors under cabergoline versus placebo. KΔDD was significantly negatively correlated with Ospan, r(23) = −.543, p = .005

Fig. 4

Relationships between Ospan and K in the four Task × Drug conditions involved in the computation of KΔDD. Ospan was significantly positively correlated with K in Condition 3 cabergoline, r(23) = .451, p < .05. Other relationships shown here were not significant (see Table 2)

Fig. 5

Relationships between KΔDD, the drug effect on the visuospatial working memory (VSWM) capacity decrement due to the presence of distractors, and FA-in errors under cabergoline (a) and under placebo (b). Data were restricted to individuals who made at least one FA-in error under placebo or cabergoline (N = 20). FA-in error rates under cabergoline were significantly negatively correlated with KΔDD, r(18) = −.535, p = .015 (a). FA-in errors under placebo were not significantly correlated with KΔDD, r(18) = .363, p = .116 (b). These correlations were significantly different from each other, Z = −3.224, p < .05, 95% CI for the difference (−1.571, −.382) (Meng et al., 1992)