Working memory (WM) refers to the ability to maintain and manipulate information ‘on-line’ in the face of disruptions such as eye movements. Examples include remembering the number you looked up to dial or the location of your coffee mug while you continue to look at your computer. WM is studied using the full experimental toolkit including neuroimaging, investigations in participants with brain lesions, brain stimulation and behavioral tasks in normal participants. Advances in molecular genetics now make it practicable to study the underlying mechanisms of WM by looking at an individual participant's genotype. In WM the focus has been on several genes that modulate the dopamine concentration. Successful WM is believed to depend on an optimal dopamine concentration and too much or too little dopamine is considered to be detrimental to executive function (reviewed in 
. Here we investigated the effects on WM of two genes that affect dopamine activity through two single nucleotide polymorphisms that are common in the general population.
One well studied genetic polymorphism codes for two versions of the catechol-O-methyltransferase (COMT) enzyme. In the prefrontal cortex (PFC), COMT is the primary enzyme that breaks down dopamine and other catecholamines 
. There is a common single nucleotide polymorphism in COMT that replaces a valine with a methionine (Val158
Met, rs4680). The rate of COMT enzymatic activity is reduced by a factor of four in the Met/Met homozygote population 
. In other words, the efficient COMT enzyme (Val/Val) breaks down dopamine quickly leaving little dopamine in the synapse whereas the less efficient COMT enzyme (Met/Met) leaves dopamine in the synapse over a longer period of time. Behavioral findings suggest that Met/Met homozygotes perform better on a number of executive function tasks including the Wisconsin Card-Sorting Task 
; reviewed in 
, Complex Working Memory Span 
, and n-back WM tasks 
; see also reviews in 
. Furthermore, differences become more apparent with age 
. A recent meta-analysis of twenty relevant neuroimaging studies clarified the link between COMT genotype, prefrontal dopamine and cognitive task performance 
. Across these studies the authors observed a consistent relationship (effect size of.73) between prefrontal activation and COMT genotype. However, other reports are inconsistent with these findings. For example, a recent study with 86 participants failed to find any effect of COMT in a change blindness WM study 
. In a second study, Bruder and colleagues tested 402 participants in four WM tasks (n-back, serial position, spatial delayed response, letter-number sequencing) and only found a Met/Met benefit for letter-number sequencing 
. Finally, a recent meta-analysis evaluating a series of cognitive tasks and COMT genotype observed no consistent relationship between performance and genotype 
. These inconsistent findings point towards an incomplete understanding of COMT effects on cognitive performance that is compounded by task differences, the need for large numbers of participants, and perhaps most importantly by unknown interactions with other genes across multiple brain regions.
There is a parallel literature investigating polymorphisms influencing striatal dopamine. There are strong frontostriatal connections and evidence supporting a role of the striatum in higher cognition, including in WM 
. Although the striatum is classically associated with habit learning it has also been associated with updating the contents of WM 
. Indeed, an individual's WM capacity predicts striatal dopamine synthesis 
. Furthermore, there is experimental evidence to suggest that WM requires the basal ganglia for gating what enters WM and the prefrontal cortex for WM maintenance 
, which is in accord with computational models 
; see also 
. In the striatum, D2 receptors are the most common dopamine receptor 
. The density of D2 dopamine receptors in the striatum is influenced by polymorphisms in the DRD2/ANKK1-Taq-Ia fragment 
, also referred to as the ANKK1 polymorphism. The presence of a single copy of the A1 allele (A1+) is associated with a 30–40% reduction in D2 receptor density 
; but see 
and reduced cognitive performance when compared to participants lacking this polymorphism (A1−). Carriers of the A1 allele perform worse on the California Verbal Learning Test of memory 
and other cognitive tasks (reviewed in 
Thus, because dopaminergic frontostriatal pathways modulate WM performance there is good reason to investigate COMT and DRD2/ANKK1-TAQ-Ia simultaneously. The COMT Val158
Met polymorphism dictates dopamine concentration in the PFC but not in the striatum 
. Likewise, there are few D2 receptors in the PFC but many in the striatum 
. One recent study explored both COMT and DRD2 effects on a series of WM tasks 
. Stelzel et al. (2009) reported that Met/Met homozygotes performed better across WM tasks, but only when they were A1−. In short, both a slow acting dopamine-catabolizing enzyme and a high concentration of dopamine receptors were associated with good WM performance 
Met and the DRD2/Ankk1-Taq-Ia polymorphisms may interact to produce differential phenotypes at the behavioral level.
In addition to WM load and genotype we were also interested in the relative contributions of COMT and DRD2/ANKK1-Taq-Ia polymorphisms with regard to interval timing. Interval timing refers to the ability to discriminate between different temporal durations. Recently, Wiener and colleagues used molecular genetics to reveal two timing circuits 
. Participants were asked to discriminate short (500 ms) and long (2000 ms) time intervals. Response time variability increased at short intervals for the A1+ DRD2/ANKK1-Taq-Ia group and at the long intervals for the COMT Val+ group. The conclusion was that separate temporal mechanisms in the striatum and the PFC are optimized for short and long intervals, respectively. However, one unresolved issue from these findings is the cause of the disruption in timing performance. One untested possibility was that the differential response for Val+ carriers for longer durations might relate to WM load rather than timing per se. The effects for longer stimuli could simply reflect a WM deficit rather than the assumed effect of stimulus duration.
To explore these issues we investigated the effects of COMT and DRD2/ANKK1-TAQ-Ia polymorphisms on WM performance in healthy adults. In addition, we included maintenance-delay and set-size manipulations to investigate differential striatal and PFC involvement for short (1000 ms) or long (5000 ms) delays with small (4-element) or large (6-element) WM maintenance requirements. With regard to COMT, we predicted that the Met/Met homozygotes would perform significantly better than the Met/Val or Val/Val groups. These predictions were based on previous findings generally reporting superior WM performance in Met/Met participants (reviewed in 
. Based on the work of Stelzel and colleagues (2009), we further predicted that in DRD2/ANKK1-TAQ-Ia A1−, COMT Met/Met participants, we would see better WM performance when compared to all other groups. We also investigated whether varying the WM maintenance delay would interact with participants' genotypes as demonstrated by Wiener and colleagues (2011)
. We had two a priori predictions. First, we predicted that COMT Val+ carriers would be disproportionately impaired at higher WM demands: longer delays and larger set sizes. Second, we predicted that the A1+ DRD2/ANKK1-TAQ-Ia carriers would be disproportionately impaired at shorter delays. Finally, set size was manipulated to avoid WM floor effects in participants with high WM capacity.