Caffeine modulates attention network function

Abstract

The present work investigated the effects of caffeine (0 mg, 100 mg, 200 mg, 400 mg) on a flanker task designed to test Posner’s three visual attention network functions: alerting, orienting, and executive control [Posner, M. I. (2004). Cognitive neuroscience of attention. New York, NY: Guilford Press]. In a placebo-controlled, double-blind study using a repeated-measures design, we found that the effects of caffeine on visual attention vary as a function of dose and the attention network under examination. Caffeine improved alerting and executive control function in a dose–response manner, asymptoting at 200 mg; this effect is congruent with caffeine’s adenosine-mediated effects on dopamine-rich areas of brain, and the involvement of these areas in alerting and the executive control of visual attention. Higher doses of caffeine also led to a marginally less efficient allocation of visual attention towards cued regions during task performance (i.e., orienting). Taken together, results of this study demonstrate that caffeine has differential effects on visual attention networks as a function of dose, and such effects have implications for hypothesized interactions of caffeine, adenosine and dopamine in brain areas mediating visual attention.

Introduction

Caffeine (1,3,7-trimethylxanthine) is the most widely consumed psychoactive stimulant in the world, found naturally in many foods and beverages, and often cited for its positive effects on vigilance and mental alertness (for reviews, see IOM, 2001, Koelega, 1993, Lieberman, 1992, Lieberman, 2001, Smith, 2002, Snel et al., 2004, Spiller, 1997). Improvements in these processes have been commonly attributed to caffeine’s antagonistic role at adenosine A₁ and A_2A receptors in dopamine-rich brain areas, ultimately stimulating dopaminergic activity and resulting in increased wakefulness and pronounced motor activity (i.e., Garrett and Griffiths, 1997, Popoli et al., 1998, Solinas et al., 2002). Indeed, many studies have demonstrated that caffeine reduces response times and error rates in tasks such as simple reaction time (Wesensten, Killgore, & Balkin, 2005), choice reaction time (Kenemans and Lorist, 1995, Lieberman et al., 2002), and visual vigilance (Fine et al., 1994, Lieberman et al., 2002). Further work suggests that caffeine may have positive influences on relatively higher-order processes such as visual selective attention (Kenemans et al., 1999, Lorist and Snel, 1997, Lorist et al., 1996, Ruijter et al., 2000), task switching (Tieges et al., 2007, Tieges et al., 2006), conflict monitoring (Tieges, Ridderinkhof, Snel, & Kok, 2004), and response inhibition (Barry et al., 2007).

Other work, however, suggests that whereas caffeine may improve overall processing speed on tasks requiring higher-order function, these improvements cannot be attributed to specific effects on response inhibition or selective visual attention (Kenemans and Verbaten, 1998, Lorist and Snel, 1997, Tieges et al., 2009). To further elucidate the locus of caffeine effects on lower- versus higher-level visual attention, we examined whether caffeine differentially affects the function of three visual attention networks in a dose–response paradigm. Specifically, we used the Attention Network Test (Fan, McCandliss, Sommer, Raz, & Posner, 2002), which is a modified flanker task that allows examination of the relative functioning of alerting, orienting and executive control networks (i.e., Posner, 1990) in a single unitary visual attention task. Our intention was to examine whether caffeine consumption ranging from 0 mg to 400 mg differentially affects lower- versus higher-order attention network functioning, and how these effects might be modulated by dose.

Caffeine is a psychoactive stimulant that is abundantly available in both natural (e.g., coffee, tea, chocolate) and supplemented (e.g., soft drinks, energy bars) food and beverages, as well as over-the-counter remedies for migraines, colds, and fatigue (Gilbert et al., 1976, James, 1991). Some studies estimate that over 80% of US adults and children habitually consume moderate daily amounts of caffeine (estimates range from 193–280 mg/day average; Barone and Roberts, 1996, Frary et al., 2005), likely due to its properties as a mild psychostimulant (Childs & de Wit, 2006). Peak plasma concentrations of caffeine occur in as few as 15 min and on average approximately 45 min after ingestion (Arnaud, 1987, Smith, 2002). A number of studies suggest that the most behaviorally-relevant role of caffeine is in blocking the inhibitory properties of endogenous adenosine (particularly at A₁ and A_2A receptors), resulting in increased dopamine, norepinephrine and glutamate release (e.g., Ferre et al., 1997, Fredholm et al., 1997, Smits et al., 1987). The effects of caffeine on physiological functions are thought to result from interactions with both adenosine and phosphodiesterase, resulting in cardiostimulatory and antiasthmatic actions (Davis et al., 2003, IOM, 2001). The result of higher dopamine and glutamate concentrations, coupled with phosphodiesterase inhibition, is a net increase in central nervous system and cardiovascular activity. In addition to affecting cognitive performance, caffeine increases perception of alertness and wakefulness (Leatherwood and Pollet, 1982, Rusted, 1999) and sometimes anxiety (particularly at high doses; Lieberman, 1992, Loke et al., 1985, Sicard et al., 1996).

Presumably as a direct result of altered CNS activity, caffeine appears to result in performance improvements on a variety of basic psychomotor tasks. For instance, performance on simple and choice reaction time tasks is faster and accuracy improves as a function of increasing doses (Kenemans and Lorist, 1995, Lieberman et al., 1987, Lieberman et al., 2002, Wesensten et al., 2005); other work suggests that these advantages diminish with very high doses of caffeine (e.g., 600 mg; Roache & Griffiths, 1987). Extended vigilance is also generally improved following caffeine consumption (Frewer and Lader, 1991, Lieberman et al., 1987, Mitchell and Redman, 1992). More recently, research has begun to examine the mechanisms responsible for these performance advantages. Lorist and Snel (1997) found that caffeine reduces stimulus evaluation times as reflected in the timing of electroencephalography (EEG) components (see also Lorist, Snel, Kok, & Mulder, 1996). Further work suggests that caffeine can shorten motor readiness potentials as measured by EEG during ergometer exercise (Barthel et al., 2001). Basic psychomotor tasks thus appear to be improved by more efficient stimulus feature analysis (i.e., Treisman & Gelade, 1980) and shorter-duration readiness potentials, leading to decreased overall response times; these effects also appear to be generally greater at higher doses.

More recently, research has identified some higher-order cognitive processes that caffeine appears to affect. In general, higher-order processes are those considered to be involved in the active monitoring, guidance, and coordination of behavior (Miller & Cohen, 2001). Tieges and colleagues have recently demonstrated that caffeine can reduce response time costs during task switching (Tieges et al., 2006, Tieges et al., 2007), and strengthen action monitoring (Tieges, Ridderinkhof, Snel, & Kok, 2004). Another component of higher-order cognitive function is inhibitory control, generally defined as the ability to inhibit inappropriate impulses and actions, and reduce the influence of interfering (and often action-incompatible) information (Shallice & Burgess, 1993). Work investigating inhibitory control suggests that caffeine can reduce interference costs during selective visual attention tasks (Lorist et al., 1994, Lorist et al., 1996) and the Stroop color-word task (Hasenfratz and Battig, 1992, Kenemans et al., 1999; but for contradictory results, see Foreman, Barraclough, Moore, Mehta, & Madon, 1989). Other work, however, suggests that caffeine (in a 3 mg/kg dose) does not significantly reduce interference on a variety of inhibitory tasks, including a cued go/no-go paradigm, a stop-signal task, and a flanker task (Kenemans and Verbaten, 1998, Tieges et al., 2009). Thus, results are mixed with regard to caffeine’s effects on higher-order control processes.

Several methodological characteristics might account for such contradictory results. First, as noted by Treisman and Gelade, 1980, Tieges et al., 2009, the caffeine doses used in previous work may not have been large enough to elicit effects on inhibitory control (p. 325). Indeed a dose approximating 200 mg may not be sufficiently high to produce changes in individuals who habitually drink 2–4 cups of coffee per day (i.e., 170–340 mg) (i.e., Kenemans and Verbaten, 1998, Tieges et al., 2009). Further, habitual coffee drinkers may be affected by both withdrawal effects and caffeine response (see James, 1994, Juliano and Griffiths, 2004), and the use of a predominantly female sample (89% female; Tieges et al., 2009) in the flanker task may limit the chances of finding caffeine effects on cognitive performance (i.e., females may be less prone to the effects of caffeine on cognitive performance; Gupta & Gupta, 1999).

As noted by Tieges et al. (2009), there is convincing evidence, however, to expect that caffeine might modulate the inhibitory control of attention, particularly on visual selective attention tasks. First, some research has demonstrated that caffeine improves conflict resolution in the classic Stroop task, which involves resolving a visual conflict between a word name and its color (i.e., Hasenfratz and Battig, 1992, Kenemans et al., 1999). Second, meta-analyses of brain activation during the Stroop task reveal a network including the anterior cingulate cortex and a number of regions in the prefrontal cortex (Bush et al., 2000, Bush et al., 1998); similar results are found with performance monitoring and conflict resolution during flanker tasks, most often implicating the anterior cingulate cortex (ACC; Botvinick et al., 2001, Casey et al., 2000, Fan, Flombaum, et al., 2003, Mac Donald et al., 2000). Given fMRI evidence that the ACC is up-regulated by caffeine, one might expect facilitation of conflict resolution either in the form of reduced response times or reduced error rates during flanker tasks (Koppelstaetter et al., 2008).

In support of this position, it should be noted that the ACC has dense dopaminergic innervation (Lumme, Aalto, Ilonen, Någren, & Hietala, 2007) and dopamine binding in this region drives executive function (Ko et al., 2009). These findings suggest a potential role of increased dopamine availability as a result of caffeine consumption in brain regions mediating executive control, and that the result of such a process may be enhanced monitoring and conflict resolution.

To further examine the locus of caffeine’s effects on lower- and higher-order visual attention, we conducted a double-blind, within-participant repeated-measures design with four levels of our treatment variable (0 mg, 100 mg, 200 mg, 400 mg caffeine). We assessed how caffeine affects visual attention in non-habitual consumers by using the Attention Network Test (ANT; Fan, McCandliss, Sommer, Raz, & Posner, 2002). The ANT simultaneously tests the individual performance of the three networks in Posner’s (1990) attention model by combining cued reaction time (Posner, 1980) and flanker tasks (Eriksen & Eriksen, 1974). Posner’s three attention networks involve alerting, orienting, and executive attention.

The alerting network allows an individual to achieve and maintain a state of alertness during task performance by using predictive cues about trial onset. Alerting cues have been found to activate the thalamus and right and left frontal and parietal brain regions, similar to results found with vigilance and sustained attention tasks (Coull et al., 1996, Fan et al., 2005, Marrocco and Davidson, 1998, Posner and Petersen, 1990). Given the dense dopaminergic innervation of the human thalamus and prefrontal cortex (García-Cabezas et al., 2007, Sawaguchi and Goldman-Rakic, 1991, Sawaguchi and Goldman-Rakic, 1994, Sánchez-González et al., 2005, Williams and Goldman-Rakic, 1995), and that caffeine is generally found to improve simple reaction times on several simple psychomotor tasks, we expected similar effects in a positive dose–response relationship. Specifically, we expect that the advantage of cued versus non-cued trials in the ANT would increase as a function of higher caffeine dose.

The orienting network allows an individual to selectively attend to regions of space by directing attention to cued areas. Orienting attention (either covertly or overtly) towards particular regions of space has been found to activate the superior parietal lobe (Corbetta et al., 2000, Fan et al., 2005). No work to date has specifically investigated caffeine’s effects on the orienting function of visual attention. However, we hypothesize that given work demonstrating relatively sparse dopaminergic innervations of the parietal lobes, the orienting network may not be specifically affected by caffeine consumption (i.e., Lidow et al., 1989, Tassin et al., 1978). As such, we do not expect that caffeine will differentially affect people’s ability to take advantage of spatial cues that orient them towards particular areas of space.

The executive attention network allows an individual to resolve a conflict among potential responses to a presented stimulus. As with the Stroop task, resolving conflict during the flanker task generally activates the anterior cingulate and lateral prefrontal cortices (Botvinick et al., 2001, Bush et al., 2000, Casey et al., 2000, Fan et al., 2005, Mac Donald et al., 2000). Given the dense dopaminergic innervation of these areas (as reviewed above), and that caffeine could reasonably be expected to enhance the executive control of attention, we expect caffeine may modulate executive control by improving conflict resolution during flanker tasks, particularly with higher doses of caffeine. This hypothesis does not run specifically counter to the results of Tieges et al. (2009); indeed it is possible that lower doses may not produce significant effects on conflict resolution. We do expect, however, that a higher dose (i.e., 400 mg) may enhance performance. Specifically, higher doses of caffeine may diminish the cost of presenting action-incompatible relative to action-compatible flankers in the ANT.