Recent behavioral modeling and pupillometry studies suggest that neuromodulatory arousal systems play a role in regulating decision formation but neurophysiological support for these observations is lacking. We employed a randomized, double-blinded, placebo-controlled, crossover design to probe the impact of pharmacological enhancement of catecholamine levels on perceptual decision-making. Catecholamine levels were manipulated using the clinically relevant drugs methylphenidate and atomoxetine, and their effects were compared with those of citalopram and placebo. Participants performed a classic EEG oddball paradigm that elicits the P3b, a centro-parietal potential that has been shown to trace evidence accumulation, under each of the four drug conditions. We found that methylphenidate and atomoxetine administration shortened RTs to the oddball targets. The neural basis of this behavioral effect was an earlier P3b peak latency, driven specifically by an increase in its buildup rate without any change in its time of onset or peak amplitude. This study provides neurophysiological evidence for the catecholaminergic enhancement of a discrete aspect of human decision-making, that is, evidence accumulation. Our results also support theoretical accounts suggesting that catecholamines may enhance cognition via increases in neural gain.
An important line of research in neuroscience aims at understanding the brain mechanisms allowing us to make reliable perceptual judgments based on noisy sensory information. Convergent data from psychophysics, computational modeling, and neurophysiology highlight the central role of “decision variable” signals in determining the timing and accuracy of perceptual reports by accumulating sensory evidence over time up to an action-triggering threshold (Kelly & O'Connell, 2015; Gold & Shadlen, 2007; Smith & Ratcliff, 2004; Smith & Vickers, 1988). Such signals have been identified both in the spiking activity of neuronal subpopulations within several sensorimotor regions of the monkey brain (Ding & Gold, 2010; Hanes & Schall, 1996) and in signals recorded noninvasively from the human brain (Newman, Loughnane, Kelly, O'Connell, & Bellgrove, 2017; Loughnane et al., 2016; Philiastides, Heekeren, & Sajda, 2014; de Lange, Rahnev, Donner, & Lau, 2013; Kelly & O'Connell, 2013; O'Connell, Dockree, & Kelly, 2012; Donner, Siegel, Fries, & Engel, 2009).
The biophysical mechanisms supporting these accumulation-to-bound dynamics remain to be fully characterized, and there is growing interest in the role of neuromodulatory arousal systems (Steinemann, O'Connell, & Kelly, 2018; Twomey, Murphy, Kelly, & O'Connell, 2015; O'Connell et al., 2012). Specifically, psychopharmacological studies in which catecholaminergic (i.e., dopamine and noradrenaline) function has been modulated demonstrate improved behavior on perceptual tasks (Steinemann et al., 2018; Twomey et al., 2015; O'Connell et al., 2012), possibly mediated by an increase in neural gain (Linssen, Sambeth, Vuurman, & Riedel, 2014).
However, our understanding of the parameters of perceptual decision-making impacted by catecholamine modulation remains limited. According to the computational modeling of behavior, pharmacological enhancement in the speed of response to stimuli could arise from several sources such as the onset or buildup rate of evidence accumulation, or a lowered decision bound (Fosco, White, & Hawk, 2017; Ratcliff & McKoon, 2008). On the other hand, neurophysiological work has demonstrated that differences in evidence accumulation rate and/or onset latency of the P3b impact the speed of perceptual reports (Loughnane et al., 2016; O'Connell et al., 2012). Here, we utilized a pharmacological–EEG approach and sought to adjudicate between these different computational- and neurophysiological-based predictions to establish which of these various signal components best accounted for pharmacologically induced improvements in RT.
We traced decision formation via the P3b component of the human ERP. In recent studies, we demonstrated that the P3b exhibits the key characteristics of a build-to-threshold evidence accumulation process. Like the decision signals that have been observed in intracranial recordings, the P3b builds gradually during decision formation, at a rate that scales with the difficulty of the perceptual decision and reaches a stereotyped amplitude immediately before response execution (Kelly & O'Connell, 2013; O'Connell et al., 2012). Previous work has demonstrated that P3b onset latency, buildup rate, and amplitude predict independent variance in RT (Harty, Murphy, Robertson, & O'Connell, 2017; Newman et al., 2017; Loughnane et al., 2016; Murphy, Robertson, Harty, & O'Connell, 2015; Kelly & O'Connell, 2013; O'Connell et al., 2012) Moreover, P3b amplitude varies in a manner consistent with adjustments in decision bounds (Steinemann et al., 2018). These dynamics have been observed during extended motion discrimination and contrast change detection judgments but are also apparent during performance of the classic oddball task (Twomey et al., 2015).
We probed catecholamine function using methylphenidate (MPH) and atomoxetine (ATM). MPH and ATM increase monoamine levels to varying extents in different brain regions. In pFC, the action of both MPH and ATM converges to increase levels of both dopamine and noradrenaline, initially via reuptake inhibition of the noradrenaline transporter (Berridge et al., 2006; Han & Gu, 2006; Bymaster et al., 2002). Within subcortical regions such as the striatum, however, their action diverges with only MPH inhibiting reuptake via the dopamine transporter (Volkow, Fowler, Wang, Ding, & Gatley, 2017; Volkow et al., 2001).
We employed a randomized, double-blinded, placebo-controlled, crossover design, manipulating monoamine levels using the clinically relevant drugs MPH and ATM and the serotonin reuptake inhibitor citalopram (CIT), in comparison with placebo (PLA). Given the paucity of evidence linking serotonin to perceptual decision-making, we made no firm predictions regarding its effect on behavior or EEG. Rather, the use of CIT acted here to establish the specificity of any catecholamine modulation. Participants performed an EEG oddball paradigm to elicit a neural correlate of this evidence accumulation process, the P3b, under each of the four drug conditions. We hypothesized first that MPH and ATM would improve RTs relative to PLA. Second, we hypothesized that enhanced perceptual decision-making with MPH and ATM would be underpinned by a decrease in the P3b peak latency, driven via an increase in the rate of evidence accumulation. Third, to confirm the specificity of catecholaminergic effects on perceptual decision formation and not stimulus processing and visual attention processes that temporally precede the evidence accumulation process, we examined whether the visual evoked potential (VEP) was modulated by our pharmacological manipulations. Given the lack of evidence in past work for effects on VEP components such as P1 and N1 (Linssen et al., 2014), we hypothesized that catecholamine manipulation would not affect these components.
Forty nonclinical volunteers completed a larger study investigating the influence of monoamine reuptake inhibitors on executive function (Dockree et al., 2017; Barnes, O'Connell, Nandam, Dean, & Bellgrove, 2014). Participants were recruited via advertisements at The University of Queensland, Queensland, Australia. All participants were male, White, and right-handed; had normal or corrected-to-normal vision; were 18–45 years old; and were prescreened by a consultant psychiatrist (L. S. N.) for suitability. Individuals were excluded from participation if they reported any history of psychiatric or neurological illness (including head injury resulting in unconsciousness), previous or current use of psychotropic medication, current smoking or use of nicotine products, or significant drug use (significant was defined here as [i] use of any illicit substances within the last month, [ii] >5 lifetime intake of any illicit drug except cannabis, or [iii] more than monthly cannabis intake and alcohol dependence [>24 units/week]). Before commencing their first session, all participants were screened by a consultant psychiatrist (L. S. N.). The psychiatrist also administered the Mini-International Neuropsychiatric Interview screen (Sheehan et al., 1998). Seven participants were excluded from the analysis because of illness (n = 4), technical data acquisition issues (n = 1), and behavioral outlier criteria (n = 2), during one session. All participants were recruited according to the principles of the Declaration of Helsinki and in accordance with the ethical guidelines of The University of Queensland.
Each participant attended four sessions, each at the same time of the day, spaced at least 1 week apart. At each session, a single blue gelatin capsule containing MPH (30 mg, mixed dopaminergic and noradrenergic action), ATM (60 mg, mixed dopaminergic and noradrenergic action), CIT (30 mg, primarily serotonergic action), or PLA (dextrose) was administered. Cognitive testing began 90 min after drug administration coinciding with peak plasma concentrations of the drugs (Müller et al., 2005; Sauer, Ring, & Witcher, 2005; Hennig & Netter, 2002).
Materials and Task Procedures
Participants performed a two-stimulus oddball task (Figure 1). Visual stimuli were presented on a dark gray background, and participants were instructed to fixate on the center of the screen. They were also asked to restrict any eye movements throughout the task, for example, large saccades or blinks. Every 2075 msec, a stimulus appeared on the screen for 75 msec. Standard stimuli consisted of a 2-cm-diameter purple circle, which appeared on 80% of the trials. Target stimuli were a slightly larger purple circle (4-cm diameter), which appeared on 20% of the trials. Participants were asked to make a speeded response to target stimuli using the response box placed in their right hand. The stimulus array was pseudorandomly designed such that between three and five standard stimuli were presented after any target stimulus. Thus, the minimum interval between P3-eliciting events was 8300 msec. Participants performed one block of this task, comprising 42 target trials.
Experimental Design and Statistical Analysis
This experiment incorporated a within-subject design, specifically using repeated-measures ANOVA and within-subject t tests, using the Benjamini–Hochberg procedure for controlling the false discovery rate of a family of hypothesis tests (adjusted p values reported in text). Multilevel mediation analysis (Baron & Kenny, 1986) was performed to investigate the possible mediation of the drug effects on behavior via our neural indices. The multilevel models involved in this analysis modeled between-subject variability in our measures as random intercepts between subjects. Additional analyses are described in the Data analysis and Results sections.
EEG Acquisition and Preprocessing
EEG was recorded using an ActiveTwo BioSemi system of 64 scalp electrodes, sampled at 1024 Hz. Data were preprocessed using MATLAB (MathWorks, Inc.) and the EEGLAB plug-in (Delorme & Makeig, 2004). Data were resampled to 512 Hz for ease of processing, filtered with a 20-Hz low-pass filter and a 0.25-Hz high-pass filter and rereferenced offline to the average of all scalp electrodes. Epochs were then taken using a window from −125 msec before stimulus onset to 750 msec poststimulus onset and baseline corrected to the prestimulus period. To minimize interaction between overlapping ERP components, the data were subjected to Current Source Density transformation (Kayser & Tenke, 2006). A further analysis of prestimulus alpha used an epoch of 1000 msec before target onset, and alpha power (8–14 Hz) was calculated using the fast Fourier transform of that time frame. Trials were excluded from analyses if EEG from any channel relevant to the study (i.e., frontal channels and central parietal channels) exceeded ±100 μV during the interval between 100 msec before target onset and the time of manual response. Critically, there were no significant differences in trial count across drug conditions, F(3, 96) = 0.65, p = .59 (M [SD]: MPH, 36.7 [7.2]; ATM, 35.9 [5.7]; CIT, 35.6 [5.7]; PLA, 38.0 [3.5]). Finally, to ensure single participants were not unduly influencing any analyses, participants were rejected from particular analyses if their data for those analyses were outside 3 SDs of the mean.
Behavioral performance measures testing the pharmacological manipulation were (a) RTs for correct detections and (b) the coefficient of variability of RTs (RT Cvar), calculated per participant as the standard deviation of RT divided by mean RT (Bellgrove, Hester, & Garavan, 2004). Hit rate was at ceiling in this task (98.7 ± 2%) and was not analyzed further. There were no false alarms for standard stimuli.
We recorded the P3b ERP component in response to the targets as our neurophysiological measure of the evidence accumulation process. This allowed us to parse four separate aspects of decision formation, which could theoretically explain the faster RTs: (i) P3b onset representing the start time of evidence accumulation, (ii) its buildup rate reflecting the rate of evidence accumulation, (iii) its amplitude at response execution representing the decision threshold, and (iv) its peak latency representing the time at which the threshold is reached (Twomey et al., 2015; O'Connell et al., 2012).
The P3b was measured from a parietal channel selected on a participant-by-participant basis, from one of nine channels centered on CPz (the peak grand-averaged channel using the 10–20 coordinate system) as the channel with the greatest amplitude at manual response. Target-locked P3b epochs were −125 to 750 msec around target onset, and response-aligned P3b epochs were −250 to 0 msec around response.
We measured P3b onset via two methods. First, we calculated a running t test against zero across time on the grand-averaged P3b waveform (collapsed across subjects and drug conditions) and identified the first time point at which signal amplitude differed significantly from zero in a positive direction, at p < .05. We then measured P3b onset as the amplitude from a 50-msec window around this time point. Second, we calculated P3b onset using a jackknife-based analysis (Miller, Patterson, & Ulrich, 1998), which calculates the onset of a signal as the intersection of two lines, one fit to a defined preonset period (0–50 msec poststimulus onset) and forced to have a slope of zero and the other fit to a postonset period (50 msec to time point of maximum amplitude of the signal). The 50-msec cutoff between preonset and postonset was chosen as the earliest possible onset of the P3b signal.
Buildup rate of evidence accumulation was calculated as the slope of a line fitted to the response-aligned P3b waveform from −150 to 0 msec relative to response (Twomey et al., 2015). The 150-msec window was chosen as one that captures the entire buildup of evidence up until manual response. P3b amplitude was measured at a participant level in a 25-msec window before response execution.
We measured the timing of stimulus-aligned P3b peak (latency) as the latency of peak amplitude in the trial-averaged stimulus-locked waveform, from grand-averaged P3b onset (100 msec) onward.
VEP components, the P1, N1, and P2, were analyzed. Here, to maximize trial counts, both standard and target epochs were included. Again, relevant electrodes for each component were chosen on a per-participant basis, with the peak electrode chosen from a group of electrodes centered around the group average peak for that component.
To examine the effect of drug on the prestimulus power spectra, we performed a Fourier transform on epochs from −1000 to 0 msec before stimulus onset. We hypothesized that prestimulus occipital alpha power (8–14 Hz) may be affected by drugs (Dockree et al., 2017). We performed an additional exploratory analysis with Drug and Region (frontal, central, parietal, and occipital) as factors for the other frequency bands—theta (3–7 Hz), beta (15–30 Hz), and gamma (30–100 Hz).
Participants completed four sessions (MPH, ATM, CIT, or PLA) of the two-stimulus oddball task, monitoring for a slightly larger-diameter target circle pseudorandomly appearing among a train of smaller standard circles (Figure 1A). Participants were faster to respond to the targets in the MPH and ATM conditions compared with the PLA condition (Figure 1B), whereas there was no significant difference in RTs between CIT and PLA (main effect of drug: F(3, 96) = 7.14, p = .0002, partial η2 = .18; MPH vs. PLA: t(32) = −2.63, p = .02, 95% CI [−40.26, −5.09]; ATM vs. PLA: t(32) = −3.03, p = .01, 95% CI [−37.76, −7.4]; CIT vs. PLA: t(32) = 0.35, p = .72). RT CVar was also significantly lower in the MPH condition compared with the PLA condition, with a similar trend in the ATM condition (Figure 1B; main effect of drug: F(3, 93) = 3.78, p = .013; MPH vs. PLA: t(31) = −2.7, p = .03; ATM vs. PLA: t(31) = −2.12, p = .06; CIT vs. PLA: t(31) = −0.49, p = .63).
Subjective side effects of the drug were measured using a visual analog scale along axes of alertness, contentedness, and calmness (Norris, 1971). Participants completed this visual analog scale three times during each session: immediately before drug administration (Time 1), +90 min (immediately before cognitive testing, Time 2), and +180 min (at the end of testing, Time 3). Repeated-measures ANOVAs were performed with Drug and Time as conditions. These revealed a main effect of Drug on alertness, F(3, 96) = 5.11, p = 2.53e-03, whereby alertness during MPH administration was greater than that in PLA, t(32) = −3.14, p = .003, and not significant for ATM or CIT (p > .3). A main effect of Time on alertness was present with decreasing alertness across time, F(2, 64) = 23.88, p = 1.79e-08. There was also a Time × Drug interaction, F(6, 192) = 4.9, p = 1.08e-04, such that alertness for MPH was not significantly different from PLA at Time 1, t(32) = 0.28, p = .77, but was so at Time 2, t(32) = −2.9, p = .006, and Time 3, t(32) = −3.56, p = .001. There were no significant differences in alertness for ATM and CIT across any time. Thus, it appeared that MPH staved off the time-on-task decrement in alertness that is typically seen on cognitive tasks compared with the other drugs (Dockree et al., 2017). An effect of Time was also found for measures of contentedness, F(2, 64) = 21.15, p = 8.87e-08, and calmness, F(2, 64) = 4.04, p = .02, with ratings for both decreasing with time. There was, however, no interaction of Drug and Time for ratings of contentedness or calmness.
We found no difference in P3b onset across pharmacological conditions using either the amplitude of a 50-msec window centered on the grand-averaged peak, F(3, 96) = 0.88, p = .46, nor the jackknife method, F(3, 78) = 0.57, p = .64 (Figure 2A; Miller et al., 1998).
We found an effect of drug condition on P3b buildup rate, F(3, 93) = 2.63, p = .05, partial η2 = .08 (Figure 2A and B), driven by a steeper P3b buildup in the ATM condition, t(31) = 2.83, p = .02, 95% CI [0.01, 0.06], and a similar trend in the MPH condition, t(31) = 1.9, p = .1, 95% CI [0, 0.06]. On the other hand, there was no effect of CIT on P3b buildup, t(31) = 0.4, p = .69. To verify that the effect of drug on behavior was related to the effect of drug on P3b buildup, we performed a multilevel mediation analysis whereby we tested the hypothesis that the effect of drug on RT was mediated by the effect of drug on P3b buildup. We found that the effect of drug on RT was indeed partially mediated by the effect of drug on P3b buildup (Figure 2C).
A similar effect of drug condition was observed on the stimulus-locked peak latency of the P3b, F(3, 90) = 3.22, p = .026, partial η2 = .10 (Figure 2A and B). This was driven by a shorter latency for both MPH and ATM, t(30) = −2.23, p = .05, 95% CI [−52.08, −53.05], and t(30) = −2.26, p = .05, 95% CI [−53.05, −2.65], respectively, compared with PLA, whereas CIT was not significantly different from PLA, t(30) = 0.19, p = .85. Similar to the analysis of P3b buildup, we performed a multilevel mediation analysis, investigating the mediation of drug effect on RT via P3b peak latency. Again, we found a partial mediation effect whereby P3b peak latency mediated the effect of drug on RT (Figure 2C).
Next, we examined the possibility that drug could impact the speed of decision-making via a lowering or raising of the bound of evidence accumulation, here represented by the peak amplitude of the response-locked P3b (Twomey et al., 2015). There was no significant effect of drug condition on response-locked P3b peak amplitudes, F(3, 93) = 0.56, p = .64.
Finally, we performed an analysis to explore any possible differences across drug condition in the relative timing of each participant's response-locked P3b peak compared with their average RT. In behavioral models of decision-making, this is reflected in a change in the nondecision time parameter along with the time between stimulus onset and evidence accumulation onset (Ratcliff & McKoon, 2008; Smith & Ratcliff, 2004). We found no difference in the relative timing of P3b peak compared with RT across drug conditions, F(3, 96) = 0.34, p = .79.
Analysis of the VEPs revealed no significant effect of drug condition on P1 or N1 amplitude. There was, however, a significant drug effect on P2 amplitude, F(3, 96) = 5.4, p = .001, partial η2 = .14, driven by a greater P2 amplitude in the MPH condition compared with the PLA condition, t(32) = 2.68, p = .03, 95% CI [0.93, 6.83]. There was, however, no significant mediation of the drug effect on RT by P2 amplitude (Figure 3).
We found no differences in prestimulus posterior alpha power (7–14 Hz) between any of the drug conditions, F(3, 96) = 0.82, p = .49 (Figure 4B).
We found a reliable decrease in theta power across the scalp in the MPH and ATM, but not CIT, conditions (main effect of drug: F(3, 96) = 5.64, p = .001). Given the inverse relationship between theta power and arousal levels (Klimesch, 1999), it is possible that this effect was because of increased arousal during the MPH and ATM conditions. Second, there was a broadband increase in beta/gamma power during the CIT condition compared with the other conditions, focused on frontotemporal areas (interaction effect of Drug × ROI: F(9, 288) = 5.66, p = 3.2e-07). The broadband nature of this effect would suggest that it could be related to muscular high-frequency artifacts (e.g., jaw clenching).
The present results provide direct neurophysiological evidence of catecholaminergic modulation of the rate of evidence accumulation during perceptual decision-making. Here, we employed a randomized, double-blinded, placebo-controlled, crossover design combined with the classic two-stimulus oddball task. We found that the introduction of MPH and ATM, but not CIT, speeded RTs to the oddball targets compared with a PLA condition. This behavioral facilitation was accompanied specifically by faster buildup rates and earlier P3b peak latencies in the MPH and ATM conditions, with no change in either the onset or peak amplitude of the P3b. This points to an effect of noradrenaline, and possibly dopamine, on the rate of evidence accumulation toward target detection in the human brain.
Catecholaminergic speeding of target detection is a well-established finding across a variety of tasks (Linssen et al., 2014; Brumaghim & Klorman, 1998; Naylor, Halliday, & Callaway, 1985). Recently, Fosco et al. (2017) used a drift diffusion model in attention deficit hyperactivity disorder (ADHD) to show that MPH increased the drift rate and nondecision time, while reducing boundary separation. This study provides neurophysiological support for one aspect of their behavioral analyses—a steeper buildup rate of the P3b by MPH, representing a faster rate of evidence accumulation toward response (Twomey et al., 2015; O'Connell et al., 2012). Fosco et al. found that MPH increased the nondecision time (the amount of RT accounted for by decision-independent processes such as sensory encoding and motor execution [Ratcliff & McKoon, 2008; Smith & Ratcliff, 2004]) component of their drift diffusion model. Given the beneficial effects of MPH on behavior, this observation was interpreted as counterintuitive. We did not find any effect of MPH or ATM on P3b onset or on the relative timing of P3b peak compared with RT; either of these effects would have been consistent with a modulation of nondecision time processes. Furthermore, Fosco et al. found decreased boundary separation in their model, an effect that could manifest itself in our task as a decrease in CPP amplitude. We propose this discrepancy is because of task differences. In contrast to our simple detection task, Fosco et al. used a two-choice discrimination task, which could have engendered a strategic need for boundary separation to discriminate accurately. Finally, the findings of Fosco et al. derive in part from individuals with ADHD and thus may not be directly comparable with the current study with neurologically healthy participants.
MPH and ATM increase levels of both dopamine and noradrenaline to varying extents in different brain regions, and it is difficult to disentangle their dopaminergic and noradrenergic effects (Berridge et al., 2006; Han & Gu, 2006; Bymaster et al., 2002). Our results are nevertheless in agreement with recent work showing an increase in the precision, or signal-to-noise ratio, of cortical representations because of catecholaminergic intervention (Warren et al., 2016). First, we show a catecholamine-related increase in the rate of evidence accumulation, which could theoretically occur because of either an increase in neural noise (and its subsequent accumulation) or a more precise representation of the target. Second, we also show a concurrent decrease in RT variability, which would imply the latter scenario of increased signal-to-noise ratio of neural representation of the target, as opposed to increased neural noise that would result in greater RT variability.
The lack of catecholaminergic effect on the early components of the VEP, namely, P1 and N1, points to a noradrenergic influence on relatively later stages of information processing. The adaptive gain theory of the locus coeruleus–noradrenaline system attributes the influence of phasic noradrenaline-related locus coeruleus responses to the translation of postthreshold activity from the decision layer to the response layer (Aston-Jones & Cohen, 2005; Nieuwenhuis, Aston-Jones, & Cohen, 2005). In our data, this would appear as a decrease between P3b peak latency and RT, which was not observed. Specifically, our results suggest an effect at the timing of the decision layer. This could occur as neural gain between the sensory and decision layers, for example, where decision-relevant neurons are more sensitive to input from sensory regions (Warren et al., 2016).
The present findings are consistent with those that have previously examined catecholaminergic effects on the P3b. Drugs that increase the amount of noradrenaline in the system have been shown to result in a faster P3b peak latency (Dockree et al., 2017; Brumaghim & Klorman, 1998; Naylor et al., 1985), whereas drugs such as clonidine that decrease noradrenaline signaling have been shown to result in a slower peak latency (Halliday et al., 1994; Swick, Pineda, & Foote, 1994). Other studies have shown an increase in P3b peak amplitude in response to MPH, which we do not confirm here (Dockree et al., 2017; Cooper et al., 2005; Brumaghim & Klorman, 1998). The absence of a drug effect on P3b amplitude in this study could arise from a ceiling effect for performance (hit rate: 98%) whereby the evidence accumulation process reliably reaches a bound regardless of pharmacological manipulation because of the ease of our task. Although our results generally agree with past studies of catecholamine modulation of target detection using EEG, our dissection of the P3b into its distinct parameters (onset, buildup rate, peak latency, and amplitude) provides unique mechanistic insight into the impact of catecholamines on the neural decision process. Such a framework may be profitably applied in future pharmacological studies to inform the pharmacology of discrete processing stages underpinning human choice behavior.
Although MPH and ATM are the most universally prescribed psychostimulants for the treatment of ADHD (Storebø et al., 2012; Garnock-Jones & Keating, 2009), we lack a clear understanding of the neurophysiological bases of their ability to enhance behavior. Such insights are critical for the identification of robust biomarkers of drug response that may ultimately facilitate personalized approaches to treatment in disorders such as ADHD. Our recent work has demonstrated the effect of MPH on attentional engagement, which was linked to a suppression of alpha-band activity as well as a reduction in its trial-to-trial variability (Dockree et al., 2017). Interestingly, we observed no such drug effects here despite the fact that our data were collected from the same participants and in the same testing session. In our view, the differences in results arise from differences in the tasks employed. The oddball task employed here is relatively easy, with performance close to ceiling, whereas the task employed by Dockree et al. was a very demanding sustained attention task designed to engender lapses of attention (hit rate was approximately 60–75%). Thus, it is plausible that the oddball task did not engender behaviorally relevant fluctuations in attentional engagement, thus leaving little scope for modulation by drug. This highlights the potential for medications such as MPH to exert task-dependent effects on neural activity and the utility of neurophysiology for dissecting these contributions.
This work was supported by grant no. 569532 from the National Health and Medical Research Council of Australia (to M. A. B.) and Young Investigator grant no. 22457 from the Brain and Behavior Research Foundation (to R. G. O.). M. A. B. is supported by a Future Fellowship (no. FT130101488) from the Australian Research Council. This project was also supported by Marie Curie International Research Staff Exchange scheme no. 612681 under the European Commission FP7 (to R. G. O. and M. A. B.). This study was registered on the Australian and New Zealand Clinical Trials Registry: “The Effect of Methylphenidate, Atomoxetine and Citalopram Versus Placebo on Behavioural and Physiological Indices of Executive Control in Healthy Individuals” (http://www.anzctr.org.au/ACTRN12609000625279.aspx; ACTRN12609000625279).
Reprint requests should be sent to Méadhbh B. Brosnan, School of Psychological Sciences and Monash Institute for Cognitive and Clinical Neurosciences, Monash University, Melbourne, Australia, or via e-mail: firstname.lastname@example.org.
These authors contributed equally to this article.