Increased ongoing neural variability in ADHD

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

Available online at www.sciencedirect.com

ScienceDirect Journal homepage: www.elsevier.com/locate/cortex

Research report

Increased ongoing neural variability in ADHD Gil Gonen-Yaacovi a, Ayelet Arazi a, Nitzan Shahar a, Anat Karmon a, Shlomi Haar b, Nachshon Meiran a and Ilan Dinstein a,b,* a b

Department of Psychology, Ben-Gurion University of the Negev, Beer-Sheva, Israel Department of Cognitive and Brain Science, Ben-Gurion University of the Negev, Beer-Sheva, Israel

article info

abstract

Article history:

Attention Deficit Hyperactivity Disorder (ADHD) has been described as a disorder where

Received 23 July 2015

frequent lapses of attention impair the ability of an individual to focus/attend in a sus-

Reviewed 14 November 2015

tained manner, thereby generating abnormally large intra-individual behavioral variability

Revised 31 December 2015

across trials. Indeed, increased reaction time (RT) variability is a fundamental behavioral

Accepted 7 April 2016

characteristic of individuals with ADHD found across a large number of cognitive tasks. But

Action editor Pia Rotshtein

what is the underlying neurophysiology that might generate such behavioral instability?

Published online 21 April 2016

Here, we examined trial-by-trial EEG response variability to visual and auditory stimuli while subjects' attention was diverted to an unrelated task at the fixation cross. Compar-

Keywords:

isons between adult ADHD and control participants revealed that neural response vari-

ADHD

ability was significantly larger in the ADHD group as compared with the control group in

Intra-individual variability

both sensory modalities. Importantly, larger trial-by-trial variability in ADHD was apparent

Noise

before and after stimulus presentation as well as in trials where the stimulus was omitted,

Sensory systems

suggesting that ongoing (rather than stimulus-evoked) neural activity is continuously more

EEG

variable (noisier) in ADHD. While the patho-physiological mechanisms causing this increased neural variability remain unknown, they appear to act continuously rather than being tied to a specific sensory or cognitive process. © 2016 Elsevier Ltd. All rights reserved.

1.

Introduction

Attention deficit hyperactivity disorder (ADHD) is a prevalent developmental disorder which is characterized by difficulties in allocating and sustaining attention, impulsivity and hyperactivity (American Psychiatric Association, 2000). One manifestation of the core ADHD symptoms is apparent in increased intra-individual behavioral variability over time as demonstrated by numerous reports of increased reaction time

(RT) variability across trials (Kofler et al., 2013; Kuntsi & Klein, 2012). This finding has been reported in a wide variety of cognitive tasks including stop-signal (Alderson, Rapport, & Kofler, 2007; Epstein et al., 2011; Klein, Wendling, Huettner, Ruder, & Peper, 2006; Lijffijt, Kenemans, Verbaten, & van € pcke, Berger, Wandschneider, & Engeland, 2005; Marx, Ho Herpertz, 2013), sustained attention to response (Bellgrove, Hawi, Kirley, Gill, & Robertson, 2005; Johnson et al., 2007; Shallice et al., 2002), choice reaction time (CRT) (Geurts et al., 2008; Gooch, Snowling, & Hulme, 2012; Leth-Steensen, Elbaz,

* Corresponding author. Department of Psychology, Ben-Gurion University of the Negev, Beer Sheva 84105, Israel. E-mail address: [email protected] (I. Dinstein). http://dx.doi.org/10.1016/j.cortex.2016.04.010 0010-9452/© 2016 Elsevier Ltd. All rights reserved.

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

& Douglas, 2000), and Go-no-go (Epstein et al., 2006; Heiser € rger, et al., 2004; Hervey et al., 2006; Kuntsi, Andreou, Ma, Bo & van der Meere, 2005; Spinelli et al., 2011). Increased RT variability in ADHD is apparent in both children (Epstein et al., 2011) and adults (Adams, Roberts, Milich, & Fillmore, 2011), is correlated with ADHD symptom severity (Kuntsi, Wood, Van Der Meere, & Asherson, 2009), and is reduced following administration of stimulants (Boonstra, Kooij, Oosterlaan, Sergeant, & Buitelaar, 2005; Epstein et al., 2006; Rosa-Neto et al., 2005; Spencer et al., 2009; Teicher, Lowen, Polcari, Foley, & McGreenery, 2004). Importantly, differences in RT variability across ADHD and control participants are more prominent and reproducible than differences in mean RT across groups (Kofler et al., 2013; Kuntsi et al., 2013). Why do individuals with ADHD exhibit such unreliable behavior over time? Executive function theories of ADHD suggest that high-level cognitive-control systems in charge of sustained attention, working memory, task-switching, and response inhibition are impaired in ADHD and fail to govern stable behavior by low-level sensory and motor systems (Alvarez & Emory, 2006; Willcutt, Doyle, Nigg, Faraone, & Pennington, 2005). Neuroimaging studies have proposed that these dysfunctions are associated with abnormal functional responses in frontal, parietal, and striatal brain areas (Giedd, Blumenthal, Molloy, & Castellanos, 2001; Rubia et al., 2014) and that weak responses in these areas are associated with larger trial-by-trial RT variability in ADHD (Spinelli et al., 2011; Suskauer et al., 2008). Alternative hypotheses have suggested that increased RT variability may result from general dysfunctions in neuroenergetic supply, where impaired supply of lactate affects the ability of neurons to fire rapidly and reliably (Killeen, Russell, & Sergeant, 2013), or the outcome of deficient dopamine release/sensitivity (Swanson et al., 2007), whereby neuromodulation of the entire brain may be altered throughout development. An additional hypothesis has suggested that neural activity in default mode brain areas, which is typically suppressed during performance of sensory and motor tasks (Raichle & Snyder, 2007), is not suppressed properly in ADHD and interferes with the reliable function of sensory and motor systems (Di Martino et al., 2008; Helps et al., 2010; SonugaBarke & Castellanos, 2007). According to these theories increased RT variability is the outcome of ongoing neuralactivity abnormalities that should be apparent continuously (regardless of the task being performed) in contrast to the cognitive theories described above, which predict that neural activity abnormalities should appear only during recruitment of the impaired cognitive process. Is increased behavioral variability in ADHD associated with increased underlying neural variability across trials? Since behavior is generated by neural activity, one might assume that such a relationship must exist, yet it is surprising that remarkably few studies have actually examined this issue (Dinstein, Heeger, & Behrmann, 2015). Two recent EEG studies have indeed reported that individuals with ADHD exhibit larger trial-by-trial variability in task-evoked EEG responses. The first study reported that P3b responses, which appear approximately 300 msec after stimulus onset and are thought to represent decision making processes (O'Connell, Dockree, & Kelly, 2012), were more variable in ADHD individuals during a

51

working memory task (Saville et al., 2015). The second study reported that frontal-midline theta oscillations, which are associated with cognitive control processes (Luu, Tucker, & Makeig, 2004), were more variable across trials in individuals with ADHD during a response-choice task (McLoughlin, Palmer, Rijsdijk, & Makeig, 2014). Both of these studies concluded that individuals with ADHD exhibit larger trial-bytrial neural variability than controls in specific cognitive processes that govern behavioral responses, which would explain the larger trial-by-trial RT variability across trials found in individuals with ADHD. To properly interpret these findings, however, it is important to examine whether larger neural variability in ADHD is apparent only in neural responses that represent task-evoked cognitive processes or also in early sensory responses to unattended stimuli and even in trials where stimuli are omitted/ absent. Answering this question is critical for determining whether the underlying pathophysiology of ADHD is associated with specific stimulus/task evoked neural mechanisms (i.e., specific cognitive processes) or with general mechanisms that govern ongoing neural fluctuations even in the absence of stimulus/task evoked responses. To address this issue we examined sensory responses in two independent visual and auditory experiments where EEG recordings were acquired from adults with ADHD and matched controls. During these sensory experiments participants performed an unrelated brightness-detection task at the fixation cross, which diverted their attention away from the sensory stimuli that were presented in the visual periphery or auditory modality. Trials containing a stimulus were interleaved with trials where the stimulus was omitted. This experimental design enabled us to compare trial-by-trial neural variability across ADHD and control groups in trials containing unattended stimuli and in trials where the stimulus was entirely absent. In addition, all subjects completed CRT and go-no-go experiments, which have commonly been used to demonstrate differences in reaction-time variability across ADHD and control groups (Kofler et al., 2013; Kuntsi & Klein, 2012). This allowed us to relate measures of trial-bytrial neural variability and behavioral measures of trial-bytrial RT variability within the same individuals.

2.

Materials and methods

2.1.

Participants

Seventeen individuals with ADHD (9 females, Mean age ¼ 25 years old; range ¼ 21e27 years) and 17 healthy controls (12 females, Mean age ¼ 24 years old; range ¼ 21e27 years) participated in the study. All participants had normal or corrected-to-normal vision and provided written informed consent according to the guidelines of the current version of the Declaration of Helsinki. The study was approved by the Ben Gurion University ethics committee. Subjects were paid 50 New Israeli Shekels per hour for participation. ADHD individuals taking stimulants were instructed to abstain from medication for at least 24 h before participation. All participants underwent a clinical interview, which was performed by two intern clinical psychologists (supervised by

52

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

an expert) using the M.I.N.I. plus structured interview (Sheehan et al., 1998). Participants with ADHD provided documentation of an ADHD diagnosis from a credited learning disability center and their diagnosis was confirmed using the M.I.N.I. plus model for ADHD, which requires the identification of symptoms before the age of 7, as well as in adulthood. All participants in the ADHD group and none in the control group met the M.I.N.I. criteria for ADHD. In addition, all participants completed the Conner's Adult ADHD Rating Scales (Conners, Erhardt, Sparrow, 1999) and all individuals with ADHD scored above the clinical cut-off (standardized score of T > 65) while control individuals scored below the cut-off. The M.I.N.I. plus was further used to identify DSM-IV Axis-I disorders. Of the ADHD group, three participants were diagnosed with generalized anxiety disorder and one of these was also diagnosed with obsessive compulsive disorder. Analyzing the data while excluding these three participants revealed similar differences in neural variability across groups (Supplementary Fig. 7). We assessed general intelligence in all subjects using the Raven's test (Raven, 1965). Raw Raven's Standard Progressive Matrices scores were similar across groups (control mean ¼ 54.7, SD ¼ 3.2; ADHD mean 54.5, SD ¼ 4.1) and did not differ significantly [t(32) ¼ .16, p > .8].

2.2.

Sensory experiments

All participants participated in visual and auditory experiments, which had equivalent event-related structures (Supplementary Fig. 1). The visual stimulus consisted of a circular, donut shaped, checkerboard with an inner radius of .6 and an outer radius of 3.7 and the auditory stimulus consisted of a tone at 1,000 Hz, which was presented through earphones to both ears simultaneously. Both experiments contained 600 trials; 400 trials with a stimulus and 200 trials where the stimulus was omitted. Each stimulus was presented for 50 msec and followed by a randomized inter-trial interval lasting 750e1200 msec. In the visual experiment a continuous white-noise auditory stimulus was presented to mask potential auditory stimulation. In both visual and auditory experiments, subjects performed a brightness-detection task at the fixation cross, which was presented in the middle of the screen throughout the entire experiment. This task diverted the attention of the subjects away from the stimuli described above and ensured that participants were attentive and awake. Participants were instructed to press a key whenever the black fixation cross changed in brightness to grey. Each experiment contained 80 random brightness changes, where the fixation cross changed brightness for 30 msec, and participants had 1 sec to respond. Correct and incorrect responses were indicated by changing the fixation cross to green or red, respectively.

2.3.

EEG recording

Continuous EEG was recorded at 1,024 Hz throughout the experiments using a Biosemi Active Two system (Biosemi, Amsterdam, Netherlands) with 64 electrodes mounted on an elastic cap according to the international 10e20 system. All channels were referenced to the mean of the two mastoids.

2.4.

Eye movement recording

During the visual experiment, monocular eye movements were recorded at 500 Hz using an eye-tracker (EyeLink 1000, SR-Research, ON, Canada). Trials in the visual experiment containing horizontal or vertical eye movements that exceeded 1.5 SD of the mean were excluded from EEG and eye tracking analyses. To ensure that trial-by-trial neural variability was not larger in the ADHD group due to differences in the quality of fixation across groups, we computed both the median absolute deviation (MAD) and the variance of gaze position across trials in the same manner as the neural variability (i.e., by computing variability across trials for each time-point and then averaging across all timepoints). Both of these measures did not differ significantly across the two groups [t(32) < 1.2, p > .1, Supplementary Fig. 2].

2.5.

Pre-processing and trial rejection

Data were analyzed using Matlab (Mathworks, Inc.) and EEGLAB (Delorme & Makeig, 2004). Pre-processing included the following steps: First, data were down-sampled to 512 Hz and filtered between 1 and 40 Hz. Second, data were segmented into trials/epochs of 700 msec (200 msec prestimulus to 500 msec post-stimulus). Trials with artifacts were identified using automated scripts, which identified trials with muscle contractions according to excessive (>25 db) power in the 20e40 Hz frequency range and trials with eye blinks according to raw absolute signal changes (>60 uv in any of the frontal electrodes; Fp1, Fp2, AF3, AF4). All trials containing button presses, eye blinks, or muscle contractions were removed from further analyses. The mean percentage of removed trials was 27% and 12% for the control and 36% and 10% for the ADHD group in the visual and auditory experiments, respectively. Note that less trials were excluded in the auditory experiment, because subjects in both groups exhibited considerably less eye blinks. The percentage of excluded trials did not differ across groups in either experiment [t(32) < .8, p > .9 in all comparisons].

2.6.

Electrode selection and identification of P100/N100

The P100 response was identified using automated scripts in each subject separately. We first computed the mean ERP across trials containing a visual stimulus for each electrode and then identified the occipital electrode with the largest positive electric potential value within a window located 80e150 msec after stimulus onset. This electrode was selected for further analysis, while noting the latency (time from stimulus onset) of the P100 response in each subject. The N100 response was identified in an analogous manner in the auditory experiment by selecting the central electrode with the largest negative electric potential value within the same time window. Individual subjects exhibited clear and robust P100 and N100 responses as apparent in the typical topography of their P100 and N100 potential maps (Supplementary Fig. 3).

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

2.7.

Independent component selection

We performed an independent component analysis (ICA) using the cleaned and segmented EEG data to identify a single independent component that best represented early sensory responses in each experiment (Makeig, Jung, Bell, Ghahremani, & Sejnowski, 1997). We computed the percent variance accounted for by each of the ICA components within the window used to identify the P100/N100 responses (i.e., 80e150 msec following stimulus onset) and selected the component that explained the highest percentage of variance within this window (i.e., the component that best captured the early sensory response) in each subject. The selected ICA components explained, on average, 50% (std ¼ 23) and 15% (std ¼ 7) of the variance in the visual and auditory experiments, respectively. Scalp maps of the individually selected components are presented in Supplementary Fig. 4.

2.8.

CRT experiments: In the visual version, a black triangle or circle was presented at the center of the screen for 300 msec and participants were instructed to press one button in response to the triangle and another button in response to the circle as quickly as possible. Each stimulus was followed by an inter-trial interval of 1,200 msec. In the auditory version two pure tone stimuli with frequencies of 200 and 2,000 Hz at a volume of 75 dB were presented instead of the two shapes and the inter-trial interval was increased to 1500 msec. A total of 200 trials were presented in each version. Go-no-go experiments: Stimuli and structure were identical to those described in the CRT experiment, except that participants were instructed to press the spacebar as quickly as possible with their index finger whenever they saw a circle or heard the high tone (i.e., ‘go’ trial) and to withhold responding when the other stimulus was presented (i.e., ‘no go’ trial). A total of 300 trials were presented in each sensory modality and 80% of the trials contained the ‘go’ stimulus.

EEG amplitude variability 2.11.

Trial-by-trial variability in the amplitude of the sensoryevoked responses was estimated by calculating the variance and MAD across trials at the time-point identified as the P100/ N100 ERP peak in the visual/auditory experiments, respectively. In additional analyses we quantified trial-by-trial neural variability in the pre-stimulus interval (i.e., from 200 msec to stimulus onset), the entire post stimulus response (i.e., from stimulus onset to 500 msec), and in intermingled trials where the stimulus was omitted (i.e., 200 to 500 msec). Here trial-by-trial variability was computed for each time-point within the examined window and then averaged across time-points to produce a single variability measure for each of the three conditions (pre-stim, post-stim, and trials without stimulus). All variability analyses were performed once using the data fromtheelectrode with thelargestP100/N100 response and again using the independent component that best captured the P100/ N100 responses in each participant (Supplementary Figs. 3&4).

2.9.

EEG latency variability

Trial-by-trial variability in the latency of the sensory-evoked responses was estimated using an Inter-Trial Phase Coherence (ITPC) analysis. We examined ITPC in the theta (4e8 Hz) and alpha (8e12 Hz) frequency bands using a sliding window with a width of 200 msec for theta band and 170 msec for the alpha band. The ITPC measure estimates the degree to which the phase at each frequency band is aligned across trials (Delorme & Makeig, 2004). Larger variability across trials in response latency yields lower ITPC values.

2.10.

53

Cognitive experiments

All of the control subjects and fifteen of the seventeen ADHD subjects also completed two cognitive experiments: CRT and go-no-go (Supplementary Fig. 1). Two ADHD subjects did not complete the cognitive experiments due to a technical problem. Participants completed one visual and one auditory version of each experiment. The order of these experiments and the responding hand were counterbalanced across participants.

Behavioral data analyses

We computed the mean accuracy, mean RT, RT variance, and RT coefficient of variation across trials for each subject in each of the cognitive experiments. The first 10 trials (considered as practice), trials with a RT below 200 msec or 2.5 standard deviations above the mean RT, trials with incorrect responses, and trials following an incorrect response (Rabbit, 1966) were excluded from all analyses.

2.12.

Statistical tests, power analysis, and Bayes factor

All statistical comparisons of behavioral, eye tracking, and EEG measures across groups were performed using two-tailed, two-sample t-tests with unequal variances and an alpha of .05. We also report the effect size for each comparison using Cohen's d. Pearson's correlation coefficients (r values) were computed to assess potential relationships between the neural and behavioral measures. Statistical significance was estimated by transforming r values into t values and performing a two tailed t-test with an alpha of .05. We did not apply correction for multiple comparisons, because all of the statistical tests in the study reproduced the same result across analyses that examined independent data segments (i.e., pre/ post stimulus intervals and trials without stimulus) and independent experiments in the visual and auditory modalities. Most comparisons of neural variability across ADHD and control groups yielded effect sizes (Cohen's d) that were larger than .7. Given our sample size (17 participants in each group), such effect sizes correspond to a statistical power of approximately .5. We also used the initial results from our analysis of the P100 and N100 responses (Fig. 3) as the prior for computing the Bayes factor (Rouder, Speckman, Sun, Morey, & Iverson, 2009) of the results when analyzing P100/N100 responses in an identical electrode across subjects (Supplementary Fig. 8) and when analyzing the pre/post stimulus intervals and trials without stimulus (Fig. 5). In all cases, the computed Bayes factors were larger than 3 indicating that these analyses revealed substantial evidence in favor of our theory that trialby-trial neural variability was larger in ADHD individuals than in controls.

54

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

Fig. 1 e Average voltage spline maps and grand ERP traces from the visual (left) and auditory (right) experiments. Top row: average voltage spline maps for the control and ADHD groups at P100 (visual experiment) and N100 (auditory experiment). Bottom row: grand ERP traces across all trials and subjects (taken from the channel with the largest response in each subject) in the ADHD (red) and control (blue) groups.

Fig. 2 e Single-trial responses to the visual and auditory stimuli in one control and one ADHD participant. Each panel presents trial-by-trial responses from the electrode with the strongest P100/N100 response (left) or the ICA component that best captured the early sensory response (right). Colors reflect positive (red) and negative (blue) changes in electric potential relative to the pre-stimulus interval (¡200 to 0 msec). The blue line at the bottom of each panel presents the average across trials. Top row: visual experiment. Bottom row: auditory experiment.

3.

Results

The grand ERP across participants in the ADHD group and the control group exhibited clear positive (P100) and negative (N100) peaks at approximately 100 msec after stimulus onset in the visual and auditory experiments, respectively (Fig. 1).

The ERP responses of individual participants were similarly robust and allowed us to identify the latency of individual subject P100 and N100 peaks and the electrode where these peaks were largest (see Methods and Supplementary Fig. 3). Stimulus evoked responses exhibited clear trial-by-trial variability when examining the data of individual subjects

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

55

Fig. 3 e Mean P100/N100 amplitude, variability across trials, and median absolute deviation (MAD) across trials in the visual and auditory experiments. Top panels: analysis with individually selected electrodes that exhibited the strongest response in each subject. Bottom panels: Analysis with the ICA component that best captured the early sensory response. Black: control, White: ADHD. Error bars represent standard error of the mean across subjects. Stars: significant difference across groups.

Fig. 4 e Inter-trial Phase Coherence in the theta (4e8 Hz, top row) and alpha (8e12 Hz, bottom row) frequency bands in the visual and auditory experiments. Solid line: mean across control subjects, Dotted line: mean across ADHD subjects. Left panels: when using selected electrode. Right panels: when using selected ICA component. The thick black line along the x axis indicates time points with significant differences across groups (p < .05, two-tailed t-tests). (Fig. 2). This was apparent when examining the electrode with the strongest P100/N100 response (as described above) and also when performing an ICA analysis and examining the ICA component that best captured the P100/N100 response in each participant (see Methods). Note that examining trial-by-trial neural variability using individually identified electrodes and components focuses the analysis on the most robust sensory responses evident in each subject. This addresses a potential weakness of group based analyses where the examined

electrode and latency are defined according to the grand ERP of the group with the potentially flawed assumption that subjects exhibit identical spatial and temporal responses.

3.1.

Response amplitude and variability

Mean P100 and N100 response amplitudes did not differ across groups in the visual and the auditory experiments, respectively when analyzing the electrode with the strongest

56

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

Fig. 5 e Trial-by-trial variability in the pre-stimulus interval (¡200 msec to stimulus onset; left column), post-stimulus interval (stimulus onset to 500 msec; middle column) and in trials where the stimulus was omitted (right column). Black: control, White: ADHD. Top row: selected electrode. Bottom row: selected ICA component. Error bars: standard error of the mean across subjects. Stars: significant difference across groups (p < .05, two tailed t-tests). response in each subject [t(32) < .37, p > .7] or the ICA component that best captured early sensory responses [t(32) < 1.89, p > .06]. Trial-by-trial P100 and N100 amplitude variability and MAD measures, however, were significantly larger in the ADHD group compared with the control group in both experiments (Fig. 3 and Supplementary Table 1) when analyzing the selected electrode [t(32) > 2.06, p < .05, Cohen's d > .67] or ICA component [t(32) > 2.1, p < .04, Cohen's d > .73]. To demonstrate the robustness of this finding we also performed the same analysis while examining the same electrode (PO8 in the visual experiment and FCz in the auditory experiment) in all subjects (Supplementary Fig. 8). In line with the previous analysis, response amplitudes did not differ across groups [t(32) < .96, p > .3], but trial-by-trial variability was significantly larger in the ADHD group in the auditory experiment [t(32) ¼ 3.16, p ¼ .003, Cohen's d ¼ 1.1, Bayes factor ¼ 52.4] and marginally significant in the visual experiment [t(32) ¼ 1.93, p ¼ .06, Cohen's d > .66, Bayes factor ¼ 3.4]. Trial-by-trial variability can be generated by variability in response amplitude or latency. To examine variability in response latency across trials, we compared ITPC in the theta (4e8 Hz) and alpha (8e12 Hz) frequency bands across groups. We computed the ITPC as a function of time by calculating it in a sliding window before and after stimulus onset (see methods). We then compared ITPC in theta and alpha bands across groups in each time-point. ITPC was significantly larger in the control group as compared with the ADHD group in multiple time-points following stimulus onset when analyzing the selected electrode or ICA component [t(32) > 2.1, p < .05, Cohen's d > .7, time-points marked in black, Fig. 4]. This indicates that response latencies in control individuals were more consistent across trials (i.e., exhibited similar phases across trials) as compared with ADHD subjects.

3.2.

Ongoing neural variability

To determine whether trial-by-trial variability was larger in the ADHD group only in stimulus-evoked P100/N100 responses or throughout the entire experiment, we estimated trial-by-trial variability in the pre-stimulus interval (i.e., 200 msec to stimulus onset), the post-stimulus interval (i.e., stimulus onset to 500 msec), and in trials where the stimuli were omitted (Fig. 5). Here, we first computed variability across trials for each time-point separately and then averaged across time-points (i.e., 200 time-points for pre-stimulus intervals, 500 time-points for post-stimulus intervals, and 700 time-points for trials without a stimulus). Trial-by-trial variability was significantly larger in individuals with ADHD as compared with controls in all cases [t(32) > 2.3, p < .05, Cohen's d > .75, Bayes factor > 7.9] when examining the selected electrode, except for the auditory post-stimulus interval where there was a marginally significant difference in the same direction [t(32) ¼ 1.8, p ¼ .07, Cohen's d ¼ .64, Bayes factor ¼ 3]. Significantly larger variability in the ADHD group was also evident in all conditions and experiments when examining the selected ICA component, which best captured the visual/auditory responses in each subject [t(32) > 2.1, p < .05, Cohen's d > .7, Bayes factor > 5].

3.3.

Brightness detection task

Participants performed an infrequent brightness-detection task at the fixation cross to divert their attention away from the examined stimuli and to ensure that they remained attentive and alert during the experiments. Control participants exhibited a trend for better accuracy, but there were no significant group differences in either accuracy (percentage of correct hits), mean RT, or RT variability in both the visual and the auditory experiments [t(32) < 1.5, p > .05; Supplementary

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

Fig. 5]. The percentage of commissions in the control group (visual mean ¼ 12.1, std ¼ 10; auditory mean ¼ 7.7, std ¼ 5.6) was larger than that of the ADHD group (visual mean ¼ 8.8, std ¼ 7; auditory mean ¼ 5.2, std ¼ 4.3), but not significantly different [t(32) < 1.43, p > .16]. The percentage of omissions in the control group (visual mean ¼ 19.5, std ¼ 10; auditory mean ¼ 15.1, std ¼ 9.1) was smaller than that of the ADHD group (visual mean ¼ 26.3, std ¼ 12.5; auditory mean ¼ 20, std ¼ 11.9), but not significantly different [t(32) < 1.76, p > .08].

3.4.

three experiments [t(30) < 1.8, p > .07, Cohen's d < .62]. RT variability was significantly larger in ADHD individuals than controls in a consistent manner across all experiments [t(30) > 2, p < .05, Cohen's d > .69, Fig. 7 and Supplementary Table 2] as was the coefficient of variation [t(30) > 2.1, p < .04, Cohen's d > .72]. Subject-by-subject RT variability values were correlated across visual and auditory modalities and across CRT and Go-no-go experiments [r(30) > .53, p < .05], demonstrating the replicability of individual RT variability measures across multiple independent experiments.

Correlated variability across sensory modalities

We correlated trial-by-trial neural variability measures between the visual and auditory experiments to determine whether individuals with larger variability in one sensory domain also exhibited larger variability in the other sensory domain. The results revealed large and significant correlations across sensory domains when examining neural variability in the pre-stimulus (control: r ¼ .76, p < .001; ADHD: r ¼ .82, p < .001) and post-stimulus (control: r ¼ .59, p ¼ .01; ADHD: r ¼ .79, p < .001) intervals of trials containing a stimulus as well as in trials where the stimulus was omitted (control: r ¼ .77, p < .001; ADHD: r ¼ .84, p < .001, Fig. 6). Similar correlations across visual and auditory modalities have been reported previously in control participants when examining visual and auditory P3b responses (Saville et al., 2012). These findings indicate that neural variability measures are a robust characteristic of individual subjects regardless of the sensory modality in which they are measured.

3.5.

57

Cognitive task performance

All control subjects and 15 of the 17 ADHD subjects also participated in a set of cognitive experiments which included visual and auditory versions of a CRT and a Go-no-go test (see Methods). Control subjects were significantly more accurate than ADHD subjects in the auditory CRT and visual Go-no-go experiments [t(30) > 2.3, p < .03, Cohen's d > .77], but not in the visual CRT and auditory Go-no-go experiments [t(30) < 1.9, p > .05, Cohen's d < .67]. Control subjects were also significantly faster than ADHD subjects in the visual CRT experiment [t(30) ¼ 3.1, p < .01, Cohen's d ¼ .98], but not in the other

3.6. Correlations between neural and behavioral variability There were no significant correlations between the different measures of trial-by-trial neural variability (P100/N100, prestimulus interval, post-stimulus interval, or across trials without stimulus) as quantified in the sensory experiments and RT variability as quantified in the CRT or Go-no-go experiments in the ADHD group [.24 < r(15) < .5, p > .06] or control group [.28 < r(17) < .45, p > .07]. Hence, while there were strong and significant differences in neural variability and RT variability across groups, subject-by-subject measures of neural and RT variability were not significantly correlated within either group. Note that neural variability (Fig. 6) and RT variability measures of individual subjects were robustly reproduced across visual and auditory experiments indicating that they are reliable characteristics of individual subjects. The lack of significant correlations between neural and RT variability may, therefore, reflect that 1) the statistical power of our sample may be too small for identifying a subtle relationship across these measures. 2) the relationship between the two measures may not be linear. 3) The intensity of RT variability in individual subjects may be related to a combination of measures rather than a single measure of neural variability.

3.7. Correlations between neural/behavioral variability and ADHD severity We did not find any significant correlations between scores of ADHD severity as assessed using the CAARS test and RT variability in the visual or auditory experiment [0 < r(17) < .31,

Fig. 6 e Correlations of trial-by-trial variability across visual and auditory experiments as quantified in the pre-stimulus interval (left panel), post-stimulus interval (middle panel), and trials where the stimulus was omitted (right panel). Black circles: control subjects, White circles: ADHD subjects. Sold line: linear fit for control subjects. Dotted line: linear fit for ADHD subjects.

58

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

Fig. 7 e Comparison of Accuracy, mean RT, RT variance, and coefficient of variation across groups in the CRT and go-no-go experiments. Black: controls, White: ADHD. Error bars: standard error of the mean across subjects. Stars: significant difference across groups (p < .05, two tailed t-tests).

p > .25] nor between ADHD severity scores and any of the neural variability measures in either experiment [0 < r(17) < .22, p > .39].

3.8. Correlations between ERP variability and electrodes offsets A potential confounding source of EEG signal fluctuations/ variability over time (measurement noise) is the quality of electrode contact, which is estimated in the Biosemi Active II system (active electrodes) with a measure called electrode offset. Correlations between neural variability in the poststimulus interval and the mean electrode offset (i.e., mean quality of electrode contact) or the electrode offset variance (i.e., variability of the electrode contact over time) were not significant [.22 < r(34) < .16, p > .2] as were correlations with neural variability apparent in trials where the stimulus was absent [.24 < r(34) < .18, p > .1]. These findings suggest that the variability differences across groups were not due to differences in electrode contact quality (Supplementary Fig. 6).

4.

Discussion

Our results reveal that sensory systems of individuals with ADHD exhibit similar response sensitivity on average, but reduced reliability across trials both in terms of the amplitude and the latency of the sensory neural responses. While mean P100 and N100 response amplitudes were similar across

ADHD and control groups in terms of amplitude, larger neural variability in ADHD was evident both in increased P100/N100 amplitude variability (Fig. 3) and in reduced ITPC (Fig. 4). Equivalent findings were apparent when analyzing data from the electrode with the strongest P100/N100 response, when analyzing a single electrode across all subjects (Supplementary Fig. 8), or when extracting a single ICA component, which best captured the P100/N100 responses in each subject. This demonstrates the robustness of the results across different analysis techniques and suggests that individuals with ADHD experience/perceive their environment in a less reliable manner despite similar sensitivity on average. Importantly, increased neural variability in ADHD was evident not only in stimulus-evoked neural responses, but also in pre-stimulus intervals and in trials where the stimulus was omitted (Fig. 5). These findings were apparent in a reproducible manner across visual and auditory experiments (Fig. 6) and suggest that ADHD is characterized by increased moment-by-moment neural variability (i.e., abnormally large ongoing neural fluctuations), which is evident continuously rather than being associated with a specific sensory or cognitive process/function. Increased behavioral variability was also evident in the same ADHD subjects across multiple cognitive tasks (Fig. 7). While it is tempting to speculate that increased neural variability might cause increased behavioral variability, we did not find significant correlations between the levels of neural variability and RT variability in the ADHD group. Nevertheless,

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

we present strong novel evidence for the existence of increased neural and behavioral variability in the same group of ADHD individuals.

4.1.

Quantitative EEG (qEEG) findings

Previous qEEG studies have reported that children, adolescents, and adults with ADHD exhibit significantly larger absolute EEG power than controls (Bresnahan, Anderson, & Barry, 1999; Bresnahan & Barry, 2002). While most of these studies have focused on differences in theta-band power or theta/beta power ratio (di Michele, Prichep, John, & Chabot, 2005; Snyder & Hall, 2006), excessive absolute EEG power is an analogous finding to increased variability (i.e., EEG signals with larger absolute power vary more over time). The results of our study demonstrate that increased neural variability is a prominent characteristic of ADHD individuals who also exhibit increased RT variability across trials. Increased neural variability was present in both sensory-evoked responses and ongoing neural activity, thereby extending previous qEEG findings (that were examined only during rest) and providing a new interpretation of their potential behavioral and pathological significance.

4.2.

Increased behavioral variability in ADHD

Prominent ADHD theories have proposed that increased behavioral variability is an outcome of dysfunctions in attentional (Leth-Steensen et al., 2000), working memory (Rapport et al., 2008), and/or behavioral inhibition (Barkley, 1997) processes, which are located in frontal and parietal brain areas and govern the activity of lower level sensory and motor processes. While some fMRI studies have indeed reported that increased RT variability in ADHD is associated with weaker responses in frontal and parietal brain areas (Spinelli et al., 2011; Suskauer et al., 2008), others have reported the opposite (Bellgrove, Hester, & Garavan, 2004). Note that these studies did not examine response variability across trials in either cognitive or sensory/motor brain areas and the apparent assumption is that abnormally weak or strong brain responses in cognitive brain areas generate trial-by-trial sensory/motor variability, which then translates into behavioral variability across trials. Two recent EEG studies have examined this issue and reported that individuals with ADHD indeed exhibit increased trial-by-trial neural variability in EEG responses associated with cognitive control processes. The first study reported that ADHD individuals exhibit larger trial-by-trial P3b response variability (Saville et al., 2015), while the second study reported that ADHD individuals exhibit larger variability in trialby-trial theta-band amplitude and phase (McLoughlin et al., 2014) in comparison to controls. These studies suggested that the neural variability driving behavioral variability was indeed specific to cognitive responses/processes as hypothesized by cognitive control theories of ADHD. Note, however, that both studies focused on specific portions of the taskevoked response as they appeared in a single principal component (Saville et al., 2015) or frequency band (McLoughlin et al., 2014) and did not examine trial-by-trial variability in raw EEG segments containing the stimulus

59

evoked response or ongoing neural activity (e.g., pre-stimulus intervals and trials without stimulus). We expect that both studies would have found similar findings to those reported here had they performed these analyses. In contrast to the cognitive theories described above, which predict that neural abnormalities should appear only during the engagement of specific cognitive processes, other theories have suggested that increased behavioral variability in ADHD may be the outcome of general, continuous neural dysfunctions. These include impaired dopamine release and/ or sensitivity, which is perhaps the strongest physiological theory of ADHD (Swanson et al., 2007). Impaired dopaminergic innervation of the cortex may lead not only to weaker neuromodulation on average, but also to more variable neuromodulation over time, thereby generating increased neural variability. Another potential mechanism is impaired neuroenergetic supply, whereby insufficient resupply of the neuron's preferred metabolite, lactate, from neighboring astrocytes, leads to neural fatigue and larger neural variability across trials (Killeen et al., 2013). A third potential mechanism is impaired regulation of the default mode system, which is a group of brain areas where neural activity is typically suppressed during the processing of external stimuli or tasks (Raichle & Snyder, 2007). Several studies have reported that this system is not suppressed properly in individuals with ADHD and proposed that this neural activity interferes with task related process, thereby generating more variable behavior across trials (Di Martino et al., 2008; Feige et al., 2013; Helps et al., 2010; Sonuga-Barke & Castellanos, 2007). While each of these theories proposes a different underlying mechanism, in all cases neural activity abnormalities should be apparent in a continuous manner regardless of what the subject is doing. Our results seem more in agreement with the later theories, because they reveal that increased neural variability in ADHD is apparent continuously in visual and auditory evoked responses as well as in ongoing neural activity measured in pre-stimulus intervals and in trials without stimulus. Note that increased neural variability was observed in auditory sensory responses of our participants even when they were engaged in an unrelated visual task. These findings, together with the resting-state qEEG results described above, suggest that increased neural variability in ADHD is not necessarily tied to a specific sensory or cognitive neural process, but rather to ongoing neural processes that are constantly fluctuating more variably in ADHD. While the specific mechanism driving these larger neural fluctuations remains unknown (and may differ across individuals), the results suggest that it is a mechanism that governs ongoing neural activity rather than being limited to a specific cognitive process.

4.3.

Neural noise in autism and ADHD

Increased neural variability is not unique to ADHD and has been reported previously in individuals with autism (Dinstein, Heeger et al., 2012; Dinstein, Thomas et al., 2010; Haigh, Heeger, Dinstein, Minshew, & Behrmann, 2015; Milne, 2011; Weinger, Zemon, Soorya, & Gordon, 2014). However, in autism increased neural variability seems to be present only in stimulus-evoked responses and not in ongoing activity

60

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

fluctuations (Dinstein et al., 2012). These findings stand in contrast to the ones reported here with ADHD individuals and suggest that the underlying pathophysiology in the two disorders may be distinct. When evaluating this issue it is important to remember that approximately 50% of individuals with autism fulfill clinical criteria for an ADHD diagnosis (Leitner, 2014) and a recent meta-analysis of behavioral studies has suggested that individuals with autism exhibit increased RT variability only if they also have ADHD symptoms (Karalunas, Geurts, Konrad, Bender, & Nigg, 2014). Dissociating the neurophysiology of the two disorders would, therefore, require performing identical experiments with subgroups of individuals who fulfill clinical criteria for autism, ADHD, and both disorders.

4.4.

24 h period preceding the experiments and assume that these instructions were followed.

4.5.

Conclusions

The results presented here demonstrate that increased neural variability is a general continuous phenomenon in ADHD, which is present across responses to unattended stimuli and even across trials where stimuli are entirely omitted. These findings suggest that ADHD neuropathology is at least in part associated with mechanisms that govern the stability of ongoing neural activity. We propose that abnormalities in such mechanisms may drive increased behavioral variability and impair sustained attention capabilities in ADHD.

Limitations

Funding EEG records electric potential changes that are generated by multiple sources including muscle contractions, eyesaccades, eye-blinks, heartbeat, and numerous neural processes. When comparing trial-by-trial neural variability across ADHD and control groups it is, therefore, critical to determine that between-group differences were due to neural sources rather than non-neural sources of the EEG signal. We took several steps to ensure that this was indeed the case. First, we identified and removed trials with muscle contractions and eye blinks from the analyses using automated signal analysis techniques. Second, we checked that eye fixation variability did not differ significantly across groups (Supplementary Fig. 2). Third we checked that both mean electrode offset and electrode offset variability throughout the experiments (measures of electrode contact quality) did not differ significantly across groups and were not correlated with neural variability measures (Supplementary Fig. 6). Finally, we performed all of our analyses once with data from the electrode with the strongest P100/N100 response, again with an identical electrode across subjects, and a third time with a single ICA component that best captured the sensory responses. ICA decomposition is useful for extracting EEG signal components that capture repeating neural processes, which have distinct spatio-temporal patterns across the EEG sensors, and separating them from other independent processes (Delorme, Palmer, Onton, Oostenveld, & Makeig, 2012; Onton, Delorme, & Makeig, 2005). In this manner we compared trial-by-trial variability across groups using the portion of the EEG data, which best represented the sensory-evoked responses in each subject while disregarding all other sources of trial-by-trial variability. The fact that results were reproducible across different types of analysis, while examining multiple independent segments of the data (i.e., pre-stimulus and poststimulus intervals and in trials without stimulus) from independent visual and auditory experiments, attests to the robustness of the finding. Nevertheless, as with any EEG study, it is still possible that our estimates of neural variability were influenced by subject-specific sources of measurement noise. Additional limitations include the relatively small sample size (17 participants in each group) and potential participant compliance issues. For example, we instructed individuals with ADHD to abstain from taking psycho-stimulants during a

This research was funded by grants from the Israeli Science Foundation (ID), Alon Fellowship (ID), and Ministry of Immigrant Absorption Fellowship (GGY).

Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.cortex.2016.04.010.

references

Adams, Z. W., Roberts, W. M., Milich, R., & Fillmore, M. T. (2011). Does response variability predict distractibility among adults with attention-deficit/hyperactivity disorder? Psychological Assessment, 23(2), 427e436. http://dx.doi.org/10.1037/a0022112. Alderson, R. M., Rapport, M. D., & Kofler, M. J. (2007). Attentiondeficit/hyperactivity disorder and behavioral inhibition: a meta-analytic review of the stop-signal paradigm. Journal of Abnormal Child Psychology, 35(5), 745e758. http://dx.doi.org/ 10.1007/s10802-007-9131-6. Alvarez, J. A., & Emory, E. (2006). Executive function and the frontal lobes: a meta-analytic review. Neuropsychology Review, 16(1), 17e42. http://dx.doi.org/10.1007/s11065-006-9002-x. American Psychiatric Association. (2000). DSM-IV-TR (American P.). Washington, DC. Barkley, R. A. (1997). Behavioral inhibition, sustained attention, and executive functions: constructing a unifying theory of ADHD. Psychological Bulletin, 121(1), 65e94. Retrieved from: http://www.ncbi.nlm.nih.gov/pubmed/9000892. Bellgrove, M. A., Hawi, Z., Kirley, A., Gill, M., & Robertson, I. H. (2005). Dissecting the attention deficit hyperactivity disorder (ADHD) phenotype: sustained attention, response variability and spatial attentional asymmetries in relation to dopamine transporter (DAT1) genotype. Neuropsychologia, 43(13), 1847e1857. http://dx.doi.org/10.1016/ j.neuropsychologia.2005.03.011. Bellgrove, M. A., Hester, R., & Garavan, H. (2004). The functional neuroanatomical correlates of response variability: evidence from a response inhibition task. Neuropsychologia, 42(14), 1910e1916. http://dx.doi.org/10.1016/ j.neuropsychologia.2004.05.007. Boonstra, A. M., Kooij, J. J. S., Oosterlaan, J., Sergeant, J. A., & Buitelaar, J. K. (2005). Does methylphenidate improve

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

inhibition and other cognitive abilities in adults with childhood-onset ADHD? Journal of Clinical and Experimental Neuropsychology, 27(3), 278e298. http://dx.doi.org/10.1080/ 13803390490515757. Bresnahan, S., Anderson, J. W., & Barry, R. J. (1999). Age-related changes in quantitative EEG in attention-deficit/hyperactivity disorder. Biological Psychiatry, 46(12), 1690e1697. Retrieved from: http://www.ncbi.nlm.nih.gov/pubmed/10624551. Bresnahan, S., & Barry, R. J. (2002). Specificity of quantitative EEG analysis in adults with attention deficit hyperactivity disorder. Psychiatry Research, 112(2), 133e144. Retrieved from: http:// www.ncbi.nlm.nih.gov/pubmed/12429359. Conners, C. K., Erhardt, D., & Sparrow, E. (1999). In M.-H. Systems (Ed.), Conners' adult ADHD rating scales (CAARS) technical manual. Toronto, Ontario. Delorme, A., & Makeig, S. (2004). EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134(1), 9e21. http://dx.doi.org/10.1016/ j.jneumeth.2003.10.009. Delorme, A., Palmer, J., Onton, J., Oostenveld, R., & Makeig, S. (2012). Independent EEG sources are dipolar. PLoS One, 7(2), e30135. http://dx.doi.org/10.1371/journal.pone.0030135. Di Martino, A., Scheres, A., Margulies, D. S., Kelly, A. M. C., Uddin, L. Q., Shehzad, Z., et al. (2008). Functional connectivity of human striatum: a resting state FMRI study. Cerebral Cortex (New York, N.Y.: 1991), 18(12), 2735e2747. http://dx.doi.org/ 10.1093/cercor/bhn041. Dinstein, I., Heeger, D. J., & Behrmann, M. (2015). Neural variability: friend or foe? Trends in Cognitive Sciences, 19(6), 322e328. http://dx.doi.org/10.1016/j.tics.2015.04.005. Dinstein, I., Heeger, D. J., Lorenzi, L., Minshew, N. J., Malach, R., & Behrmann, M. (2012). Unreliable evoked responses in autism. Neuron, 75(6), 981e991. http://dx.doi.org/10.1016/ j.neuron.2012.07.026. Dinstein, I., Thomas, C., Humphreys, K., Minshew, N., Behrmann, M., & Heeger, D. J. (2010). Normal movement selectivity in autism. Neuron, 66(3), 461e469. http://dx.doi.org/ 10.1016/j.neuron.2010.03.034. Epstein, J. N., Conners, C. K., Hervey, A. S., Tonev, S. T., Arnold, L. E., Abikoff, H. B., et al. (2006). Assessing medication effects in the MTA study using neuropsychological outcomes. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 47(5), 446e456. http://dx.doi.org/10.1111/j.14697610.2005.01469.x. Epstein, J. N., Langberg, J. M., Rosen, P. J., Graham, A., Narad, M. E., Antonini, T. N., et al. (2011). Evidence for higher reaction time variability for children with ADHD on a range of cognitive tasks including reward and event rate manipulations. Neuropsychology, 25(4), 427e441. http://dx.doi.org/10.1037/ a0022155. Feige, B., Biscaldi, M., Saville, C. W. N., Kluckert, C., Bender, S., Ebner-Priemer, U., et al. (2013). On the temporal characteristics of performance variability in attention deficit hyperactivity disorder (ADHD). PLoS One, 8(10), e69674. http:// dx.doi.org/10.1371/journal.pone.0069674. , S., Oosterlaan, J., Geurts, H. M., Grasman, R. P. P. P., Verte Roeyers, H., van Kammen, S. M., et al. (2008). Intra-individual variability in ADHD, autism spectrum disorders and Tourette's syndrome. Neuropsychologia, 46(13), 3030e3041. http:// dx.doi.org/10.1016/j.neuropsychologia.2008.06.013. Giedd, J. N., Blumenthal, J., Molloy, E., & Castellanos, F. X. (2001). Brain imaging of attention deficit/hyperactivity disorder. Annals of the New York Academy of Sciences, 931, 33e49. Retrieved from: http://www.ncbi.nlm.nih.gov/pubmed/ 11462751. Gooch, D., Snowling, M. J., & Hulme, C. (2012). Reaction time variability in children with ADHD symptoms and/or dyslexia.

61

Developmental Neuropsychology, 37(5), 453e472. http:// dx.doi.org/10.1080/87565641.2011.650809. Haigh, S. M., Heeger, D. J., Dinstein, I., Minshew, N., & Behrmann, M. (2015). Cortical variability in the sensoryevoked response in autism. Journal of Autism and Developmental Disorders, 45(5), 1176e1190. http://dx.doi.org/10.1007/s10803014-2276-6. Heiser, P., Frey, J., Smidt, J., Sommerlad, C., Wehmeier, P. M., Hebebrand, J., et al. (2004). Objective measurement of hyperactivity, impulsivity, and inattention in children with hyperkinetic disorders before and after treatment with methylphenidate. European Child & Adolescent Psychiatry, 13(2), 100e104. http://dx.doi.org/10.1007/s00787-004-0365-3. Helps, S. K., Broyd, S. J., James, C. J., Karl, A., Chen, W., & SonugaBarke, E. J. S. (2010). Altered spontaneous low frequency brain activity in attention deficit/hyperactivity disorder. Brain Research, 1322, 134e143. http://dx.doi.org/10.1016/ j.brainres.2010.01.057. Hervey, A. S., Epstein, J. N., Curry, J. F., Tonev, S., Eugene Arnold, L., Keith Conners, C., et al. (2006). Reaction time distribution analysis of neuropsychological performance in an ADHD sample. Child Neuropsychology: a Journal on Normal and Abnormal Development in Childhood and Adolescence, 12(2), 125e140. http://dx.doi.org/10.1080/09297040500499081. Johnson, K. A., Robertson, I. H., Kelly, S. P., Silk, T. J., Barry, E.,  ibhis, A., et al. (2007). Dissociation in performance of Da children with ADHD and high-functioning autism on a task of sustained attention. Neuropsychologia, 45(10), 2234e2245. http://dx.doi.org/10.1016/j.neuropsychologia.2007.02.019. Karalunas, S. L., Geurts, H. M., Konrad, K., Bender, S., & Nigg, J. T. (2014). Annual research review: reaction time variability in ADHD and autism spectrum disorders: measurement and mechanisms of a proposed trans-diagnostic phenotype. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 55(6), 685e710. http://dx.doi.org/10.1111/jcpp.12217. Killeen, P. R., Russell, V. A., & Sergeant, J. A. (2013). A behavioral neuroenergetics theory of ADHD. Neuroscience and Biobehavioral Reviews, 37(4), 625e657. http://dx.doi.org/10.1016/ j.neubiorev.2013.02.011. Klein, C., Wendling, K., Huettner, P., Ruder, H., & Peper, M. (2006). Intra-subject variability in attention-deficit hyperactivity disorder. Biological Psychiatry, 60(10), 1088e1097. http:// dx.doi.org/10.1016/j.biopsych.2006.04.003. Kofler, M. J., Rapport, M. D., Sarver, D. E., Raiker, J. S., Orban, S. A., Friedman, L. M., et al. (2013). Reaction time variability in ADHD: a meta-analytic review of 319 studies. Clinical Psychology Review, 33(6), 795e811. http://dx.doi.org/10.1016/ j.cpr.2013.06.001. € rger, N. A., & van der Meere, J. J. Kuntsi, J., Andreou, P., Ma, J., Bo (2005). Testing assumptions for endophenotype studies in ADHD: reliability and validity of tasks in a general population sample. BMC Psychiatry, 5, 40. http://dx.doi.org/10.1186/1471244X-5-40. Kuntsi, J., Frazier-Wood, A. C., Banaschewski, T., Gill, M., Miranda, A., Oades, R. D., et al. (2013). Genetic analysis of reaction time variability: room for improvement? Psychological Medicine, 43(6), 1323e1333. http://dx.doi.org/10.1017/ S0033291712002061. Kuntsi, J., & Klein, C. (2012). Intraindividual variability in ADHD and its implications for research of causal links. Current Topics in Behavioral Neurosciences, 9, 67e91. http://dx.doi.org/10.1007/ 7854_2011_145. Kuntsi, J., Wood, A. C., Van Der Meere, J., & Asherson, P. (2009). Why cognitive performance in ADHD may not reveal true potential: findings from a large population-based sample. Journal of the International Neuropsychological Society;: JINS, 15(4), 570e579. http://dx.doi.org/10.1017/ S135561770909081X.

62

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

Leitner, Y. (2014). The co-occurrence of autism and attention deficit hyperactivity disorder in children e what do we know? Frontiers in Human Neuroscience, 8, 268. http://dx.doi.org/ 10.3389/fnhum.2014.00268. Leth-Steensen, C., Elbaz, Z. K., & Douglas, V. I. (2000). Mean response times, variability, and skew in the responding of ADHD children: a response time distributional approach. Acta Psychologica, 104(2), 167e190. Retrieved from: http://www.ncbi. nlm.nih.gov/pubmed/10900704. Lijffijt, M., Kenemans, J. L., Verbaten, M. N., & van Engeland, H. (2005). A meta-analytic review of stopping performance in attention-deficit/hyperactivity disorder: deficient inhibitory motor control? Journal of Abnormal Psychology, 114(2), 216e222. http://dx.doi.org/10.1037/0021-843X.114.2.216. Luu, P., Tucker, D. M., & Makeig, S. (2004). Frontal midline theta and the error-related negativity: neurophysiological mechanisms of action regulation. Clinical Neurophysiology: Official Journal of the International Federation of Clinical Neurophysiology, 115(8), 1821e1835. http://dx.doi.org/10.1016/ j.clinph.2004.03.031. Makeig, S., Jung, T. P., Bell, A. J., Ghahremani, D., & Sejnowski, T. J. (1997). Blind separation of auditory event-related brain responses into independent components. Proceedings of the National Academy of Sciences of the United States of America, 94(20), 10979e10984. Retrieved from: http://www. pubmedcentral.nih.gov/articlerender.fcgi? artid¼23551&tool¼pmcentrez&rendertype¼abstract. € pcke, C., Berger, C., Wandschneider, R., & Marx, I., Ho Herpertz, S. C. (2013). The impact of financial reward contingencies on cognitive function profiles in adult ADHD. PLoS One, 8(6), e67002. http://dx.doi.org/10.1371/ journal.pone.0067002. McLoughlin, G., Palmer, J. A., Rijsdijk, F., & Makeig, S. (2014). Genetic overlap between evoked frontocentral theta-band phase variability, reaction time variability, and attentiondeficit/hyperactivity disorder symptoms in a twin study. Biological Psychiatry, 75(3), 238e247. http://dx.doi.org/10.1016/ j.biopsych.2013.07.020. di Michele, F., Prichep, L., John, E. R., & Chabot, R. J. (2005). The neurophysiology of attention-deficit/hyperactivity disorder. International Journal of Psychophysiology: Official Journal of the International Organization of Psychophysiology, 58(1), 81e93. http://dx.doi.org/10.1016/j.ijpsycho.2005.03.011. Milne, E. (2011). Increased intra-participant variability in children with autistic spectrum disorders: evidence from single-trial analysis of evoked EEG. Frontiers in Psychology, 2, 51. http:// dx.doi.org/10.3389/fpsyg.2011.00051. Onton, J., Delorme, A., & Makeig, S. (2005). Frontal midline EEG dynamics during working memory. NeuroImage, 27(2), 341e356. http://dx.doi.org/10.1016/j.neuroimage.2005.04.014. O'Connell, R. G., Dockree, P. M., & Kelly, S. P. (2012). A supramodal accumulation-to-bound signal that determines perceptual decisions in humans. Nature Neuroscience, 15(12), 1729e1735. http://dx.doi.org/10.1038/nn.3248. Rabbit, P. M. (1966). Errors and error correction in choice reaction tasks. Journal of Abnormal Psychology, 71, 264e272. Raichle, M. E., & Snyder, A. Z. (2007). A default mode of brain function: a brief history of an evolving idea. NeuroImage, 37(4), 1083e1090. http://dx.doi.org/10.1016/ j.neuroimage.2007.02.041. discussion 1097e9. Rapport, M. D., Alderson, R. M., Kofler, M. J., Sarver, D. E., Bolden, J., & Sims, V. (2008). Working memory deficits in boys with attention-deficit/hyperactivity disorder (ADHD): the contribution of central executive and subsystem processes. Journal of Abnormal Child Psychology, 36(6), 825e837. http:// dx.doi.org/10.1007/s10802-008-9215-y. Raven, J. C. (1965). Advanced progressive Matrices, sets I and II (H. K. Lewi.). London.

Rosa-Neto, P., Lou, H. C., Cumming, P., Pryds, O., Karrebaek, H., Lunding, J., et al. (2005). Methylphenidate-evoked changes in striatal dopamine correlate with inattention and impulsivity in adolescents with attention deficit hyperactivity disorder. NeuroImage, 25(3), 868e876. http://dx.doi.org/10.1016/ j.neuroimage.2004.11.031. Rouder, J. N., Speckman, P. L., Sun, D., Morey, R. D., & Iverson, G. (2009). Bayesian t tests for accepting and rejecting the null hypothesis. Psychonomic Bulletin & Review, 16(2), 225e237. http://dx.doi.org/10.3758/PBR.16.2.225. Rubia, K., Alegria, A. A., Cubillo, A. I., Smith, A. B., Brammer, M. J., & Radua, J. (2014). Effects of stimulants on brain function in attention-deficit/hyperactivity disorder: a systematic review and meta-analysis. Biological Psychiatry, 76(8), 616e628. http:// dx.doi.org/10.1016/j.biopsych.2013.10.016. Saville, C. W. N., Feige, B., Kluckert, C., Bender, S., Biscaldi, M., Berger, A., et al. (2015). Increased reaction time variability in attention-deficit hyperactivity disorder as a response-related phenomenon: evidence from single-trial event-related potentials. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 56(7), 801e813. http://dx.doi.org/10.1111/ jcpp.12348. Saville, C. W. N., Shikhare, S., Iyengar, S., Daley, D., Intriligator, J., Boehm, S. G., et al. (2012). Is reaction time variability consistent across sensory modalities? Insights from latent variable analysis of single-trial P3b latencies. Biological Psychology, 91(2), 275e282. http://dx.doi.org/10.1016/ j.biopsycho.2012.07.006. Shallice, T., Marzocchi, G. M., Coser, S., Del Savio, M., Meuter, R. F., & Rumiati, R. I. (2002). Executive function profile of children with attention deficit hyperactivity disorder. Developmental Neuropsychology, 21(1), 43e71. http://dx.doi.org/ 10.1207/S15326942DN2101_3. Sheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weiller, E., et al. (1998). The Mini-International Neuropsychiatric Interview (M.I.N.I.): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. The Journal of Clinical Psychiatry, 59(Suppl 2), 22e33. quiz 34e57. Retrieved from: http://www.ncbi.nlm. nih.gov/pubmed/9881538. Snyder, S. M., & Hall, J. R. (2006). A meta-analysis of quantitative EEG power associated with attention-deficit hyperactivity disorder. Journal of Clinical Neurophysiology: Official Publication of the American Electroencephalographic Society, 23(5), 440e455. http://dx.doi.org/10.1097/01.wnp.0000221363.12503.78. Sonuga-Barke, E. J. S., & Castellanos, F. X. (2007). Spontaneous attentional fluctuations in impaired states and pathological conditions: a neurobiological hypothesis. Neuroscience and Biobehavioral Reviews, 31(7), 977e986. http://dx.doi.org/10.1016/ j.neubiorev.2007.02.005. Spencer, S. V., Hawk, L. W., Richards, J. B., Shiels, K., Pelham, W. E., & Waxmonsky, J. G. (2009). Stimulant treatment reduces lapses in attention among children with ADHD: the effects of methylphenidate on intra-individual response time distributions. Journal of Abnormal Child Psychology, 37(6), 805e816. http://dx.doi.org/10.1007/s10802-009-9316-2. Spinelli, S., Vasa, R. A., Joel, S., Nelson, T. E., Pekar, J. J., & Mostofsky, S. H. (2011). Variability in post-error behavioral adjustment is associated with functional abnormalities in the temporal cortex in children with ADHD. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 52(7), 808e816. http://dx.doi.org/10.1111/j.1469-7610.2010.02356.x. Suskauer, S. J., Simmonds, D. J., Fotedar, S., Blankner, J. G., Pekar, J. J., Denckla, M. B., et al. (2008). Functional magnetic resonance imaging evidence for abnormalities in response selection in attention deficit hyperactivity disorder: differences in activation associated with response inhibition but not habitual motor response. Journal of Cognitive

c o r t e x 8 1 ( 2 0 1 6 ) 5 0 e6 3

Neuroscience, 20(3), 478e493. http://dx.doi.org/10.1162/ jocn.2008.20032. Swanson, J. M., Kinsbourne, M., Nigg, J., Lanphear, B., Stefanatos, G. A., Volkow, N., et al. (2007). Etiologic subtypes of attention-deficit/hyperactivity disorder: brain imaging, molecular genetic and environmental factors and the dopamine hypothesis. Neuropsychology Review, 17(1), 39e59. http://dx.doi.org/10.1007/s11065-007-9019-9. Teicher, M. H., Lowen, S. B., Polcari, A., Foley, M., & McGreenery, C. E. (2004). Novel strategy for the analysis of CPT data provides new insight into the effects of methylphenidate on attentional states in children with ADHD. Journal of Child

63

and Adolescent Psychopharmacology, 14(2), 219e232. http:// dx.doi.org/10.1089/1044546041648995. Weinger, P. M., Zemon, V., Soorya, L., & Gordon, J. (2014). Lowcontrast response deficits and increased neural noise in children with autism spectrum disorder. Neuropsychologia, 63, 10e18. http://dx.doi.org/10.1016/ j.neuropsychologia.2014.07.031. Willcutt, E. G., Doyle, A. E., Nigg, J. T., Faraone, S. V., & Pennington, B. F. (2005). Validity of the executive function theory of attention-deficit/hyperactivity disorder: a metaanalytic review. Biological Psychiatry, 57(11), 1336e1346. http:// dx.doi.org/10.1016/j.biopsych.2005.02.006.