Edited by: Lena Palaniyappan, University of Nottingham, UK
Reviewed by: Christian Sorg, Klinikum rechts der Isar der Technischen Universität München, Germany; Rajeev Krishnadas, University of Glasgow, UK; Thomas P. White, King’s College London, UK
This article was submitted to Neuropsychiatric Imaging and Stimulation, a section of the journal Frontiers in Psychiatry.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Abundant research has identified abnormal frontal-striatal brain function in attention-deficit hyperactivity disorder (ADHD) (
Multiple functional imaging studies have shown abnormal activation or metabolic effects during rest and/or sub-executive operations in ADHD (
Our research in this domain begins with the precept that complex task-directed actions are likely to rely on a specific manner of sensory information processing that facilitates fast categorical parsing of sensory data. To illustrate, if a person wants to find a pen on a cluttered countertop in order to sign a document, it is task-adaptive to quickly identify (i.e., categorize) that stimulus using the minimal sensory detail required. Here, the pen’s esthetic details and any surrounding content are task-extraneous. Alternatively, if an artist wants to paint a still-life portrait of this pen, they should indulge as much detail as possible. One approach seeks to identify a stimulus using the minimal sensory detail required. The other seeks to indulge as much sensory detail as possible in order to produce a prolonged sensory-immersive experience. We theorize that ADHD involves a reduced capacity for the former mode of processing.
This task-specialized manner of sensory information processing likely depends on the coordinated function of multiple distributed brains systems that get dynamically integrated in service to task-directed actions [for full description of this model see Ref. (
Evidence from our previous behavioral laterality studies in ADHD adults supports the presence of a right-hemisphere bias. These demonstrated greater RH contribution to processing task stimuli, associated left hemisphere (LH) linguistic impairments, and abnormal interhemispheric interaction (
Although not yet widely understood, this pattern of findings is well aligned with extant ADHD literature. As noted, slow naming speed is identified in ADHD, which is consistent with impoverished LH contribution to sensory encoding. Previous behavioral laterality studies of ADHD have also indicated increased RH contribution (
A similar pattern of reduced LH and increased RH contributions is evident during more complex tasks; however, this literature is more variable, showing diffuse effects consistent with multiple weaknesses across distributed brain-systems (
Thus, the literature strongly implicates some form of increased weighting of non-verbal sensory processing in ADHD. We hypothesize that this stems from variable impairments to task-directed brain functions that otherwise facilitate fast/efficient identification and verbal encoding of task relevant stimuli (for model description: (
Methods for the direct analysis of EEG asymmetry are well developed, and as noted, have consistently shown R > L patterns in ADHD. However, related fMRI methods to assess the asymmetry of BOLD signal have only recently begun to overcome methodological difficulties involving thresholding techniques (
Furthermore, since we theorize that asymmetry in low-level perceptual processing is directly related to abnormalities in higher-level processing, we sought to understand the relationship between perceptual asymmetry and activity in other brain networks. Patterns of intrinsic functional connectivity in the brain have revealed multiple networks of brain regions whose activity is correlated during rest (
Subjects were recruited from Los Angeles County and the surrounding regions using a database of participants from previous UCLA studies who indicated they were willing to participate in future studies. Subjects were also recruited through flyers posted near UCLA, and advertisements on focus group websites (e.g., parenting blogs). Given our interest to examine ADHD-specific asymmetry effects, we chose to limit possible variability in brain-laterality due to gender, handedness, and/or variation in pubertal onset (
After receiving verbal and written explanations of study requirements a parent and the participating child provided written informed consent/assent, as approved by the UCLA Institutional Review Board. To screen for ADHD and other psychiatric disorders using DSM-IV criteria, participating children and their mothers were interviewed using the semi-structured interview of the Schedule for Affective Disorder and Schizophrenia for School-Age Children-Present and Lifetime Version (K-SADS-PL) (
Handedness was assessed with a shortened version of the Edinburgh Handedness Inventory (
Clinical variables | Controls |
ADHD |
Statistic |
---|---|---|---|
IQ | |||
Age | |||
SES | |||
ADHD type | N/a | 11C, 10I | N/a |
Anxiety | 0 affected | 1 affected (GAD) | N/a |
Mood | 0 affected | 0 affected | N/a |
ODD | 0 affected | 6 affected | fe: |
CD | 0 affected | 1 affected | N/a |
Handedness score | |||
Vocabulary | |||
Phonology | |||
Reading | |||
Spelling |
Thirty-one ADHD and 25 typically developing right-handed male children between the ages of 11 and 16 underwent fMRI procedures. Ten ADHD subjects were excluded (5 = motion, 2 = non-compliance, 1 = sleep, 1 = non-tolerant of fMRI environment, and 1 = image distortion from permanent retainer). Four control subjects were excluded (1 = medical problem that impacted brain development, 1 = father diagnosed with ADHD, 1 = borderline ADHD, and 1 = non-tolerant of fMRI environment). The final sample consisted of 21 ADHD and 21 control subjects. The ADHD sample was 81% Caucasian, 14% African American, and 5% Hispanic. The control sample was 62% Caucasian, 9.5% African American, 19% Hispanic, and 9.5% Asian.
The fMRI task was adapted from a previous block-design study that uncovered robust laterality differences for “letter” versus “spatial” processing in healthy adults (
Stimuli were generated using the MRC Psycholinguistic Database (
Two data collection runs were performed. Each presented eight task-blocks (four location and four letter) interspersed with seven baseline conditions. Within runs, the order of task-blocks was randomized, with pre-block instruction screens indicating which task to perform. Task-blocks contained 12 2-s randomly jittered trials (±250 ms) – 6 with target “A”s, 6 without, and among these sets, an equal number of targets in the second or third position. The order of trial types was randomized within blocks. During trials, words were presented centrally for 150 ms in all capital 48-point black-font (except for the red target letter). A central fixation cross was displayed between stimulus presentations. Baseline conditions contained eight trials and used the same stimulus presentation parameters. The 192 task-stimuli were newly randomized for each subject, with no stimuli repeating across both runs. Stimulus presentation and response collection were controlled using MATLAB (The Mathworks, Inc.) and the Psychophysics Toolbox (
fMRI procedures were a component of a broader protocol. On the first day, subjects underwent clinical, cognitive, and EEG assessments. On the second day, they underwent fMRI consenting, safety screening, training, and testing. The mean time difference between days 1 and 2 was 91.5 days for ADHD subjects, and 56.3 days for controls (no statistical group difference). Before fMRI scanning, task training occurred via a standardized computer program implemented using E-prime software (Psychology Software Tools, Inc.). Although the program was designed to operate automatically, research staff read aloud the instructions and prompted subjects to repeat any training module not performed above chance. Task training was performed to reduce the likelihood of capturing brain activation associated with task learning during scanning procedures.
The program first introduced subjects to each of the task conditions. This required active participation as subjects learned about stimuli and associated response mappings for each condition (location, letter, and baseline). Each of these modules ended with a practice that provided trial-by-trial and overall performance feedback. Next, the program portrayed the intermixing of blocked-conditions and associated instruction screens that signaled which task to perform. The instruction screens were identical to those used in the scanner. These screens were designed to signal which task to perform next, and provide prompts to help children remember condition-specific response mappings (i.e., instruction screen graphics displayed associated response mappings). This task-mixing practice section also ended with a brief practice that provided overall performance feedback. Finally, subjects underwent a mock run of the experiment exactly as presented in the scanner, barring a few differences (different word stimuli, keyboard responses, and overall performance feedback).
Task training took approximately 30 min, after which, subjects and a parent walked to the fMRI facility, where they waited in a lounge during set-up. During this time, subjects were encouraged to explore a nearby mock-scanner, listen to recordings of MRI and fMRI scanner noises, and practice inserting earplugs. After fMRI equipment and software set-up was complete, subjects entered the scanner control room and were given an opportunity to become familiar/comfortable with the environment, as well as select a movie to watch during set-up and structural imaging. Upon entering the scanner-room, subjects were instructed to use the critical equipment (response box, head phones, goggles, and emergency button) and were allowed to watch their selected movie (via fMRI goggles) during additional set-up and shimming procedures. Throughout these and subsequent scanning procedures, a concerted effort was made to keep children actively engaged and comfortable.
Before running the fMRI task, children were shown a “start screen.” This reminded them what each of the instruction screens looked like and repeated general task-instructions. A research staff read the instructions to the subject and prompted them to demonstrate button presses associated with each condition. This assured us that children were using the button box correctly, and it made the subjects aware that we were able to monitor their button presses in real-time.
This study was conducted at the Staglin IMHRO Center at UCLA. MRI recording was performed with a standard 12-channel head coil on a Siemens 3T Trio Magnetic Resonance Imaging System with TIM. Two functional runs including 195 volumes each were acquired. These images were collected over 33 axial slices covering the whole cerebral volume using a T2*-weighted gradient-echo sequence (TR = 2000 ms, TE = 30 ms, flip angle = 78°, matrix size 64 × 64, 3-mm in-plane resolution, 3-mm thick slices, and 0.75-mm gap). For each participant, a high-resolution MP-RAGE structural volume was also acquired (TR = 1900, TE = 2.26, and flip angle = 9°) with 176 sagittal slices, each 1 mm thick with 1 mm × 1 mm in-plane resolution.
Analysis was carried out using FSL’s FMRI Expert Analysis Tool (FEAT) Version 5.1 (FMRIB’s Software Library,
Each subject’s statistical data were then warped into a standard-space based on the MNI-152 atlas. We used FLIRT to register the functional data to the atlas space in three stages. First, functional images were aligned with the high-resolution co-planar T2-weighted image using a six-degrees-of-freedom rigid-body warping procedure (
After analyzing each functional run for each subject, the two functional runs were combined using a fixed-effects analysis. Data from each subject were then passed into a higher-level analysis, which allowed comparisons within and between groups. Higher-level analysis was carried out using FLAME (FMRIB’s Local Analysis of Mixed Effects), such that group-level effects were modeled using random effects (
The purpose of our asymmetry analysis was twofold. First, we intended to characterize brain asymmetry in patients and controls in visual processing regions of the brain, i.e., those regions involved with the perceptual processing of the stimuli during the task. Second, we intended to probe how asymmetry in visual areas was related to processing in several key networks throughout the brain, several of which are suspected to play a role in ADHD. Recent work in neuroimaging has shown that the brain can be parceled into distinct networks based on intrinsic functional connectivity at rest, and that these networks may represent meaningful cognitive units (
The most common approach to quantifying asymmetries of functional brain activation in the neuroimaging literature is to compute an AI as the ratio of the difference between left hemisphere activation (LHA) and right-hemisphere activation (RHA) and the sum of activation in both hemispheres:
Two types of strategy for dealing with thresholding issues have recently emerged: (1) AI values are computed across a range of threshold values instead of a single threshold, and laterality curves are presented (
Asymmetry indexes were computed using the iBrain Laterality Toolbox (
Group differences in the adaptive-threshold based VN-AI were examined for each condition (all–baseline, letter–baseline, and location–baseline) using univariate ANOVA (adjusted for age), and are considered our primary analyses of visual network asymmetry. AI-curves are included mainly for visual inspection; however, a “principal components analysis” (PCA) based assessment of AI-curves is also presented as a secondary statistical approach.
Note that contrasts used in these asymmetry analysis (letter–baseline, location–baseline, and all–baseline) involved comparison of conditions that had identical visual stimuli, and thus produced modest visual network activation. The primary adaptive-threshold approach contends with this by normalizing each subject’s AI score to their own mean signal strength within the visual network. However, with AI-curves, the maximum
Group differences across the 20
Finally, a key goal of the current study was to examine the relationship between hypothesized visual processing asymmetries in ADHD and identified functional networks suspected to play a role in the disorder (
Group differences in letter- and location-task behavioral performance (accuracy and response time) were tested using univariate ANOVA (adjusted for age). Two additional analyses used partial correlations (adjusted for age) to examine the relationship between letter-task performance and VN-AI values, and mean signal intensity across the six extra-visual networks examined in this study.
Where relevant, we used Fisher’s
Controls exhibited better accuracy during the letter task, and a trend suggested the same during the location task (Table
Behavior measure | Controls |
ADHD |
Statistic |
||||
---|---|---|---|---|---|---|---|
SE | SE | df | |||||
Letter Accuracy | 0.91 | 0.018 | 0.86 | 0.018 | 4.4 | 2, 39 | 0.043 |
Letter RT | 590 | 21 | 590 | 21 | 0.007 | 2, 39 | 0.93 |
Location Accuracy | 0.93 | 0.017 | 0.88 | 0.017 | 3.29 | 2, 39 | 0.08 |
Location RT | 490 | 17 | 500 | 17 | 0.141 | 2, 39 | 0.71 |
Both groups exhibited significant activation of the occipital cortex (extending into fusiform regions), but in opposite hemispheres (RH in ADHD, LH in controls). ADHD and controls also exhibited several overlapping activations in LH brain regions that included supplementary motor, pre-central gyrus (superior lateral, inferior medial, plus inferior lateral in controls), and post-central gyrus bordering the supramarginal gyrus (extending into superior parietal cortex in controls). Lastly, ADHD subjects showed additional unique activations in the brain stem and cerebellum (Table
Control | ADHD | |||
---|---|---|---|---|
Region | Hem. | MNI | ||
Supplementary motor cortex | L | −10, 2, 50 | 2.93 | 3.4 |
Superior lateral pre-central gyrus | L | −40, −12, 66 | 6.03 | 5.63 |
Inferior lateral pre-central gyrus | L | −44, −2, 28 | 4.05 | None |
Inferior medial pre-central gyrus | L | −24, −12, 50 | 5.02 | 4.34 |
Supramarginal gyrus | L | −38, −36, 38 | 4.83 | 3.93 |
Occipital cortex | L | −36, −94, −4 | 5.62 | None |
Occipital cortex | R | 40, −82, −4 | None | 4.29 |
Brain stem | Mid | −4, −22, −16 | None | 3.72 |
Brain stem | Mid | 0, −38, −24 | None | 4.07 |
Cerebellum lobule VI | R | 26, −46, −28 | None | 5.04 |
Supplementary motor cortex | L | −10, 2, 50 | 2.37 | None |
Superior lateral pre-central gyrus | L | −40, −12, 66 | 5.24 | 4.87 |
Inferior lateral pre-central gyrus | L | −44, −2, 28 | 3.80 | None |
Inferior medial pre-central gyrus | L | −24, −12, 50 | 4.20 | 3.90 |
Supramarginal gyrus | L | −38, −36, 38 | 4.44 | 3.93 |
Occipital cortex | L | −36, −94, −4 | 5.17 | None |
Thalamus | L | −18, 20, 6 | None | 2.93 |
Supplementary motor cortex | L | −12, 2, 46 | 2.53 | None |
Superior lateral pre-central gyrus | L | −40, −12, 66 | 5.16 | 4.90 |
Inferior lateral pre-central gyrus | L | −46, −2, 26 | 3.60 | None |
Inferior medial pre-central gyrus | L | −24, −12, 50 | 5.14 | 4.34 |
Supramarginal gyrus | L | −36, −36, 36 | 4.50 | 3.66 |
Occipital cortex | L | −32, −92, −4 | 4.98 | None |
Occipital cortex | R | 34, −86, −4 | None | 4.0 |
Brain stem | Mid | 2, −26, −18 | None | 3.31 |
Brain stem | Mid | 0, −34, −24 | None | 3.47 |
Cerebellum lobule VI | R | 24, −48, −28 | None | 3.91 |
Hippocampus | L | −32, −26, −14 | None | 3.53 |
Hippocampus | R | 32, −34, −4 | None | 2.60 |
This contrast showed the same basic pattern as all–baseline except for the following: only controls activated supplementary motor cortex; significant occipital, brain stem, or cerebellum activations were not present in ADHD subjects; and ADHD subjects showed a unique activation in the left thalamus (Table
This contrast showed the same basic pattern as all–baseline, as well as additional unique hippocampal activations among ADHD subjects (Table
Direct comparison between groups did not show significant differences.
There were no significant effects for this contrast.
Attention-deficit hyperactivity disorder and control subjects exhibited several overlapping activations in brain regions associated with the DMN (medial prefrontal, medial parietal, and inferior parietal cortices). Additional unique activations were evident in subcortical regions among ADHD subjects, and within somatomotor regions among in controls (Figure
Analysis of group differences in the adaptive-threshold based VN-AI showed that controls had significantly greater leftward asymmetry for the all–baseline and letter–baseline contrasts (Table
fMRI contrasts | Controls |
ADHD |
Statistic |
||||
---|---|---|---|---|---|---|---|
SE | SE | df | |||||
All–baseline | 0.18 | 0.06 | −0.06 | 0.06 | 6.8 | 1,39 | 0.013 |
Letter–baseline | 0.21 | 0.06 | −0.03 | 0.06 | 6.5 | 1,39 | 0.014 |
Location–baseline | 0.09 | 0.08 | −0.08 | 0.08 | 2.2 | 1,39 | 0.15 |
Although ADHD subjects did not differ on vocabulary and phonological measures, a trend effect (
For AI-curve values that contained the full sample (
Consistent with primary VN-AI analysis, analysis of PCA asymmetry components derived from thresholds
fMRI contrasts | Controls |
ADHD |
Statistic |
||||
---|---|---|---|---|---|---|---|
SE | SE | df | |||||
All–baseline | 0.36 | 0.20 | −0.36 | 0.20 | 6.3 | 1,39 | 0.016 |
Letter–baseline | 0.34 | 0.20 | −0.34 | 0.20 | 5.5 | 1,39 | 0.024 |
Location–baseline | 0.23 | 0.22 | −0.22 | 0.22 | 2.1 | 1,39 | 0.16 |
Please note that additional analyses involving visual network asymmetry exclusively utilized the letter–baseline adaptive-threshold VN-AI metric. Moreover, in the assessment of VN-AI association with extra-visual networks, extra-visual-network values were derived exclusively from the letter–baseline condition where group differences in visual network asymmetry occurred. Also, note from Figure
Partial correlation analysis (adjusted for age) showed that VN-AI in ADHD subjects was generally and positively correlated with signal in extra-visual networks during the letter task, with the effect surviving Bonferroni correction for the somatomotor, ventral-attention, and DMN. In contrast, controls showed a pattern of negative (but mostly non-significant) associations between VN-AI and signal in extra-visual networks. One effect in controls (i.e., VN-AI correlation to LIM) was significant and survived Bonferroni correction. Fisher’s
Asymmetry | SOM* | DAN* | VAN* | LIM* | FPN* | DMN* | |
---|---|---|---|---|---|---|---|
A: VN-AI | 0.66 |
0.49 |
0.66 |
0.40 |
0.52 |
0.57 |
|
C: VN-AI | −0.29 |
−0.22 |
−0.19 |
−0.63 |
−0.27 |
−0.31 |
There were no significant effects in controls. ADHD subjects showed exclusive positive association between VN-AI and BOLD response across multiple extra-visual brain regions, the majority of which fell within the DMN. The ADHD-exclusive associations produced significant group differences (Table
ADHD | ||||
---|---|---|---|---|
Region | Hem | MNI | ||
Frontal pole (lateral) | L | −28, 52, 34 | 3.6 | 2.5 |
Frontal pole (lateral) | R | 46, 34, −8 | 3.8 | 2.9 |
Frontal pole (mid) | L | −10, 72, 6 | 3.5 | 3.7 |
Frontal pole (mid) | R | 12, 64, 20 | 3.0 | 3.0 |
Superior-frontal gyrus (lat) | L | −20, 16, 46 | 3.6 | |
Superior-frontal gyrus (mid) | Mid | 4, 50, 36 | 3.0 | 2.4 |
Inferior frontal gyrus | R | 48, 16, 16 | 3.4 | 3.7 |
Inferior frontal gyrus | R | 58, 34, 10 | 3.6 | 3.0 |
Frontal operculum cortex | R | 44, 16, 8 | 3.4 | 2.9 |
Frontal operculum cortex | R | 46, 0, 14 | 3.7 | 3.9 |
Paracingulate gyrus | Mid | −2, 48, 14 | 3.1 | 2.9 |
Paracingulate gyrus | Mid-R | 8, 48, 14 | 3.5 | 2.9 |
Pre-central gyrus (Inf) | R | 62, 4, 14 | 3.2 | |
Post-central gyrus (Inf) | R | 64, −10, 24 | 3.1 | 2.8 |
Post-central gyrus (Sup) | R | 58, −10, 48 | 2.8 | 3.3 |
Temporal pole | R | 46, 8, −40 | 3.0 | 2.8 |
Middle temporal gyrus (ant) | R | 48, −2, −28 | 3.8 | 3.8 |
Middle temporal gyrus (Inf) | R | 56, −28, −14 | 4.0 | 3.7 |
Inferior temporal gyrus | R | 52, −28, −22 | 2.6 | 3.3 |
Middle temporal gyrus (post) | R | 64, −22, −8 | 3.6 | 3.0 |
Temporal-occipital cortex | L | −66, −52, −8 | 3.9 | 3.3 |
Temporal-occipital cortex | R | 60, −44, 6 | 3.0 | 2.7 |
Superior temporal gyrus | L | −64, −10, 6 | 4.3 | 3.0 |
Superior temporal gyrus (Lat) | R | 62, −30, 2 | 3.4 | 3.2 |
Superior temporal gyrus (post) | L | −60, −40, 10 | 4.1 | 3.6 |
Superior temporal gyrus (post) | R | 44, −34, 4 | 4.1 | 3.2 |
Angular gyrus | L | −62, −58, 30 | 4.0 | 3.5 |
Angular gyrus | R | 62, −50, 28 | 3.7 | 3.9 |
Precuneus cortex | Mid-L | −6, −52, 46 | 4.4 | |
Precuneus cortex | Mid | 4, −52, 48 | 3.5 | |
Superior parietal lobule | Mid-L | −6, −56, 66 | 4.1 | |
Superior parietal lobule | Mid | 4, −56, 62 | 3.2 |
These analyses are performed exclusively in ADHD subjects, as there was insufficient variability in symptoms metrics in controls to justify examination. Symptom data reflect DSM-IV criteria for inattentive and hyperactive subscales, obtained during K-SAD-PL semi-structured interviews (with mother and child participant). Due to the relative importance of symptom effects in ADHD, we have provided scatter plots to help guide interpretation of correlation findings.
Partial correlations (adjusted for age) indicated no relationship between symptoms and VN-AI during the letter–baseline condition.
Partial correlations (adjusted for age) indicated no relationship between symptoms and extra-visual networks during letter task (i.e., letter–baseline). However, trend level effects suggested possible associations between inattention and limbic (
For all–baseline and location–baseline contrast, ADHD subjects exhibited several positive associations between inattentive symptoms and BOLD signal in medial prefrontal brain regions. There were no associations for hyperactive symptoms (Table
ADHD |
|||
---|---|---|---|
Region | Hem | MNI | |
Frontal pole | L | −14, 56, 8 | 3.70 |
Frontal pole | Mid | −4, 62, 8 | 3.51 |
Frontal pole | Mid-R | 8, 58, 10 | 3.06 |
Frontal medial cortex | Mid | 2, 48, 14 | 2.73 |
Paracingulate gyrus | L | −16, 46, −2 | 4.42 |
Paracingulate gyrus | Mid | 0, 46, 8 | 3.33 |
Paracingulate gyrus | R | 16, 54, 2 | 4.11 |
Frontal pole | L | −14, 56, 10 | 4.06 |
Frontal pole | Mid | −4, 64, 10 | 2.90 |
Frontal pole | R | 14, 60, 4 | 4.36 |
Frontal pole | R | 10, 62, 30 | 3.08 |
Frontal medial cortex | L | −18, 48, −4 | 4.08 |
Cingulate gyrus (anterior) | Mid-L | −8, 36, 6 | 3.58 |
Partial correlation analysis (adjusted for age) demonstrated that neither group showed any association between letter-task accuracy and VN-AI. Both groups showed non-significant associations between letter-task RT and VN-AI [ADHD (
In ADHD subjects, letter-task performance was not significantly correlated with extra-visual networks. In controls, accuracy was negatively correlated with the limbic (
In ADHD subjects, location-task performance was not significantly associated with extra-visual networks. In controls, accuracy was negatively correlated with DMN activation (
Partial correlations (adjusted for age) indicated no relationship between ADHD symptoms and behavioral performance.
Controls exhibited several positive associations between response time and BOLD signal in somatomotor brain regions. There were no group differences (see Part 4 in Supplementary Material for details).
Controls showed negative associations between tasks accuracy and BOLD signal in brain regions understood to reflect DMN activation. These associations also produced significant group differences (Table
Controls | ||||
---|---|---|---|---|
Region | Hem | MNI | ||
Frontal pole | L | −18, 54, 40 | −3.20 | |
Frontal pole | Mid | 2, 54, −24 | 3.01 | |
Frontal orbital cortex (inferior) | L | −10, 6, −20 | 3.83 | |
Frontal medial cortex | R | 10, 42, −16 | 3.31 | |
Subcallosal cortex | Mid-L | −6, 28, −12 | −3.45 | 3.42 |
Subcallosal cortex | Mid | 4, 26, −14 | −3.13 | 3.50 |
Cingulate gyrus (anterior) | Mid | 0, 42, 8 | −3.41 | 4.05 |
Paracingulate gyrus | L | −14, 38, 18 | −3.24 | |
Occipital pole | L | −26, −100, −16 | 3.11 | |
Cerebellum | L | −12, −80, −48 | 3.72 | |
Cerebellum | Mid-R | 6, −86, −42 | 3.43 | |
Frontal pole (inferior) | R | 14, 36, −26 | −4.16 | 4.47 |
Frontal pole (superior) | R | 22, 44, 38 | −3.95 | |
Frontal pole | Mid-L | −8, 64, −2 | −3.37 | 2.94 |
Frontal pole | L | −22, 58, 24 | −2.98 | |
Frontal orbital cortex | L | −12, 22, −24 | 4.05 | |
Frontal orbital cortex | R | 10, 30, −22 | −4.88 | 4.71 |
Subcallosal cortex | Mid-R | 8, 26, −22 | −4.68 | 4.83 |
Middle frontal gyrus | L | −34, 24, 44 | −3.28 | |
Superior-frontal gyrus | L | −18, 26, 52 | −3.19 | |
Frontal medial cortex | Mid | −2, 52, −26 | −3.02 | 2.74 |
Cingulate gyrus (anterior) | Mid | 0, 28, 14 | −3.60 | |
Cingulate gyrus (anterior) | Mid | 2, 36, 20 | −3.44 | |
Paracingulate gyrus (ant) | R | 12, 38, 24 | −3.36 | |
Paracingulate gyrus (dorsal) | Mid | 2, 20, 48 | −3.06 |
For the purpose of data interpretation, two additional
The current study used recently developed fMRI methods to replicate and further examine identified abnormal rightward biased information processing in ADHD. Our task presented four-letter word stimuli and required subjects to detect a uniquely colored red letter and decide whether it was an “A” or not (letter task), or whether it was on the left or right (location task). Initial within-group analyses revealed a pattern of left-lateralized visual cortical activity in controls, but right-lateralized visual cortical activity in ADHD children. Our primary direct analyses of visual network asymmetry (VNA) confirmed that atypical rightward VNA was present in ADHD children and significantly different from controls in the letter task and overall. This finding adds to the growing literature that identifies abnormal information processing to be a key factor in ADHD. Moreover, in conjunction with our previous work (see
Through our secondary aim, we additionally demonstrated that ADHD subjects’ rightward VNA during the letter task was atypically associated with reduced DMN activation. Recall that positive VNA scores reflect leftward asymmetry, with the reverse also true (i.e., negative VNA scores reflect rightward asymmetry). We found, in two separate analyses of BOLD signal in the letter task that ADHD subjects exhibited an atypical positive correlation between VNA and DMN signal. This indicates that leftward VNA is associated with greater DMN activation, and that rightward VNA is associated with reduced DMN activation. Given ADHD subjects’ atypical rightward VNA during the letter task, we focus our discussion of DMN findings on the link between rightward VNA and reduced DMN activation in ADHD. Regardless of the directionality, this and additional network findings importantly demonstrate that atypical rightward VNA in ADHD is associated with multiple distributed brain-systems, including previously implicated large-scale networks and frontal brain regions.
Functional abnormalities in the visual cortex have proven to be a key feature of ADHD (
Hemispheric specialization of visual cortical functions notably includes LH specialization for linguistic stimuli and RH specialization for faces (for review see Ref. (
The above noted right-lateralized brain functions reflect two classes of sensory information processing: self-directed top-down and automatic bottom-up. Within these domains, we can further distinguish processing that supports fast stimulus identification (i.e., categorization) versus in-depth sensory analysis. In the top-down domain, this reflects applied effort to identify/categorize a stimulus, or to scrutinize a stimulus’s details (
We have previously hypothesized (
To examine this thesis, our current study was designed so that task conditions differentially engaged the task-specialized manner of visual information processing, but were otherwise perceptually identical. The letter task required subjects to identify a nominated target “A,” and distinguish it from other letters. This was expected to tax RH mechanisms that support top-down selective attention (
The above discussion addresses possible sources of atypically increased RH contribution to visual processing, however, our current study did not, strictly speaking, uncover such an effect. We demonstrated increased rightward VNA in ADHD. This indicates a
Consistent with this, both developmental and adult studies show a transfer of right to LH processing of visual information that coincides with the learning of new visual items and their name codes (
Evidence of increased rightward VNA in ADHD is well aligned with identified reduced posterior corpus callosum size (
Another aspect of collosal functioning that is perhaps relevant to ADHD and our current finding has to do with the directionality of interhemipsheric transfer. Although previously considered symmetric, a recent study showed that a greater proportion of splenial collosal fibers project right-to-left than left-to-right (
Our previous work indicated that atypical rightward asymmetry in ADHD is sensitive to top-down modulation of attention and brain-state orientation (
Default mode network function has been widely investigated in recent years, with multiple studies linking it to ADHD (
The current study showed that VNA in ADHD was more generally and robustly associated with extra-visual networks compared to controls. The generality of these associations may fit with the above view insomuch as abnormal DMN function in ADHD might be synonymous with having a less stable task-directed neural architecture (
In addition to the above noted general effects, a critical role for DMN function in ADHD was also directly indicated. VNA association with DMN signal was one of the three effects that survived Bonferroni correction for multiple testing. Moreover, DMN signal showed unique abnormal association to both inattentive symptoms and behavioral performance in ADHD subjects. Inattentive symptoms showed a positive association with medial anterior aspects, while ADHD subjects showed no behavioral association with DMN function, with controls exhibiting the expected pattern of greater accuracy with reduced DMN activation (also involving medial anterior aspects). Moreover, and consistent with the above discussion, these effects occurred mainly during the more difficult letter-task condition, which ostensibly placed greater demands on internal task processing, possibly including an increased requirement for DMN modulation of task-positive networks (
With regard to the directionality of effects, our findings showed a pattern of positive association between the VN-AI metric and all extra-visual networks examined. This means that among ADHD subjects leftward VNA was associated with stronger network signal, while rightward VNA was associated with reduced network signal. Given our primary finding of increased rightward VNA in ADHD, the latter aspect is most relevant. That is, atypical increased rightward VNA in ADHD during the letter task was associated with reduced network signal, most notably for the default mode and VAN. Reduced DMN activation occurs with active externally oriented processing (
As noted, medial prefrontal aspects of the DMN network were associated with ADHD inattentive symptoms and task performance. This brain region has been identified as a source of top-down regulation of the brain-stem locus coeruleus (
Default mode network influence over applied attention may also occur. As noted, Wang et al. (
The current study demonstrated rightward VNA in ADHD during a simple letter discrimination task. This result, in conjunction with our previous findings, adds an important novel consideration to the growing literature identifying abnormal visual sensory information processing in ADHD. We expect rightward VNA reflects increased perceptual engagement of task-extraneous content, and that this occurs with any form of reduced ability for top-down task-directed visual sensory information processing. The current study also identified that rightward VNA in ADHD was atypically and robustly associated with multiple extra-visual network systems, namely the DMN and VAN. Rightward VNA in ADHD was associated with reduced activation in these networks, possibly indicating some form of task-adaptive compensatory processing. Moreover, we also identified abnormal DMN associations with ADHD inattentive symptoms and behavioral performance during our letter task. We postulate that abnormal DMN function in ADHD may index a general reduced capacity to induce and/or maintain a task-adaptive neural architecture, with negative cascading effects resulting in less efficient task-directed perceptual encoding of visual stimuli, and associated increased rightward VNA.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at
This work was funded by National Institute of Mental Health Grant by the National Institute of Mental Health Grant MH082104 (PI Hale).