Targeting working memory to modify emotional reactivity in adult attention deficit hyperactivity disorder: a functional magnetic resonance imaging study

Abstract

Understanding the neural mechanisms of emotional reactivity in Attention-Deficit/Hyperactivity Disorder (ADHD) may help develop more effective treatments that target emotion dysregulation. In adult ADHD, emotion regulation problems cover a range of dimensions, including emotional reactivity (ER). One important process that could underlie an impaired ER in ADHD might be impaired working memory (WM) processing. We recently demonstrated that taxing WM prior to the exposure of emotionally salient stimuli reduced physiological and subjective reactivity to such cues in heavy drinkers, suggesting lasting effects of WM activation on ER. Here, we investigated neural mechanisms that could underlie the interaction between WM and ER in adult ADHD participants. We included 30 male ADHD participants and 30 matched controls. Participants performed a novel functional magnetic resonance imaging paradigm in which active WM-blocks were alternated with passive blocks of negative and neutral images. We demonstrated group-independent significant main effects of negative emotional images on amygdala activation, and WM-load on paracingulate gyrus and dorsolateral prefrontal cortex activation. Contrary to earlier reports in adolescent ADHD, no impairments were found in neural correlates of WM or ER. Moreover, taxing WM did not alter the neural correlates of ER in either ADHD or control participants. While we did find effects on the amygdala, paCG, and dlPFC activation, we did not find interactions between WM and ER, possibly due to the relatively unimpaired ADHD population and a well-matched control group. Whether targeting WM might be effective in participants with ADHD with severe ER impairments remains to be investigated.

Keywords: Adult Attention-Deficit/Hyperactivity Disorder (ADHD); Emotional dysregulation; Emotional reactivity; Functional magnetic resonance imaging (fMRI); Working memory.

Fig. 1

Study design: A) Interleaved active n-back blocks and passive emotional stimuli blocks. The effect of WM-blocks preceding emotional stimuli was assessed by subdividing the emotional stimuli into four conditions: neutral-after-0-back (0N), emotional-after-0-back (0E), neutral-after-2-back (2N), emotional-after-2-back (2E). B) The working memory (WM) blocks consisted of 15 trials lasting 2s each. The emotional images blocks (EMO) consisted of 8 trials lasting 3s each. Every condition was shown twice, resulting in 8 WM and 8 EMO-blocks randomly interleaved, which results in a total task duration of 8:53 min

Fig. 2

Behavioral data: A) Task performance; No significant differences between participants with ADHD (blue) and controls (grey). The performance of the 2-back task was lower than the 0-back task. B) Recognition task; Neutral images were recognized less correctly than negative images. There is a trend towards participants with ADHD recognizing negative images better than controls. C) Validation task; Both groups rated negative images as more negative than neutral images. Error bars represent the standard error

Fig. 3

Whole brain activation maps calculated with permutation tests: BOLD signal for negative vs. neutral images (top row) and the 2-back vs. 0-back task (bottom row). Values shown are corrected and therefore correspond to the statistically significant regions. Colors correspond to 1-p results; with red = 0.95 (corresponding to p = 0.05) and yellow = 1.00

Fig. 4

Region of Interest analysis: A) Amygdala activity during negative vs. neutral images (left) and emotional images preceded by working memory blocks (right) divided into the four conditions (0N, 2N, 0E, 2E). B) PaCG activity during 2-back vs. 0-back blocks (left) and emotional images preceded by working memory blocks (right) divided into the four conditions (0N, 2N, 0E, 2E). C) Left (left) and right (right) dlPFC activity during 2-back vs. 0-back blocks. paCG = paracingulate gyrus; dlPFC = dorsolateral prefrontal cortex; Error bars represent the standard error