Attention bias modification for major depressive disorder: Effects on attention bias, resting state connectivity, and symptom change

Abstract

Cognitive theories of depression posit that selective attention for negative information contributes to the maintenance of depression. The current study experimentally tested this idea by randomly assigning adults with Major Depressive Disorder (MDD) to 4 weeks of computer-based attention bias modification designed to reduce negative attention bias or 4 weeks of placebo attention training. Findings indicate that compared to placebo training, attention bias modification reduced negative attention bias and increased resting-state connectivity within a neural circuit (i.e., middle frontal gyrus and dorsal anterior cingulate cortex) that supports control over emotional information. Further, pre- to post-training change in negative attention bias was significantly correlated with depression symptom change only in the active training condition. Exploratory analyses indicated that pre- to post-training changes in resting state connectivity within a circuit associated with sustained attention to visual information (i.e., precuenus and middle frontal gyrus) contributed to symptom improvement in the placebo condition. Importantly, depression symptoms did not change differentially between the training groups-overall, a 40% decrease in symptoms was observed across attention training conditions. Findings suggest that negative attention bias is associated with the maintenance of depression; however, deficits in general attentional control may also maintain depression symptoms, as evidenced by resting state connectivity and depression symptom improvement in the placebo training condition.

Figure 1

Baseline resting-state connectivity within the attention control network (ACN). The underlying red color reflects the activation in 55 depressed and control participants when required to shift attention away from cue stimuli (invalid > valid cue condition). Seven critical nodes were identified from the statistical Z-max of each region within this network. Significant node-to-node pathways (p < .01) are indicated in yellow with their associated partial correlations.

Figure 2

Correlation between pre-training negative attention bias and pre-training resting state connectivity between the dACC and rMFG nodes (r = −.37, n = 37).

Figure 3

Scatterplot of self-reported depression symptoms (BDI-II) at pre-training (Pre), training sessions 1–8 (s1-s8), post-training (Post), and 1-month follow-up (FU). Lines indicate mean change in depression symptoms for each training group. BDI-P indicates data points for the placebo condition. BDI-A indicates data points for the active training condition.