A new set of cognitive techniques, known as Cognitive Bias Modification (CBM), has been developed to modify cognitive biases in emotional disorders. In a seminal experiment, healthy participants were randomly assigned to receive training designed to direct their attention either toward or away from threat-related words [
31]. The training procedure was an adaptation of the dot-probe paradigm for measuring attentional bias for threat [
32]. In the original paradigm, a pair of words (e.g., one neutral and the other threat-related) appeared on a computer, one above center screen and one below center screen. After 500 ms, the word pair vanished and a dot appeared in the location vacated by one of the words. Participants were told to push a button to indicate the dot’s position as soon as they detected the dot. A quicker response to the dot when it occurs in the previous location of a threatening stimulus is interpreted as vigilance for threat. Authors modified this paradigm to train attention away from threat by having the dot repeatedly replacing neutral words and toward threat by having it repeatedly replacing threatening words. Then participants were exposed to a laboratory stressor. It was found that the group trained to attend to neutral stimuli was significantly less emotionally reactive to the stressor than was the group trained to threat stimuli. The researchers concluded that attentional bias for threat heightens anxiety proneness, and that reducing such bias diminishes anxiety proneness. This experiment inspired the development of Attention Bias Modification (ABM) procedures as well as indicated the causal effects of cognitive processes on emotion [
33,
34].
Because of its conceptual appeal (i.e., the possibility of changing emotions through pure cognitive interventions), its reduced costs and its acceptability, ABM procedures have been proposed as valuable clinical tools [
34,
35]. Yet, only a few ABM studies have been conducted in depression so far and, with rare exceptions [
36,
37], most involved healthy participants scoring high on questionnaire measures of depression [
38‐
42]. Moreover, literature reviews of ABM efficacy in emotional disorders have found inconsistent results. Some meta-analyses have found moderate or large effects for change in attention bias after ABM [
43,
44], whereas others have found small effect sizes [
45,
46]. Similar inconsistency has been found for symptom improvements after training. Medium effect sizes have been found in some meta-analyses [
44], whereas small effects sizes were found in others [
43,
47]. In a more refined meta-analysis, which specifically focused on ABM in distinct clinical conditions (i.e., anxiety, depression, and substance abuse), the authors concluded that attentional biases and symptom changes were successfully reduced in anxiety (both in clinical and healthy samples) although the average effect size was small. Yet, the authors did not find significant changes for other clinical conditions, such as depression [
46]. A similar negative conclusion was reached in a recent meta-analysis on CBM procedures in anxiety and depression [
45]. Although this meta-analysis did not separate the results by different types of procedures (e.g., interpretation bias training as opposite to ABM), as in the Mogoase et al.’s meta-analysis [
46], their conclusions were rather pessimistic on the utility of CBM procedures. In the case of depression, the moderate significant effect sizes found when comparing normal controls to subclinical samples (g = 0.43) or clinical samples (g = 0.33) disappeared when outliers and publication bias were considered. The authors of this meta-analysis concluded that CBM is not as promising as many had hoped [
34].