Abstract
The goal of attention bias modification (ABM) for anxiety is to systematically reduce the anxiety-related attention bias (AB) and ameliorate anxiety severity. Yet, few studies have examined patterns of learning processes over the course of ABM, and whether these patterns influence treatment effects. This study aimed to address this gap by first quantifying changes in training performance (training gains indicating improvements in training performance, and training losses indicating declines in training performance measured via increases in reaction times) over a four-week course of ABM and examining their associations with post-treatment anxiety severity. Eighty-one anxious adults were randomly assigned to either four weekly sessions of ABM or Placebo training (PT). Training gains and losses during each session were quantified as online (i.e., training changes that emerged within a training block or a training session) or offline changes (i.e., training changes that emerged between training sessions). ABM, versus PT, showed greater offline gains between training sessions. Offline gains between sessions were associated with reliable reductions in anxiety severity, but this association did not differ between the conditions. Implications for creating more targeted and personalized ABM protocols are discussed.