Martijn van Teffelen

Chapter 4 90 unvalenced) cognitions (Beck, 1985). It also aligns with our clinical impression that participants in the I-CR condition reported increased compassion or sympathy for their original source of frustration, while participants in the CR condition would merely show a quantitative decrease of angry cognitions. This clinical impression is supported by (paradoxical) evidence showing that hostility and compassion are positively related constructs as they both reflect vigilant prevention focused self-regulation (Keller & Pfattheicher, 2013). Nonetheless, the clinical impression remains to be experimentally tested. I-CR was not only efficacious in changing hostile beliefs but also in reducing aggressive tendencies (H2) and state anger (H3). Here I-CR only partially differentiated from CR. That is, one session of I-CR did not significantly differ from CR in reducing aggressive tendencies and state anger. However, for I-CR but not for CR efficacy was retained at 1-week follow-up in terms of aggressive tendencies. An explanation for our finding that I-CR was more efficacious in reducing aggressive tendencies than CR, is that only under mental imagery conditions hostile beliefs were able to mediate the effect of treatment on aggressive tendencies. This fits with previous work showing that a tendency to interpret emotionally ambiguous stimuli in a hostile way increases the likelihood of aggressive behavior (Crick & Dodge, 1994; Epps & Kendall, 1995). During mental imagery, all senses are activated. Holmes and Mathews (2010) convincingly showed that memories are stored multi-modally (i.e., including all senses), rather than solely verbally. Therefore, during mental imagery and not so much during a verbal treatment technique like CR, reduction in hostile beliefs could have more strongly reduced aggressive tendencies. An alternative explanation is that the mental imagery condition increased state anger, which made it easier to modulate aggressive tendencies. Indeed, a recent review on depression treatment suggests that in-session affect mediates treatment effects in CBT for depression (Aafjes-van Doorn & Barber, 2017). Given the strong impact of mental imagery on emotional memory (Holmes & Mathews, 2010), one might expect that I-CR increases in-session anger, which in turn mediates efficacy. However, this explanation is not supported given that the present results indicated that the difference between I-CR and CR in terms of state anger was marginally non-significant ( p = .08). Findings on the sustained and more ‘long term’ efficacy of I-CR demonstrated that reductions in hostile beliefs and aggressive tendencies were (partially) sustained at one- week follow-up (T2a) and after an emotional stressor (T2b). However, our intervention conditions were not superior to the AC condition in terms of change in hostility trait levels. Probably we observed a ‘regression to the mean’ here, implying that the observed decrease in hostility traits reflects random chance. Alternatively, it may be possible that one session of I-CR or CR is just not potent enough to impact change-resistant hostility traits and we set the expectancy of a one-session intervention too high. Indeed, another study that also used a short cognitive intervention (interpretation bias modification training