The Reflective-Impulsive Model (RIM; Strack & Deutch, 2004) is a dual-working model that specifically identifies the role for implicit cognition. The model comprises two systems that are believed to co-exist: a reflective system and an impulsive system. The former is more consistent with explicit methods for assessing aggression which, involve conscious deliberation and appraisal (Hoffman, Friese & Wiers, 2008). The reflective system generates behavioural decisions that are based on factual and value-based knowledge, while the impulsive system elicits behaviour through associative links and motivational orientations. In other words, in the reflective system, the elicited behaviour is the consequence of a decision process that activates the appropriate behavioural schemata. By contrast, the impulsive system activates behavioural schemata through spreading activation i.e., a search process for associative or semantic networks, which may originate from perceptual input or from reflective processes. Here, behaviour may be elicited without the person’s intention or goal. Ireland and Adams (2015) argued that this model has not yet been applied to aggression research even though it has valuable components, with clear utility for further understanding of distinct cognitive processing for aggressive behaviour.
The social-cognition models outlined above present a case for the empirical investigation of implicit processing in aggression since it is a relatively neglected phenomenon. In particular, this applies to the RIM since it proposes distinct cognitive processes which are thought to occur concurrently. Thus, the use of both implicit and explicit testing methods will allow for the theoretical consideration of the RIM to further understand the role of cognitive processes, in both reflective and impulsive systems in relation to anger, and its relationship with aggression.
92 5.1.4. A framework to measure the cognitive dimensions of anger in reactive
aggression
Examining the implicit and explicit cognitive processes in relation to anger and in the context of inpatient aggression is of current interest. Previous studies that have identified anger as a predictor of inpatient aggression (e.g., Doyle & Dolan, 2006) have not only used explicit testing methods, but have rarely defined and measured anger as a construct distinct from hostility, or from aggression (for a review see: Reagu, Jones, Kumari & Taylor, 2013). Thus, the term anger has often been used synonymously with the terms hostility and even aggression. As a result, these individual elements of anger have been treated as a global phenomenon.
A model which identifies the mechanisms involved in the relationship between anger and aggression from a cognitive perspective dimension is Wilkowski and Robinson’s (2010) Integrative Cognitive Model (ICM) (Figure 5.1.). The ICM specifies three cognitive components with hypothesised relationships between each, and in relation to anger and aggression. The model also considers the contribution of both automatic and controlled cognitive processes. The ICM is proposed to be influenced by relevant prior models, including the GAM (Anderson & Bushman, 2002) and SIP (Crick & Dodge, 1994), but attempts to represent the mechanisms that are involved in more specific cognitive terms. Namely, the cognitive components within the ICM are: hostility, rumination and effortful control. The authors describe how each component is linked and provide empirical support for each, sourced from evidence using implicit testing methods.
Figure 5.1. The Integrative Cognitive Model of trait anger and reactive aggression. (Wilkowski & Robinson, 2010).
Note: Solid lines depict pathways by which anger and aggression are increased, whereas dotted lines depict pathways by which anger and reactive aggression are decreased.
93 The model depicts an individual’s habitual cognitive processing tendencies as intervening between hostile situational input and resultant tendencies toward anger and reactive aggression. In the first cognitive process, the model specifically suggests that certain individuals are automatically biased toward hostile interpretations of situational stimuli, and this, in turn, leads to the more frequent elicitations of anger. Next, it is suggested that ruminative aspects reinforce interpretation-related biases that amplify anger and increase the likelihood of reactive aggression. Finally, the model postulates that effortful control processes are effective in counteracting the incipient tendencies (i.e., hostility bias, rumination) toward anger and reactive aggression (Wilkowski & Robinson, 2010). Three specific pathways are described by which effortful control would be useful for an individual: to enable reappraisal of situational stimuli in favour of a non- hostile interpretation; to interrupt ruminative attention processes, thus enabling a person to distract themselves from hostile thoughts; and to suppress tendencies toward
expressive behaviour indicative of anger arousal and aggressive behaviour.
The ICM clarifies and delineates the constituent construct, for which terms have previously been used interchangeably, and has the potential to illuminate aspects of the relationship between anger and inpatient aggression. Use of the model as a framework to aid understanding of the components of anger in inpatient aggression will allow for inferences about which of its cognitive components are most relevant including in associated incidents of aggression. Also, consistent with the RIM, the use of both implicit and explicit methods of testing for each of the cognitive components will aid further understanding in terms of the type of cognitive processing. The relevance of these cognitive components could potentially inform measurement, risk assessments and targeted treatment interventions.
Research studies involving the use of implicit cognitive measures of each of the cognitive components of anger (i.e., hostility, rumination and effortful control) included in the ICM do, however, provide mixed findings and are reviewed in the following sections.
94 5.1.4.1. Implicit cognitive processing: hostility
Measurement of hostile interpretations, an arguably automatic process, as identified in the ICM, has been assessed in different ways. Linder, Werner and Lyle (2010) administered vignettes which contained an ambiguous sentence describing a harmful act in which the intent of the perpetrator was unclear, and a target sentence which the victim in the ambiguous sentence retaliates with an act of aggression. Respondents indicated on a 4-point scale the extent to which they believed the perpetrator in the ambiguous sentence intended to do harm, and whether the actions were justified in the target sentence. Because of the poor (Cronbach’s alpha) internal reliability of these variables they were not included in analyses. However, a reading time response across the vignettes for the target sentence was subsequently used to
measure automaticity; whereby faster reading times were predicted to be associated with aggression. This implicit measure was not associated with a cognitive controlled (explicit) measure, which was the assessment for the belief about the acceptability of aggression (Werner & Nixon, 2005), but both variables independently predicted
aggression. This suggests that, despite the incongruence between implicit and explicit measures, they do demonstrate predictive utility for the outcome of interest. These findings could be understood as the distinct information processing patterns, as outlined in the previously mentioned models, such as the concept of concurrent cognitive
systems in the RIM.
Crouch et al., (2012) found support for the utility of a word game as a procedure for assessing implicit information processes; win/loss experience was differentially related to schema accessibility in parents with either low or high risk of committing child physical abuse. The results revealed that parents who were grouped as high risk
responded faster to negative words and slower to positive words after losing (compared to winning) rounds in the game. These findings were interpreted by the study authors such that, following a negative interpersonal experience, the negative schemata of those at high risk of committing abuse becomes more accessible as opposed to positive schemata. In the same vein of schema accessibility, it was found that exposure to songs with violent lyrics (Anderson, Canagey & Eubanks, 2003) and violent video games (Anderson et al., 2004) increased aggressive thoughts. This was demonstrated by more aggressive interpretations of ambiguously aggressive words, an increase in the
95 relative speed in which participants read aggressive (vs. non aggressive) words, and an increase in the proportion of aggressive word completions. More specifically,
Lobbestael, Cima and Arntz (2013) used a clinical sample to examine the predictive values of antisocial personality disorder and of hostile interpretation bias for aggression. Vignettes and pictorial stimuli depicting provocative ambiguous situations were used to measure interpretative bias where participants provided open and closed responses. These responses were coded in mutually exclusive categories as hostile, negative, positive and neutral. The negative responses differed from hostile responses, as the category included a heterogeneous mix of non-hostile themes such as social exclusion and illness. It was found that the relationship between reactive aggression and hostile interpretation bias was specific, meaning that other interpretation indices did not correlate with reactive aggression. Collectively, these studies indicate that access to cognitive schemata can be examined with the use of an implicit measure; however, its relation to observable aggression remains inconclusive. The evidenced link between implicit information processing, in particular hostility, and the reliance on recall of
incidents of aggression or laboratory-based aggression questions the ecological validity of these studies.