Emotionally triggered involuntary violent behaviour not attributed to a mental disorder: conceptual criteria and their reliability. This study conceptualised emotionally triggered involuntary violent behaviour (ETIVB), developed criteria for the identification of ETIVB, and examined their validity and reliability. South Africa criminal courts allow for a defence called “non-pathological criminal incapacity”. It refers inter alia to extreme emotional arousal triggering involuntary violent behaviour. This is usually called an “emotional storm” or a so-called “psychological blow automatism”. Psychiatrists are often called to give expert testimony in this regard, but there is no conceptual clarity or criteria upon which to base it. This difficulty is compounded by the requirement set by the criminal courts that the behaviour may not be attributed to a mental disorder (in this thesis the terms mental disorder and mental illness are used interchangeably). The first advance in obtaining clarity was afforded by focusing on the behaviour as distinct from 1) legal and jurisprudential considerations as well as 2) whether the behaviour is (not) attributed to mental disorder. The subsequent research questions driving the study were, “what counts as ETIVB?” In particular, what counts respectively as „emotional triggered‟, „involuntary‟, „violent‟, and „behaviour‟. Through conceptual methods suggested in the work of J.L. Austin, draft ETIVB-criteria were developed and then repetitively applied to a set of 28 cases. Twenty seven of these cases involved a charge of murder. The other one had behaved violently, but was not charged. The criteria were refined and their content validity derived by a repetitive to-and-fro process between comparative exploration of the concepts and their empirical application to the set of cases. The criteria were formulated into an instrument by which a psychiatrist can identify ETIVB. In addition provision was made to record whether ETIVB is, or is not, attributed to specific causes including mental disorder, but only as an attribution that follows after ETIVB has been identified. The ETIVB-instrument was subjected to reliability testing among 14 psychiatrists and 10 psychiatrists in training. They applied the ETIVB-criteria to a set of 5 externally validated case summaries. The participants had first been trained in the use of the instrument by considering other example cases. Statistical analyses of inter-rater reliability were performed. The modified kappa agreement ranged from 0.388 (fair) to 0.636 (substantial) across study cases for all criteria A.1 – H.1. The criteria for violence had exceptionally high inter-rater agreement, all being 0.947 and higher. The complexity of the ETIVB-construct transpired when inter-rater reliability analyses were compared between cases in that different cases posed different challenges for inter-rater agreement. The reliability testing across cases however, showed which less agreed-upon items would benefit from refinement. / Thesis (PhD)--University of Pretoria, 2015. / gm2015 / Psychiatry / PhD / Unrestricted