The defence of sane automatism and non-pathological criminal incapacity in South Africa: an epistemological intersection between psychology and law

Criminal liability

An accused person can only be found criminally liable if it can be proven that the individual had criminal capacity at the time of the commission of the crime; that the individual committed an unlawful act (actus reus), which was accompanied by fault (mens rea) either in the form of intention or negligence. Hence, the three elements of criminal liability are both distinct from one another (each having its own requirements) – and interconnected – in the determination of criminal liability (Kaliski, 2006). More simply put, a mental illness (for example a delusional disorder where a caregiver murders an infant while believing the infant is a devil) can impact upon an individual’s ability to form intention as well as her ability to appreciate the wrongfulness of the act. Similarly, a mental illness secondary to a medical condition (such as epilepsy) can impact upon an accused person’s ability to act in a voluntary manner. Hence, the elements of criminal liability are both distinct and dimensional in criminal law. Also noteworthy is the fact that the legal test for capacity is subjective and contextually determined, with no normative appraisal of reasonable standards of behaviour (Kaliski, 2006).

Where the court finds that the actions of an accused person were involuntary (not wilful), the accused will not be found to be criminally liable. In this case, the defence of automatism (unwilled act) may be raised. It must be noted that the law draws a distinction between pathological and non-pathological defences, and in relation to the automatism defence, a distinction is made between automatism attributable to mental illness/insane automatism and sane automatism (not attributable to mental illness) – this article is concerned with the latter form of automatism (see Joubert & van Staden, 2016; Kaliski, 2006; Snyman, 2002 for a detailed explication of this distinction). Examples of case law, which have employed the defence of automatism, include S v Schoonwinkle (1953) and R v Mkize (1959) (epilepsy); S v Chretien (1981) and S v Ramdass (2017) (intoxication); S v Eadie (2002) (provocation/emotional stress).

The defence of provocation and severe emotional distress

While the majority of legal systems do not regard provocation or severe emotional stress to be exculpatory since individuals are expected to control their emotions, a decision by the Appellate Division in 1925 that section 141 of the Transkeian Penal Code of 1886 reflected the South African law on provocation allowed for provocation to become a partial excuse for unlawful behaviour (Burchell & Milton, 1997). The Transkeian Penal Code states that ‘any wrongful act or insult of such a nature as to be sufficient to deprive any ordinary person of the power of self-control may be provocation’ (cited in Burchell & Milton, 1997, p. 279). The provocation defence, therefore, required an objective assessment to determine whether under a particular set of conditions, the reasonable person would have lost his or her self-control. Legal scholars have pointed out that the adoption of section 141 of the Transkeian Penal Code mitigated the severity of the mandatory death sentence for murder prescribed by the Criminal Procedure Evidence Act (1917), and in 1935, the extenuating circumstances rule allowed judges (in the face of extenuating circumstances) to not impose the death sentence for murder (Burchell & Milton, 1997; Singh, 2000). During the mid-1930s the use of the provocation defence in line with the Transkeian Penal Code declined (Burchell & Milton, 1997).

Hoctor (2011) points out that while Roman-Dutch law conceded that anger could be a factor mitigating punishment, Roman-Dutch writers worked on the premise that an individual consciously falls in love/becomes angry and this individual cannot later claim in the face of a strong emotion that he or she was acting unconsciously. This logic was also applied in the matter of voluntary intoxication, even if the accused’s actions were outside of his or her control, since the individual consciously used an intoxicating substance. However, following the assassination of Prime Minister Verwoerd in 1966 by a parliamentary messenger (Dimitri Tsafendas), a commission of enquiry into mental illness and criminal responsibility was established under Judge Rumpff. The Rumpff Commission was heavily reliant on the work of Wiersma (a Dutch psychiatrist), which was influential in the establishment of the cognitive/conative legs of the test for criminal capacity (in particular, the definition of self-control), which were included in section 78 of the CPA (Hoctor, 2011).

The Chretien (1981) decision, which held that severe intoxication could exclude criminal liability (by the negation of any of the three elements of criminal liability), paved the way for a re-emergence of the defence of provocation/severe emotional stress in South African courts. This judgement established that the determination of criminal capacity is not bound to mental illness and an accused person found to be lacking in insight or self-control (on any basis) could not be found to be criminally liable (Kaliski, 2006). While the defence of mental illness or mental defect is statutory, the defence of non-pathological incapacity is a common law defence that emerged in the 1980s out of an increasing recognition by the courts of (temporary) forms of extreme stress and provocation as a defence (see for example S v Arnold, 1985; S v Campher, 1987; Nelson, 2012). South African courts recognised that the criteria developed in section 78 of the CPA could also be used to assess non-pathological factors relating to incapacity (Hoctor, 2011). The term ‘NPCI’ emerged during the late 1980s in the matter of S v Laubscher (1988) and the defence continues to be prominent. The conflation of the terms NPCI and sane automatism was highlighted in S v Francis (1999), where Judge JA Schutz noted that ‘He contends that he acted in a state of non-pathological criminal incapacity (“sane automatism”) with the results that either he was unable to distinguish right from wrong or, if he could, that he was unable to control his actions’ (p. 2).

A landmark case for the defence of NPCI followed an Appellate Division decision in S v. Eadie (2002), which conflated the defence of NPCI with that of sane automatism (in particular, the conative test for criminal incapacity). In this judgement, the court determined that no distinction can be made between the exercise of conscious control over one’s behaviour (automatism test) and the lack of self-control (incapacity test). The court’s decision in this regard has been the subject of robust debate (see Hoctor, 2011; Snyman, 2002).

Non-pathological criminal incapacity: insight and self-control

The NPCI defence is characterised by the actor in a crime being motivated by severe emotional states (like anger, jealousy, or fear), which deprive an individual of their ability to appreciate wrongfulness or act in accordance with this appreciation (Burchell & Milton, 1997). The NPCI defence, therefore, relates to emotional factors that impair capacity, whereas the defence of mental illness or mental defect relates to biological/pathological factors. Hence, the defence of NPCI relates to a temporary absence of criminal capacity in respect of the cognitive (insight) and/or conative legs of the test for criminal capacity (self-control). Hoctor (2011) notes that the definition of self-control penned by Wiersma (and heavily relied upon by the Rumpff Commission) as ‘. . .the force which insight into the unlawfulness of the proposed act can exercise in that it constitutes a counter-motive’ was particularly significant in relation to the test for conative capacity (p. 81).

The term NPCI appears to be used mainly in South Africa and is not cited to any significant degree in the literature in other Anglophone countries. NPCI is a legal term and not a psychological concept; however, the cognitive and conative tests used in a determination of criminal capacity relate broadly to concepts of insight and self-control – which are described in psychological theory. Inasmuch as clinicians routinely refer to their patients’ level of insight, few pause to consider its exact definition and significance and a precise and agreed definition is elusive (Calcedo-Barba, Castelli Candia, Conejo Galindo, & Garcı’a Solano, 2004). Scholarship in insight arose particularly in the context of an individual’s awareness of symptoms of mental illness (cognitive or functional), notably in schizophrenia. Insight can be defined from a number of perspectives (biological, psychological, and social) and it has been described as a ‘patients’ degree of awareness and understanding that they are ill’ (Kaplan & Sadock, 1994, p. 279).

In the context of cognitive psychology and impaired insight, Sackheim (1998, cited in Calcedo-Barba et al., 2004, p. 398), reported six basic elements including the following:

Sackheim’s description of impairments in insight implies that deficits in insight are dimensional rather than distinct. Calcedo-Barba et al. (2004) note that ‘insight should not be taken in clinical and forensic practice as a present-or-absent sign’ and caution the clinician to carefully determine the clinical significance of impairments in insight (p. 400). Kaplan and Sadock (1994) highlight the distinction between intellectual insight and emotional insight, with the former referring to an individual’s understanding that she is ill, but being unable to use this knowledge to adjust future behaviour, and the latter referring to an awareness of motives and feelings and the ability to use this knowledge to alter personality or behaviour. The psychological literature on insight points to degrees of insight, rather than a dichotomous conceptualisation. While the courts require a definitive determination of insight being present or absent, this is inconsistent with the dimensional nature of insight in the psychological literature.

The concept of self-control relates broadly in the psychological theory to one’s ability to regulate behaviour, thoughts, and emotions in pursuit of long-term goals. Self-control has been defined as ‘the tendency to avoid acts whose long-term costs exceed their momentary advantages’ (Hirschi & Gottfredson, 1994, p. 3). In the context of the criminology literature, Gottfredson and Hirschi (1990) argue that offenders have little control over their desires, and that individuals who lack self-control are inclined to be impulsive, insensitive, preferring physical rather than cognitive tasks, risk-taking, short sighted, and non-verbal – these traits (which influence an individual’s ability to weigh up costs) tend to converge in the same people, and hence self-control is seen by Gottfredson and Hirschi (1990) as a unitary disposition towards offending behaviour. There has been much scholarly debate on whether self-control is a unidimensional or multidimensional construct, and it has been pointed out that Gottfredson and Hirschi’s six elements can be classified into one or two of the five personality domains (extraversion, openness to experience, neuroticism, conscientiousness, and agreeableness) described in Costa and McCrae’s Five Factor Model taxonomy for personality traits (Ward, Nobles, & Fox, 2015), suggesting that self-control is a multidimensional construct.

Research from the developmental neuroscience literature supports this notion. In the context of adolescent risk-taking behaviour and decision-making, Steinberg et al. (2008) note that risk-taking behaviour and impulsivity follow different neurological timetables and have different neural underpinnings – describing a dual systems model. The authors describe a temporal gap in the maturation of the socio-emotional and cognitive control system in adolescents. Hence, risk-taking behaviour (related to the early maturation of the socio-emotional system, which increases adolescent’s propensity towards seeking out novel and pleasurable activities/reward sensitivity) increases in adolescence in the face of the relatively immature cognitive control system (which allows for response inhibition/self-control) (Steinberg, 2010; Steinberg et al., 2008). This research has supported approaches that disaggregate self-control into sensation-seeking behaviour and impulsivity (Ward et al., 2015). Functional imaging of the brain demonstrates a correlation between self-control and areas of the prefrontal cortex that are involved in executive brain functioning (such as planning of complex cognitive behaviour, reasoning, decision-making, and judgement). The psychological literature, therefore, points to a nuanced multidimensional rather that unitary understanding of the construct of self-control, which is frequently discordant with the requirements of the court.

In S v Eadie (2002), Griesel (in the Cape Town High Court) and Navsa (in the Supreme Court of Appeal) argued that one can only lack self-control when one is in a state of automatism (unwilled behaviour). This view has gained traction among scholars in mental health, and Kaliski (2006) contends that ‘in the Eadie judgement the defence of temporary non-pathological incapacity has been brought in line with a psychiatric (and neurological) understanding of automatism’, questioning the notion that an individual can be ‘consciously lacking self-control’ (p. 53). The conflation of self-control with automatism has, however, been contested among legal scholars (Hoctor, 2011; Snyman, 2002), with Snyman (2002) pointing out that in the case of impaired criminal capacity, the mental power of resistance/the actor’s ability to resist the temptation to commit a crime is lacking, whereas with automatism, the power to control physical movements/wilfully subject bodily movements is impaired.

In the legal literature, there does not appear to be a consensus as to whether the two components of capacity (cognitive/conative) exist exclusively (so that having capacity in one area undoubtedly results in a finding of capacity in the other area) or whether they are interdependent. From a psychological perspective, insight is a prerequisite of self-control, one cannot exert self-control over one’s actions without some level of insight, and an assessment of self-control in the absence of insight seems illogical. Similarly, an assessment of criminal capacity seems unwarranted if an act is unwilled, and an individual cannot intend something that she did not will (see Coles, 2000a). In S v Cunningham (1996) the judge in this matter noted that ‘criminal responsibility presupposes a voluntary act (or omission) on the part of the wrongdoer. Automatism therefore necessarily precludes criminal responsibility’ (p. 13).

However, one can have insight without exerting self-control and the presence of insight and self-control assumes that the actor is behaving in a willed manner. In the assessment of criminal capacity in children between the ages of 10 and under 14 years (in terms of the Child Justice Act no. 75 of 2008b), the courts make allowance for the possibility that a young person may lack insight or self-control (hence, the child lacks criminal capacity), thereby making allowances for willed behaviour in the absence of either insight or self-control (see Snyman, 2002). Put another way, an individual without insight is not necessarily unconscious, they can still act in a willed manner, just as someone who is mentally ill can act in a willed manner while lacking criminal capacity. In case law, the courts have recognised that an accused person can have insight into the wrongfulness of an act, while also lacking self-control (S v Moses, 1996; S v Nursingh, 1995).

Sane automatism

South African courts (as well as mental health practitioners) use a variety of terms to refer to sane automatism, notably, ‘emotional storm’, ‘acute catathymic crisis’, ‘emotional flooding of the mind’, and psychogenic amnesia (Joubert & van Staden, 2016). While it is well understood in law that a person cannot be held criminally responsible for an involuntary act, the law (and the courts) have thus far had some difficulty applying a common workable definition of the concept. This difficulty is reflected in both the national and international literature. Yeo (2001) points out that a common misrepresentation regarding the concept of automatism relates to the common perception that automatism or an involuntary act occurs in a state of unconsciousness or impaired consciousness, and the act is, therefore, not purposeful or as a result of mindful deliberation. Yeo (2001) points out that automatism is characterised by a ‘total inability to control one’s conduct’ (p. 390), hence the central aspect relates to the absence of the ability to exercise the will to act or control one’s actions – rather than a lack of knowledge or consciousness. Hence, the legal literature on automatism has referred to loss of control/involuntariness as a key defining feature. Automatism has been understood as relating to behaviour over which the individual is unable to exercise voluntary control. For the defence of automatism to succeed, it needs to be demonstrated that an individual had no control over his or her actions (as opposed to partial control or an impairment in this regard).

There are a number of disparities in the literature relating to a common understanding of the psychiatric definition of automatism. Broadly these differences relate to behaviour performed without awareness; behaviour performed while consciousness is impaired; behaviour of which the individual has no knowledge, and finally, behaviour that is not planned/purposeful or intentionally exhibited and for which cognitive functions are absent (see Joubert & van Staden, 2016, p. 11). Fenwick and Cantab (1987) describe an automatism as

an involuntary piece of behaviour over which an individual has no control. The behaviour is usually inappropriate to the circumstances, and may be out of character for the individual. It can be complex, coordinated, and apparently purposeful and directed, though lacking in judgement. Afterwards, the individual may have no recognition or only partial and confused memory for his actions. (p. 349)

Automatism has also been defined as behaviour performed without conscious knowledge; however, it has been noted in the psychological literature that goal-directed behaviour is possible in states of impaired consciousness where partial awareness is maintained, and McSherry (2004) argues that ‘evidence of dissociation in itself is not sufficient to lead to the conclusion that an individual’s behaviour is involuntary’ (p. 453). In the context of Canadian law, Coles (2000a) points out that in courtroom testimony, mental health experts have confused the legal understanding of automatism by arguing that complex and deliberate acts are possible despite an individual functioning automatically (in the absence of willed behaviour). While the courts make a distinction between sane automatism (which arises from an external factor – such as a blow to the head, or too much insulin) and insane automatism (which arises from an internal factor – such as an epileptic seizure or brain tumour), this distinction does not make sense medically (Fenwick & Cantab, 1987) since the mere occurrence of an automatism is suggestive of disordered brain functioning. Coles (2000a) notes that

from a psychological point of view, however, it must be asked whether it is, in fact, possible to commit a crime under such circumstances- whether it is possible to act involuntarily, without conscious awareness of what is being done; and, if so what actions are possible under these conditions? (p. 35)

It has been argued (Joubert & van Staden, 2016; Kaliski, 2006) that mental health professionals have applied differing understandings of the ‘unconscious’ in relation to automatism in legal testimony, with the origins of the medical underpinnings of the term being applied to patients with complex partial epilepsy who demonstrate complex, co-ordinated behaviour though lacking in judgement. Increasingly, Freudian, psychodynamic applications of the term ‘unconscious’ have made their way into the courtrooms (Kaliski, 2006; S v Nursingh, 1995) thereby obfuscating the original medical definition of automatism.

Despite the lack of consensus in the definition of automatism, in assessing for automatism, some consensus has been reached. The Fenwick criteria remain prominent in this assessment and both the mental health and legal disciplines make allowances for the impact of emotional factors in automatism. Kaliski (2006) notes (1) that a clear cause or trigger event for the automatism should be determined (either medical or psychological). In the defence of sane automatism involving heightened emotions, the individual would likely have been exposed to enduring stress (usually emanating from intense interpersonal conflict, which leaves the individual feeling humiliated or abused) and culminating in a climax prior to the offence; (2) there should not be evidence of pre-meditation (as the automatism arises spontaneously); (3) behaviour should be inappropriate to the circumstances; (4) following the automatism, the individual should appear bewildered, making no attempt to leave the crime scene or efforts to avoid detection by the authorities; rather the perpetrator may make attempts to assist the victim. In cases of heightened emotions, the individual should have amnesia for the period of the automatism but be able to give an account of what happened just prior to the incident, and subsequent to the incident, as well as a description of the trigger event.

Disinhibition and diminished responsibility

In the context of Canadian law, Coles (2000a) argues that many crimes that were determined by the courts to have been perpetrated while the accused was in an automatistic state (acting without volition/consciousness) occurred in a state of diminished consciousness. Coles (2000a, p. 35) describes five situations under which an individual can be said to be acting without conscious control, namely the following:

Pivotal to Coles’ (2000a) argument is the assertion that ‘an unusual event or stimulation will serve to restore or redirect conscious awareness, and permit the individual to stop his/her behaviour should it be undesired’ (p. 36). Apart from behaviour in relation to pathology (mental disorder/substance use), Coles (2000a) argues that individuals’ conscious awareness will in the face of environmental stimulation (for example a victim screaming) be restored. He proposes that through a process of disinhibition and the loosening of cognitive controls, behaviour that the individual would usually be able to control finds expression – suggestive of diminished responsibility. Coles (2000a) notes, however, that disinhibition cannot account for complex, goal-directed and novel behaviour that is so characteristic of the rape and murder crimes of our time.

Drawing from the neuroscience literature on voluntary movement, Bennett (2009) points out that electrical activity in the brain precedes a physical movement, suggesting that ‘voluntary movement is not a movement caused by a felt urge, any more than to refrain voluntarily from moving is to feel an urge not to move that prevents one from moving’ (p. 296). This has led cognitive neuroscientists to question whether voluntary acts in fact require a wilful act (Bennett, 2009). On the other hand, movements that are caused by a feeling or desire are not voluntary, and Bennett (2009) highlights this with an example: in the case of an individual with an urge to take a drug, the felt urge does not draw her hand uncontrollably to the drugs. In other words, the desire to take the drug is not the cause of this action, rather, the act of taking the drug is to relieve the craving for the drug. Notwithstanding the criticisms of reductionism directed at the use of neuroscience in the courtroom (see for example Maoz & Yaffe, 2015), much scholarly work is still required in this field to elucidate the underlying neurological mechanisms that play a part in criminal responsibility.

Non-pathological criminal incapacity, sane automatism, and amnesia

Bourget, Gagne and Wood (2017) report on a number of studies that demonstrated an association between amnesia, dissociation, and crimes that were characterised by (1) heightened emotional states, (2) an emotional connection to the victim, (3) a lack of planning and premeditation, and (4) alcohol use. The authors point out that ‘identifiable stressful events and heightened emotional states in an individual with personal vulnerabilities preceding the crime are more likely to trigger a dissociative state and amnesia’ (p. 154). Nevertheless, the authors point out that ‘noninsane automatism is rare’ (p. 155).

Kaliski (2006) indicates that crime-related amnesia may be due to (1) dissociation (due to the stress of the crime); (2) neuropsychiatric causes (intoxication is frequently cited; hypoglycemia), (3) the individual being diagnosed with a disorder that causes amnesia (such as dementia), or (4) malingering (particularly if there is no clear cause of the amnesia). Alcohol-induced ‘blackouts’ usually result in complete memory loss for the duration of intoxication; however, there may be (partial) fragmentary memories and recall can be aided by cueing. Acute intoxication is characterised by difficulties in episodic memory (memory for events encoded with time, spatial, and social context).

It must be noted that amnesia in and of itself does not indicate an automatism. Moreover, as reported in Arboleda-Florez (2002), ‘. . .amnesia is crucial to an understanding of automatism, because whereas there is no automatism without amnesia, not every case of amnesia amounts to automatism’ (p. 572). It is well known in the psychological literature that intense emotions can interfere with memory processes, such that a particular intensely emotional/traumatic event is not encoded; similarly, it is possible to recall information in the short term (after a traumatic event) and then lose this ability (retrograde amnesia). In a state of automatism, no memories can be encoded due to interferences in cognitive processes. This was reflected in S v Moses (1996) where it was held that ‘automatism is followed by total amnesia because of the fact that the person is not able to register his conduct during the state of automatism’.

Conclusion

In a scathing editorial titled ‘The emperor in the courtroom: psychology and pseudo-science’, Coles (2000b) argues that although psychologists (in particular) describe their discipline as the ‘scientific study of behaviour’, their court testimony (and that of other mental health professionals) frequently falls short of scientific requirements and routinely refers to possibilities rather than probabilities (p. 1). Coles (2000b) notes that

The courtroom testimony of many health experts fails to meet one or more of the criteria of science by lacking an open, sceptical mind, presenting idiosyncratic data or theories, failing to follow established methods in drawing conclusions, and/or the inappropriate presentation, or quantification, of data. Among these criteria, the first – the lack of an open, sceptical mind – is foremost. (p. 2)

The law frequently requires that psychologists (and other forensic mental health experts) provide absolute and unequivocal opinions relating to an accused’s state of mind at the time of the commission of a crime; however, the following considerations should be borne in mind:

While mental health experts can assist the courts in imparting relevant specialised knowledge, they should be cautious of equating psychological constructs to legal concepts and using reductionistic approaches to circumscribe psychological knowledge onto these legal concepts.