Overall, then, Experiment One revealed significant similarities as well as differences between the TBI and control participants. In particular, on the sarcasm experiment, although TBI patients did not show decreased performance on sarcastic contexts, they did show impairments when processing literal items in comparison to the control group. Performance on the TASIT for the TBI group was suggestive of difficulties in emotion recognition and broader features of social cognition, importantly including perception and comprehension of sarcasm. For the TBI group, the TASIT SI-M subtest also showed impairment on sincere items as well as sarcastic items, suggesting that they also experience difficulties interpreting literal information from the perspective of another person, indicating theory of mind impairments. The results suggest that sarcasm
comprehension is facilitated for the TBI participants by identification of sarcastic prosody and contextual cues, even though there may be an impairment of theory of mind. In addition, there is evidence to suggest relationships between such theory of mind abilities and observed aggressive behaviour.
4.2 Experiment Two
Experiment Two was conducted to explore further whether non-literal language processing, and in particular sarcasm, plays a role in the modulation of impulsive aggression in young healthy adults who were divided into two groups, low and moderate aggression, according to their self-rated BPAQ scores.
The results of this experiment indicated that context, question type and question order all led to significant effects. Increased facilitation, reflecting greater ease of processing, was observed when the story context was literal rather than sarcastic, and when the question was factual rather than attitude, in line with the hypotheses that literal contexts and factual questions would facilitate processing. These findings are consistent with findings from Experiment One, and will be discussed in the general discussion section.
In addition, although there was no associated hypothesis, questions which appeared second in order were answered significantly faster than those which appeared first. This finding mirrored the performance of the control participants in Experiment One and is thought to reflect a priming effect; the first question sets up the second question, allowing it to be answered faster, regardless of whether it is a factual or an attitude question.
Considering the performance between the two groups, no statistically significant differences were observed between the low aggression and moderate aggression groups on RTs or error rates. Interestingly, the pattern of the RT data (as seen in Figure 12) shows slightly longer RTs in the moderate aggression group compared to the low aggression group, however this difference did not present on statistical analysis. With regards to errors, overall very few errors were made by participants in Experiment Two, with most participants performing at or near ceiling, making it difficult to identify any effects of context, question type or question order on error rates, if indeed there were any.
Whilst existing research in non-clinical adult groups has demonstrated an association between poor verbal skills and aggressive behaviour in healthy populations (Barratt et al., 1997; Stanford et al., 1997), no such association can be inferred from the performance on the current task as group differences were not large enough to reach statistical
significance. Barratt et al., (1997) used the Wechsler Adult Intelligence Scale (WAIS; Wechsler, 1955) prorated verbal IQ score to reflect verbal skills, and Stanford et al. (1997) used the Controlled Oral Word Association Test (COWAT; Lezak, 1995). These previous studies therefore considered verbal skills quite generally, whereas, Experiment Two aimed to shed more light on the specific nature of the poor verbal skills in sarcasm comprehension. Whilst the current study did not identify a significant effect, this does not rule out a potential difficulty with figurative language comprehension, and more
specifically sarcasm comprehension, in more aggressive individuals. It is possible that a population with higher aggression scores might be necessary to capture such differences in sarcasm comprehension. Furthermore, it is possible that the task developed and completed by this healthy young adult sample may not have been sensitive enough to detect subtle differences in performance between the groups; after all, the task was developed with a clinical population in mind and thus had to be pitched appropriately.
Interestingly, across the two groups, self-rated BPAQ scores significantly positively correlated with self-rated sarcasm identity score; the more aggressive characteristics people reported, the more sarcastic they identified as. In line with this pattern, the moderate aggression group rated themselves as more sarcastic than the low aggression group, and this finding approached statistical significance. Interestingly, however, despite the relationship between the level of identification as a sarcastic person and self-rated aggression, sarcasm comprehension as assessed by the current task was not different between the low and moderate aggression groups.
4.2.1 Limitations and future research
Due to the recruitment process mainly targeting university students, the resultant sample cannot be considered broadly representative of young healthy adults as there is a lack of diversity with regards to educational attainment and therefore associated socioeconomic status. In addition, there was not an even gender split in the sample. There was, however, a fair distribution of males between the two groups with six males in the low aggression group and seven in the moderate group.
With regards to gender differences in aggression, past research indicates that males are more directly aggressive than females (Archer, 2004), and mean BPAQ scores in past research have indicated higher scores for men than women (77.8 and 68.2 respectively; Buss & Perry, 1992). However, although there were not equal numbers of males and females in the current experiment, the mean BPAQ scores did not differ significantly between males and females within each group, and on the whole with groups collapsed (63.31 and 59.71 respectively). Gender differences were not considered in the analysis of the sarcasm experiment data; this was not deemed necessary since BPAQ scores did not differ significantly between males and females and, due to the relatively small number of participants, splitting the data into additional groups by gender would have further limited the statistical power of the study. However, it would be interesting to consider gender differences in any future studies of sarcasm comprehension, especially since gender has been reported to play a role in the perception of sarcasm, in that men are considered more sarcastic than women (Katz et al., 2004).
Moreover, although there was a statistically significant difference between the BPAQ scores of the two groups, this experiment was limited by relatively low to average aggression scores reported on the whole; as can be seen by the scores outlined in the above paragraph, the BPAQ scores in this study were lower than those reported by Buss and Perry (1992) in the development of the measure. This was, however, not surprising given the sample targeted. In an attempt to further explore the research question in a healthy adult population, it would be important to recruit participants who self-report much higher aggression scores, thus allowing a greater variation between low and high aggressive groups, which may result in more notable differences on the experimental task between the groups.
Although all healthy participants completed the MoCA and scored within the normal range, they were not screened with any further measures of executive control. Future
research, in particular, should consider the role of executive function in contributing to aggressive behaviour. Executive weaknesses have been related to aggressive behaviour in healthy adults in past research and it has been suggested that aggression impulses may be sensitive to subclinical deficits of executive function (for a review see Hawkins & Trobst, 2000). It would be interesting for future research to explore the differential and combined effects of non-literal language difficulties and executive dysfunction, in particular in relation to impulse control, on overt aggressive behaviour.