There have been increasing concerns about violence perpetrated by patients with schizophrenia (1–3). Although the evidence on the causality remains inconclusive, several large population-based studies have shown a significant association between violence and schizophrenia (4, 5). A major source of violent acts in patients with schizophrenia is often related to the levels of hostility and aggression, which lay a substantial burden upon their closest caregivers in daily living (6–8). Hostility refers to an enduring attitude of ill will and negative evaluation of others (9), and aggression is defined as any behavior directed toward another individual that is carried out with the proximate intent to cause harm (10). Both hostility and aggression could worsen clinical outcomes of patients with schizophrenia, including lengthening of hospitalization and increasing costs (11), stigmatization (12), and poor treatment adherence (6).

Several neuroimaging studies have suggested that hostility and aggression in patients with schizophrenia are related to dysfunction of certain brain networks (13). A structural brain magnetic resonance imaging (MRI) study revealed that patients with schizophrenia who had a history of violence showed significantly smaller gray matter (GM) volumes in the right inferior temporal area compared to those with no history of violence (14). In studies of chronic patients with schizophrenia or schizoaffective disorder, higher levels of aggression were found to be correlated with larger GM and white matter (WM) volumes in the orbitofrontal cortex (15) and larger caudate volumes (16). Functional brain MRI studies found that violent patients showed relative hyperactivity in the medial frontal and anterior cingulate regions during negative emotion processing, compared to non-violent patients (17, 18). Although the neurobiological pathology underlying hostility and aggression is still not clear (19, 20), it has been consistently found that structural and functional abnormalities in the fronto-temporo-limbic circuits are significantly associated with hostility and aggression in patients with schizophrenia.

Meanwhile, deficits in sensory-perceptual processing, consistently observed in patients with schizophrenia (21–23), have been considered an important precursor to violent behaviors (24). Aberrant processing of sensory stimuli, particularly of a visuospatial nature, could make them prone to develop biased emotional and behavioral responses (25–28). Given that the visual system is the most dominant source of sensory inputs from our environment, brain circuits recruited for the visuospatial function are expected to play a substantial role in hostility and aggression in patients with schizophrenia. The visuospatial function network, including the fronto-parieto-occipital cortices and thalamus, involves emotional processing of visual stimuli, motivated attention to the significance, and sensory distinctiveness of the stimuli (29–31). The WM regions in the fronto-parieto-occipital cortices and thalamus contain the posterior limb of the internal capsule, posterior corona radiata (PCR), posterior thalamic radiation (PTR), corpus callosum (CC) splenium, superior longitudinal fasciculus (SLF), sagittal stratum, and superior fronto-occipital fasciculus (SFOF) (32). These regions are composed of neuronal fibers interconnecting the thalamus, visual association area near the occipital lobe, and sensory motor cortex, which play an important role in behavioral decisions in response to external stimuli by integrating and transmitting the visual stimuli and influencing the autonomic emotional responses in the amygdala, hypothalamus, and midbrain (32–36). In a recent study of healthy individuals, high trait anger was associated with aberrant neural responses to unpleasant emotional stimuli in brain regions consisting of the visuospatial function network (31). However, whether and how hostility and aggression are related to abnormalities in the visuospatial function network have never been investigated in patients with schizophrenia.

The present study aimed to investigate the WM correlates of hostility and aggression in patients with schizophrenia, focusing on the visuospatial function network. Based on the notion of schizophrenia as a disorder of aberrant brain connectivity (37), hostility and aggression in schizophrenia would be better understood by exploring miscommunication between several brain regions rather than specific abnormalities in a single distinct area. Diffusion tensor imaging (DTI) was used to assess WM connectivity at a microscopic scale by quantifying the degree of water diffusion along neuronal fibers (38). We hypothesized that (1) structural connectivity within WM tracts consisting of the visuospatial function network would be associated with the levels of hostility and aggression in patients with schizophrenia and that (2) WM aberrations associated with hostility and aggression would be correlated with the poorer visuospatial ability and more severe clinical symptoms.

To the best of our knowledge, this is the first study to investigate the WM correlates of hostility and aggression within the visuospatial function network in patients with schizophrenia. Increased FA values of the CC splenium, left PTR, and left PCR and decreased RD values extracted from the CC splenium were associated with higher levels of aggression in the schizophrenia group. Furthermore, FAs in these regions showed positive correlations with the severity of clinical symptoms measured using the PANSS. Our findings suggest that dysconnectivity in WM tracts consisting of the visuospatial function network may underlie aggression in patients with schizophrenia.

Compared to HCs, participants with schizophrenia had lower FA values of the WM tracts in the CC splenium, PTR, right of the PCR, right of the posterior limb of the internal capsule, and right of the SFOF. This result is grossly in line with that of a previous ENIGMA study with large-size schizophrenia samples, which found decreased FAs in WM regions consisting of the visuospatial function network, including the SLF, PTR, and SFOF (51). However, the FA values of the right of the PTR, right of the PCR, right of the posterior limb of the internal capsule, and right of the SFOF were not associated with hostility and aggression in participants with schizophrenia. Instead, FAs in the CC splenium, left PTR, and left PCR regions were positively correlated with not only higher aggression but also more severe clinical symptoms in our participants. Although the reason for these nonoverlapping findings cannot be established here, it may be assumed that differences in WM connectivity found in patients with schizophrenia are not necessarily associated with aggressive behaviors. Instead, WM connectivity associated with aggression may contribute to psychotic exacerbation in schizophrenia (52). Since aggression and violence significantly contribute to stigmatization and rejection of mentally ill people from their social communities (53), understanding the neurobiological origin of aggression in patients with schizophrenia is crucial to reduce biases against them.

At first glance, it seems counterintuitive that higher FA values were associated with higher aggression, poorer visuospatial performance, and more severe clinical symptoms. However, higher FA values do not always suggest enhanced brain function (54–56). In patients with schizophrenia, excessive excitation in the WM pathway due to gamma-aminobutyric acid (GABA) system dysfunction and hyperdopaminergic activity modulates synaptic maps, leading to deficient axonal pruning (57). Deficient axonal pruning may induce redundant networks and reduce the efficiency of information processing, as reflected by increased FA values in patients with schizophrenia (58). Reduced efficiency in information processing can lead to lower cognitive performance, deficits in behavioral regulation, and worsening symptoms along with difficulties in filtering sensory inputs and distorted perception (59–61). Therefore, we speculated that the aberrant WM connectivity due to the overexcited neuronal activity, which is shown as high FA and low RD values, would be associated with brain malfunctioning, including aggression, poor visuospatial performance, and severe clinical symptoms (62).

The CC is the most extensive WM tract that provides reciprocal signal transfer between the left and right hemisphere. The CC supports the information processing of input and output signals to facilitate coordination in thoughts and behaviors (63). Recent evidence suggests that interhemispheric inhibition mediated by the CC is related to anger and aggression (64). Increased left-to-right transcallosal inhibition was associated with a stronger attentional bias for angry faces (65). Although the role of CC in aggression is unclear, we found increased FA and decreased RD values of the CC splenium showed significant association with the levels of aggression in participants with schizophrenia. FA is a sensitive measure of water diffusion along WM tracts, determined by the structural features of the myelin sheath, axonal membrane, and neurofibrils, while RD is associated with myelin compactness. Although we cannot determine the neurobiological meaning of DTI measures, our current findings suggest that structural changes in the CC splenium related to the increased risk of aggression may partly be related to excessive myelination in patients with schizophrenia.

We also observed a negative correlation between FAs in the CC splenium and visuospatial ability measured using the R-CFT. The posterior part of the CC, including the splenium, is known to play an important role in the integration of perception and action within a subcortico-cortical network, promoting a unified experience between the two hemispheres (66). Successful copying of the complex figure reflects sufficient capacity for visuospatial perception and visual-motor function (67). The immediate recall condition is used to assess encoding and short-term retention of visual information, and its performance is strongly predicted by the figure copy scores (43). In this regard, our findings suggest that dysconnectivity of the CC splenium that was related to visuospatial perception and short-term visual memory may contribute to the increased risk for aggression in patients with schizophrenia.

The PTR connects the thalamus with the posterior visual brain areas. The thalamus is involved in processing and integrating sensory information that enters the brain (68). Visuospatial information, initially processed in the thalamus, is transferred to the parieto-occipital cortex through the PTR (69). In this study, we found that higher aggression was correlated with increased FA in the left PTR in schizophrenia. Given that deficits in early stages of sensory and perceptual processing lead to impaired higher-order cognitive functioning in patients with schizophrenia, our finding suggests that excessive thalamo-visual connections may affect abnormal responses in the association cortices, causing aggressive behaviors by the process of disposing immediately without high-level judgment (27, 70).

The corona radiata, a tract that carries ascending and descending information from the cerebral cortex, is related to different elements of cognitive control, including visuospatial memory (71, 72). In this study, the GM area near the left PCR that was significantly correlated with aggression was the inferior parietal lobule (IPL). This finding is consistent with that of a previous study that showed an increased activation in the left-lateralized IPL on exposure to unpleasant images in individuals with high trait anger (31). The IPL is concerned with multiple aspects of sensorimotor integration (73). Taken together, it may be assumed that increased connectivity in the WM regions of the visuospatial function network causes misinterpretation of neutral stimuli, sensitivity to minor stimuli, or induction of negative emotions to aggressive behaviors in patients with schizophrenia. However, it is unclear how the “relatively” increased connectivity in these regions is related to heightened aggression in schizophrenia. There might be possible explanations for these findings. First, disruption of the inhibitory GABA-ergic system in these regions might lead to overactivity in the WM regions (74, 75). Second, increased WM connectivity within visuospatial function network could be related to the hyperdopaminergic status in schizophrenia, which may cause excessive neuronal activity in the downward WM pathway. In addition, hostility did not significantly correlate with the mean FA values of the visuospatial function-related WM regions in participants with schizophrenia. Hostility is commonly defined as general dislike and negativity toward others over a range of situations. In contrast, aggression is defined as a willingness to inflict harm, regardless of whether this is behaviorally expressed and physical harm is sustained (76). Our findings may indicate that aggression related to a motive to inflict harm reflects a pathophysiological process different from hostility related to attitudes and negative cognitive bias. A previous study has shown that biased negative attitudes toward other group members are associated with increased activation in the lateral orbitofrontal cortex, while positive attitudes toward affiliated group members are related to activation in the medial orbitofrontal cortex (77, 78). According to the defense motivation system (79, 80), aggression may be more likely influenced by interactions between the thalamus, amygdala, and hypothalamic area, which mediate the autonomic emotional response based on sensory input processing.

There are some limitations to be considered in this study. First, hostility and aggression we measured may reflect state rather than trait characteristics of patients with schizophrenia. Although it takes a long enough period of time for certain factors to elicit structural changes in the brain, WM tracts are known to possess relatively higher neuroplastic properties than GM regions do. Future longitudinal studies will be required to figure out the causal relationship between state and trait aggression and neurobiological changes in the visuospatial function network. Second, antipsychotic medications might affect WM integrity, although there was a relatively short time when participants underwent MR scans (6.70 ± 5.85). Although most participants with schizophrenia (80.7%) were first exposed to antipsychotic medication in this study, future studies are needed to examine drug-naïve patients. Third, a crossing fiber problem may have influenced the results. The tensor model represents an independent, dominant direction; thus, the estimated orientation for voxels with a complex fiber structure may be ambiguous or misleading.

In conclusion, we found that increased WM integrity in the CC splenium, left PTR, and left PCR was correlated with higher levels of aggression in schizophrenia. Furthermore, a negative correlation was observed between FA in the CC splenium and visuospatial ability. Our main finding suggests that WM abnormality-related aggression is not only found in the well-known fronto-temporo-limbic area but also in the visuospatial function network and that aggression of patients with schizophrenia may be associated with poor visuospatial ability and underlying WM dysconnectivity.

The data supporting the findings of this study are not publicly available due to ethical restrictions for protecting participants' confidentiality and privacy but are accessible from the corresponding author on reasonable request with the approval of the Institutional Review Board of CHA Bundang Medical Center. Requests to access the datasets should be directed to Sang-Hyuk Lee (drshlee@cha.ac.kr).

The studies involving human participants were reviewed and approved by the Institutional Review Board of CHA Bundang Medical Center. The patients/participants provided their written informed consent to participate in this study.

S-HL designed the study. IA, TC, and MB managed the participant recruitment and data acquisition and compiled the database. IA and MB conducted the data preprocessing and statistical analysis. S-HL and IA implemented the literature reviews and interpretation of data. IA wrote the first draft of the manuscript. S-HL, TC, and MB provided the critical revision of the manuscript. All authors contributed to and approved the final manuscript.

This research was supported by the Basic Science Research Program through the National Research Foundation funded by Ministry of Science and ICT, Republic of Korea (Grand no. NRF-2019M3C7A1032262 to S-HL) and (Grand no. 2021R1C1C1012901 to MB).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.