Suicidal behavior is strongly linked to impaired decision-making, particularly in challenging situations. Both deficits in value-based decision-making and self-reported impulsivity are associated with suicidal behavior. Previous research has indicated disrupted ventromedial prefrontal cortex (vmPFC) value encoding in suicide attempters, with this disruption scaling with trait impulsivity, suggesting overlapping neural mechanisms. Computational models of decision-making tasks can reveal precise disruptions in decision processes. In suicidal behavior, these models have revealed difficulties in choosing between similarly valued options and impaired encoding of recent reinforcement. Importantly, these decision-making alterations appear specific to those who have engaged in suicidal behavior, not merely suicidal ideation or depression. Disrupted neural processing of value may relate to elevated impulsivity in suicide attempters. While impulsivity is linked to suicidal behavior, self-reported impulsivity alone doesn't distinguish ideation from attempts. Understanding the connection between disruptions in neural value processing, impulsivity, and suicidal behavior is crucial. Facets of impulsivity, such as negative urgency (acting rashly during negative mood), may best map onto neural correlates of disrupted value in suicide. In the general population, impulsive individuals show weaker frontoparietal connectivity, potentially reflecting reduced cognitive control. During decision-making, reduced connectivity between frontoparietal regions and areas representing expected reward value (vmPFC) is linked to impulsive decision-making. Reduced vmPFC-frontostriatal connectivity may lead to prioritizing immediate over long-term consequences, similar to suicidal crises. However, whether this impulsivity-related connectivity pattern characterizes suicidal behavior remains unknown. This study aimed to examine decision processes during a simulated suicidal crisis in impulsive and non-impulsive suicide attempters, focusing on disrupted value signals, investigating correlates of suicidal behavior beyond ideation, and measuring impulsivity facets relevant to suicidal crises. The study particularly focused on older adults, as they are a more representative group for death by suicide and display a wider range of impulsivity.

Existing literature suggests a strong correlation between suicidal behavior and impaired decision-making, particularly in dynamic, high-stakes situations. Studies have shown that both impaired value-based decision-making and high levels of impulsivity are linked to suicidal behavior. Neuroimaging research has previously identified disrupted ventromedial prefrontal cortex (vmPFC) value encoding in individuals who have attempted suicide. This disruption appears to be related to levels of impulsivity, suggesting a shared neural vulnerability. Computational models of decision-making have been employed to identify specific deficits in learning from reinforcement and in choosing between similarly valued options in individuals with suicidal behavior. These deficits appear to be more pronounced in those who have made suicide attempts, as opposed to those who only experience suicidal ideation. Impulsivity, particularly in its negative urgency aspect (the tendency to act rashly during negative mood states), plays a significant role. While impulsivity is associated with suicidal behavior, it alone doesn't differentiate between ideation and actual attempts. Previous research has also established a link between impulsivity and reduced connectivity among frontoparietal brain regions, indicating a potential role of cognitive control deficits. Studies have shown that reduced connectivity between frontoparietal regions and the vmPFC is associated with impulsive decision-making, potentially contributing to the poor choices made during a suicidal crisis. However, the specific neural mechanisms linking impulsivity, disrupted value processing, and suicidal behavior remain unclear.

This study used a sample of middle-aged and older adults (n=116) divided into four groups: suicide attempters, suicide ideators, depressed non-suicidal controls, and non-depressed non-suicidal controls. Participants underwent comprehensive assessments, including diagnostic interviews (SCID), depression severity measures (HRSD-17), and impulsivity assessments (UPPS). The core of the study involved a three-armed bandit task administered during fMRI scanning. This task presented participants with choices among three stimuli with dynamically changing reward probabilities, requiring continuous learning and value updating. Behavioral choices were modeled using reinforcement learning algorithms to isolate value-based decision-making processes. fMRI data were preprocessed and analyzed using general linear models (GLMs) and psychophysiological interaction (PPI) analyses. GLMs were used to assess vmPFC activation during value processing, comparing activation between groups. PPI analyses examined the functional connectivity between vmPFC and other brain regions during value updating, specifically investigating the moderating effect of impulsivity. An independent functional mask was created from impulsivity-modulated connectivity patterns in nonpsychiatric controls to test these patterns in the patient groups. Finally, a hierarchical linear regression was used to examine the behavioral impact of vmPFC-frontoparietal connectivity on the quality of choices made by participants.

Participants with a history of suicide attempts showed significantly reduced activation to expected value in the vmPFC compared to nonpsychiatric controls (p<0.05). This reduction was not significantly different from ideators or depressed controls though the overall effect across all groups was approaching significance (p=0.08). In nonpsychiatric controls, vmPFC-frontoparietal connectivity during value updating was negatively moderated by impulsivity (p<0.05). This moderation was preserved in comparison patient groups, but it was significantly abolished in suicide attempters (p<0.001). This altered connectivity pattern impacted behavior. Suicide attempters exhibited a disrupted relationship between vmPFC-frontoparietal connectivity, impulsivity, and reinforcement on choice quality (p<0.001). These effects were specific to the vmPFC and not observed in the striatum. Specifically, in non-suicidal depressed participants, higher vmPFC-frontoparietal connectivity was associated with higher-valued choices after both rewarded and non-rewarded trials. However, in suicide attempters, high connectivity predicted lower-valued choices after rewarded trials. These results suggest a distinct pattern of vmPFC-frontoparietal connectivity in impulsive individuals with suicidal behavior, different from non-suicidal impulsive individuals, and linked to disrupted choice processes.

The findings support the hypothesis that disrupted vmPFC value signals contribute to suicidal behavior. The study extended previous research by demonstrating that impulsivity's modulation of vmPFC-frontoparietal connectivity is disrupted in suicide attempters. Unlike non-suicidal impulsive individuals who show reduced connectivity, suicide attempters display abnormal connectivity, impacting their ability to learn from rewards and make optimal choices. While blunted vmPFC value signals were found, as in previous studies, the differences between suicide attempters and other patient groups were less pronounced, potentially requiring larger sample sizes or accounting for additional factors. The regional specificity of the observed effects (impulsivity's impact on frontoparietal connectivity, not limbic connectivity) is consistent with existing models of vmPFC function. The study's results suggest a neurobiologically distinct subtype of suicidal behavior characterized by abnormal vmPFC-frontoparietal connectivity, distinct from general impulsivity. This abnormal connectivity may interfere with integrating recent experiences into decision-making, potentially distorting the appraisal of stressors and increasing the appeal of escape during a crisis.

This study reveals a unique pattern of neural value processing and connectivity in older adults with a history of suicide attempts. Disrupted vmPFC value signals and the absence of impulsivity's typical modulation of vmPFC-frontoparietal connectivity were key findings. These neural alterations negatively affect learning from reinforcement and decision-making, potentially contributing to the heightened risk of suicidal behavior in these individuals. Future research should investigate the longitudinal course of these neural alterations and explore potential therapeutic targets for intervening in this distinct neurobiological subtype of suicidal behavior.

The cross-sectional design limits causal inferences. The exclusion of more severely cognitively impaired participants due to fMRI scanning requirements might limit the generalizability of the findings. The specific task used might have influenced the strength or type of value-related signals observed, potentially highlighting different aspects of value processing in stable versus volatile environments.