Suicide is one of the most significant global mental health problems with high prevalence and mortality rate. It is a complex act, comprising a continuum from intention to attempt at death (Klonsky et al., 2016). More than 800,000 individuals die from suicide annually worldwide, accounting for 1.4% of global deaths (WHO, 2014). Suicides are mostly accompanied by a mental disorder, particularly depression (Dong et al., 2015). In China, the prevalence of depression is approximately 6%; there are ∼30 million people diagnosed with depression and the general standard suicide ratio in this group is 10%–15% (Lu et al., 2020). Clinically, the comorbidity between suicide and depression is of particular importance. Given the high rates of disability and mortality in patients with major depressive disorder (MDD) of high suicide risk, it is necessary to predict and prevent the occurrence of suicidal behavior. Therefore, it is essential to identify the neuropsychological markers that are associated with suicidality in patients with MDD.

Executive dysfunction might be one potential cognitive mechanism of suicidal behaviors (Arioli et al., 2018; Brand et al., 2007). Patients with suicidality risk exhibited various degrees of executive functions impairment such as flawed social cognitive function and response inhibition (Ho et al., 2018). Specifically, it may involve not only the generation of false assumptions that one is a burden to relatives but the failure to control inappropriate thoughts and behaviors. Reward-learning decision-making (DM) is a crucial component of executive function; its dysfunction is also a potential mechanism of suicidality risk. DM can be defined as a series of cognitive processes that evaluates and selects the advantageous option based on values of previous decisions (Saperia et al., 2019). In at least some cases, committing suicide could be considered as a unique decision to avoid unbearable pain in the short-term (Shneidman, 1993; Verrocchio et al., 2016). Nevertheless, the individuals showed impaired cognitive function when they perceive suicide as the only means of escaping from constant psychological pain for a long duration. Dombrovski and colleagues found that older suicide attempters displayed impaired reward-based DM, discounted past successful experiences, and made present-focused unfavorable decisions (Dombrovski et al., 2010). Indeed, it is plausible that suicidal acts may be viewed as an outcome of DM dysfunction (Sastre-Buades et al., 2021; Song et al., 2019). However, few studies have explored the relationship between suicidal ideation (SI) and DM as a part of the suicidal risk continuum (Huang et al., 2020). While SI does not cause physical harm, it is associated with increased future suicide risk (Giletta et al., 2015). In addition, suicidality itself is not an invariant characteristic, and it is very likely to develop from intentions and plans to behavioral implementation in a short period of time (Caine and Conwell, 2001). Therefore, it is necessary to increase the observation of suicidal ideation at the time of its emergence and to prevent suicidal behavior immediately, which may help reduce suicide rates. Considering that the DM process occurs along a time course, the event-related potential (ERP) technique could potentially elucidate the neural dynamics of the DM process in MDD patients with or without SI.

The measured ERP components during the Iowa Gambling Task (IGT) can provide a high-time-precision means for monitoring the whole DM process. The IGT is a DM task performed measuring participants’ reward learning ability under uncertainty (Bechara et al., 2013). Successful performance in IGT requires participants to reasonably distribute attention and evaluate reward precisely. An attention deficit is one of the consistently reported cognitive dysfunctions during the DM process in patients with MDD. Patients with depression tend to have negative attentional bias and are unable to shift attention to highly relevant events (Keilp et al., 2008; Posner and Petersen, 1990). Early ERP components are regarded as a manifestation of the attention phase. Specifically, the N1 component (at ∼80–200 ms of stimulus onset) is sensitive to visual feedback (Dan et al., 2020). During the process of feedback evaluation, two ERP components play an important role in neural response to outcomes. The feedback-related negativity (FRN), which reflects the neural responses differences in reward learning differences and shows association with the anterior cingulate cortex (ACC) activity, is distributed over frontocentral areas (Clark et al., 2011; Holroyd et al., 2002). Typically, it is a relatively negatively deflected brainwave that appears in a time window of approximately 200–350 ms after the presentation of the feedback stimulus (Angus et al., 2017). Another feedback-locked component is P3, which generally peaks between 300–600 ms and is usually recorded in the centroparietal brain regions (Wang et al., 2017; Zheng et al., 2020). The P3 during the feedback evaluation stage has been shown to reflect incentive salience individuals pay to a reward and punishment (Broyd et al., 2012). Previous studies have shown that these ERP components are relatively stable in the IGT task (Garrido-Chaves et al., 2021; Na et al., 2019). Therefore, the present study adopted this task to measure behavioral performance, and observe ERP components of attention and feedback evaluation in patients with MDD.

Using the IGT, Smoski et al. showed that participants with depression adopted an advantageous strategy compared to healthy controls (Smoski et al., 2008). However, this DM advantage may be undermined by the impact of suicide risk. Patients with MDD and a history of violent suicide attempt tend to perform poorly compared to non-attempters, which has been demonstrated using the IGT in many studies (Gorlyn et al., 2013; Richard-Devantoy et al., 2016). Their behaviors focus on immediate possible outcomes, while neglecting the establishment of long-term advantage strategies, and lead to impulsive DM patterns (Alacreu-Crespo et al., 2020; Hegedűs et al., 2018). The findings from ERP studies on suicide risk have been influenced by study paradigms, study objectives, and patient population heterogeneity, and no consensus has been reached (Gallyer et al., 2021). Tavakoli et al. found that there was an overall enhanced ERP component that reflected the capturing of attention in the suicidal behavior group compared to the healthy control group (Tavakoli et al., 2018). Using a similar auditory oddball task, the amplitude of N2 evoked by attentional processing was slightly, but nonsignificantly, increased in adolescents with suicide attempt (Tavakoli et al., 2021). During a feedback evaluation process, patients with SI or a history of suicidal attempts showed smaller P3 amplitudes to positive feedback than healthy controls, but no group difference was found in FRN amplitudes (Song et al., 2019). In another study of children, the SI group showed blunted FRN amplitudes (Tsypes et al., 2019). Thus, previously published studies on the neural dynamics of attention and feedback evaluation processes in patients with suicidal thoughts or behaviors measured by N1, FRN, and P3 are inconsistent and require further elucidation.

The aim of this study was to explore reward learning DM impairments in the early stage of suicide risk in patients with MDD. Behavioral performances and neurophysiological activity were measured in patients with MDD, with and without suicidal ideation (SI), using the IGT. It was hypothesized that 1) patients with SI would exhibit impaired learning compared to patients without SI during the IGT, and 2) patients with SI would show abnormal ERP results during the early attention and later reward evaluation process of DM, measured by N1, FRN, and P3 components.