In daily life, people confront various situations requiring decision making. Decision making involves the ability to choose between competing behaviors associated with uncertain benefits and penalties (van Leijenhorst, Crone, & Bunge 2006). Each decision holds different risks and consequences. Risk refers to a condition that a certain benefit can be obtained along with the possibility of damage or danger (Leigh 1999). Risky decision making is a complex process that involves weighing different options in terms of their likelihood of potential rewards and risks (He, Xue, Chen, Dong, & Chen 2014). It has been proposed by a classical theoretical model that risky decision making relies on the net assets of the outcome (Machina 1982). Importantly, there are clear individual differences in risk-taking behaviors in various risk conditions (Bruine de Bruin, Parker, & Fischhoff 2007; Parker & Fischhoff 2005). Thus, studies considering individual difference may provide new insight on the neural bases of risky decision.
Balloon Analogue Risk Task (BART), designed by C. W. Lejuez et al. (2002), is a widely used paradigm to investigate risky decision making. In the BART task, participants need to virtually pump balloons to earn as many points as they can. The balloon can either grow larger with the pump or explode. Every pump accumulates points, but the risk of balloon explosion increases as the number of pumps goes up. If the balloon explodes, the participant loses all the points acquired from the balloon. Thus, the participant needs to decide whether to pump for more points or to stop pumping to save the current point for that balloon. BART task performance has been shown to significantly correlate with risk-related variables such as impulsivity, substance abuse, gambling behavior, and risky behavior (r = 0.20 ~ 0.44), establishing the reliability of the BART task (C. W. Lejuez et al. 2002). In line with these findings, other studies also reported correlation between the risk preferences measured by BART and scores on risks-related constructs (Hunt, Hopko, Bare, Lejuez, & Robinson 2005; C. Lejuez, Aklin, Zvolensky, & Pedulla 2003; C. W. Lejuez et al. 2003). Many BART experimental designs were based on a single risk condition. (Cazzell, Li, Lin, Patel, & Liu 2012; Gu, Zhang, Luo, Wang, & Broster 2018; C. W. Lejuez et al. 2002; Mata, Hau, Papassotiropoulos, & Hertwig 2012; Rao, Korczykowski, Pluta, Hoang, & Detre 2008; Juan Yang, Li, Zhang, Qiu, & Zhang 2007). Few studies examined the differences related to risk level (Hupen, Habel, Schneider, Kable, & Wagels 2019). J. Yang and Zhang (2011) measured event-related potential (ERP) in low- and high-risk conditions and found high-risk condition evoked a more negative N400 (time window of 300–500 ms) in the frontocentral areas than low-risk condition. Additionally, Juan Yang et al. (2007) found high-risk condition evoked greater N500 than low-risk condition, thus N500 was proposed to be related to responses in risky decision making (Juan Yang et al. 2007). Theses ERP studies indicated that high- and low- risky decision-making may have differential neural bases, but lacked information on which specific brain regions were related to such difference.
Functional magnetic resonance imaging (fMRI) has been widely used to examine brain mechanism of cognitive functions. Schonberg et al. (2012) observed brain activation during the BART task and found that activities in bilateral anterior insula, anterior cingulate cortex, and right dorsolateral prefrontal cortex brain were correlated with the mean number of pumps, a measure for risk-taking tendency. These brain regions were commonly known to be linked to risk processing and risk-taking. Moreover, the same study found that ventromedial prefrontal cortex (vmPFC) and bilateral medial temporal lobe (MTL) decreased with the mean number of pumps (Schonberg et al., 2012). Tisdall et al. (2020) analyzed neuroimaging data from a subsample of the Basel–Berlin Risk Study which included two widely used risk-taking tasks, the BART and monetary gambles task. They found associations between the risky choice and activations in the nucleus accumbens (NAcc) and anterior cingulate cortex (ACC) in both tasks. Specifically, there were negative associations between the mean number of pumps in the BART and activation in ACC and NAcc, and negative associations between the proportion of accept decisions in monetary gambles task and activations in ACC, NAcc, and anterior insula cortex (AIC). Rao et al. (2008) compared active choice mode and passive no-choice mode brain activation during the BART using fMRI. The authors found that a wide network of regions such as midbrain, insula, dorsal lateral prefrontal cortex (DLPFC), striatum, and anterior cingulate/medial frontal cortex (ACC/MFC) were associated with the voluntary risk. Voluntary risk showed higher activation in insula, DLPFC, ACC/MFC, and striatum compared with the involuntary risk. These studies demonstrated that distributed brain regions, such as DLPFC, vmPFC, MTL, ACC, and AIC, are associated with risky decision making using the BART.
In addition to task-related fMRI, several studies using resting-state fMRI have shown complementary and consistent findings of the neural correlates of decision making. For example, Gentili et al. (2022) found that resting state amplitude of low-frequency fluctuation (ALFF) in the right inferior parietal lobule and the left caudate lobe was positively correlated the brain activity evoked during BART execution. In addition, total earning of the BART was correlated with the ALFF in the ACC/MPFC, and the Hurst Exponent, a measure of efficient online information processing, in the IFG/insula was correlated with total earnings (Gentili et al. 2020). Moreover, the connectivity between vmPFC and dorsomedial prefrontal cortex (dmPFC) in the resting state was associated with the choice of high reward card in the Cambridge Gambling Task (CGT) and number of pumps in the BART across all the participants (including young and old participants) (Yu et al. 2017). However, to our knowledge, no studies have examined decision-making in the brain under both high- and low-risk situations.
The present study aimed to investigate neural correlates contributing to individual differences in decision making with low- and high- risk level using resting state fMRI. To do it, we associated resting state brain activity features with behavioral performance of the BART, ALFF (defined as the total power within the frequency range between 0.01 and 0.1 Hz), fALFF (the ALFF of a given frequency band as a fraction of the sum amplitudes across the whole frequency range), and ReHo (the evaluation of the similarities or coherence of intra-regional spontaneous low-frequency (< 0.08 Hz) Blood Oxygen Level-Dependent (BOLD) signal fluctuations in voxel-wise analysis across the entire brain). Using the brain regions identified by the previous association analysis as seeds, we further checked the functional connectivity associated with the behavioral performance of the BART. In the BART, we picked four behavioral measurements to evaluate various aspects of behavioral performance in low- and high-risk situations: mean adjusted number of pumps, number of explosions, mean adjusted pumps following an explosion, and the coefficient of variation of adjusted pump number. Mean adjusted pump number is the mean number of pumps on trials where the balloon did not explode, this was preferred than absolute number of pumps because explosions artificially restrict the range of pumping behavior (Pleskac, Wallsten, Wang, & Lejuez 2008). Thus, mean adjusted pump is sensitive to risk-taking tendency. Number of explosions serves as a measure of the propensity for continued risk-taking behavior after experiencing a prior balloon explosion (Leslie, Leppanen, Paloyelis, Nazar, & Treasure 2019). Mean adjusted pumps following an explosion may reflect greater risk propensity because of their chronological association with a failure (DeMartini et al. 2014). Coefficient of variation of adjusted pump numbers reflects intra-individual variability of adjusted pumps (Blair, Moyett, Bato, DeRosse, & Karlsgodt 2018), and is a strong indicator of risky decision-making(Weber, Shafir, & Blais 2004). These behavioral measures can capture different aspects of individual variability in task performance and serve as indices to measure risk-taking propensity. (DeMartini et al. 2014; C. W. Lejuez et al. 2002). We expect brain regions related to decision making and risk-taking to show differences between low- and high-risk conditions.