DIFFERENTIATING RISK FROM REWARD MOTIVATION IN DECISION MAKING A Thesis Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Master of Arts in Developmental Psychology by Xinyi Deng August 2023 © 2023 Xinyi Deng ABSTRACT Risk tolerance and reward sensitivity are important phenotypes in decision making tactics and mental health. While related, these traits are distinct. When people engage in similar risky behaviors, their decision-making strategies may be reflected by different neural correlates of reward sensitivity and risk tolerance. However, few neuroimaging studies have directly compared rewards with and without risk in the same context. To fill this gap, we modified a Balloon Analogue Risk Task (BART) to include reward-only trials and modeled neural correlates of reward motivation with and without risk. During multi-echo fMRI scans of mesolimbic regions, 48 participants (28 females; M ± SD: 20.57 ± 1.92 years old ) inflated balloons across three conditions: 1) earn monetary reward but risk loss if a balloon popped (RWR); 2) earn reward without risk (RWOR); and 3) no loss or gain (neutral). Multi-echo data were preprocessed with AFNI and Tedana and sequentially analyzed with FSL GLM and permutation testing (threshold-free cluster enhancement with 5000 permutations), using the number of pumps as parametric modulator. Compared to the neutral, RWR in the decision-making phase showed greater activation in the insula, ventral- (VS) and dorsal striatum, brainstem (BS), thalamus (TH), prefrontal (PFC) and visual cortices (VC). Similarly, RWOR coincided with greater activation in the right frontal pole and middle temporal gyrus, and VC, surprisingly, not in VS. The contrast [RWR > RWOR] showed increased activity in the insula, right caudate, BS, TH, PFC, and VC. Our results suggest that RWR decision-making recruits canonical reward areas and insula, but risk involves greater recruitment of prefrontal and midbrain regions in the same task. It implies that rewards with risk recruit greater neural correlates of reward and learning. Our neuroimaging study is a first step in differentiating neural responses during risky versus simply rewarding decision making. We will further model risk-related differences in the consummatory phase to better differentiate risk tolerance from reward sensitivity. Keywords: risk tolerance, reward sensitivity, striatum, decision making 1 BIOGRAPHICAL SKETCH Xinyi Deng was born in Sichuan and grew up in Chongqing, China. In 2021, she received her bachelor’s degrees in psychology from Australian National University, Australia and Southwestern University, China. Under the supervision of Prof. Xu Lei, she worked as a full- time research assistant at the Sleep and NeuroImaging Center, SWU and focusing on sleep and bipolar disorder. Then Xinyi started her master’s degree in developmental psychology at the Department of Psychology, Cornell University in August 2022. Under the supervision of Prof. Marlen Z. Gonzalez, Xinyi completed multi-echo MRI analysis and wrote her master’s thesis in the summer of 2023. Currently, Xinyi is a research coordinator at Life History Lab and Action Research Collaborative, Cornell. Her research interests are to investigate how people's affective disorders influence their decision-making strategies, and the underlying neural mechanisms (e.g., reward circuits). She has been involved in translation research and volunteer activities to improve people’s mental health. 2 ACKNOWLEDGMENTS I would like to acknowledge and give my warmest thanks to my supervisor, Prof. Marlen Zoraida Maria Gonzalez Caraballo, who made this work possible. Her guidance and advice carried me through the entire writing process. Talking with her brightens my day and inspires my research career. I would also like to thank my committee member, Prof. Adam Anderson, for making my defense an enjoyable brainstorm moment and for your brilliant comments and suggestions. I would also like to thank Dr. Minwoo Lee and Dr. Elizabeth Riley for their generous advice on my MRI analysis. Besides, I would like to express my gratitude to my family and partner for their encouragement and support all through my life. 3 TABLE OF CONTENTS BIOGRAPHICAL SKETCH ........................................................................................................... 1 ACKNOWLEDGMENTS ............................................................................................................... 2 DIFFERENTIATING RISK FROM REWARD MOTIVATION IN DECISION MAKING .......... 5 RISK TOLERANCE AND REWARD SENSITIVITY AS TRANSDIAGNOSTIC PHENOTYPES ................................................................................................................................ 5 THE RELATIONSHIP BETWEEN RISK TOLERANCE AND REWARD SENSITIVITY ......... 6 DISTINCT MECHANISMS UNDER DECISION MAKING ....................................................... 7 NEURAL CORRELATES OF RISK AND REWARD PROCESSING ........................................ 10 CURRENT STUDY ...................................................................................................................... 10 METHOD ...................................................................................................................................... 13 PARTICIPANTS ............................................................................................................................ 13 MEASURES .................................................................................................................................. 14 SELF-REPORTED SURVEYS .......................................................................................................... 14 MODIFIED BART PARADIGM ...................................................................................................... 15 PROCEDURE ............................................................................................................................... 19 DATA ACQUISITION ................................................................................................................... 19 DATA ANALYSIS ......................................................................................................................... 21 DATA PRE-PROCESSING ............................................................................................................... 21 LOWER-LEVEL ANALYSES ........................................................................................................... 21 4 HIGHER-LEVEL ANALYSES ........................................................................................................... 22 RESULTS ...................................................................................................................................... 23 RWR > NEUTRAL ....................................................................................................................... 23 RWOR > NEUTRAL .................................................................................................................... 25 RWR > RWOR .............................................................................................................................. 26 DISCUSSION ................................................................................................................................ 28 DECISION MAKING IN THE CONTEXT OF REWARDS WITH RISK .................................. 29 DECISION MAKING IN THE CONTEXT OF REWARDS WITHOUT RISK .......................... 30 RISK TOLERANCE VS. REWARD SENSITIVITY ................................................................... 31 IMPLICATIONS OF CURRENT STUDY ................................................................................... 31 LIMITATIONS AND FUTURE DIRECTIONS ........................................................................... 32 CONCLUSION ............................................................................................................................. 33 REFERENCE ................................................................................................................................ 34 5 Differentiating Risk from Reward Motivation in Decision Making When people make decisions, risk and reward coexist in most of their daily lives. Both risk tolerance and reward sensitivity constitute individual differences in how individuals react to these environmental affordances. Previous research has shown that risk tolerance and reward sensitivity are important phenotypes in decision-making tactics (Charness et al., 2013; Chick, 2015; Gentili et al., 2022; Korucuoglu, Harms, Kennedy, et al., 2020) as well as in healthy behavior (Ackerman et al., 2015; Nelson et al., 2013; Scott-Parker & Weston, 2017). Decision- makers who chose immediate large rewards followed by potential penalties appear to be more reward-sensitive and risk-prone , particularly in anxiety-inducing situations (Rivalan et al., 2009). Although the two phenotypes are positively correlated in decision behavior, they are distinct, as supported by previous studies (FeldmanHall et al., 2019; Rudebeck et al., 2008; Talmi et al., 2009; Trepel et al., 2005). However, neuroimaging studies rarely distinguish between these two concepts at the neural level. This is particularly important when considering interactions between individual traits (e.g., reward-sensitive) and a particular context (e.g., high risk and high reward). Furthermore, neuroimaging provides a way to consider how similar outcomes can nonetheless be arrived at with different strategies. To this end, we modified a risk-taking paradigm, the Balloon Analogue Risk Task (BART), by creating identical trials that differed only in whether the reward is associated with risk or not. Risk tolerance and Reward Sensitivity as Transdiagnostic Phenotypes As decision-making strategies, risk tolerance and reward sensitivity are closely related to physical and mental health. Risk tolerance is defined as the degree of risk that an individual is willing to tolerate to pursue a goal (Charness et al., 2013; Hertwig & Erev, 2009), while reward sensitivity is defined as the degree to which a person's behavior is motivated by reward-related 6 stimuli (Carver & White, 1994; Torrubia et al., 2001). In terms of physical health, a systematic review showed that greater reward sensitivity coincided with speeding, red light running, and self-reported risky driving (Scott-Parker & Weston, 2017). In terms of mental health, depressed adults show more risk-averse performance (i.e., the majority of choices are from non-risky decks) when choosing a card from the high-risk or low- risk decks (Smoski et al., 2008). Blunted reward sensitivity is a characteristic of unipolar depression, whereas increased reward sensitivity is more associated with hypomania (Alloy et al., 2016). Moreover, people with suicide attempts also showed an increased risk tolerance in a gambling task compared to healthy controls (Ackerman et al., 2015). Meanwhile, young females with suicide attempts are more intolerant to delayed reward (i.e., more reward-sensitive) compared to the control group when facing a trade-off between immediate smaller gains and delayed larger rewards (Mathias et al., 2011). Therefore, understanding an individual's risk tolerance and reward sensitivity provides insight into decision-making processes and behaviors that hinder or promote well-being. The Relationship between Risk Tolerance and Reward Sensitivity Risk tolerance and reward sensitivity are often related in an everyday practical sense, as with the behaviors and conditions above, and from a theoretical perspective. For instance, Kahneman and Tversky’s (2013) model suggests that higher reward sensitivity is associated with higher risk tolerance, as individuals with a strong desire for rewards may be more willing to take risks to pursue them (2013). Another financial theory also suggests that investors are generally inclined to take on higher levels of risk when they expect or anticipate higher potential returns (Lundblad, 2007; Weber, 2010). 7 Non-human animal research suggests that there is an entanglement between higher risk tolerance and higher reward sensitivity (Rivalan et al., 2009). Rats that preferred immediate food pellet rewards over 50% of the time in a gambling task, despite longer and unpredictable timeout penalties, reached the food box faster in a runway. They also explored more unfamiliar areas in the elevated plus maze in the dark than rats that preferred delayed rewards. These outcomes indicate that poor decision-making is because of hypersensitivity to both risk and reward. Importantly, some of this entanglement may be due to measurement. If the ultimate measure is the likelihood of choosing a risky behavior, we cannot tell if that is driven by a strong desire for the reward or by high-level tolerance for the risk. However, neuroimaging may help us look at how different neural strategies could underlie the same behavior. For example, in a study on surgeons, a previous study observed that, when determining the next safest surgical maneuver, novices had significantly more activation in the dorsolateral prefrontal cortex (dlPFC) than attendings and residents, even if they made the same decision (Leff et al., 2017). The finding implies attendings use a habitual decision system, whereas novices use an effortful approach in a decision context. Thus, neuroimaging is a good tool to measure neural decision strategies, as demonstrated by Leff et al (2017). For example, the BART paradigm is widely used in neuroimaging studies of decision making, active risky decisions for monetary reward showed greater activation in mesolimbic- frontal areas compared to no-choice decisions made to continuously inflate balloons (Korucuoglu, Harms, Kennedy, et al., 2020; Rao et al., 2008; Van Leijenhorst et al., 2010). Distinct Mechanisms under Decision Making Reward sensitivity and risk tolerance have shared some neural correlates, but human and animal neuroscientific evidence suggests they could be differentiated. A recent study presented 8 the differential neural correlates of risk and reward processing in a decision-free context in a modified monetary incentive delay task (MID; Sun et al., 2022). In this MID task, people are instructed to detect a neutral target in a limited time to secure a lottery that includes a large and small monetary reward, each with an equal 50% chance of being won. The lottery includes two reward values that were sampled from different means and standard deviations (SD) of the possible outcomes, separately representing expected reward and risk. The researchers found that the main regional effects of increasing expected reward were in the left caudate, and increased risk was associated with the anterior insula. Results also revealed distinct neural connectivity associated with reward processing (the striatal-limbic-frontal network) and risk processing (i.e., insula-ventral medial prefrontal cortex; vmPFC, anterior cingulate cortex; ACC). The study demonstrated distinct neural representations of reward and risk in a non-decision context. However, this study cannot help us distinguish between the neural correlates of risk and reward motivations in actual decision-making because participants' keystroke performance was not correlated with the magnitudes of risk and reward, nor could it determine the outcome. Another neuroimaging study also supported that increased activation in the OFC was associated with greater variance in reward (i.e., risk) and that expected reward was positively related to striatal activity (Tobler et al., 2007). This previous research also suggested that increased OFC activity correlated with individual risk tolerance, as measured by the sensitivity of change to variance in pleasantness within button press trials. It demonstrates the selective influence of high or low risk tolerance on uncertain outcomes, but it didn't consider the concurrent effect of reward sensitivity in this situation, and the sample size was limited (16 participants). A recent review article also supports the related but distinct mechanism by synthesizing the findings from multiple studies (Haber & Knutson, 2010; Mohr et al., 2010; Van 9 Duijvenvoorde et al., 2016; Wu et al., 2021). They suggest increased risk tolerance is associated with greater blood oxygen level-dependent (BOLD) signals in the posterior parietal cortex, anterior insula, medial prefrontal cortex (mPFC), ventral striatum (VS), and amygdala, while higher reward sensitivity is linked to greater activations in the mPFC and VS (Van Duijvenvoorde et al., 2022). Importantly, non-human studies also suggest some distinction between risk and reward in frontal regions. Using similar definitions of risk and reward, an animal study observed a monotonic-risk signal in the orbitofrontal cortex (OFC) neurons that did not necessarily display a monotonic reward signal (O’Neill & Schultz, 2010). The researchers studied the electrophysiological activity of two monkeys in OFC while they made a saccade from a fixation point to a risk or reward cue. The reward cues were three monotonically increasing level bars (i.e., low, medium, and high-level) that predicted the amount of juice after each successful saccade. Each risk cue consists of two bars at low and high vertical positions, representing low and high equivalent reward magnitudes, respectively (i.e., a monotonically increasing range of positive and negative SD with constant probability and the medium reward as the mean). The monkeys' actual saccade points in this range of risk cues indicated the amount of juice. Though this neural study designs a saccade task to suggest that the encoding of risk by orbitofrontal neurons is largely distinct from the encoding of reward value, the non-human sample size and human generalizability are limited. Moreover, the firing of the neuron corresponding to risk cues could not fully reflect risk tolerance because these cues involve the simultaneous encoding of risk and reward. 10 Neural Correlates of Risk and Reward Processing Previous studies provide moderate evidence for neural correlates of reward and risk processing, respectively. For example, the mOFC-to-ventral striatum pathway supports potential high-value choices, while the mOFC-to-dorsal striatum circuitry is beneficial for tracking changes in different conditions to obtain flexible rewards (Jenni et al., 2022). Other findings have also demonstrated the close association between robust activations in the nucleus accumbens (Nacc) with the BART paradigm and high self-reported reward sensitivity in the cross-sectional study (Braams et al., 2015; Van Duijvenvoorde et al., 2014). This neural comparison provides insight into the neural mechanism underlying reward sensitivity. Furthermore, prior investigations have suggested that ACC predominantly encodes risk processing (Rangel & Hare, 2010). The lesions of the insula and mPFC resulted in increased gambling behavior among patients, and this lesion study implies the corresponding function of risk adjustment (Clark et al., 2008). Greater activity in the amygdala has been shown to reflect the level of risk and may be related to risk avoidance (Mohr et al., 2010). Current study The above investigations provide moderate evidence of distinct neural mechanisms. However, there has been no direct comparison. One of the most common behavioral and neuroimaging paradigms for the study of risky decision-making is the BART. The BART is a simple balloon popping game that involves pumping a virtual balloon to earn monetary rewards while balancing the risk of the balloon popping and losing all money. Unfortunately, researchers using the BART do not always agree on what the BART measures. Some individuals classify it as an in-lab measure of reward sensitivity (Harden et al., 2018), while others say it measures risk (Clark et al., 2008; Kohno et al., 2016; Rao et al., 2008; Van Duijvenvoorde et al., 2016). In a 11 sense, it could be both since active trials offer both risk and reward, and no trial offers rewards alone. One study tried to remedy this by using the MID paradigm. They used the MID task and the computation technique to measure neural responses to different magnitudes of monetary rewards (mean) and risk (SD of possible outcomes; Sun et al., 2022). The study showed that processing reward coincided with greater striatal activation, while processing risk was associated with the greater activation in the anterior insula in a decision-free context. While important in establishing the connection between risk tolerance versus reward sensitivity, the reliance on decision-free context means that we are not explicitly testing risk and reward in a decision- making process. The present study distinguishes itself from previous work in three ways. First, to differentiate the neural correlates of risk tolerance from those of reward sensitivity in the same context, we modified the BART paradigm to include reward-only trials while preserving uncertainty across trials (see methods). This resulted in three possible conditions: 1) rewards with risk (RWR; the traditional BART balloon); 2) rewards without risk (RWOR); and 3) neutral. Comparing the brain activity of people in the RWR and RWOR conditions allows differentiating the neural correlates of risk tolerance and reward sensitivity in the same experimental context. Furthermore, comparing the RWR and neural conditions allows us to model the neural response to reward (reward sensitivity). Second, to achieve a higher signal-to-noise ratio in subcortical areas, we scanned participants using a multi-echo fMRI configuration (Lynch et al., 2021). Third, to achieve greater spatial precision in the heterogeneous mesolimbic region, we targeted imaging towards the center of the brain, but still including most of the PFC. Richard and colleagues (2013) summarized that across paradigms, reward-related risky decisions in the choice phase showed greater activation in 12 dopamine-rich reward regions, including the subcortical limbic structure (e.g., striatum, OFC, ACC, IFG, insula, and lPFC). This greater activation of the mesolimbic-frontal area during risky decision-making with reward has been demonstrated by several meta-analyses and recent human MRI studies (Bartra et al., 2013; Korucuoglu, Harms, Kennedy, et al., 2020; Rao et al., 2008; Rolls et al., 2022; Sescousse et al., 2013; Wu et al., 2021). However, these studies used whole brain scanning to obtain relatively coarse results during risky choices with reward. Our partial brain scanning with multi-echo MRI can refine brain activity in the mesolimbic-frontal area with high resolution, especially in the insula, striatum, ACC, and OFC. The current study primarily investigates what risk and reward motivation happen in the decision-making period from cue presentation to button pressing. This decision-making process indicates weighting competing options with different expected values, gains, and losses (Richards et al., 2013). Based on previous human and non-human animal research (Clark et al., 2008; O’Neill & Schultz, 2010; Rao et al., 2008; Rudebeck et al., 2008; Sun et al., 2022; Tobler et al., 2007; Wu et al., 2021), we held three predictions of the neural responses during the modified BART for the decision-making phase. We first hypothesized that the RWR versus neutral contrast would coincide with increased BOLD signals in the anterior insula, ACC/mPFC, mOFC, striatum, and amygdala in line with other traditional BART findings (Korucuoglu, Harms, Kennedy, et al., 2020; Mohr et al., 2010; Rao et al., 2008; Wu et al., 2021). We also hypothesized that the RWOR versus neutral contrast would coincide with greater BOLD in the medial OFC and striatum, especially in the nucleus accumbens (NAcc), based on previous research using reward paradigms (Rangel & Hare, 2010; Rudebeck et al., 2008; Talmi et al., 2009; Van Duijvenvoorde et al., 2016). Finally, we hypothesized that there would be greater salient BOLD signals in the insula, mPFC/ACC, and amygdala in the RWR versus RWOR 13 contrast based on previous lesion and gambling studies (Rangel & Hare, 2010; Clark et al., 2008; Mohr et al., 2010). Taken together, the hypotheses suggest that risk tolerance should be distinguished from reward sensitivity via greater recruitment of regions putatively related to error tracking and processing, even if sharing similar responses in canonical reward regions (e.g., striatum). Method Participants Forty-eight healthy undergraduate participants (28 females) recruited from Cornell University completed the current study. Their ages ranged from 18 to 22 years (M ± SD: 20.57 ± 1.92), 6 participants declined to provide their age. In this undergraduate sample, 27% of participants self-identify as White (of European descent), 35% are Asian. Approximately 37% of participants self-identify as one of the above ethnicities (10 Hispanic, 3 Black, 1 Middle Eastern, and 4 Mixed race). Participants’ socioeconomic status includes their parents’ income and the highest level of education their parents received. The distribution of their family income was as follows: except for two who did not report this information, 17.39% (8) of the participants’ parents had an income below $40,000, 15.22% (7) between $40,000 - $59,999, 10. 87% (5) between $60,000 - $99,999, 28% (13) between $100,000 - $174,999, 19.57% (9) between $175,000 - $299,999, 4.35% (2) between $500,000 - $749,999, and 4.35% above $750,000. Of the participants, two had parents who did not complete high school, 22.92% (11) had parents with a high school diploma or GED, 14.58% (7) had parents with a bachelor’s degree, 27.08% (13) had parents with a master’s degree, 16.67% (8) had parents with a doctorate, 6.25% (3) had parents with an advanced professional degree (MD, DOS, OD, JD), and 8.33% (4) had parents who attended some college. 14 The Institutional Review Board for Human Participant Research (IRB) approval was obtained for all procedures. Participants were initially screened through a phone interview. Eligibility criteria included being right-having, no history of neurological conditions (e.g., stroke, seizure, brain tumor, or closed head injury), having no current or past episodes of psychosis, and not currently using psychotropic drugs. Furthermore, to adhere to safety standards for MRI scanning, participants were excluded if they had any metal in their body (such as pacemakers, neural implants, metal plates or joints, shrapnel, or surgical staples), experienced claustrophobia, or were pregnant. A written informed consent was obtained by participants after a detailed explanation of this study. Out of the 48 participants initially enrolled, fMRI analyses were conducted on 46 participants who had complete data. One participant had no corresponding brain imaging for the negative feedback condition and another had neuroimaging data that could not be preprocessed successfully. Measures Self-reported Surveys Demographic Questionnaire. The demographic questionnaire is designed to collect information about the participants' background and identity. The 13 items include age, race and ethnicity, year in school, transfer student status, college at Cornell, parents' income and education, number of people in household, sex, gender, living on/off campus, minority status, and childhood home cross streets. These questions allowed us to observe potential group effects. Sensitivity to Punishment and Sensitivity to Reward Questionnaire (SPSRQ). The SPSRQ (Torrubia et al., 2001; Vandeweghe et al., 2016) is a 36-item behavioral assessment to measure individuals’ reactions to specific cues of punishment (e.g., “I am troubled by punishments”) and reward (e.g., “I like to compete and do everything I can to win”). Participants 15 are asked to select their responses using a 5-point scale, with options ranging from "Never" to "Always." Higher scores indicate greater sensitivity to punishment and reward. Behavioral Inhibition System/Behavioral Activation System (BIS/BAS) Scales. The BIS/BAS scales (Vandeweghe et al., 2016) are a 14-item behavioral measure of an individual's responses to the general stimuli of punishment (BIS) and reward (BAS). For example, for the BIS scale, “Criticism or scolding hurts me a lot”, and for the BAS scale, “I go out of my way to get things I want”. All response options are on a 4-point Likert scale ranging from "totally disagree" to "totally agree”. High scores on the BIS scale indicate greater responsiveness to punishment cues (i.e., greater punishment sensitivity), while high scores on the BAS scale indicate greater responsiveness to reward cues (i.e., greater reward sensitivity). Although we collected information on individual differences in punishment and reward sensitivity, self- reported data from the SPSRQ and BIS/BAS will be analyzed in further studies. Modified BART Paradigm The Balloon Risk Analogue Task (BART) paradigm is to measure risk tolerance, risk- taking selection and execution, and the experience of win or loss outcomes in the risky decision- taking procedure (Lejuez et al., 2002). It was widely adapted in MRI studies (Korucuoglu, Harms, Kennedy, et al., 2020; Rao et al., 2008). We modify the scanner version of the BART paradigm with three conditions: a reward-and-risk condition (RWR), a reward-only condition (RWOR), and the neural condition (see Figure 1). Participants were instructed that there were three colors of virtual balloons (i.e., red, white, and gray) that they could inflate by pressing a button and that the colors of the balloons corresponded to three different BART conditions (i.e., RWR, RWOR, and neutral condition) in a 16 game. Participants were also told that their goal was to make as much money as possible, and this money was then given to them at the end of the study. In the RWR condition, participants had the option to inflate a red balloon to earn a monetary reward but faced the risk of losing money if the balloon popped. Participants could choose to pump or stop. If they chose to pump, each pump would inflate the balloon once and put money ($0.05) into a bank that would then be at risk in the subsequent trial. Participants were instructed that the more times they pumped, the higher the wager and the greater the potential reward, but the greater the chance that a balloon would pop, and they would lose their entire wager for that round. If participants chose to stop pumping the balloon, the wager for that round could be saved in their bank account where they would eventually receive the same amount of cash. In the RWOR condition, participants performed the same tasks and received rewards ($0.05 per pump) for white balloons, but there was no risk of financial loss or balloon explosion. In the neutral condition, participants were asked to inflate a gray balloon without gaining or losing money. There was no risk of the balloon popping. In RWOR and neutral conditions, participants had to pump these balloons until they disappeared. The maximum number of pumps for each condition was randomly chosen between 2 and 12. Once the number of pumps reached the maximum, the red balloon popped or the gray and white balloon disappeared, and then participants moved on to the next trial. While undergoing multi-echo fMRI, participants engaged in two 10-minute BART runs. In each run, the computer randomly generates 40 balloons of different colors and condition sequences. The actual number of balloons participants pumped depended on how fast they completed the BART within ten minutes. Each trial started with a virtual balloon, during which participants had unlimited time to press response buttons using their right thumbs. Following 17 obtaining the maximal pumps or choosing to save money (only in the RWR condition), the pre- feedback interstimulus interval (ISI) was presented between 0.5 and 3 seconds. Then feedback (e.g., negative, positive, and neutral feedback) was presented for 2 seconds, depending on different conditions. There was a 2-second jitter after the feedback to be ready for the next trial. 18 19 Fig. 1. (A) A flow diagram of the BART paradigm: (a) the diagram of the RWR condition; (b) the diagram of the RWOR condition; (c) the diagram of the neutral condition. (B) the details of the BART procedure. Procedure Based on the exclusion criteria, all participants received a thorough phone screening. In the pre-scanning session, selected participants completed safety screening and informed consent in a test room. Female participants received a pregnancy test. They then practiced this modified BART paradigm on the computer to make sure they understood three different BART conditions with red, white, and gray balloons. During scanning session, participants remained motionless during a 6-minute anatomical scan and then performed three tasks while undergoing a multi-echo fMRI scan, including two runs of the modified BART (each run lasts 10 minutes). After the entire fMRI scan, participants received cash rewards based on their performance on the BART and completed a series of surveys including demographic questionnaire, SPSRQ, BIS/BAS, etc. Data Acquisition Neuroimaging data were acquired using a General Electric (GE) Discovery MR750 3.0T MRI scanner at Cornell University. Task stimuli were projected on a screen at the back of the MRI’s bore, so participants viewed the stimuli with a 32-channel phased-array head coil. Before anatomical scanning, three-plane localizer images were acquired, and then ASSET (i.e., Array Spatial Sensitivity Encoding Technique) calibration was performed. . One hundred and seventy- six high-resolution anatomical T1-weighted images were acquired in 6 minutes, using the sagittal plane of imaging and the magnetization-prepared rapid acquisition gradient-echo (MP-RAGE) sequence (TR = 7 ms, TE = 3.42 ms, flip angle = 7°, field of view (FOV) = 256 mm, 256 × 256 matrix, 176 axial slices, voxel size = 1 × 1 × 1 mm3, 1-mm slice thickness). 20 Thirty interleaved functional multi-echo (ME) Echo Planar images (EPI’s) sensitive to BOLD (blood oxygenation level-dependent) contrast on GE were obtained during two runs of the BART, each run lasting 10 min (TR = 2600 ms; TE1 = 13.4 ms, TE2 = 35.8 ms, TE3 = 58.5 ms; 80° flip angle; FOV = 192 mm; matrix = 96 × 96; 30 axial slices; 3 echoes; 2 × 2 × 1.5 mm voxels; 240 volumes, slice thickness = 1.5 mm). To focus on mesolimbic brain activity, the partial-brain functional images were selected from the edge between midbrain and pons to the top of the corpus callosum across 30 slices and were aligned with the AC-PC axis, as shown in Figure 2. To control data quality, all participants’ head positions were stabilized by foam pillows, and they used foam earplugs to diminish the scanning noise. They were also informed of the importance of staying still during scanning, and all neuroimaging scans at the Cornell MRI Facility were performed by a well-trained MR technician working with a standardized protocol. Heart rate was recorded by an oximeter placed on the left index finger, and respiratory information was monitored by a sensor belt. 21 Fig. 2. Acquisitions of MRI scans providing partial brain coverage at very high resolution in order to identify the functional changes in mesolimbic-frontal area across time under decision making. Data Analysis Data Pre-processing Imaging data were preprocessed using the Analysis of Functional NeuroImaging (AFNI) software suite (Version 23.0.01; https://afni.nimh.nih.gov/; Cox & Hyde, 1997)). Using afni_proc.py, we remove spikes across the fMRI time series (despike) and slice-timing differences were adjusted (tshift). The EPI images were aligned with the corresponding anatomical one (align), and then the anatomical and functional data were normalized to the standard Montreal Neurological Institute brain template with a 2 × 2 × 2 mm3 voxel size (i.e., MNI152_T1_2009c+tlrc_2mm.BRIK; tlrc). The motion correction was applied using volreg. Then we created a binary brain mask from the aligned functional volumes (mask). Moreover, using TE-dependent analysis (tedana), preprocessed BOLD images with three echoes were denoised via principal component analysis (PCA), and independent component analysis (ICA) and non-brain materials were removed from denoised data (DuPre et al., 2021). Using FMRIB Software Library (FSL) software (Version 6.0.6.2; www.fmrib.ox.ac.uk/fsl), signal-to-noise ratio was enhanced via a high-pass filter with a cutoff point of 100 s and we used a 2-mm full width at half-minimum (FWHM) Gaussian kernel, and grand-mean intensity normalization for spatial smoothing via the fMRI Expert Analysis Tool (FEAT; Woolrich et al., 2009). Lower-level analyses After preprocessing, we used FSL’s FEAT query to model reward and risk processing during decision-making phase (i.e., the 1-second duration from the onset of presenting the virtual 22 balloon stimulus). The number of pumps was entered as a parametric modulator in the decision- making model. To model risk-reward decision making, the BOLD signals from stimulus presentation to button press in the RWR condition were subtracted from those in the neutral condition. To model reward sensitivity, we contrasted the neural signals of decision-making process in the RWOR condition with those in the neutral conditions. To model risk tolerance, we compared the neural correlates of decision making in the RWR and RWOR conditions during the same period. To improve the signal-to-noise ratio, we performed a fixed-effects model at the level two analysis in FEAT. For each participant, the average of the parameter estimates across two BART runs was calculated. This approach allowed us to enhance the reliability of the estimates by reducing the impact of noise. Higher-level analyses The aggregated activations in the contrasts of RWR > neutral, RWOR > neutral, and RWR > RWOR were estimated using FSL’s non-parametric permutation testing and threshold- free cluster enhancement (TFCE) with 5000 permutations per contrast at the group-level analyses. Using TFCE, the cluster-like activations were enhanced by considering the magnitude and spatial extent of activations simultaneously, but these images remained fundamentally voxel- wise (Smith & Nichols, 2009). Controlling for false positives and multiplicity, the family-wise error rate (FWER) for the aggregated images’ contrast were corrected, and the individual voxel significance level was set at p<0.05. The locations of significant voxel-scanned activations in three contrasts were determined using the Harvard-Oxford Cortical and Subcortical Atlas. For each activation, the specific brain region and its corresponding coordinates were reported below. The statistically significant activation information was generated using FSL’s atlasquery. 23 Results RWR > Neutral Replicating the previous BART MRI studies (Braams et al., 2015; Korucuoglu, Harms, Kennedy, et al., 2020; Rao et al., 2008), we observed greater brain activations in the lateral orbital frontal cortex (OFC) extending to the frontal operculum cortex and then the insula, striatum, including bilateral caudate and nucleus accumbens as well as the left putamen in very small clusters, the lateral occipital cortex, cerebellum, right frontal pole, right temporal gyrus, right brainstem, left cingulate gyrus, and left thalamus, as shown in Table 1 and Figure 3. Table 1 Local maxima for TFCE significantly activated for RWR> Neutral contrast. Local Maxima Coordinates (mm) Voxels X Y Z Right Occipital Pole 6943 19.5 -97.5 -4.5 Right Frontal Orbital Cortex, Right Insula 2449 33 27 -3 Right Frontal Pole 2036 42 55.5 12 Right Caudate, Right Accumbens 1216 10.5 3 10.5 Left Frontal Operculum Cortex, Left Insula 1065 -42 15 0 Right Cerebral White Matter 719 3 -34.5 3 Left Caudate, Left Accumbens 371 -9 3 9 Right Brain Stem (midbrain) 348 4.5 -13.5 -13.5 Right Middle Temporal Gyrus 250 63 -34.5 -13.5 Right Inferior Temporal Gyrus 46 55.5 -48 -15 Right Frontal Pole 28 21 66 -7.5 24 Right Temporal Occipital Fusiform Cortex 26 46.5 -58.5 -16.5 Right Frontal Pole 12 43.5 54 -7.5 Right Middle Temporal Gyrus 10 58.5 -21 -7.5 Left Thalamus 6 -1.5 -7.5 9 Left Thalamus 4 -16.5 -10.5 7.5 Left Putamen 3 -21 9 -3 Left Cingulate Gyrus 3 -1.5 -24 27 Right Inferior Temporal Gyrus 3 58.5 -58.5 -10.5 Left Putamen 1 -19.5 16.5 -1.5 Right Frontal Pole 1 16.5 54 -19.5 Left Thalamus 1 -13.5 -27 16.5 Left Thalamus 1 -16.5 -24 18 Fig. 3. Visualization of areas with greater activation in the RWR > Neutral contrasts at the group level (NAcc = Nucleus Accumbens). Images were FWER corrected (p=0.05) and overlaid on to the MNI 152 T1-weighted high resolution anatomical image. Sagittal (A), Coronal (B), and Axial (C) slices located at X = 10.5, Y = 3, Z = 10.5. 25 RWOR > Neutral The decision-making motivation contrast between RWOR and Neutral showed greater activation in right frontal pole, right middle temporal gyrus and bilateral occipital cortex including right occipital pole, lingual gyrus, occipital fusiform gyrus and lateral occipital cortex, and left intracalcarine cortex. However, we did not observe the greater activity of striatum in the RWOR condition, compared to the neutral condition against our hypothesis (see results in Table 2 and Figure 4). Table 2 Local maxima for TFCE significantly activated for RWOR> Neutral contrast. Local Maxima Coordinates (mm) Voxels X Y Z Right Occipital Pole 5804 12 -94.5 -1.5 Right Frontal Pole 455 42 58.5 4.5 Right Lateral Occipital Cortex 240 31.5 -70.5 34.5 Right Middle Temporal Gyrus 227 64.5 -28.5 -1.5 Left Intracalcarine Cortex 72 -19.5 -63 9 Right Lateral Occipital Cortex 28 42 -85.5 -9 Right Middle Temporal Gyrus 19 66 -22.5 -12 Left Intracalcarine Cortex 10 -16.5 -76.5 9 Right Frontal Pole 6 43.5 46.5 16.5 Right Lingual Gyrus 5 3 -75 -13.5 Right Lateral Occipital Cortex 4 28.5 -79.5 40.5 Left Intracalcarine Cortex 3 -22.5 -69 12 26 Right Lateral Occipital Cortex 1 36 -90 -7.5 Right Occipital Fusiform Gyrus 1 24 -70.5 3 Fig. 4. Visualization of areas with greater activation in the RWOR > Neutral contrasts at the group level. Images were FWER corrected (p = 0.05) and overlaid on to the MNI 152 T1- weighted high resolution anatomical image. Sagittal (A), Coronal (B), and Axial (C) slices were located at X = 13, Y = -80, Z = -1.5. RWR > RWOR Consistent with the third hypothesis, we found that the RWR condition coincided with a stronger hemodynamic response in the bilateral insula, lateral occipital cortex, caudate, thalamus and brain stem, left anterior cingulate cortex (ACC), right OFC, right putamen, and right frontal and occipital poles, based on Table 3 and Figure 4. Table 3 Local maxima for TFCE significantly activated for RWR> RWOR contrast. Local Maxima Coordinates (mm) Voxels X Y Z Right Occipital Pole 3501 18 -97.5 -4.5 27 Right Frontal Orbital Cortex, Right Insula 1501 46.5 19.5 -6 Left Insula 998 -40.5 18 -3 Right Frontal Pole 718 40.5 54 13.5 Right Caudate 450 10.5 1.5 12 Right Brain stem 384 6 -28.5 -6 Right Brain stem 240 4.5 -13.5 -13.5 Left Caudate 164 -9 3 9 Right Putamen 148 18 16.5 -6 Right Thalamus 119 4.5 -18 16.5 Right Frontal Pole 81 31.5 64.5 -10.5 Left Brain Stem 15 -4.5 -33 -15 Right Thalamus 15 3 -25.5 10.5 Left Brain Stem 12 0 -27 -18 Left Caudate 11 -15 18 -1.5 Left Anterior Cingulate Gyrus 7 0 -31.5 9 Right Caudate 4 15 -12 24 Left Brain stem 4 -4.5 -19.5 -18 Left Thalamus 2 -3 -18 0 Left Inferior Lateral Occipital Cortex 1 -45 -81 1.5 Right Frontal Pole 1 25.5 70.5 1.5 Right Frontal Pole 1 27 58.5 -9 Right Lateral Occipital Cortex 1 45 -82.5 -9 Right Thalamus 1 6 -24 13.5 28 Fig. 5. Visualization of areas with greater activation in the RWR > RWOR contrasts at the group level. Images were FWER corrected (p = 0.05) and overlaid on to the MNI 152 T1-weighted high resolution anatomical image. Sagittal (A), Coronal (B), and Axial (C) slices were located at X=10.5, Y=1.5, Z=12. Discussion We investigated the neural correlates of risk tolerance and reward sensitivity during decision-making using the modified BART paradigm. The results partially support our hypotheses. As expected, decision-making involving reward and risk was associated with the activity of mesolimbic-frontal areas. Surprisingly, decision-making involving just reward coincided with activation outside of the hypothesized mOFC and striatum. Finally, we also found decision-making involving reward and risk coincided with greater BOLD signals in the insula, ACC, and putamen, but not the amygdala, when contrasted with reward-only trials. Taken together, findings suggest that while decision-making in the context of rewards with and without risk coincides with increased activation in expected neural regions, rewards with risk involve greater neural engagement, especially in regions involved in error detection. 29 Decision Making in the Context of Rewards with Risk In line with our hypothesis and consistent with previous studies, the RWR condition coincided with greater activation in bilateral insula extending to the right lateral OFC, left frontal operculum cortex, bilateral striatum, and left cingulate gyrus (Korucuoglu, Harms, Kennedy, et al., 2020; Mohr et al., 2010; Rao et al., 2008; Wu et al., 2021). We also observed increased BOLD in the visual cortex, right frontal pole, temporal gyrus, and subcortical structures, including the thalamus, brainstem, and cerebellum, but not the amygdala. These findings are unique when compared to other BART studies but are also supported by decision neuroimaging research (Guo et al., 2013; Korucuoglu, Harms, Kennedy, et al., 2020; Rao et al., 2008). All BART trials in our modified paradigm involve a level of uncertainty, as the participant never knows when the balloon will pop or disappear. In line with our findings, a prior study also showed that greater activation in the occipital and visual cortex, posterior cerebellar lobe, and brainstem occurred during uncertain-reward decision-making when playing a poker game (Guo et al., 2013). Furthermore, the thalamus exhibits notable hemodynamic responses in decision-making, which aligns with the observed activity seen during the inflation of an active balloon in other BART studies (Kohno et al., 2016; Rao et al., 2008). Cho and colleagues (2013) found notable patterns in thalamic activity during uncertain decision-making and suggested that the thalamus “alerts” and this signal is combined with interoceptive details which subsequently are projected to the striatum immediately before a decision-making behavior. The cerebellum, too, has been implicated in decision making. Researchers found that left cerebellum gray matter volume was negatively and robustly associated with individual risk tolerance measured by BART (Quan et al., 2022). Taken together, this indicates that decision-making under risky rewards involves 30 greater calculations on the part of subcortical regions beyond the canonical striatum. We hypothesize that the use of multi-echo and zooming in on the midbrain regions provided greater signal clarity from these regions that might be seen using more common EPI scanning parameters. Further research focused on these regions provides exciting new ways to think about risky decision making that go beyond the rewarding component. We expected to see greater amygdala activation in the RWR condition, but the data did not yield this result. A previous study indicated that the amygdala may reflect the magnitude of risk (Mohr et al., 2010). In our analysis plan, we used a parametric modulator accounting for the number of pumps already completed and therefore the amount of money that was risked. Given this, it’s possible that amygdala signals were washed out. We suspect that comparing earlier versus later trials in any one block would coincide with increasing amygdala BOLD, perhaps tracking the magnitude of risk as other researchers have suggested. Decision Making in the Context of Rewards Without Risk The RWOR condition was a unique addition to the canonical BART paradigm. Like with other human and non-human studies on rewards, RWOR coincided with greater BOLD signals in the right frontal pole (Rushworth et al., 2011; Tsujimoto et al., 2010), right middle temporal gyrus (Murty et al., 2016), and bilateral occipital cortex (Yang & Shadlen, 2007). However, contrary to previous work and our own hypothesis, we saw no significantly greater activation in the mOFC or striatum. Medial OFC is thought to calculate the reward value of choices, and greater activation in this region is associated with more reward-seeking behavior (Rangel & Hare, 2010; Rudebeck et al., 2008). One explanation for the lack of mOFC in our RWOR condition could be that the participant’s behavior did not matter much in this condition. The balloon would continue to inflate and have no negative consequences. This is unique to this 31 paradigm, as even the MID reward condition depends on the participant’s timing of their button press for them to receive the reward. Lack of significant BOLD signals in both mOFC and striatum may reflect that individuals did not need to evaluate their potential choices or track changes within each pump as others have found (Jenni et al., 2022). Risk Tolerance vs. Reward Sensitivity The ultimate goal of this study was to differentiate the neural correlates of risky reward (RWR) from general reward sensitivity (RWOR). As expected, the RWR > RWOR contrast coincided with larger BOLD responses in bilateral insula, caudate, lateral occipital cortex, brain stem, and thalamus, as well as the right OFC, putamen, and frontal pole, and left anterior gyrus. The insular cortex is involved in signaling the probability of aversive outcomes, which could only occur in a risky condition (Clark et al., 2008). Furthermore, only the risky condition demanded encoding of risk costs, perhaps supported by increased BOLD signal in OFC, ACC (Rangel & Hare, 2010) and dorsal striatum which could enhance tracking changes in different conditions to gain flexible reward in the risk-and-reward condition (Jenni et al., 2022). The greater caudate BOLD signal also aligns with previous research, which indicates that the caudate nucleus is involved in shifting motivational contexts (Korucuoglu, Harms, Kennedy, et al., 2020; Qu et al., 2015), and there is high test-retest reliability of these caudate signals during risky decision making in the BART paradigm (Korucuoglu, Harms, Astafiev, et al., 2020). Implications of Current Study We modified the traditional BART paradigm to study neural correlate differences between risk tolerance and reward sensitivity within the same context. We found that the appearance of risk during decision making coincided with greater neural activation in regions generally related to computing risk, rewards, and errors, but not in the striatum, an area 32 canonically related to reward processing before and during receipt of a reward. Interpreted through the perspective of prospect theory, these findings may suggest the relatively larger importance of potential loss over potential gain in managing risk (Kahneman & Tversky, 2013). In other words, risk tolerance and not reward sensitivity per se may drive risky decisions and this is instantiated through greater recruitment of these neural regions. This is important when considering individual differences in reward sensitivity alone as a predictor of impulsive or risky behavior since this only captures a correlate and not a driving force. It also suggests that risk tolerance may be behind some of the impulsivity issues seen in certain neurodivergent populations, such as with ADHD. Risk itself may be stimulating dopamine-rich neural regions (e.g., caudate), resulting in some relief from under stimulation which rewards alone cannot provide. Future studies will need to engage with how individual differences in neural response to risky versus non-risky rewards may help us understand both healthy decision-making processes and those associated with psychopathology. Limitations and Future Directions There are limitations associated with the current task design. Although we refer to the period from stimulus presentation to button response in all three conditions as the decision phase, only the risk condition (i.e., RWR) involves a truly consequential decision. In RWOR and neutral conditions, participants could choose not to press the button, but this would just result in the screen not changing. The only “real” option was to continue to press the button until the balloon went away. This deviates from other reward paradigms in that the participant’s behavior was seemingly inconsequential to the receipt of reward. Unlike in MID, for example, which requires easy but accurate timing of the button press, performance in the RWOR trials was certain, and there was truly no risk of loss or opportunity cost. Though perhaps a limitation, it may also 33 elucidate how the utility or meaning of a behavior rather than just the reward itself is what stimulates striatum activation. In future studies, we will examine the different neural correlates of reward and risk processing in the anticipatory and consummatory phases of the BART decision context. This may help us comprehensively understand how environmental inputs, including risk and reward, calibrate people's anticipatory and feedback-related neural systems. Furthermore, we also did not consider individual differences in a trait of risk and reward motivations measured by self- reported questionnaires and how they influence people’s decision-making behavior in the current study (Hertwig et al., 2019). We will further utilize the self-reported BAS/BIS scales and SPSRQ to investigate what the neural correlates of individuals’ sensitivities toward risk and reward are (Carver & White, 1994; Torrubia et al., 2001). Conclusion In summary, our study investigated the neural distinction between risk tolerance and reward sensitivity using a modified BART paradigm with multi-echo MRI partial brain scanning. Task modification revealed preliminary evidence that risky decision-making recruits canonical reward areas and the insula, but that risk involves greater recruitment of prefrontal and midbrain regions within the same task. Curiously, reward-only trials did not coincide with significantly increased activation in the striatum, perhaps as a consequence of the reward being certain and the participants having little impact on the outcomes. We will complete further analysis of the impact of parametric modulators and model risk-related differences in the consummatory phase of the task to better differentiate risk tolerance from reward sensitivity. 34 Reference Ackerman, J. P., McBee-Strayer, S. M., Mendoza, K., Stevens, J., Sheftall, A. H., Campo, J. V., & Bridge, J. A. (2015). Risk-sensitive decision-making deficit in adolescent suicide attempters. Journal of Child and Adolescent Psychopharmacology, 25(2), 109–113. https://doi.org/10.1089/cap.2014.0041 Alloy, L. B., Olino, T., Freed, R. D., & Nusslock, R. (2016). Role of reward sensitivity and processing in major depressive and bipolar spectrum disorders. Behavior Therapy, 47(5), 600–621. https://doi.org/10.1016/j.beth.2016.02.014 Bartra, O., McGuire, J. T., & Kable, J. W. (2013). The valuation system: A coordinate-based meta-analysis of BOLD fMRI experiments examining neural correlates of subjective value. NeuroImage, 76, 412–427. https://doi.org/10.1016/j.neuroimage.2013.02.063 Braams, B. R., van Duijvenvoorde, A. C. K., Peper, J. S., & Crone, E. A. (2015). Longitudinal changes in adolescent risk-taking: A comprehensive study of neural responses to rewards, pubertal development, and risk-taking behavior. Journal of Neuroscience, 35(18), 7226– 7238. https://doi.org/10.1523/JNEUROSCI.4764-14.2015 Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: The BIS/BAS Scales. Journal of Personality and Social Psychology, 67(2), 319–333. https://doi.org/10.1037/0022- 3514.67.2.319 Charness, G., Gneezy, U., & Imas, A. (2013). Experimental methods: Eliciting risk preferences. Journal of Economic Behavior & Organization, 87, 43–51. https://doi.org/10.1016/j.jebo.2012.12.023 35 Chick, C. F. (2015). Reward processing in the adolescent brain: Individual differences and relation to risk taking. The Journal of Neuroscience, 35(40), 13539–13541. https://doi.org/10.1523/JNEUROSCI.2571-15.2015 Cho, Y. T., Fromm, S., Guyer, A. E., Detloff, A., Pine, D. S., Fudge, J. L., & Ernst, M. (2013). Nucleus accumbens, thalamus and insula connectivity during incentive anticipation in typical adults and adolescents. NeuroImage, 66, 508–521. https://doi.org/10.1016/j.neuroimage.2012.10.013 Clark, L., Bechara, A., Damasio, H., Aitken, M. R. F., Sahakian, B. J., & Robbins, T. W. (2008). Differential effects of insular and ventromedial prefrontal cortex lesions on risky decision-making. Brain, 131(5), 1311–1322. https://doi.org/10.1093/brain/awn066 Cox, R. W., & Hyde, J. S. (1997). Software tools for analysis and visualization of fMRI data. NMR in Biomedicine, 10(4–5), 171–178. https://doi.org/10.1002/(sici)1099- 1492(199706/08)10:4/5<171::aid-nbm453>3.0.co;2-l DuPre, E., Salo, T., Ahmed, Z., Bandettini, P., Bottenhorn, K., Caballero-Gaudes, C., Dowdle, L., Gonzalez-Castillo, J., Heunis, S., Kundu, P., Laird, A., Markello, R., Markiewicz, C., Moia, S., Staden, I., Teves, J., Uruñuela, E., Vaziri-Pashkam, M., Whitaker, K., & Handwerker, D. (2021). TE-dependent analysis of multi-echo fMRI with tedana. Journal of Open Source Software, 6(66), 3669. https://doi.org/10.21105/joss.03669 FeldmanHall, O., Glimcher, P., Baker, A. L., NYU PROSPEC Collaboration, & Phelps, E. A. (2019). The functional roles of the amygdala and prefrontal cortex in processing uncertainty. Journal of Cognitive Neuroscience, 31(11), 1742–1754. https://doi.org/10.1162/jocn_a_01443 36 Gentili, C., Di Rosa, E., Podina, I., Popita, R., Voinescu, B., & David, D. (2022). Resting state predicts neural activity during reward-guided decision making: An fMRI study on Balloon Analogue Risk Task. Behavioural Brain Research, 417, 113616. https://doi.org/10.1016/j.bbr.2021.113616 Guo, Z., Chen, J., Liu, S., Li, Y., Sun, B., & Gao, Z. (2013). Brain areas activated by uncertain reward-based decision-making in healthy volunteers. Neural Regeneration Research, 8(35), 3344–3352. https://doi.org/10.3969/j.issn.1673-5374.2013.35.009 Haber, S. N., & Knutson, B. (2010). The Reward Circuit: Linking Primate Anatomy and Human Imaging. Neuropsychopharmacology, 35(1), 4–26. https://doi.org/10.1038/npp.2009.129 Harden, K. P., Mann, F. D., Grotzinger, A., Patterson, M. W., Steinberg, L., Tackett, J. L., & Tucker-Drob, E. M. (2018). Developmental differences in reward sensitivity and sensation seeking in adolescence: Testing sex-specific associations with gonadal hormones and pubertal development. Journal of Personality and Social Psychology, 115(1), 161–178. https://doi.org/10.1037/pspp0000172 Hertwig, R., & Erev, I. (2009). The description–experience gap in risky choice. Trends in Cognitive Sciences, 13(12), 517–523. https://doi.org/10.1016/j.tics.2009.09.004 Hertwig, R., Wulff, D. U., & Mata, R. (2019). Three gaps and what they may mean for risk preference. Philosophical Transactions of the Royal Society B: Biological Sciences, 374(1766), 20180140. https://doi.org/10.1098/rstb.2018.0140 Jenni, N. L., Rutledge, G., & Floresco, S. B. (2022). Distinct medial orbitofrontal–striatal circuits support dissociable component processes of risk/reward decision-making. The Journal of Neuroscience, 42(13), 2743–2755. https://doi.org/10.1523/JNEUROSCI.2097-21.2022 37 Kahneman, D., & Tversky, A. (2013). Prospect Theory: An analysis of decision under risk. In L. C. MacLean & W. T. Ziemba, World Scientific Handbook in Financial Economics Series (Vol. 4, pp. 99–127). World Scientific. https://doi.org/10.1142/9789814417358_0006 Kohno, M., Nurmi, E. L., Laughlin, C. P., Morales, A. M., Gail, E. H., Hellemann, G. S., & London, E. D. (2016). Functional genetic variation in dopamine signaling moderates prefrontal cortical activity during risky decision making. Neuropsychopharmacology, 41(3), 695–703. https://doi.org/10.1038/npp.2015.192 Korucuoglu, O., Harms, M. P., Astafiev, S. V., Kennedy, J. T., Golosheykin, S., Barch, D. M., & Anokhin, A. P. (2020). Test-retest reliability of fMRI-measured brain activity during decision making under risk. NeuroImage, 214, 116759. https://doi.org/10.1016/j.neuroimage.2020.116759 Korucuoglu, O., Harms, M. P., Kennedy, J. T., Golosheykin, S., Astafiev, S. V., Barch, D. M., & Anokhin, A. P. (2020). Adolescent decision-making under risk: Neural correlates and sex differences. Cerebral Cortex, 30(4), 2691–2707. https://doi.org/10.1093/cercor/bhz269 Leff, D. R., Yongue, G., Vlaev, I., Orihuela-Espina, F., James, D., Taylor, M. J., Athanasiou, T., Dolan, R., Yang, G.-Z., & Darzi, A. (2017). “Contemplating the Next Maneuver”: Functional neuroimaging reveals intraoperative decision-making strategies. Annals of Surgery, 265(2), 320–330. https://doi.org/10.1097/SLA.0000000000001651 Lejuez, C. W., Read, J. P., Kahler, C. W., Richards, J. B., Ramsey, S. E., Stuart, G. L., Strong, D. R., & Brown, R. A. (2002). Evaluation of a behavioral measure of risk taking: The Balloon Analogue Risk Task (BART). Journal of Experimental Psychology: Applied, 8(2), 75–84. https://doi.org/10.1037/1076-898X.8.2.75 38 Lundblad, C. (2007). The risk return tradeoff in the long run: 1836–2003. Journal of Financial Economics, 85(1), 123–150. https://doi.org/10.1016/j.jfineco.2006.06.003 Lynch, C. J., Elbau, I., & Liston, C. (2021). Improving precision functional mapping routines with multi-echo fMRI. Current Opinion in Behavioral Sciences, 40, 113–119. https://doi.org/10.1016/j.cobeha.2021.03.017 Mathias, C. W., Dougherty, D. M., James, L. M., Richard, D. M., Dawes, M. A., Acheson, A., & Hill-Kapturczak, N. (2011). Intolerance to delayed reward in girls with multiple suicide attempts: Girls’ suicide attempts. Suicide and Life-Threatening Behavior, 41(3), 277– 286. https://doi.org/10.1111/j.1943-278X.2011.00027.x Mohr, P. N. C., Biele, G., & Heekeren, H. R. (2010). Neural processing of risk. The Journal of Neuroscience, 30(19), 6613–6619. https://doi.org/10.1523/JNEUROSCI.0003-10.2010 Murty, V. P., LaBar, K. S., & Adcock, R. A. (2016). Distinct medial temporal networks encode surprise during motivation by reward versus punishment. Neurobiology of Learning and Memory, 134, 55–64. https://doi.org/10.1016/j.nlm.2016.01.018 Nelson, B. D., McGowan, S. K., Sarapas, C., Robison-Andrew, E. J., Altman, S. E., Campbell, M. L., Gorka, S. M., Katz, A. C., & Shankman, S. A. (2013). Biomarkers of threat and reward sensitivity demonstrate unique associations with risk for psychopathology. Journal of Abnormal Psychology, 122(3), 662–671. https://doi.org/10.1037/a0033982 O’Neill, M., & Schultz, W. (2010). Coding of reward risk by orbitofrontal neurons Is mostly distinct from coding of reward value. Neuron, 68(4), 789–800. https://doi.org/10.1016/j.neuron.2010.09.031 Qu, Y., Galvan, A., Fuligni, A. J., Lieberman, M. D., & Telzer, E. H. (2015). Longitudinal changes in prefrontal cortex activation underlie declines in adolescent risk taking. 39 Journal of Neuroscience, 35(32), 11308–11314. https://doi.org/10.1523/JNEUROSCI.1553-15.2015 Quan, P., He, L., Mao, T., Fang, Z., Deng, Y., Pan, Y., Zhang, X., Zhao, K., Lei, H., Detre, J. A., Kable, J. W., & Rao, H. (2022). Cerebellum anatomy predicts individual risk-taking behavior and risk tolerance. NeuroImage, 254, 119148. https://doi.org/10.1016/j.neuroimage.2022.119148 Rangel, A., & Hare, T. (2010). Neural computations associated with goal-directed choice. Current Opinion in Neurobiology, 20(2), 262–270. https://doi.org/10.1016/j.conb.2010.03.001 Rao, H., Korczykowski, M., Pluta, J., Hoang, A., & Detre, J. A. (2008). Neural correlates of voluntary and involuntary risk taking in the human brain: An fMRI Study of the Balloon Analog Risk Task (BART). NeuroImage, 42(2), 902–910. https://doi.org/10.1016/j.neuroimage.2008.05.046 Richards, J. M., Plate, R. C., & Ernst, M. (2013). A systematic review of fMRI reward paradigms used in studies of adolescents vs. adults: The impact of task design and implications for understanding neurodevelopment. Neuroscience & Biobehavioral Reviews, 37(5), 976–991. https://doi.org/10.1016/j.neubiorev.2013.03.004 Rivalan, M., Ahmed, S. H., & Dellu-Hagedorn, F. (2009). Risk-prone individuals prefer the wrong options on a rat version of the Iowa Gambling Task. Biological Psychiatry, 66(8), 743–749. https://doi.org/10.1016/j.biopsych.2009.04.008 Rolls, E. T., Wan, Z., Cheng, W., & Feng, J. (2022). Risk-taking in humans and the medial orbitofrontal cortex reward system. NeuroImage, 249, 118893. https://doi.org/10.1016/j.neuroimage.2022.118893 40 Rudebeck, P. H., Behrens, T. E., Kennerley, S. W., Baxter, M. G., Buckley, M. J., Walton, M. E., & Rushworth, M. F. S. (2008). Frontal cortex subregions play distinct roles in choices between actions and stimuli. The Journal of Neuroscience, 28(51), 13775–13785. https://doi.org/10.1523/JNEUROSCI.3541-08.2008 Rushworth, M. F. S., Noonan, M. P., Boorman, E. D., Walton, M. E., & Behrens, T. E. (2011). Frontal cortex and reward-guided learning and decision-making. Neuron, 70(6), 1054– 1069. https://doi.org/10.1016/j.neuron.2011.05.014 Scott-Parker, B., & Weston, L. (2017). Sensitivity to reward and risky driving, risky decision making, and risky health behaviour: A literature review. Transportation Research Part F: Traffic Psychology and Behaviour, 49, 93–109. https://doi.org/10.1016/j.trf.2017.05.008 Sescousse, G., Caldú, X., Segura, B., & Dreher, J.-C. (2013). Processing of primary and secondary rewards: A quantitative meta-analysis and review of human functional neuroimaging studies. Neuroscience & Biobehavioral Reviews, 37(4), 681–696. https://doi.org/10.1016/j.neubiorev.2013.02.002 Smith, S., & Nichols, T. (2009). Threshold-free cluster enhancement: Addressing problems of smoothing, threshold dependence and localisation in cluster inference. NeuroImage, 44(1), 83–98. https://doi.org/10.1016/j.neuroimage.2008.03.061 Smoski, M. J., Lynch, T. R., Rosenthal, M. Z., Cheavens, J. S., Chapman, A. L., & Krishnan, R. R. (2008). Decision-making and risk aversion among depressive adults. Journal of Behavior Therapy and Experimental Psychiatry, 39(4), 567–576. https://doi.org/10.1016/j.jbtep.2008.01.004 Sun, S., Cai, C., & Yu, R. (2022). Behavioral and neural representation of expected reward and risk. NeuroImage, 264, 119731. https://doi.org/10.1016/j.neuroimage.2022.119731 41 Talmi, D., Dayan, P., Kiebel, S. J., Frith, C. D., & Dolan, R. J. (2009). How humans integrate the prospects of pain and reward during choice. The Journal of Neuroscience, 29(46), 14617–14626. https://doi.org/10.1523/JNEUROSCI.2026-09.2009 Tobler, P. N., O’Doherty, J. P., Dolan, R. J., & Schultz, W. (2007). Reward value coding distinct from risk attitude-related uncertainty coding in human reward systems. Journal of Neurophysiology, 97(2), 1621–1632. https://doi.org/10.1152/jn.00745.2006 Torrubia, R., Ávila, C., Moltó, J., & Caseras, X. (2001). The Sensitivity to Punishment and Sensitivity to Reward Questionnaire (SPSRQ) as a measure of Gray’s anxiety and impulsivity dimensions. Personality and Individual Differences, 31(6), 837–862. https://doi.org/10.1016/S0191-8869(00)00183-5 Trepel, C., Fox, C. R., & Poldrack, R. A. (2005). Prospect theory on the brain? Toward a cognitive neuroscience of decision under risk. Cognitive Brain Research, 23(1), 34–50. https://doi.org/10.1016/j.cogbrainres.2005.01.016 Tsujimoto, S., Genovesio, A., & Wise, S. P. (2010). Evaluating self-generated decisions in frontal pole cortex of monkeys. Nature Neuroscience, 13(1), 120–126. https://doi.org/10.1038/nn.2453 Van Duijvenvoorde, A. C. K., Op De Macks, Z. A., Overgaauw, S., Gunther Moor, B., Dahl, R. E., & Crone, E. A. (2014). A cross-sectional and longitudinal analysis of reward-related brain activation: Effects of age, pubertal stage, and reward sensitivity. Brain and Cognition, 89, 3–14. https://doi.org/10.1016/j.bandc.2013.10.005 Van Duijvenvoorde, A. C. K., Peters, S., Braams, B. R., & Crone, E. A. (2016). What motivates adolescents? Neural responses to rewards and their influence on adolescents’ risk taking, 42 learning, and cognitive control. Neuroscience & Biobehavioral Reviews, 70, 135–147. https://doi.org/10.1016/j.neubiorev.2016.06.037 Van Duijvenvoorde, A. C. K., Van Hoorn, J., & Blankenstein, N. E. (2022). Risks and rewards in adolescent decision-making. Current Opinion in Psychology, 48, 101457. https://doi.org/10.1016/j.copsyc.2022.101457 Van Leijenhorst, L., Moor, B. G., Op de Macks, Z. A., Rombouts, S. A. R. B., Westenberg, P. M., & Crone, E. A. (2010). Adolescent risky decision-making: Neurocognitive development of reward and control regions. NeuroImage, 51(1), 345–355. https://doi.org/10.1016/j.neuroimage.2010.02.038 Vandeweghe, L., Matton, A., Beyers, W., Vervaet, M., Braet, C., & Goossens, L. (2016). Psychometric properties of the BIS/BAS Scales and the SPSRQ in Flemish adolescents. Psychologica Belgica, 56(4), 406. https://doi.org/10.5334/pb.298 Weber, E. U. (2010). Risk attitude and preference. WIREs Cognitive Science, 1(1), 79–88. https://doi.org/10.1002/wcs.5 Woolrich, M. W., Jbabdi, S., Patenaude, B., Chappell, M., Makni, S., Behrens, T., Beckmann, C., Jenkinson, M., & Smith, S. M. (2009). Bayesian analysis of neuroimaging data in FSL. NeuroImage, 45(1), S173–S186. https://doi.org/10.1016/j.neuroimage.2008.10.055 Wu, S., Sun, S., Camilleri, J. A., Eickhoff, S. B., & Yu, R. (2021). Better the devil you know than the devil you don’t: Neural processing of risk and ambiguity. NeuroImage, 236, 118109. https://doi.org/10.1016/j.neuroimage.2021.118109 Yang, T., & Shadlen, M. N. (2007). Probabilistic reasoning by neurons. Nature, 447(7148), 1075–1080. https://doi.org/10.1038/nature05852