Neuropsychologia 67 (2015) 100–110

Contents lists available at ScienceDirect

Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia

The neural basis of social risky decision making in females with major depressive disorder Robin Shao a,b,1, Hui-jun Zhang a,b,1, Tatia M.C. Lee a,b,c,d,n a

Laboratory of Neuropsychology, The University of Hong Kong, Hong Kong Laboratory of Cognitive Affective Neuroscience, The University of Hong Kong, Hong Kong c The State Key Laboratory of Brain and Cognitive Sciences, The University of Hong Kong, Hong Kong d Institute of Clinical Neuropsychology, The University of Hong Kong, Hong Kong b

art ic l e i nf o

a b s t r a c t

Article history: Received 4 September 2014 Received in revised form 29 November 2014 Accepted 7 December 2014 Available online 9 December 2014

Recent evidence indicates that Major Depressive Disorder (MDD) may be associated with reduced tendency of committing noncompliant actions during social decision-making even when the risk of being punished is low. The neural underpinnings of this behavioral pattern are unknown, although it likely relates to compromised functioning of the lateral prefrontal-striatal/limbic networks implicated in executive control, emotion regulation and risk/value-based instrumental behaviors. We employed a modified trust game (TG) that provided explicit information on the risk levels of cheating behaviors being detected and punished. Behavioral and neuro-image data were acquired and analyzed from 14 first-episode female MDD patients and 15 age- and gender-matched controls performing the role of trustee in the TG. Relative to controls, MDD patients exhibited less behavioral switching to making cheating choices under low risk, and reduced activity in the dorsal putamen, anterior insula and dorsolateral prefrontal cortex (DLPFC) during making low-risk cheating versus benevolent choices, with limited evidence indicating abnormal bilateral inferior frontal gyrus activities of patients when making high-risk cheating versus benevolent choices. Patients' left dorsal putamen/anterior insular signals correlated positively with their frequency of low-risk cheating. MDD patients' symptom severity correlated positively with their signals in the lateral prefrontal networks during decision-making. A psychophysiological interaction analysis provided tentative evidence for the recruitment of IFG-striatal/limbic circuitry among the control participants, but greater frontopolar-striatal/limbic connectivity among the MDD patients, during low-risk decision-making. We propose that making risky social decisions based on the balancing of self-gain and other's welfare relies on the functioning of the integrated lateral prefrontal-striatal/limbic networks, which are less efficient and dysregulated among MDD patients compared with controls, impacting negatively on the patients' social capacity and highlighting a key therapeutic target for MDD. & 2014 Elsevier Ltd. All rights reserved.

Keywords: Major Depressive Disorder Social cognition Risky decision making Anterior insula Dorsal putamen Prefrontal regions

1. Introduction Major Depressive Disorder (MDD) is a devastating condition accompanied by impaired social functioning (Segrin, 2000) and less fulfilling social interactions (Nezlek et al., 1994), with

Abbreviations: MDD, Major Depressive Disorder; TG, Trust Game; DLPFC, dorsolateral prefrontal cortex; IFG, inferior frontal gyrus; VLFPC, ventrolateral prefrontal cortex; BDI-II-C, the Chinese version of Beck Depression Inventory; BOLD, Blood-Oxygen-Level-Dependent; SVC, small volume correction; VBM, voxel-based morphometry; PSC, percentage signal change; PPI, psycho-physiological interaction n Correspondence to: Room 656, Jockey Club Tower, Pokfulam Road, The University of Hong Kong, Hong Kong. E-mail address: [email protected] (T.M.C. Lee). 1 Both authors contributed equally to this work. http://dx.doi.org/10.1016/j.neuropsychologia.2014.12.009 0028-3932/& 2014 Elsevier Ltd. All rights reserved.

patients often benefiting from good social support during symptom recovery (Brugha et al., 1997). It is hence of high theoretical and clinical significance to understand the neural underpinnings of the impaired social decision-making process in MDD. One component of social interaction that plays important roles in evolutionary human survival is trust, which is based on presumed reciprocity and social promise among multiple parties (Rilling and Sanfey, 2011). Nevertheless, a trust relationship is unstable and can be broken by cheating behaviors from any involved parties, which may in turn be detected and punished by the (adversely) affected parties (de Quervain et al., 2004). Therefore, adaptive social functioning involves the balancing of self-gain and other's welfare based on considerations of the risk of cheating detection, which processes might be altered in MDD (Zhang et al., 2012).

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

Previous research conducted on healthy populations has employed the multi-round trust game (TG), in which 2 players performed the roles of ‘investor’ and ‘trustee’ while engaging in a number of interactions involving investment and return (KingCasas et al., 2005). Focusing on the trustee, the TG measures both social reciprocity (benevolence), when the trustee returns a satisfactory share to the investor and keeps a social promise, and social non-reciprocity (cheating), when the trustee returns less than the investor’s expectation and breaks a social promise. While benevolence or compliance might be the normative response, cheating or defiance could be motivated by interest in self-gain and/or in punishing the investor for ungenerous behavior (de Quervain et al., 2004). In a recent study involving both healthy and MDD samples (Zhang et al., 2012), different risk levels were incorporated into the modified TG by systematically manipulating the probability that the trustee's cheating behavior would be detected by the investor and punished, since the capacity to accurately evaluate risk and act accordingly is critical to adaptive decision-making (Shao and Lee, 2014). Zhang et al.'s (2012) findings indicate that while non-depressed participants switched to giving more cheating responses when the risk was low, MDD patients showed less of such behavioral switching, suggesting (relative) lack of integration of risk information to social decision-making. Past research has generated valuable findings on neural responses elicited by social decision-making on trust and reciprocity, among both healthy participants and patients with emotion or affective disorders (Delgado et al., 2005; Krueger et al., 2007; KingCasas et al., 2008; Sripada et al., 2009). Making trust decisions in healthy participants were associated with activations of the striatum and dorsal prefrontal regions (Delgado et al., 2005; Krueger et al., 2007), and dorsal striatal activities in the trustees delivered prediction-error-like signals that encoded the behavioral patterns of the investor in a multi-round TG (King-Casas et al., 2005). These findings are consistent with existing evidence indicating important roles of the dorsal striatal regions in instrumental learning and actions (Balleine et al., 2007; Haruno and Kawato, 2006). On the other hand, patients with social anxiety disorders exhibited reduced prefrontal activities in the superior and inferior frontal gyri (IFG) when making trust decisions compared with control participants (Sripada et al., 2009), implicating possible deficits in executive control and affect regulatory processes (Grecucci et al., 2013; Miller and Cohen, 2001; Yamasaki et al., 2002). Specifically, the inferior frontal gyri (IFG) is a critical area for action implementation based on emotive and value information, through connections with limbic areas via the insula and with motor cortices (Carr et al., 2003). Thus, IFG dysfunction may lead to compromised capacity in making goaldirected decisions under inhibitory and regulatory control of prefrontal networks (Gotlib et al., 2005). Furthermore, both healthy participants and patients with borderline personality disorder exhibited signals in the anterior insula that negatively predicted their reciprocity in a trust game (King-Casas et al., 2008), consistent with evidence implicating this region in risky decision-making and behaviors (Clark et al., 2008; Kuhnen and Knutson, 2005; Paulus et al., 2003). The IFG, anterior insula and dorsolateral prefrontal (DLPFC) areas were all involved in social promise breaking (Baugmgartner et al., 2009), and in non-reciprocal trustee behavior in a multiround anonymous TG (Bereczkei et al., 2013), possibly due to the greater demand on executive control and emotion regulatory processes when committing the more risky actions that violate social norm (Spencer et al., 2001). MDD is associated with functional deficiencies in the striatal/ limbic circuitries (Dichter et al., 2009; Gotlib et al., 2010; Smoski et al., 2009) and associated lateral prefrontal networks (Gotlib et al., 2005; Mayberg, 1997; Ochsner and Gross, 2005). The rostral dorsal putamen is a key region involved in instrumental actions based on expected reward value (Haruno and Kawato, 2006;

101

Lerchner et al., 2007; McClure et al., 2003) and punishment (Bjork et al., 2008; Voon et al., 2010), through functional connections with lateral prefrontal, sensori-motor, limbic and other striatal networks (Postuma and Dagher, 2006). MDD patients show reduced left putamen and insular activities during reward anticipation and selection (Gotlib et al., 2010; Smoski et al., 2009), with the former normalized following treatment (Dichter et al., 2009). As such MDD individuals typically exhibit altered sensitivity to reward and punishment (Mogg et al., 1995; Pizzagalli et al., 2008; Von Gunten et al., 2011), as well as to risk (Corwin et al., 1990; Smoski et al., 2008), during decision-making. Previous research also indicates that MDD individuals show abnormal activities of the lateral prefrontal networks, such as the IFG, during emotion regulatory processes (Gotlib et al., 2005; Ochsner and Gross, 2005). As making defiant social decisions requires regulating prosocial emotions such as guilt and empathy (Elliott et al., 2011; O’Connor et al., 2007), MDD individuals who are less capable of resolving those emotions may experience difficulties in deciding to cheat, even when the risk of such behavior being detected is low (Thayer et al., 2003; Pulcu et al., 2013; Zhang et al., 2012). Furthermore, MDD is associated with impaired executive functioning such as working memory (Landro et al., 2001), mental flexibility (Beats et al., 1996) and inhibitory control (Dunkin et al., 2000), along with abnormal DLPFC activities during executive performance (Fales et al., 2008; Okada et al., 2003; Siegle et al., 2007). It was proposed that MDD individuals' elevated DLPFC activations compared with controls may serve as a compensatory mechanism for their reduced prefrontal functional efficiency (Harvey et al., 2005; Wagner et al., 2006). In the present study, we aimed to investigate the neural basis of MDD patients' reduced tendency of making low-risk cheating choices in the modified TG. Such investigation would give insights on a social cognitive/affective model of MDD, elucidate on the underlying neural mechanism of the disorder, as well as informing intervention techniques targeted at improving social efficacy and/ or regulating the functioning of key neural circuitries (e.g. Dichter et al., 2009; Young et al., 2006). We included only patients with first MDD episode in order to avoid the confounding effect of disorder recurrence and longitudinal disorder severity (Harvey et al., 2004). Only female participants were included, as females are more representative of the total MDD population (Kessler et al., 2003), and previous evidence showed gender differences in risk-taking tendency (Byrnes et al., 1999) and in social decisionmaking preference (Andreoni and Vesterlund, 2001). We hypothesized that, based on previous findings (Zhang et al., 2012), MDD individuals would exhibit less behavioral switching from benevolent to cheating responses when the risk level changed from high to low. At the neural level, depressed individuals would show reduced functional activities in neural circuitries implicated in risk/value-based instrumental behaviors such as the anterior insula and dorsal putamen, as well as in DLPFC networks implicated in executive control, during making the critical low-risk cheating versus benevolent responses compared with non-depressed controls, corresponding to their behavioral differences. Also, MDD patients would show elevated IFG activities during making cheating relative to benevolent responses compared to controls, particularly when the risk was high, which situation was expected to elicit the maximal level of negative affect.

2. Materials and methods 2.1. Participants Ethical approval was granted by The University of Hong Kong and the Hospital Authority of Hong Kong West Cluster. Thirty-four

102

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

Table 1 Demographic and psychometric details as well as the proportions (in % of the total number of choices) of cheating task choices under high and low risk conditions of 14 MDD patients and 15 healthy controls matched on age, years of education and IQ scores who completed a modified trust game in a standard fMRI protocol. Both mean and standard error of the mean are presented, as well as 2-tailed independent t-test statistics and associated p values. Note that the proportions of benevolent choices were (100% – % cheating choices).

Age (in years) Years of education BDI-II-Ca CAS positiveb CAS negativeb ERQ (R-S)c BIS-Cd RSPM-rawe High-risk cheating Low-risk cheating

MDD Patients

Controls

t Value

p Value

39.717 2.12 11.86 7 0.65 25.50 7 3.36 11.79 7 1.28 22.077 1.80 12.96 7 1.46 64.507 2.48 47.317 1.86 14.25 7 3.02% 29.80 7 7.80%

41.007 2.71 13.80 7 1.26 7.737 1.92 22.007 1.07 11.53 7 1.92 16.077 1.51 60.337 3.37 44.93 7 3.26 21.69 7 4.34% 53.36 7 6.79%

 0.37  1.34 4.59  6.14 3.99  1.52 0.98  0.63  1.41  2.29

0.71 0.19 p o 0.001 p o 0.001 p o 0.001 0.14 0.33 0.54 0.17 0.03

a BDI-II-C: the Chinese version of Beck Depression Inventory (Byrne et al., 2004). b CAS Positive/negative: the positive/negative affect scale of the Chinese Affect Scale (Hamid and Cheng, 1996). c ERQ (R  S): The Chinese version of the Emotion Regulation Questionnaire (Gross and John, 2003; Yan and Yu, 2009): Reappraisal subscale score minus Suppression subscale score. d BIS-C: the Chinese version of the Barratt Impulsiveness Scale (Yao et al., 2007). e RSPM-Raw: Raw scores of the Raven's Standard Progressive Matrices Test (Raven, 2000).

Chinese females were recruited (age range ¼25–55 years), of which eighteen were clinically diagnosed patients in first episode of MDD according to the criteria in DSM-IV (APA, 1994) (age¼25– 55 years). The patients were free of any history of organic brain disorders, traumatic brain injuries, substance abuse/dependence disorders, psychotic disorders, or affective disorders other than MDD that might affect cognitive functioning. No patient had received formal diagnosis of anxiety disorders. All patients were taking antidepressant medications at the time of experiment. All but one patient were taking selective serotonin reuptake inhibitors (SSRIs) combined with other antidepressant/anxiolytic drugs such as benzodiazepine and imidazopyridine. One patient was taking imidazopyridine only. No patient had any history of abusing substances including alcohol. Except from one patient who was an exsmoker, no patient had had any history of nicotine use. The remaining sixteen participants were healthy controls who matched with the patients on age, years of education, and intellectual abilities (Table 1). All control participants were free of any history of major physical illnesses, neurological or psychological disorders. All participants gave written informed consent for participation. The data of five participants (four MDD patients and one control participant) were subsequently excluded due to unsuitability for the MRI scanner or no cheating response given (see Section 2.2), leaving fourteen MDD patients and fifteen controls in the data analysis. Further details about participant recruitment are provided in Supplementary materials. 2.2. Experimental task The participant played the role of trustee who made decisions on the proportion of appreciated investment to return to another female player (the investor) at another computer terminal (whose responses were actually delivered by a computer program) (Zhang et al., 2012). Anonymity was maintained by instructing the participant that she would be playing with a number of other female participants with unknown identities in the role of investors

during the task, such that she might be faced with a different investor on any given trial. This manipulation was to prevent the perception of the need for long-term relationship formation or reputation building, which could exert additional influences on participants' task responses (e.g. Wardle et al., 2013). The participant was also informed explicitly that she would not be having contacts with the counterparts either before or after the experimental task, so as to remove potential confounding influences on the participant's task responses towards the investor. The participant was provided with explicit information about both the requested return of the investor and the risk associated with returning less than requested, which, if detected by the investor, would result in the confiscation of the earnings for that trial. The modified TG task was implemented using the Eprime v1.1 software. At each trial onset, the participant was notified with the high- or low-risk level which carried 65% or 35% probability of being detected, respectively. Following a variable duration of 3, 4 or 5 s (Fig. 1), the participant was informed about the total appreciated investment (indicated by the number to the right of the upper bar) and the requested return by the investor, as indicated by the number inside the green bar (70%, 80% or 90%). Both the total appreciated investment and the requested return by the investor were balanced across the high- and low-risk conditions. The participant could adjust the repayment via button-press within a 5-s interval. If the repayment was equal to or greater than requested, such response was considered ‘benevolent’ (see Supplementary materials). Returning less than requested was classified as ‘cheating’. The participant was instructed to only start adjusting the repayment after having reached a decision. Following the choice, the 2-s outcome phase was delivered after a 1-s anticipatory phase (Fig. 1). If cheating was detected, the participant earned nothing for that trial. The task consisted of 120 trials in total delivered in 5 scan sessions. All task parameters are balanced across sessions. Further details about the modified TG task are presented in Supplementary materials. 2.3. Experimental procedure Participants completed the Chinese version of the Beck Depression Inventory (BDI-II-C) (Byrne et al., 2004), the Chinese Affect Scale (CAS) (Hamid and Cheng, 1996), the Chinese version of the Barratt Impulsiveness Scale (BIS-C) (Yao et al., 2007) and the Raven’s Standard Progressive Matrices Test (Raven, 2000). Negative affect as measured by the CAS was previously found to correlate positively with neuroticism, negative self-appraisal, stress and pessimism, and the positive affect scale of the CAS correlated positively with extraversion, positive self-appraisal, optimism and self-esteem (Hamid and Cheng, 1996). The participant also completed the Chinese version of the Emotion Regulation Questionnaire comprised of 2 subscales: reappraisal and suppression (Gross and John, 2003; Yan and Yu, 2009). As previous evidence indicates that emotion reappraisal was associated with greater positive emotion, lesser negative emotion and better social functioning whereas emotion suppression was associated with the reverse (Gross and John, 2003), we subtracted participants' suppression scores from their appraisal scores to generate an ERQ score that reflects the capacity to regulate emotions positively (Table 1). The participant then sat in front of a computer that delivered task instructions and practice tasks. To reinforce the participant's belief that she was playing with human partners, a slide was presented to the participant showing the different computer terminals being connected to each other via a central server. The same slide was also displayed at the beginning of scanning to simulate the network connecting process. The participant was debriefed after scanning.

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

103

Fig. 1. The modified Trust Game (TG). Participants first saw a display indicating the level of risk that cheating response would be detected by the investor (High ¼65% or Low¼ 35%), which lasted for 3, 4 or 5 s. They were then notified with the amount of total appreciated investment indicated by the number to the right of the upper bar (60, 80, 120 or 160), and the proportion of return requested by the investor (70%, 80% or 90%), indicated by the number inside the lower red bar. Participants had 5 s to increase (right button press) or decrease (left button press) the lower red bar to make either benevolent (returning equal to or more than the requested proportion) or cheating (returning less than the requested proportion) choices, within a restricted range of 60–100%. Following a 1-s anticipatory phase, participants were presented with 1 of 4 possible outcomes (detected benevolence, undetected benevolence, detected cheating, undetected cheating), which lasted for 2 s. Participant gained the amount difference between total appreciated investment and the returned amount unless her cheating response was detected, in which case participant gained no money. The task consisted of 120 trials grouped into 5 sessions.

2.4. Behavioral data analysis MDD patients and control participants were compared on demographic and psychometric measures using independent-samples t-tests. The participant's proportions of choices (relative to the total number of choices), reaction times and their mean amounts of actual relative to requested repayments (see Section 3) were analyzed with repeated-measures ANOVA, including the withinsubjects factors of choice and risk, and the between-subjects factor of group. Simple-effects tests were carried out to clarify higherlevel ANOVA results where applicable. The influence of Beck Depression Inventory (BDI-II-C) scores on MDD patients' choices was assessed as a covariate variable using the same repeated-measures ANOVA model within the patient group. Statistical significance level for between-group comparisons based on a priori hypotheses was set at p o0.05, two-tailed. 2.5. Image acquisition, processing and analysis Data were acquired with a 3-Tesla Phillips scanner equipped with a standard eight-channel head coil. A total of 750 volumes of functional data (150 volumes per session) were collected as 3  3  3.5 mm3 T2-weighted echo-planar images (slice number/TR/ TE/flip angle ¼40/2000 ms/30 ms/90°, matrix¼ 128  128, FOV ¼ 230  230 mm2) along the anterior–posterior plane. Anatomical images were acquired with a T1-weighted spin-echo pulse sequence with spatial resolution of 1  1  1 mm3. Image preprocessing was carried out using the SPM8 software (Wellcome Trust Center for Neuroimaging, UCL, UK). Functional images were high-pass filtered at 128 s and realigned to the first image of the

scan session in order to correct for head motion artifacts. Each volume was corrected for slice acquisition timing and smoothed with a Gaussian filter (full-width-half-maximum ¼6 mm). The functional images were co-registered to high-resolution T1 images at the participant level and normalized to the Montreal Neurological Institute (MNI) template (resolution ¼ 3  3  3 mm3) using unified segmentation T1 images. Events of interest were convolved with a gamma hemodynamic-response-function (HRF) for the modeling of Blood-Oxygen-Level-Dependent (BOLD) signals. Analysis focused on brain signals elicited by making decisions on the repayment (Fig. 1), employing a general linear model with a factorial design involving within-subjects factors of risk (high versus low) and choice (benevolent versus cheating), and the between-subjects factor of group (MDD patients versus controls). BOLD signals during decisionmaking were modeled with a duration equivalent to participant's reaction time, i.e. the duration between the onset of the choice phase and the time of the first button press. At the participant level, individual event contrasts of interest were evaluated with one-sample t-tests. At the group level, each contrast was assessed with one-sample t-test for the MDD patients and control participants separately, and with independent-samples t-tests to evaluate the interactive effect of group and event contrast. The effects of participants' mean repayment difference scores (RDS) when making benevolent and cheating choices (see Section 3), as well as those of MDD patients' BDI-II-C scores, on BOLD signals were also assessed as regressors-of-interest. We collapsed benevolence- and cheating-related RDS across risk levels for the imaging analysis as RDS associated with the same type of choice under high and low risk showed considerable inter-

104

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

correlations across participants (|r|s 40.3, ps o0.27). Whole-brain activation threshold was set at po0.001 at peak level, and p o0.01 at cluster level, corresponding to a cluster size of Z 15 according to the Monte-Carlo simulation. Activations within areas of a priori interest, i.e., the putamen, insula and IFG, were further evaluated with Small-Volume-Correction (SVC) analysis using bilateral automated anatomical labeling (AAL) masks, cluster-corrected at p o0.05. Voxel-Based-Morphometry (VBM) analyzes were carried out to minimize confounding effect of anatomical volumetric differences between the patients and controls. We linked the image and behavioral data by extracting the percentage-signal-change (PSC) values within key areas of interest using Marsbar toolbox, and correlating those data with participants' task responses. Statistical significance level was set at p o0.05. Finally, psycho-physiological interaction (PPI) analysis was carried out to elucidate the functional connectivity networks during key task events. Activation threshold for the exploratory PPI analysis was set at p o0.001 at peak level, uncorrected.

3. Results 3.1. Behavioral analysis The clinical and control participants were matched on age, years of education and IQ score (ps 40.1). Relative to controls, MDD patients reported higher BDI-II-C scores, as well as greater negative affect and lower positive affect as measured by the CAS (Table 1). The patients also showed a trend of having lower emotion regulation capacity than the controls (Table 1), which correlated negatively with their BDI-II-C scores (r (14) ¼  0.59, p o0.05). MDD patients and controls did not differ in their levels of impulsiveness as measured by the BIS-C. Participants overall made more benevolent than cheating choices, F (1, 27) ¼38.552, p o0.05, consistent with the former being the socially normative response. Participants' choices were also significantly modulated by risk, F (1, 27) ¼ 22.329, p o0.05. Participants made substantially greater numbers of benevolent choices under high risk, t (28) ¼ 11.723, p o0.05, but comparable benevolent and cheating choices under low risk (p 40.1) (Fig. 2). The 3-way group  risk  choice effect just failed to reach the conventional statistical threshold (F (1, 27) ¼ 2.605, p ¼0.118). Further analyzes revealed that MDD patients and controls did not differ in choices under high risk (p 40.1) (Fig. 2a). However, MDD patients made fewer cheating choices than control participants under low risk, t (27) ¼2.29, p o0.05 (Table 1 and Fig. 2b). Similar results were obtained on participants' reaction times (see Supplementary materials). We also analyzed participants' mean repayment difference scores (RDS) calculated by subtracting the investor's requested repayments from the participant’s actual repayments on each trial, which were then averaged within each of the 4 choice  risk conditions for the patient and control groups (Supplementary Table S1). Thus, more positive and negative RDS indicate greater over-repayments and under-repayments, respectively. ANOVA analysis revealed a significant main effect of choice (F (1, 27) ¼ 266.513, p o0.05), risk (F (1, 27) ¼9.273, p o0.05) and choice  risk interaction (F (1, 27) ¼4.733, p o0.05). Participants overall chose to repay less under low risk, particularly when giving cheating responses (Table S1). The factor of group did not interact with any of the within-subjects factors (ps 4 0.1). As an additional check, we tested for group difference in RDS in each of the 4 choice  risk conditions, and found no significant effect (ps 40.05). MDD patients' BDI-II-C scores had no significant influence on patients' choices or on the effect of risk on choices (ps 40.01). Further details of behavioral analysis are presented in Supplementary materials.

Fig. 2. MDD patients' and control participants' proportions (in percentage) of benevolent and cheating choices under (a) high (65% detection rate) and (b) low risk (35% detection rate) conditions. Error bars indicate one standard error of the mean.

3.2. Image analysis 3.2.1. Events of interest Table 2 displays the activation clusters elicited by individual event contrasts. We will focus only on contrasts in which we had a priori theoretical interest (i.e. choice  risk), and those showing significant signal differences between MDD patients and the controls. Other results are included in Supplementary Table S2. When the detection risk was high, making cheating relative to benevolent choices was associated with bilateral IFG and right posterior insular activations in the MDD patients (Fig. 3a), but with dorsal caudate activations in the control participants. Interestingly, relative to the control participants, a region of the right middle frontal gyrus extending into the IFG (local maxima¼45, 27, 24, t¼4.56, cluster size ¼ 30, surviving SVC test) exhibited activities that were modulated more positively by inter-participant variations in mean cheating-related RDS among MDD patients. When the risk was low, the same contrast elicited significantly greater activities in the left superior and middle frontal gyri, right posterior insula and left anterior dorsal putamen among the control participants, compared with MDD patients (Fig. 3b). Control participants also exhibited activities in a region of the left superior frontal gyrus (local maxima¼  24, 51, 30, t¼5.80, cluster size¼21) that were modulated significantly and positively by cheating-related RDS. Direct comparison between different risk levels revealed that making benevolent choices under high versus low risk elicited significantly greater signals in the right IFG among the controls compared with MDD patients. Making cheating choices under high versus low risk elicited significantly greater activities in the parietal, occipital and frontal motor areas among the MDD patients relative to the controls. The control participants, but not the

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

Table 2 The Montreal Neurological Institute (MNI) coordinates and associated t statistics of the local maxima of the activation clusters identified by group-level analysis on BOLD signals elicited by events of interest, based on the factorial design involving within-subjects factors of choice (benevolent versus cheating) and risk (high versus low), and the between-subjects factor of group (MDD patients versus controls). Activation threshold was set at p o 0.001 at peak-level and p o 0.01 at cluster level (minimal cluster size ¼ 15). Activations within regions of a priori interest (putamen, insula, inferior frontal gyrus) were additionally assessed with Small Volume Correction (SVC) test using bilateral automated anatomical labeling (AAL) masks, thresholded at p o 0.05 at cluster level. BAa

Area

High-Risk: Benevolent4 Cheating Control Precentral gyrusc 4

Side KEb Max t X

Y

Z

L

15

4.52 4.00

 30  27

 24 51  15 51

L

28

6.57

 36 33

9

11/47 R

10

13

R L

15 18

4.57 4.09 5.74 4.95

33 42 39  18

 12 9 18 27

3

L L

21 18

5.34 5.88

 27 9 6  42  27 51

46

L

17

4.49

 27 48

18

13

R L

25 15

4.22 4.10

45 6  30 6

6 6

47

R

19

5.32

30

33

6

19

R

27

6.40

24

 84 21

17/18 R

60

2

L

54

5.52 3.58 4.64 3.87

12 0  60  51

 84  3  81  3  21 33  21 24

6

R

35

4.56

6

9

2/4

R

34

4.24

63 60

 15 36  24 30

L

18

4.88

 30 12  39 15

9 3

(Low – High Risk) L 15 4.73

 39 12

 12

(Low – High Risk)d L 24 6.42 4.74 4.22 L 33 4.84 3.80

 27  30  33  33  39

6 3 3 6 6

d

High-Risk: Cheating 4 Benevolent MDD Inferior frontal gyrusc,e Inferior frontal gyruse Insulac Control Caudatec Low-Risk: Benevolent4Cheating MDD Putamenc,e Control Postcentral gyrusc Low-Risk: Cheating 4Benevolentd Control4 Superior/Middle MDD frontal gyrusc Insulac Putamenc,e Benevolence: High 4Low Risk Control4 Inferior frontal MDD gyrusc Cheating: High 4Low Risk MDD Superior occipital cortexc MDD 4 Lingual gyrus/ Control Calcarine sulcusc Postcentral/Supramarginal gyrusc Supplementary motor areac Postcentral/Supramarginal gyrusc Cheating: Low4High Risk Control Insulac

47

Benevolent (Low – High Risk) 4Cheating Control Inferior frontal 47 gyrus/Insulac Cheating (Low – High Risk) 4 Benevolent Control Insulae/Putamenc

Control4 MDD

Insulae/ Putamenc,e

33 36  18 3

12 24 3 6 0

48

a

BA: Brodmann area. b KE: cluster size. c Cluster surviving whole-brain cluster-level correction. d Contrasts of a priori hypotheses. e Cluster surviving the SVC test.

MDD patients, showed significant left anterior insular activations prior to making cheating responses under low versus high risk. We generated two composite contrasts corresponding to two opposite task response patterns: (i) making more benevolent choices under low risk and/or more cheating choices under high risk (i.e. Benevolent (Low – High Risk)4Cheating (Low – High Risk), Table 2) and (ii) making more cheating choices under low risk and/ or more benevolent choices under high risk (i.e. Cheating (Low – High Risk)4Benevolent (Low – High Risk), Table 2). The second

105

pattern is more risk-adaptive and is associated with a greater mean expected reward value than the first. The results showed that the first response pattern elicited activations in a region of left posterior IFG extending into the insula among control participants, whereas the second response pattern was associated with significant activations in the left anterior insula extending into the dorsolateral and dorsomedial anterior putamen among the control participants only, which activities were also significantly greater than those among the MDD patients (Fig. 3c). For the second response pattern, a region of the left anterior insula (local maxima¼  39, 6,  9, t¼5.38, cluster size¼16, surviving SVC test) showed activities that were positively modulated by cheating-related RDS among control participants only (Supplementary Fig. S1). 3.2.2. The influence of BDI-II-C scores BDI-II-C scores had a positive influence on left middle frontal gyrus activities prior to making low-risk cheating versus benevolent choices among the MDD patients (local maxima ¼  51, 30, 30, t ¼5.23, cluster size ¼16). BDI-II-C scores also had positive influence on BOLD signals in the left IFG (local maxima¼  51, 24, 9, t¼5.76, cluster size ¼16) elicited by the more risk-adaptive pattern of making low-risk cheating choices and/or high-risk benevolent choices (Fig. 4). 3.2.3. VBM analysis Whole-brain VBM analysis revealed no significant clusters that survived family-wise-error-rate (FWE) correction. Also, MDD patients and controls did not differ in volumes within key regions of activations, including the putamen, insula and IFG (ps 40.1). 3.2.4. Linking behavioral and neural responses We further tested whether participants' propensity for making cheating versus benevolent choices under low risk could be predicted by their BOLD amplitudes within the left dorsal putamen/ anterior insula region (local maxima ¼ 33, 6, 6, Fig. 5a) that were more positively activated in controls than in MDD patients, as described above. Percentage-signal-changes were extracted and correlated with participants' task choices under high and low risk levels (Fig. 5c, d). As expected, control participants showed greater PSCs elicited by making cheating versus benevolent choices under low risk (t (27) ¼4.049, po 0.05), and by making benevolent versus cheating choices under high risk, compared with the patients (t (27) ¼2.114, po 0.05) (Fig. 5b). Under low risk, patients' mean PSCs prior to making cheating versus benevolent choices correlated positively with their proportions of cheating choices (r (14) ¼0.62, p o0.05) (Fig. 5c). Interestingly, the same analysis on controls revealed a marginal negative correlation (r (15) ¼ 0.507, p ¼0.054) (Fig. 5d). 3.2.5. PPI analysis This was carried out based on a 6-mm-sphere VOI mask centered on the local maxima of the left dorsal putamen/anterior insula activation cluster ( 33, 6, 6). We specifically tested for brain areas whose activities correlated with those of the seed region differentially when participants made cheating versus benevolent choices under low risk. Among the controls, a small region of the medial right IFG (local maxima ¼36, 18, 18, cluster size ¼ 5) showed activities that correlated more positively with those of the seed region when participants made low-risk cheating relative to benevolent choices. No such pattern was observed among MDD patients. Direct comparison between groups revealed a region of the right frontal pole (local maxima¼ 30, 45, 21, cluster size ¼12) whose activities correlated with those of the seed region more positively among MDD patients relative to controls.

106

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

Fig. 3. Activations elicited by (a) cheating compared with benevolent choices under high risk among MDD patients in bilateral inferior frontal gyrus (BA 47) (b) cheating compared with benevolent choices under low risk among control participants relative to patients in left superior/middle frontal gyrus (c) the combined events of cheating relative to benevolent choice under low risk and benevolent relative to cheating choice under high risk among control participants relative to patients in left anterior insula/ dorsal putamen, rendered onto standard brain templates (ch2bet). Signals were thresholded at p o 0.001 at peak level and p o 0.01 at cluster level (minimal cluster size ¼ 15). Regions of a priori interest (putamen, insula, inferior frontal gyrus) were further assessed with small volume correction test with a threshold of p o 0.05 at cluster level. Montreal Neurological Institute (MNI) coordinates are provided, as well as a color scale indicating t statistics.

4. Discussion During making low-risk cheating versus benevolent choices, relative to the controls, MDD patients exhibited weaker BOLD signals in executive networks of the DLPFC, and in the anterior

insula and dorsal putamen implicated in value- and risk-based instrumental behaviors. We also obtained limited evidence that MDD patients showed abnormal IFG activities when making highrisk cheating choices. Such functional deficiencies in MDD patients could not be attributed to demographic factors, difference in IQ or

Fig. 4. MDD patients' scores on the Chinese version of the Beck Depression Inventory (BDI-II-C)51 showed a positive association with their BOLD signals elicited by the combined events of making cheating relative to benevolent choices under low risk and benevolent relative to cheating choices under high risk within a region of the left inferior frontal gyrus. (a) Activation cluster involving a region of the left IFG. (b) The significantly positive correlation between MDD patients' scores on BDI-II-C and their percentage-signal-change (PSC) values. LC: low-risk cheating choice. LB: low-risk benevolent choice. HC: high-risk cheating choice. HB: high-risk benevolent choice.

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

107

Fig. 5. BOLD percentage-signal-changes (PSC) elicited by making benevolent and cheating choices under high and low risk conditions, extracted from a region of interest derived from the group-level analysis on the group  choice  risk interactive effect encompassing the left dorsal putamen and anterior insula (local maxima¼  33, 6, 6), among MDD patients and controls, and their association with participants' task choices. (a) The region of interest. (b) PSCs exhibited by MDD patients and controls during making benevolent and cheating choices under high and low risk conditions. Error bars indicate one standard error of the mean. n indicates significant group difference (po 0.05). Note that both the PSCs of the low-risk cheating versus benevolent contrast and those of the high-risk benevolent versus cheating contrast for control participants were more positive than those of the patients (ps o0.05). (c) MDD patients' PSCs of the low-risk cheating versus benevolent contrast showed a positive correlation with their proportions of cheating choices under low risk. (d) Control participants' PSCs of the low-risk cheating versus benevolent contrast showed a marginal negative correlation with their proportions of cheating choices under low risk.

anatomical volumetric abnormalities. Importantly, we obtained novel direct evidence that MDD patients' deficient activities in the left dorsal putamen/anterior insula circuitry predicted their reduced behavioral tendency in switching to cheating responses when the risk was low. Our findings provide new theoretical insights into the neural mechanisms of risky decision-making processes among MDD individuals in social contexts, implicating possible functional deficiencies in the lateral PFC-striatal/limbic networks that are critical for risk-adaptive social responses. Such findings advance our understanding of, and promote intervention techniques targeted at, the social impairments of MDD individuals. Replicating previous findings (Zhang et al., 2012), we found that, while both MDD patients and controls made predominantly benevolent choices when the detection risk was high, control participants switched to making cheating choices when the risk was low. This transition was less apparent among the patients. By including risk in our task, we demonstrated that healthy participants took risk level into consideration when deciding whether to comply with rigorous requests from another, whereas MDD patients tended to exhibit invariable compliance behaviors regardless of risk level. This finding is consistent with behavioral models of depression which posit that MDD is characterized by reduced capacity in directing actions for generating intra- and inter-personal rewards in social contexts (Libet and Lewinsohn, 1973; Rehm, 1977), which is at least partly accounted for by the patient’s blunted sensitivity to rewards, heightened reactivity to punishment and altered risk processing (Mogg et al., 1995; Pizzagalli et al., 2008; Smoski et al., 2008). During making cheating versus benevolent choices under high risk, MDD patients showed significant recruitment of the bilateral IFG and posterior insula, implicating elevated visceral and emotive

responses to potential punishment (Naqvi and Bechara, 2009), as well as increased efforts in regulating the negative affect that arose from anticipated punishment and from the guilty feelings and emphatic distress elicited by cheating (Carr et al., 2003; Elliott et al., 2011; Grecucci et al., 2013). This finding is also consistent with past evidence indicating abnormal IFG activities during making trust decisions among patients with affective disorders (Sripada et al., 2009). Nevertheless, this result needs to be treated cautiously given the absence of significant between-group difference. MDD patients also exhibited activities in the middle and inferior frontal gyri that were more positively modulated by cheating-related RDS than those among the control participants. We speculate that individual difference in emotion regulatory functioning might determine the extent of cheating particularly among the depressed individuals, who typically experience greater difficulties in controlling and regulating negative emotions (Thayer et al., 2003; Gotlib et al., 2005). In contrast, control participants showed significant caudate activation, possibly reflecting a focus on the instrumental aspect of cheating and the anticipation of (negative) consequences based on action-outcome learning (Delgado et al., 2005; Haruno and Kawato, 2006; King-Casas et al., 2005). Under low risk when making cheating versus benevolent choices, the controls showed greater activations in the right posterior insula, left dorsal putamen and DLPFC than the patients, possibly reflecting their efforts in evaluating the value and risk associated with different response options in the low-risk context and acting accordingly (Clark et al., 2008; Haruno and Kawato, 2006). Making cheating rather than benevolent choices under low risk might place great cognitive demand, due to the conflict arisen from breaking a social promise for the sake of self-interest (Baugmgartner et al., 2009), the anticipation of potential

108

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

punishment (albeit low-likelihood), and the deviation from the requested repayment and deciding the alternative response, which may explain the greater DLPFC recruitment (Bereczkei et al., 2013). Consistent with this, DLPFC activities among control participants also predicted individual variations on cheating related RDS, possibly reflecting the recruitment of cognitive resource necessary for balancing the incentive of maximizing self-gain (i.e. repaying the lowest possible amount) and the cost of violating social norm to the greatest extent (Ruff et al., 2013). Thus, deficiencies in activities within the DLPFC, as well as in the associated networks implicated in risk signaling and instrumental actions, may have contributed to MDD patients' reduced tendency in switching from benevolent to cheating responses when the risk level decreased. Such interpretation is also consistent with previous research associating affective disorders with reduced DLPFC activities during making trust decisions (Sripada et al., 2009). When exposed to different risk levels, the controls, relative to MDD patients, showed stronger right IFG activities when making benevolent choices under high versus low risk, suggesting that control participants may have had greater capacity in selectively directing their attention to the more conservative and adaptive benevolent choice, while inhibiting the alternative response that carried a high chance of punishment (Hampshire et al., 2009; Vidal et al., 2012). In contrast, cheating committed by control participants when the likelihood of punishment was lower was associated with activation in the left anterior insula, which delivers interoceptive signals indicative of risk (Naqvi and Bechara, 2009). These findings, although not explicitly hypothesized, are consistent with the interpretation that, the IFG-insula circuitry critical for risk-based action implementation (Carr et al., 2003) differentiated between social contexts associated with different probabilities of reward and punishment among the control participants, enabling them to make flexible, contextual adjustments to their responses. The functioning of such circuitry may be compromised in people with MDD, resulting in their relatively invariable behavioral patterns across risk levels. Making value-based, risk-adaptive choices in the current TG task involved giving benevolent responses under high risk and cheating responses under low risk, which behavioral pattern was associated with significant and greater activations in the left dorsal putamen/anterior insula among the control participants, compared with MDD patients. Left insular activity also predicted cheating-related RDS among control participants. These findings are generally consistent with previous evidence indicating the involvement of anterior insula in making socially defiant or uncooperative decisions (King-Casas et al., 2008; Rilling et al., 2008; Sanfey et al., 2003). We also observed a significant positive correlation between neural signals in this brain region and the proportion of low-risk cheating choices among MDD patients, confirming that the dorsal putamen is a primal site for implementing instrumental actions, based on information about risk and value from the adjacent anterior insula (Naqvi and Bechara, 2009). The deficient striatal/limbic activities of MDD patients during risky decision-making might be related to their altered sensitivity to rewards and punishments, which are in turn associated with anhedonia (Pizzagalli et al., 2008). MDD individuals' reduced anterior insular activities might be associated with their altered risk processing (Smoski et al., 2008), as well as with impaired capacity in adjusting responses as trustee during dynamic social reciprocations (King-Casas et al., 2008). The same analysis also revealed a marginal negative correlation among the controls. It might be that control participants exhibiting very high levels of dorsal putamen activations had already formed context-independent, habitual benevolent responses (Balleine et al., 2007). The BDI-II-C scores of MDD patients positively correlated with left DLPFC activities when making cheating versus benevolent

choices under low risk. Given previous findings indicating a negative relationship between MDD severity and executive functioning (Cella et al., 2010), this observation is consistent with our hypothesis that MDD is associated with reduced prefrontal functioning efficiency in executive control during making the cognitively more demanding low-risk cheating decisions (Harvey et al., 2005; Wagner et al., 2006). MDD severity also positively correlated with left IFG activities elicited by the pattern of low-risk cheating/ high-risk benevolent responses. This result could potentially indicate greater efforts in inhibitory control and emotion regulation for more severe cases of MDD when making risk-adaptive cheating and benevolence decisions (Grecucci et al., 2013; Froeliger et al., 2012). Psycho-physiological interaction analyzes revealed limited evidence for stronger positive functional connectivity between the right IFG and the dorsal putamen/anterior insula circuitries during making cheating relative to benevolent choices under low risk among the control participants but not MDD patients, which finding, albeit tentative, is nevertheless consistent with our proposal that risk-adaptive decision-making and responses depend on the integrated IFG-striatal/limbic functional networks. Relative to controls, MDD patients exhibited elevated connectivity between the right frontal pole and dorsal putamen/anterior insula. The frontopolar areas are previously implicated for internal focus and self-reflection (Moayedi et al., 2014), as well as for the moral emotion of guilt (Takahashi et al., 2004; Kedia et al., 2008). It might be that MDD patients engaged in greater self-focused negative ruminative processes centered on guilt when deciding to cheat the partner (Watkins and Moulds, 2005), which reduced their tendency of engaging in such behavior even when the risk of being punished was low. Taking all findings together, we propose that, during social risky decision-making, brain signals that code for value and risk are generated in the anterior insula and dorsal striatal regions. Such signals are delivered to the dorso- and ventro-lateral prefrontal areas that perform executive functions necessary for monitoring and evaluating the value/risk analysis, as well as for resolving conflicts that arise from incompatible goals, such as breaking social norm for self-gain. Action-related signals are then generated in the IFG-striatal/limbic circuitries for implementing instrumental behaviors, as well as for inhibiting the interference of negative affects that may arise from emphatic distress, guilty feelings or anticipated punishment. The functioning of these networks are critical for making flexible risk-adaptive decisions in social contexts based on balancing intra-personal monetary and interpersonal (pro)social rewards, and additionally predict individual differences in the magnitude of cheating others for maximizing self-gain. Such networks are compromised in MDD individuals, resulting in their reduced tendency in adjusting responses flexibly with the ever-changing risk conditions, as well as in exhibiting behavioral patterns that promote self-gain. Our proposals are also consistent with existing evidence on the effects of serotonergic and glutamatergic challenges on both lateral prefrontal and limbic-striatal network functioning, and social and risky decision-making processes of healthy individuals and patients with affective disorders (Bearer et al., 2009; Macoveanu et al., 2013a,b; Sachdev and Malhi, 2005). On the other hand, MDD patients' social decision-making can be considered as more prosocial and altruistic than non-depressed controls (e.g. Fujiwara, 2009). However, there might be greater mismatch between the negative emotive response and the subsequent compliance action among MDD patients compared with controls (Harle et al., 2010), with the consequence that the negative emotions might be strengthened and accumulated over time, accentuating the MDD symptom (Segrin, 2000). Also, one recent study suggested that MDD individuals' prosocial decision-making

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

might serve to avoid social rejection (Destoop et al., 2012), highlighting a focus on the negative aspects of social interactions characteristic of MDD. Future research could extend the current findings by comparing MDD individuals and controls on tasks that involve manipulation of social moral emotions and risk, such that risk-taking affects the gain of the partner but not that of oneself, combined with measures that specifically test for value/risk processing, executive control and/or emotion regulation. The present study has several limitations. First, our samples are relatively small and contained females only, which limit the generalizability of the findings to the general MDD and male populations. For example, previous evidence suggests that females might be more altruistic than men when the cost to oneself is high (Andreoni and Vesterlund, 2001). Future research may replicate and extend the current findings in a larger sample comprised of males or both genders. Second, the current data set does not allow for testing of the influence of previous outcome histories and the expected values of choices on the behavioral and neural measures. Future research employing larger participant samples and extended task paradigms might resolve the issue. Third, for ethical reasons our patients were continuing taking antidepressant/anxiolytic medications at the time of study. Future research involving remitted MDD patients and/or preclinical, medication-free samples would help minimize the confounding effect of medication. Fourth, both strict compliance (returning as exactly requested) and over-generous behaviors (returning more than requested) were grouped together. Future research should test the commonalities and distinctions between these behaviors. Finally, future research could additionally include a baseline risk-free social decisionmaking task that allows separation of individuals' risk-taking and prosocial tendencies.

5. Conclusion Making flexible and risk-adaptive social decisions based on the balancing of self-gain and other's welfare, which relies on the functioning of integrated lateral PFC and striatal/limbic networks, is a critical component of social functioning. We show that MDD individuals exhibited reduced behavioral tendency to engage in cheating behaviors even when the risk of being punished was low, coupled with functional deficiencies in neural networks implicated in executive control, emotion regulation and instrumental behaviors based on risk and value. Our findings shed light on a cognitive and affective model of social decision-making in both healthy populations and MDD patients. They also provide insight for interventions for MDD with therapies that enhance the functioning of lateral prefrontal-striatal/ limbic circuitries (e.g. Dichter et al., 2009; Schaefer et al., 2006) and/or improve the patients’ psychosocial and emotion regulatory capacities (e.g. Young et al., 2006; Kovacs et al., 2006).

Conflicts of interests None declared.

Disclosure Author R.S. was involved substantially in data analysis and manuscript preparation and revision; Author H.-J. Zhang was involved substantially in study design, data acquisition and manuscript revision. Author TMC Lee was involved substantially in study design, manuscript preparation and revision. All authors have given final approval for the current manuscript.

109

Acknowledgements This work was supported by the Research Grants Council Humanities and Social Sciences Prestigious Fellowship Scheme (HSSPFS) (Ref: HKU703-HSS-13). The funding agency has no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Appendix A. Supplementary Information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.neuropsychologia. 2014.12.009.

References American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th ed. American Psychiatric Press, Washington, DC. Andreoni, J., Vesterlund, L., 2001. Which is the fair sex? Gender differences in Altruism. Q. J. Econ. 116 (1), 293–312. Balleine, B.W., Delgado, M.R., Hikosaka, O., 2007. The role of the dorsal striatum in reward and decision-making. J. Neurosci. 27 (31), 8161–8165. Baugmgartner, T., Fischbacher, U., Feierabend, A., Lutz, K., Fehr, E., 2009. The neural circuitry of a broken promise. Neuron 64, 756–770. Bearer, E.L., Zhang, X., Janvelyan, D., Boulat, B., Jacobs, R.E., 2009. Reward circuitry is perturbed in the absence of the serotonin transporter. NeuroImage 46 (4), 1091–1104. Beats, B.C., Sahakian, B.J., Levy, R., 1996. Cognitive performance in tests sensitive to frontal lobe dysfunction in the elderly depressed. Psychol. Med. 26, 591–603. Bereczkei, T., Deak, A., Papp, P., Perlaki, G., Orsi, G., 2013. Neural correlates of Machiavellian strategies in a social dilemma task. Brain Cognit. 82, 108–116. Bjork, J.M., Smith, A.R., Hommer, D.W., 2008. Striatal sensitivity to reward deliveries and omissions in substance dependent patients. Neuroimage 42, 1609–1621. Brugha, T.S., Bebbington, T.E., Stretch, D.D., MacCarthy, B., Wykes, T., 1997. Predicting the short-term outcome of first episodes and recurrences of clinical depression: a prospective study of life events, difficulties, and social support networks. J. Clin. Psychol. 58, 298–306. Byrnes, J.P., Miller, D.C., Schafer, W.D., 1999. Gender differences in risk-taking: a metaanalysis. Psychol. Bull. 125 (3), 367–383. Byrne, B.M., Stewart, S.M., Lee, P.W.H., 2004. Validating the beck depression inventory-II for Hong Kong community adolescents. Int. J. Test. 4 (3), 199–216. Carr, L., Iacoboni, M., Dubeau, M.-C., Mazziotta, J.C., Lenzi, G.L., 2003. Neural mechanisms of empathy in humans: a relay from neural systems for imitation to limbic areas. Proc. Natl. Acad. Sci. USA 100 (9), 5497–5502. Cella, M., Dymond, S., Cooper, A., 2010. Impaired flexible decision-making in major depressive disorders. J. Affect. Disord. 124 (1–2), 207–210. 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, 1311–1322. Corwin, J., Peselow, E., Feenan, K., Rotrosen, J., Fieve, R., 1990. Disorders of decision in affective disease: an effect of beta-adrenergic dysfunction? Biol. Psychiatry 27 (8), 813–833. de Quervain, D.J., Fischbacher, U., Treyer, V., Schellhammer, M., Schnyder, U., Buck, A., et al., 2004. The neural basis of altruistic punishment. Science 305 (5688), 1254–1258. Delgado, M.R., Frank, R.H., Phelps, E.A., 2005. Perceptions of moral character modulate the neural systems of reward during the trust game. Nat. Neurosci. 8, 1611–1618. Destoop, M., Schrijvers, D., De Grave, C., Sabbe, B., De Bruijn, E.R.A., 2012. Better to give than to take? Interactive social decision-making in severe major depressive disorder. J. Affect. Disord. 137, 98–105. Dichter, G.S., Felder, J.N., Petty, C., Bizzell, J., Ernst, M., Smoski, M.J., 2009. The effects of psychotherapy on neural responses to rewards in major depression. Biol. Psychiatry 66 (9), 886–897. Dunkin, J.J., Leuchter, A.F., Cook, I.A., Kasl-Godley, J.E., Abrams, M., Rosenberg-Thompson, S., 2000. Executive dysfunction predicts nonresponse to fluoxetine in major depression. J. Affect. Disord. 60, 13–23. Elliott, R., Zahn, R., Deakin, J.F., Anderson, I.M., 2011. Affective cognition and its disruption in mood disorders. Neuropsychopharmacology 36 (1), 153–182. Fales, C.L., Barch, D.M., Rundle, M.M., Mintun, M.A., Snyder, A.Z., Cohen, J.D., et al., 2008. Altered emotional interference processing in affective and cognitive-control brain circuitry in major depression. Biol. Psychiatry 63 (4), 377–384. Froeliger, B., Modlin, L.A., Kozink, R.V., Wang, L., McClernon, F.J., 2012. Smoking abstinence and depressive symptoms modulate the executive control system during emotional information processing. Addict. Biol. 17 (3), 668–679. Fujiwara, T., 2009. Is altruistic behavior associated with major depression onset? PLOS One 4 (2), e4557. Gotlib, I.H., Hamilton, J.P., Cooney, R.E., Singh, M.K., Henry, M.L., Joorman, J., 2010. Neural processing of reward and loss in girls at risk for major depression. Arch. Gen. Psychiatry 67 (4), 380–387. Gotlib, I.H., Sivers, H., Gabrieli, J.D.E., Whitfield-Gabrieli, S., Goldin, P., Minor, K.L., et al., 2005. Subgenual anterior cingulate activation to valenced emotional stimuli in major depression. Neuroreport 16 (16), 1731–1734. Grecucci, A., Giorgetta, C., Bonini, N., Sanfey, A.G., 2013. Reappraisal social emotions: The role of inferior frontal gyrus, temporo-parietal junction and insula in interpersonal emotion regulation. Front. Hum. Neurosci. 7, 523.

110

R. Shao et al. / Neuropsychologia 67 (2015) 100–110

Gross, J.J., John, O.P., 2003. Individual differences in two emotion regulation processes: implications for affect, relationships, and well-being. J. Personal. Soc. Psychol. 85 (2), 348–362. Hamid, P.N., Cheng, S.-T., 1996. The development and validation of an index of emotional disposition and mood state: the Chinese affect scale. Edu. Psychol. Meas. 56 (6), 995–1014. Hampshire, A., Thompson, R., Duncan, J., Owen, A.M., 2009. Selective tuning of the right inferior frontal gyrus during target detection. Cognit. Affect. Behav. Neurosci. 9 (1), 103–112. Harle, K.M., Allen, J.J.B., Sanfey, A.G., 2010. The impact of depression on social economic decision-making. J. Abnorm. Psychol. 119 (2), 440–446. Haruno, M., Kawato, M., 2006. Different neural correlates of reward expectation and reward expectation error in the putamen and caudate nucleus during stimulus-action-reward association learning. J. Neurophysiol. 95, 948–959. Harvey, P.O., Fossati, P., Pochon, J.-B., Levy, R., Lebastard, G., Lehericy, S., et al., 2005. Cognitive control and brain resources in major depression: an fMRI study using the n-back task. Neuroimage 26, 860–869. Harvey, P.O., Le Bastard, G., Pochon, J.B., Levy, R., Allilaire, J.F., Dubois, B., et al., 2004. Executive functions and updating of working memory in unipolar depression. J. Psychiatr. Res. 38 (6), 567–576. Kedia, G., Berthoz, S., Wessa, M., Hilton, D., Martinot, J.L., 2008. An agent harms a victim: a functional magnetic resonance imaging study on specific moral emotions. J. Cognit. Neurosci. 20 (10), 1788–1798. Kessler, R.C., Berglund, P., Demler, O., Jin, R., Koretz, D., Merikangas, K.R., et al., 2003. The epidemiology of major depressive disorder: results from the National Comorbidity Survey Replication (NCS-R). J. Am. Med. Assoc. 289 (23), 3095–3105. King-Casas, B., Sharp, C., Lomax-Bream, L., Lohrenz, T., Fonagy, P., Montague, P.R., 2008. The rupture and repair of cooperation in borderline personality disorder. Science 321, 806–810. King-Casas, B., Tomlin, D., Anen, C., Camerer, C.F., Quartz, S.R., Montague, P.R., 2005. Getting to know you: reputation and trust in a two-person economic exchange. Science 308, 78–83. Kovacs, M., Sherrill, J., George, C.J., Pollock, M., Tumuluru, R.V., Ho, V., 2006. Contextual emotion-regulation therapy for childhood depression: description and pilot testing of a new intervention. J. Am. Acad. Child Adolesc. Psychiatry 45 (8), 892–903. Krueger, F., McCabe, K., Moll, J., Kriegeskorte, N., Zahn, R., Strenziok, M., et al., 2007. Neural correlates of trust. Proc. Natl. Acad. Sci. 104 (50), 20084–20089. Kuhnen, C.M., Knutson, B., 2005. The neural basis of financial risk taking. Neuron 47, 763–770. Landro, N.I., Stiles, T.C., Sletvold, H., 2001. Neuropsychological function in nonpsychotic unipolar major depression. Neuropsychiatry Neuropsychol. Behav. Neurol. 14, 233–240. Lerchner, A., La Camera, G., Richmond, B., 2007. Knowing without doing. Nat. Neurosci. 10 (1), 15–17. Libet, J.M., Lewinsohn, P.M., 1973. Concept of social skill with special reference to the behavior of depressed persons. J Consult. Clin. Psychol 40 (2), 304–312. Macoveanu, J., Hornboll, B., Elliott, R., Erritzoe, D., Paulson, O.B., Siebner, H., et al., 2013a. Serotonin 2A receptors, citalopram and tryptophan-depletion: a multimodel imaging studies of their interactions during response inhibition. Neuropsychopharmacology 38, 996–1005. Macoveanu, J., Rowe, J.B., Hornboll, B., Elliott, R., Paulson, O.B., Knudsen, G.M., et al., 2013b. Playing it safe but losing anyway—serotoninergic signaling of negative outcomes in dorsomedial prefrontal cortex in the context of risk-aversion. Eur. Neuropsychopharmacol. 23 (8), 919–930. Mayberg, H.S., 1997. Limbic-cortical dysregulation: a proposed model of depression. J. Neuropsychiatry Clin. Neurosci. 9, 471–481. McClure, S.M., Berns, G.S., Montague, P.R., 2003. Temporal prediction errors in a passive learning task activate human striatum. Neuron 38, 339–346. Miller, E.K., Cohen, J.D., 2001. An integrative theory of prefrontal cortex function. Annu. Rev. Neurosci. 24, 167–202. Moayedi, M., Salomons, T., Dunlop, K.A., Downar, J., Davis, K.D., 2014. Connectivity-based parcellation of the human frontal polar cortex. Brain Struct. Funct. http://dx.doi.org/ 10.1007/s00429-014-0809-6 (Epub ahead of print). Mogg, K., Bradley, B.P., Williams, R., 1995. Attentional bias in anxiety and depression: The role of awareness. Br. J. Clin. Psychol. 34 (1), 17–36. Naqvi, N.H., Bechara, A., 2009. The hidden island of addiction: The insula. Trends Neurosci 32 (1), 56–67. Nezlek, J.B., Imbrie, M., Shean, G.D., 1994. Depression and everyday social interaction. J. Pers. Soc. Psychol. 67, 1101–1111. O’Connor, L.E., Berry, J.W., Lewis, T., Mulherin, K., Crisostomo, P.S., 2007. Empathy and depression: the moral system on overdrive. In: Farrow, T.F.D., Woodruff, P.W.R. (Eds.), Empathy in Mental Illnesses. Cambridge University Press, Cambridge, pp. 49–75. Ochsner, K.N., Gross, J.J., 2005. The cognitive control of emotion. Trends Cognit. Sci. 9 (5), 242–249. Okada, G., Okamoto, Y., Morinobu, S., Yamawaki, S., Yokota, N., 2003. Attenuated left prefrontal activation during a verbal fluency task in patients with depression. Neuropsychobiology 47 (1), 21–26. Paulus, M.P., Rogalsky, C., Simmons, A., Feinstein, J.S., Stein, M.B., 2003. Increased activation in the right insula during risk-taking decision making is related to harm avoidance and neuroticism. Neuroimage 19, 1439–1448.

Pizzagalli, D.A., Iosifescu, D., Hallet, L.A., Ratner, K.G., Fava, M., 2008. Reduced hedonic capacity in major depressive disorder: Evidence from a probabilistic reward task. J. Psychiatr. Res. 43 (1), 76–87. Postuma, R.B., Dagher, A., 2006. Basal ganglia functional connectivity based on a metaanalysis of 126 positron emission topography and functional magnetic resonanceimaging publications. Cereb. Cortex 16 (10), 1508–1521. Pulcu, E., Zahn, R., Elliott, R., 2013. The role of self-blaming moral emotions in major depression and their impact on socio-economical decision making. Front. Psychol. 4, 310. Raven, J., 2000. The Raven's progressive matrices: change and stability over culture and time. Cognit. Psychol. 41, 1–48. Rehm, L.P., 1977. A self-control model of depression. Behav. Ther. 8 (5), 787–804. Rilling, J.K., Goldsmith, D.R., Glenn, A.L., Jairam, M.R., Elfenbein, H.A., Dagenais, J.E., et al., 2008. The neural correlates of the affective response to unreciprocated cooperation. Neuropsychologia 46, 1256–1266. Rilling, J.K., Sanfey, A.G., 2011. The neuroscience of social decision-making. Annu. Rev. Psychol. 62, 23–48. Ruff, C.C., Ugazio, U., Fehr, E., 2013. Changing social norm compliance with noninvasive brain stimulation. Science 342, 482–484. Sachdev, P.S., Malhi, G.S., 2005. Obsessive–compulsive behaviour: a disorder of decisionmaking. Aust. N. Z. J. Psychiatry 39 (9), 757–763. Sanfey, A.G., Rilling, J.K., Aronson, J.A., Nystrom, L.E., Cohen, J.D., 2003. The neural basis of economic decision-making in the Ultimatum game. Science 300, 1755–1758. Schaefer, H.S., Putnam, K.M., Benca, R.M., Davidson, R.J., 2006. Event-related functional magnetic resonance imaging measures of neural activity to positive social stimuli in pre- and post-treatment depression. Biol. Psychiatry 60 (9), 974–986. Segrin, C., 2000. Social skills deficits associated with depression. Clin. Psychol. Rev. 20 (3), 379–403. Shao, R., Lee, T.M.C., 2014. Aging and risk taking: toward an integration of cognitive, emotional, and neurobiological perspectives. Neurosci. Neuroecon. 3, 47–62. Siegle, G.J., Thompson, W., Carter, C.S., Steinhauer, S.R., Thase, M.E., 2007. Increased amygdala and decreased dorsolateral prefrontal BOLD responses in unipolar depression: related and independent features. Biol. Psychiatry 61, 198–209. Smoski, M.J., Felder, J., Bizzell, J., Green, S.R., Ernst, M., Lynch, T.R., et al., 2009. FMRI of alterations in reward selection, anticipation, and feedback in major depressive disorder. J. Affect. Disord. 118 (1–3), 69–78. 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. J. Behav. Ther. Exp. Psychiatry 39 (4), 567–576. Spencer, S.A., Farrow, T.F., Herford, A.E., Wilkinson, I.D., Zheng, Y., Woodruff, P.W., 2001. Behavioral and functional anatomical correlates of deception in humans. Neuroreport 12, 2849–2853. Sripada, C.S., Angstadt, M., Banks, S., Nathan, P.J., Liberzon, I., Phan, K.L., 2009. Functional neuroimaging of mentalizing during the trust game in social anxiety disorder. Neuroreport 20 (11), 984–989. Takahashi, H., Yahata, N., Koeda, M., Matsuda, T., Asai, K., Okubo, Y., 2004. Brain activation associated with evaluative processes of guilt and embarrassment: an fMRI study. NeuroImage 23, 967–974. Thayer, J.F., Rossy, L.A., Ruiz-Padial, E., Johnsen, B.H., 2003. Gender differences in the relationship between emotional regulation and depressive symptoms. Cognit. Ther. Res. 27 (3), 349–364. Vidal, J., Mills, T., Pang, E.W., Taylor, M.J., 2012. Response inhibition in adults and teenagers: spatiotemporal differences in the prefrontal cortex. Brain Cognit. 79, 49–59. Von Gunten, A., Herrmann, F.R., Elliott, R., Duc, R., 2011. Abnormal sensitivity to negative feedback in late-life depression. Psychiatry Clin. Neurosci. 65 (4), 333–340. Voon, V., Pessiglione, M., Brezing, C., Gallea, C., Fernandez, H.H., Dolan, R.J., et al., 2010. Mechanisms underlying dopamine-mediated reward bias in compulsive behaviors. Neuron 65, 135–142. Wagner, G., Sinsel, E., Sobanski, T., Kohler, S., Marinou, V., Mentzel, H.J., et al., 2006. Cortical inefficiency in patients with unipolar depression: an event-related fMRI study with the stroop task. Biol. Psychiatry 59 (10), 958–965. Wardle, M.C., Fitzgerald, D.A., Angstadt, M., Sripada, C.S., McCable, K., Phan, K.L., 2013. The caudate signals bad reputation during trust decisions. PLOS One 8 (6), e68884. Watkins, E., Moulds, M., 2005. Distinct modes of ruminative self-focus: Impact of abstract versus concrete rumination on problem solving in depression. Emotion 5 (3), 319–328. Yamasaki, H., LaBar, K.S., McCarthy, G., 2002. Dissociable prefrontal brain systems for attention and emotion. Proc. Natl. Acad. Sci. USA 99 (17), 11447–11451. Yan, D., Yu, G., 2009. Assessing two emotion regulation processes in Chinese adolescents. In: IEEE Procedings of the Fifth International Conference on Natural Computation, ICNC’09, vol. 5, pp. 112–116). Yao, S., Yang, H., Zhu, X., 2007. An examination of the psychometric properties of the Chinese version of the Barratt impulsiveness Scale, 11th version in a sample of Chinese adolescents. Percept. Mot. Skills 104, 1169–1182. Young, J.F., Mufson, L., Davies, M., 2006. Efficacy of interpersonal psychotherapy-adolescents skills training: an indicated preventive intervention for depression. J. Child Psychol. Psychiatry 47 (12), 1254–1262. Zhang, H., Sun, D., Lee, T.M.C., 2012. Impaired social decision making in patients with major depressive disorder. Brain Behav. 2 (4), 415–423.