The neural bases underlying social risk perception in purchase decisions

NeuroImage 91 (2014) 120–128

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

The neural bases underlying social risk perception in purchase decisions Ryoichi Yokoyama a,b,c,⁎, Takayuki Nozawa d, Motoaki Sugiura a,e, Yukihito Yomogida b,g, Hikaru Takeuchi f, Yoritaka Akimoto a, Satoru Shibuya h, Ryuta Kawashima a,d,f a

Department of Functional Brain Imaging, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan Japan Society for the Promotion of Science, Tokyo, Japan University of California, Berkeley, Haas School of Business, CA, USA d Smart Ageing International Research Center, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan e International Research Institute of Disaster Science, Tohoku University, Sendai, Japan f Division of Developmental Cognitive Neuroscience, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan g Tamagawa University Brain Science Institute, Tokyo, Japan h Graduate School of Economics and Management, Tohoku University, Sendai, Japan b c

a r t i c l e

i n f o

Article history: Accepted 21 January 2014 Available online 26 January 2014 Keywords: Social risk Anterior insula Neuroeconomics Social decision making

a b s t r a c t Social considerations significantly influence daily purchase decisions, and the perception of social risk (i.e., the anticipated disapproval of others) is crucial in dissuading consumers from making purchases. However, the neural basis for consumers' perception of social risk remains undiscovered, and this novel study clarifies the relevant neural processes. A total of 26 volunteers were scanned while they evaluated purchase intention of products (purchase intention task) and their anticipation of others' disapproval for possessing a product (social risk task), using functional magnetic resonance imaging (fMRI). The fMRI data from the purchase intention task was used to identify the brain region associated with perception of social risk during purchase decision making by using subjective social risk ratings for a parametric modulation analysis. Furthermore, we aimed to explore if there was a difference between participants' purchase decisions and their explicit evaluations of social risk, with reference to the neural activity associated with social risk perception. For this, subjective social risk ratings were used for a parametric modulation analysis on fMRI data from the social risk task. Analysis of the purchase intention task revealed a significant positive correlation between ratings of social risk and activity in the anterior insula, an area of the brain that is known as part of the emotion-related network. Analysis of the social risk task revealed a significant positive correlation between ratings of social risk and activity in the temporal parietal junction and the medial prefrontal cortex, which are known as theory-of-mind regions. Our results suggest that the anterior insula processes consumers' social risk implicitly to prompt consumers not to buy socially unacceptable products, whereas ToM-related regions process such risk explicitly in considering the anticipated disapproval of others. These findings may prove helpful in understanding the mental processes involved in purchase decisions. © 2014 Elsevier Inc. All rights reserved.

Introduction Traditional economics has focused on how to maximize people's desire. According to the traditional view, people make economic decisions, such as purchase decision – the most fundamental economic decision in daily life – on the basis of their personal preferences. Neuroscience studies have revealed the neural circuits of personal preferences. For example, the dopamine (DA) network is believed to be critical for regulating personal preferences (Knutson et al., 2007). Abbreviations: fMRI, functional magnetic resonance imaging; ToM, theory of mind; mPFC, medial prefrontal cortex; TPJ, temporal parietal junction; MNI, Montreal Neurological Institute; RT, response time; OFC, orbitofrontal cortex; BOLD, blood oxygenation level-dependent; S, social risk; PI, purchase intention. ⁎ Corresponding author at: Department of Functional Brain Imaging, Institute of Development, Aging, and Cancer, Tohoku University, 4-1 Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan. Fax: +81 22 717 7988. E-mail address: [email protected] (R. Yokoyama). 1053-8119/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.neuroimage.2014.01.036

However, contrary to the traditional economics view, human beings are naturally able to inhibit personal preference in economic decisions (Henrich et al., 2001). This is how human society is able to maintain public order, social norms, and morals (Elster, 1989). For example, the process of purchase decision making is affected by subjective or social norms, i.e., consumers' perceptions of what other people approve of (Berns et al., 2010; Childers and Rao, 1992; Lascu and Zinkhan, 1999; Mason et al., 2009; Peter and Olson, 1996; Wänke, 2009). More specifically, fearing criticism from family or friends, consumers might stop buying a preferred product (Rook and Fisher, 1995), particularly products that might be considered controversial (e.g., those made of alligator skin) (Xu et al., 2004). These previous studies indicate that purchase decisions are not always regulated by personal preference (i.e., DA network activity) alone; instead, perceptions of social risk, which, in this study, is defined as the anticipated disapproval of others, influence purchase decisions independent of personal preference. However, no studies have identified

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

121

Fig. 1. Task. Subjects viewed the images of T-shirts and rated each of them from a 1–8 levels, from completely disagree (level 1) to completely agree (level 8). This rating was based on image brightness, purchase intention, social risk, and preferences. A statement corresponding to each session was displayed once before the scanning in each session.

the psychological and neural bases of social risk perceptions regarding purchase decision making. Therefore, in this study, we aimed to clarify the neural basis of social risk perception during purchase decisions. Neuroimaging studies have identified various brain structures that are involved in ensuring compliance with social and moral norms. These neural substrates are mainly located in two regions of the brain. The first comprises the theory of mind (ToM)-related regions, such as the medial prefrontal cortex (mPFC) and temporal parietal junction (TPJ) (Lieberman, 2007; Saxe and Kanwisher, 2003). The second comprises emotion-related regions, such as the anterior insula (Craig, 2009; Harenski and Hamann, 2006; Kurth et al., 2010; Montague and Lohrenz, 2007), amygdala (Adolphs et al., 1994, 1998; Buckholtz and Marois, 2012), and lateral orbitofrontal cortex (OFC) (Spitzer et al., 2007). We assume that, in ordinary cases, social risk is dominantly processed implicitly and emotionally in purchase decision making to make consumer behavior efficient. Otherwise, individuals would have to explicitly think about the opinions of others before any purchase by using a ToM-related neural process. Therefore, we expect that our social risk perception during purchase decisions will be related to social– emotional related regions of the brain (Kurth et al., 2010). To verify this, firstly, we will investigate the neural basis of social risk perception during purchase decision making, and secondly, we will focus on the differences in brain activity between explicit and implicit evaluations of social risk, i.e., purchase decision making. Taking these points together, we formulated the following two hypotheses: H1. Consumers' social risk perceptions in purchase decisions are processed in the social–emotional regions of the brain. H2. The neural basis of social risk perception during purchase decision making is different from that of the explicit evaluation of social risk.

studies (Haugtvedt et al., 2008; Peter and Olson, 1996; Robertson and Kassarjian, 1990). Our fMRI studies extended over four sequenced rating sessions: (1) product brightness rating, (2) purchase intention rating, (3) social risk rating, and (4) product preference rating. The order of sessions (3) and (4) was counterbalanced among subjects. The brightness judgment task functioned as a control to elicit neural activity (fMRI signals) related to motor and simple cognitive processes as a result of button pressing common to all subsequent experimental tasks. This task was performed before subjects were informed of the purchase intention, product preference, and social risk sessions, with the assurance that neural patterns would not be influenced by purchase deliberation. In addition, we conducted a purchase intention rating session before the social risk and preference sessions to assure that the neural process and ratings of purchase intention would not be biased by the social risk and preference ratings. Participants Observing the Declaration of Helsinki (1991), we obtained written informed consent from participants before the study. The Tohoku University School of Medicine Ethics Committee approved the study protocol. A total of 30 healthy, right-handed individuals (17 male, 13 female) participated in the fMRI experiment. Their mean age was 20.87 years (19–24 years). Handedness was evaluated using the Edinburgh Handedness Inventory (Oldfield, 1971). Four participants' data were excluded because of greater than two millimeter movements and response rates below 90% during the fMRI task. The mean age of the 26 remaining subjects was 20.92 years (19–24 years). All subjects had normal or corrected-to-normal vision and no history of neurological or psychiatric illness. Stimuli

Methods Overview Our fMRI experiment was governed by two criteria. First, the object stimulating a neural response should be a product that consumers can or do refuse to purchase because of social pressure. To satisfy the first condition, following Izuma and Adolphs (2013), we selected T-shirts as the stimulus because T-shirts can feature designs that trigger the perception of social risk. Second, the analyses should be performed on fMRI data acquired during the purchase intention rating to assure that observed brain activity relates specifically to purchase decisions. To satisfy the second condition, we recorded blood oxygenation leveldependent (BOLD) contrast fMRI data during subjects' rating of purchase intention for our main analysis. This common method to make subjects deliberate purchases is taken from consumer behavior

A total of 63 photos of T-shirts chosen from one online vendor (http://www.graniph.com/) were used as the stimuli to control for price, style, brand, and quality. We conducted a preliminary experiment to select T-shirts that elicited a range of social risk ratings to assure the collection of a sufficient number from each subgroup for statistical analysis. This preliminary experiment involved 10 healthy university students who did not participate in the fMRI experiment. Before the purchase intention session, we instructed the participants that all T-shirts were of identical price, brand, and quality to exclude these as confounding factors as preferences. Experimental tasks Participants completed a practice trial immediately before the fMRI sessions, after which they entered the MRI scanner. All the images of

122

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

Fig. 2. Experiment overview. The upper line explains the session and the kind of data acquired in each session. The bottom line explains fMRI analysis using the data.

the T-shirts were rear-projected onto a semi-lucent screen, which participants viewed via a mirror attached to the MRI scanner. All images were presented at the same viewing angle (b5°) against a gray background. Images appeared for 4 s, after which a small fixation cross appeared at the center of the screen (Fig. 1). This inter-stimulus interval lasted for 2–6 s before the next T-shirt appeared. Every eighth stimulus was interspersed with a null task (a prolonged inter-stimulus fixation period) of 4 s to acquire no-task baseline images. An inclusion of such baseline data enhanced the sensitivity of model testing during the analysis. A statement was presented, after which each image was displayed during each session (Fig. 1). A Likert-like scale ranging from one (completely disagree) to eight (completely agree) was presented under each image. Participants indicated their degree of agreement/disagreement with the statement using an MRI-compatible keypad attached to their hands. Fingers from left small to right small were assigned the ratings 1–8, respectively. The statements were intended to elicit participants' purchase intentions, social risk, and preference ratings. The four statements were as follows: (1) “This T-shirt is bright” (brightness judgment task), (2) “I would like to buy this T-shirt” (purchase intention task), (3) “Friends will say I should not wear this T-shirt” (social risk task), and (4) “This T-shirt is attractive” (preference task). Each question was presented only once before scanning (Fig. 1), and each T-shirt was rated only once per session. The order in which T-shirts appeared was randomized among participants in all sessions. There were 63 trials in each session. After the fMRI scanning was completed, we conducted a post-study manipulation check. An interview analysis confirmed their disregard of price, brand, and quality. To control for effects of novelty and familiarity, we also confirmed that participants had not previously seen any of the T-shirts. Consumer behavior studies consider purchase intention as a good indicator of purchase behavior (Haugtvedt et al., 2008). Therefore, we asked participants about their purchase intentions to evoke relevant neural processes. As a measure of social risk, we chose the following question on the basis of the results of our preliminary experiment on 210 university students: “Friends will say I should not wear this T-shirt.” The procedure of the preliminary experiment is described in Supplementary Material 1. Note that this preliminary experiment for selecting the questions was conducted separately from the preliminary experiment for selecting the stimuli which is described in the Stimuli section.

Behavioral data analysis We conducted behavioral data analysis to test our hypothesis about the relationship between social risk perception and purchasing behavior; we built the following multiple regression model to confirm that perception of social risk discouraged purchases. PI ¼ β1ð PrÞ þ β2ðSÞ þ ω; where PI denotes the subjective ratings of purchase intention, Pr denotes the subjective ratings of preference (confounding factor), S denotes the subjective ratings of social risk, and ω denotes the residual. Image acquisition and data analysis MRI data acquisition All MRI data were acquired with a 3 T Philips Achieva scanner. Functional images were acquired using the echo-planar functional images sensitive to blood oxygenation level-dependent contrast (64 × 64 matrix, TR = 2000 ms, TE = 30 ms, flip angle = 80°, FOV = 192 mm, 32 slices, 4 mm slice thickness). Preprocessing of fMRI data Preprocessing and statistical analyses for all images were performed using SPM8 (Wellcome Department of Cognitive Neurology, London) implemented in MATLAB (www.mathworks.com). During preprocessing, images were corrected for slice-timing and head motion, spatially normalized to the Montreal Neurological Institute (MNI) template, and smoothed using a Gaussian kernel of 8 mm full width at half maximum.

Table 1 Behavioral results of multi-regression analysis. Adjusted R2

Regression coefficient (β) Preference (β1)

Social risk (β2)

0.460⁎⁎

−0.248⁎⁎

0. 3795⁎⁎

Regression coefficients of subjects' preference ratings and predictions of social risk were significant (p b 0.01). In particular, the regression coefficient of social risk is negative, indicating that it inhibits purchase intentions. ⁎⁎ p b 0.01.

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

123

Table 2 Correlation of local increases in neural activity and social risk in the PI and S tasks. Area

MNI peak coordinates (mm) x

y

z

t value

k

PI session with Social risk Occipital inferior cortex Fusiform gyrus Anterior insula mPFC TPJ

L R L n.s. n.s.

−24 36 −42

−73 −70 23

−14 −17 −11

6.06 5.60 4.69

379 287 106

S session with Social risk mPFC TPJ Anterior insula

L L n.s.

−12 −51

44 −34

22 22

5.51 5.35

182 334

Local increases in neural activity with each session are presented as MNI coordinates (cluster peak expressed as the t value). k indicates the cluster size. The statistical threshold at the voxel level was set at p b 0.001 and corrected for multiple comparisons at the cluster level (FWE, p b 0.05).

Fig. 3. Scatter plot of preference (Pr) and social risk (S) rating. The X-axis indicates the number of Pr ratings; the Y-axis indicates the number of S ratings. The gradation of each box indicates the number of counts. The distribution of Pr and S ratings shows a negative correlation.

Overview of fMRI data analysis Functional MRI analyses focused on data from the purchase intention session and social risk session, while fMRI data from the brightness

judgment session were used to create a mask (described below) (Fig. 2). Moreover, fMRI data from the preference session were ignored because fMRI data from this session will be used in another study. Statistical fMRI analyses were performed at the subject and group levels. In subject-level fixed-effect analyses, trial-related activity was modeled by convolving a vector of trial onsets (duration set at 0 s) using a canonical hemodynamic response function with time and dispersion derivatives within the context of the General Linear Model. Confounding factors (head motion, magnetic field drift) were included in the model. Trials in which no response was made were modeled separately by another regressor. fMRI data analysis of the purchase intention task For fMRI data from the purchase intention task, we inserted two parametric regressors into the model: we regressed the subjective attractiveness (preference: confounding factor) of products and employed a social risk regressor to identify regions of increased neural activity as a function of subjective social risk rating. Ratings for each participant were entered as first-order parametric modulators. For the group-level random-effect analysis, we identified social risk activations that were consistent across participants using one-sample t tests of contrast images from the subject-level, fixed-effect analyses. The statistical threshold at the voxel level was set at p b 0.001 and corrected for multiple comparisons at the cluster level (FWE, p b 0.05).

Fig. 4. Results of reaction time data. Participants rated their purchase intentions from 1 to 8. The purchase intention session (A) produced no relationship between response time (RT) and social risk (S). The graphical relationship between response RT and S in the social risk session (B) formed an inverted U, with longer RTs for intermediate ratings (2–7) and shorter RTs for extreme ratings. All error bars represent the standard error of the mean.

fMRI data analysis of the social risk task For fMRI data from the social risk task, we also inserted two regressors (preference and social risk) as first-order parametric modulators to identify regions of increased neural activity associated with each participant's social risk rating. For the group-level, random-effect analysis, we identified social risk activations that were consistent across participants using one-sample t tests of contrast images from the subject-level, fixed-effect analyses. However, some of the observed brain activity was unavoidably associated with motor responses because the social risk ratings in social risk task, unlike the purchase intention task, was totally correlated to button pressing. Therefore, we identified those brain regions (motor-responserelated regions) by using the brightness judgment (control) task, and excluded the possibility of obtaining results from those regions in the social risk task (Fig. 2). More specifically, we identified regions of increased neural activity as a function of brightness rating using fMRI data from brightness judgment (control) task by the same parametric modulation analysis procedure explained above. We inserted brightness judgment ratings as well as preference and social risk regressors into the model as first-order parametric modulators to identify regions

124

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

Fig. 5. Activation correlated to social risk in the purchase intention task. Brain regions showing task-specific activation were identified by parametric modulation coupling the social risk parameter with fMRI images from the purchase intention task. Regional activation is superimposed on a sagittal section (top left), a coronal section (top right), and an axial section (bottom left) of the T1-weighted anatomical images provided by SPM8. The red–yellow color scale indicates t values. The statistical threshold at the voxel level was set at p b 0.001 and corrected for multiple comparisons at the cluster level (FWE, p b 0.05).

of increased neural activity as a function of brightness judgment rating at the subject level. We then identified brightness judgment activations that appeared consistently among participants using one-sample t tests of contrast images from subject-level, fixed-effect analyses for the group-level, random-effect analysis. A mask image was generated from the obtained contrast map by using the statistical threshold of peak level set at p b 0.05 (uncorrected) based on the results of previous studies (Mano et al., 2011; Sugiura et al., 2012). We then excluded the regions in the mask image from the result of the social risk task. Importantly, the mask image successfully excluded the motor-related regions (Supplementary Fig. 1). In our study, the advantage of using the mask compared to other analyses (e.g., subtraction analysis) is to eliminate the possibility of capturing brain regions deactivating in the control session. Furthermore, this mask image was not applied as an explicit mask (limiting search volume) but only visually masked the result. Therefore, the mask did not have an impact on the statistical analysis, but visually excluded the motor-related regions (clusters in yellow in Supplementary Fig. 1). Finally, we obtained results using the statistical threshold at

the voxel level set at p b 0.001 and corrected for multiple comparisons at the cluster level (FWE, p b 0.05). Post-hoc analysis of the purchase intention task and social risk task As reported in the results, we observed activity in the left anterior insula during the purchase intention session, and in the mPFC and left TPJ during the social risk task. Therefore, in a post-hoc analysis, we focused on these regions and directly compared brain activation in these three regions between social risk perception during the purchase intention task and the social risk task. This additional analysis enabled us to clarify whether a difference exists in the neural basis of social risk perception underlying participants' purchase decisions and their explicit evaluations of social risk. More specifically, we computed average beta values at the left anterior insula, mPFC, and left TPJ for each subject. The beta values at the left anterior insula were extracted from the cluster identified in the parametric modulation analysis conducted using subjective ratings of social risk on the fMRI data in the purchase intention task, whereas the beta values at the mPFC and left TPJ were extracted from the

Fig. 6. Activity correlated to social risk in the social risk task. Brain regions showing task-specific activation were identified by parametric modulation combining the social risk parameter with fMRI images from the social risk task. Regional activation is superimposed on a sagittal section (top left), a coronal section (top right), and an axial section (bottom left) of the T1-weighted anatomical images provided by SPM8. The red–yellow color scale indicates the t value. The statistical threshold at the voxel level was set at pb0.001 and corrected for multiple comparisons at the cluster level (FWE, pb0.05). A indicates TPJ activity. B shows activity in the mPFC. This image was masked by an image from the control task to exclude motor-related activation.

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

125

126

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

clusters identified in the analysis using subjective ratings of social risk in the social risk task. We then performed paired t tests on those average beta values. Additional analysis between regressors in the purchase intention task As reported in the results, we observed anterior insula activity, which is one of the social–emotional regions, in the purchase intention session. Therefore, to ascertain that the identified brain region that correlated with the perception of social risk did not simply respond to (un)pleasant feelings about products in the purchase intention task, we computed and plotted the average beta values within the cluster in the anterior insula. These average beta values showed correlations between the anterior insula and preference ratings (confounding factor), and between the anterior insula and social risk ratings. We expected no negative correlation between the anterior insula and preference ratings. Results Behavioral results Results of multiple regression were positive for β1, a preference weighting, and negative for β2, a social risk rating (Table 1). In other words, preferences enhance the intentions to purchase; however, perception of social risk inhibits purchases. This result confirms that social risk has an effect toward purchase decision-making, independent of personal preference. Social risk ratings were widely dispersed, yielding a sufficient number of trials rated in each of the eight levels of the social risk scale (Fig. 3) for statistical analyses. Social risk and preference ratings were negatively correlated (r = −0.45, p b 0.01); however, the data scatter was consistent with our selection of heterogeneous stimuli (Fig. 3). We observed no linear relationship between response time (RT) and social risk (S) during the purchase intention session (r = 0.04, p = 0.27) or the social risk session (r = − 0.04, p = 0.22) (Fig. 4). fMRI results: perceiving social risk fMRI data from the purchase intention task indicated that activity in the left anterior insular cortex increased monotonically as a function of the social risk rating (Table 2 and Fig. 5). Activity in the left inferior occipital cortex and right fusiform gyrus also correlated positively and significantly with social risk rating (Table 2 and Fig. 5). fMRI data from the social risk task indicated that activity in the mPFC and left TPJ increased as a function of social risk ratings (Table 2 and Fig. 6). To understand the activation in the occipital inferior cortex and fusiform gyrus in the purchase intention task, we conducted additional analyses. We surmise that activity in these structures might relate to processing facial features. Many images of the T-shirts featured faces that resulted in high social risk scores. Therefore, we divided the images (T-shirts) into two groups—T-shirts with and without faces—and conducted a t-test using social risk scores. The average social risk score for T-shits with faces (mean = 5.03, SD = 2.07) was significantly higher than that of T-shirts without faces (mean = 3.19, SD = 1.79) (p b 0.001).

Fig. 2). There were significant differences between tasks in terms of the average beta values in the mPFC (t = 2.32, p = 0.03, two-tailed paired t test) and left TPJ cluster (t = 2.40, p = 0.02, two-tailed paired t test). fMRI results: between regressors in the purchase intention task The purchase intention task displayed a positive significant correlation between activity in the left anterior insula cortex and social risk ratings and a positive correlation (not significant) between activity in the left anterior insula cortex and preference ratings (Supplementary Fig. 3). There was a marginally significant difference between the average beta value in the left anterior insular cortex cluster between the regressors (t = 2.00, p = 0.05, two-tailed paired t test). Discussion To the best of our knowledge, this study is the first to determine the neural processes underlying social risk perception during purchase decisions. Our fMRI experiment revealed that activity in the left anterior insula positively correlated with subjective ratings of social risk during the purchase intention task. On the other hand, analysis of the social risk task revealed a significant positive correlation between the social risk ratings and activity in the left TPJ and mPFC. Social risk perceptions during purchase decisions are processed in social– emotional regions of the brain In this study, the activity in the anterior insula was positively correlated with subjective ratings of social risk during the purchase intention task, which is consistent with the findings of previous studies. Activity in the anterior insula has been associated with social interactions that involve violations of social norms (Sanfey et al., 2003; Spitzer et al., 2007; Xiang et al., 2013). Furthermore, the anterior insula is active when people perceive that their behavior might transgress social norms (King-Casas et al., 2008; Montague and Lohrenz, 2007; Rilling and Sanfey, 2011), when they perceive socially unacceptable behavior in others (Hsu et al., 2008; Izuma, 2013; Montague and Lohrenz, 2007; Singer et al., 2004; Spitzer et al., 2007; Xiang et al., 2013), and when they experience social–emotional pain (Eisenberger et al., 2003; Sanfey et al., 2003). Activity in the anterior insula may be due to unpleasant feelings toward the product, as cautioned by Royet et al. (2003) and Small et al. (2003). However, we were able to dismiss this possibility because of our inclusion of subjective preference ratings of products used in the fMRI analysis; the anterior insula showed no negative correlation with preference ratings (Supplementary Fig. 4). We could also dismiss possible effects of correlation between social risk and preference ratings in the fMRI analysis, since SPM makes regressors in the model orthogonal in the parametric modulation analysis. Therefore, the anterior insula activity found in this study reflects social–emotional processes. Taken together, our first hypothesis (H1) is supported: Consumers' social risk perceptions in purchase decisions are processed in the social–emotional regions of the brain. The neural basis of social risk perception during purchase decisions differs from that during the explicit evaluation of social risk

fMRI results: between purchase intention task and social risk task Activity in the left anterior insular cortex correlated significantly with the perception of social risk in the purchase intention task, but not in the social risk task (Supplementary Fig. 2). There was a marginally significant difference between tasks with reference to the average beta value in the left anterior insular cortex cluster (t = 1.86, p = 0.08, two-tailed paired t test). Activity in the mPFC and left TPJ correlated significantly with the perception of social risk in the social risk task, but the same was not found for the purchase intention task (Supplementary

In the fMRI results from the social risk task, the subjective rating of social risk did not show a significant correlation with anterior insular activity, although it correlated with activity in the mPFC and TPJ. In contrast, in the fMRI results from the purchase intention task, the subjective rating of social risk did not significantly correlate with mPFC or TPJ activity, but did correlate with activation in the anterior insula (see fMRI results: between purchase intention task and social risk task section). This result corroborates the findings of previous studies, which indicated that explicit anticipation of the reactions of others

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

involves the mPFC and TPJ, both of which are ToM-related regions (Lieberman, 2007; Saxe and Kanwisher, 2003). Moreover, a direct comparison of the beta values for the anterior insula, mPFC, and TPJ revealed that there was a (marginally) significant difference in neural activity between the purchase intention task and the social risk task (Supplementary Fig. 3). Therefore, our second hypothesis (H2) was marginally supported and, thus, confirming that the neural basis of social risk perception during purchase decision making is different from that of the explicit evaluation of social risk. Study limitations Overall, there was one limitation in our study: we presented the same T-shirts several times to the subjects. This might have created a habituation effect, resulting in different brain activation patterns. For example, different activity patterns in the anterior insula during the purchase intention and social risk tasks might be caused by a habituation effect because social risk had already been processed by the anterior insula in the preceding purchase intention task. Potential confounding factors As noted, results of parametric modulation analysis from the social risk task included brain activity in regions associated with motor responses (pushing buttons) and perceptual decisions. However, the activity observed, except for that in the mPFC and left TPJ, was successfully masked by the control task (Supplementary Fig. 1). As expected, no monotonic relationship was observed between response times and ratings in the purchase intention task. However, slower responses (approximately 25% reduction in response time) coincide with intermediate ratings (2–7) compared with extreme ratings (1–8) in the social risk task. No significant difference appeared within the 2–7 ratings or between the two extremes (Fig. 4). Therefore, differences in task difficulty are unlikely to have significantly contributed to the regional activation patterns we observed. Unexpectedly, we found that activity in the occipital inferior cortex and fusiform gyrus correlated with the perception of social risk in the purchase intention task. We surmise that activity in these structures might relate to the processing of facial features. Additionally, our finding on the effect of faces on the social rating, as described in the fMRI results: perceiving social risk section, indicates that the occipital inferior cortex and fusiform gyrus may be responsible for processing faces, similar to the findings of previous studies (Grill-Spector et al., 2004; Hesselmann et al., 2008).

127

brain activity may have occurred when both the social risk and preference scores were high (i.e., in the condition where a subject likes a product but considers its purchase socially risky). Thus, we conducted an additional parametric modulation analysis by using a parameter representing the magnitude of the conflict, obtained by multiplying social risk ratings and preference ratings (Supplementary Material 5). Therefore, the value of this parameter increased when both the social risk and preference scores were high. However, we did not identify any significant brain regions correlated with the conflict parameter. This result suggests that there is no brain region that responds to the interaction between social risk and preference. Alternatively, possibly not enough trials were presented that triggered this conflict in the brain. Study implications Implicit and explicit social risk processes in purchase decisions Our results suggest that the anterior insula processes consumers' social risk implicitly, whereas ToM-related regions process it explicitly. These results are an extension of previous studies. Activity in the anterior insula has been found to encode representations of risk and uncertainty about decisions in nonsocial contexts (Berns et al., 2006; Bossaerts, 2010; Critchley et al., 2001; Huettel et al., 2006; Kuhnen and Knutson, 2005; Mohr et al., 2010; Paulus et al., 2003; Preuschoff et al., 2006; Simmons et al., 2008), and previous studies indicate a significant role of the anterior insula in processing risk implicitly (Clark et al., 2008; Cunningham et al., 2004; Herwig et al., 2011; Ishii et al., 2012; Morris, 2002; van den Bos et al., 2013). Thus, our observation of activity in the anterior insula implies that it is involved in implicitly processing other people's disapproval of one's purchases. On the other hand, the mPFC and TPJ are known to be involved in situations in which people violate social norms (Berthoz et al., 2002). Thus, our observation of activity in the mPFC and TPJ implicates these regions in the explicit consideration of other people's disapproval of one's purchases.

Previous studies have indicated that the left insula and left TPJ have specific functions (Bahnemann et al., 2010; Oppenheimer et al., 1996). In particular, the left anterior insula has wider structural connectivity to other brain regions as compared to the right anterior insula (Jakab et al., 2012). Furthermore, a previous meta-analysis of the function of the insula indicated that the left insula is more robustly involved than the right insula in social–emotional processes (Kurth et al., 2010). Thus, the left insula may play a more important role in the perception of implicit social risk. However, we found bilateral activation in the insula and TPJ by using a more liberal threshold (p b 0.001 uncorrected at the voxel level, without cluster level FWE correction). Thus, laterality is inconsequential for our results.

Neural processes underlying purchase decisions involve the calculation of social risk Previous neuroeconomic studies indicate that social information modulates activity in the ventral striatum, a brain region associated with reward processes (Erk et al., 2002; Fliessbach et al., 2007; Izuma, 2012; Izuma et al., 2010; Klein and Platt, 2013; Tricomi et al., 2010). Furthermore, a previous study suggested that the ventral striatum encodes purchase intentions (Knutson et al., 2007). The present study suggests that purchase intentions are suppressed when consumers perceive social risk (Table 1) and that the anterior insula encodes this social risk. Furthermore, an additional analysis regarding preference revealed that personal preference is processed in the midbrain, which is located in the DA network (Supplementary Material 6). Combining our results with that of previous studies that examined the ventral striatum, we suggest that two neural mechanisms underlie purchase decisions. First, the midbrain encodes personal preference, and the anterior insula responds to social risk. Second, the ventral striatum integrates prospective personal preferences and social risk to determine the strength of purchase intentions. Combined with previous neuroeconomics studies (Hsu et al., 2008; Sanfey, 2007; Tricomi et al., 2010), our results accentuate the importance of social factors in economic behavior and challenge traditional economic theory favoring a rational and narrowly self-interested Homo economicus.

Additional analysis on the conflict between social risk and preference

Conclusion

Based on the behavioral results, purchase intention could be predicted by preference and social risk. In addition, preference and social risk were negatively correlated (Fig. 3). Therefore, it is likely that conflict-related

This study provided direct neurobiological identification of the brain structures that are associated with consumers' perceptions of social risk when making purchase decisions. Specifically, it demonstrated a strong

Laterality of regional activation patterns

128

R. Yokoyama et al. / NeuroImage 91 (2014) 120–128

positive correlation between anterior insula activity and experimental subjects' subjective ratings of social risk in terms of purchases. This finding suggested that the anterior insula evaluates social risk and emits signals, which prompts consumers to not buy socially unacceptable products. Acknowledgments We would like to thank Dr. N. Terui and Dr. T. Nakagawa for their invaluable advice and expertise. We thank T. Abe, Y. Otaka, and M. Hanihara for data collection; T. Fancy, C. S. Kikuchi, and S. Michael for proofreading the language of this manuscript, and all participants and colleagues at IDAC, Tohoku University, for their support. A grant from the Japan Society for the Promotion of Science and JSPS KAKENHI supported this study (Grant Number 23300080). Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.neuroimage.2014.01.036. References Adolphs, R., Tranel, D., Damasio, H., Damasio, A., 1994. Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature 372, 669–672. Adolphs, R., Tranel, D., Damasio, A.R., 1998. The human amygdala in social judgment. Nature 393, 470–474. Bahnemann, M., Dziobek, I., Prehn, K., Wolf, I., Heekeren, H.R., 2010. Sociotopy in the temporoparietal cortex: common versus distinct processes. Soc. Cogn. Affect. Neurosci. 5, 48–58. Berns, G.S., Chappelow, J., Cekic, M., Zink, C.F., Pagnoni, G., Martin-Skurski, M.E., 2006. Neurobiological substrates of dread. Science 312, 754–758. Berns, G.S., Capra, C.M., Moore, S., Noussair, C., 2010. Neural mechanisms of the influence of popularity on adolescent ratings of music. Neuroimage 49, 2687–2696. Berthoz, S., Armony, J.L., Blair, R.J., Dolan, R.J., 2002. An fMRI study of intentional and unintentional (embarrassing) violations of social norms. Brain 125, 1696–1708. Bossaerts, P., 2010. Risk and risk prediction error signals in anterior insula. Brain Struct. Funct. 214, 645–653. Buckholtz, J.W., Marois, R., 2012. The roots of modern justice: cognitive and neural foundations of social norms and their enforcement. Nat. Neurosci. 15, 655–661. Childers, T.L., Rao, A.R., 1992. The influence of familial and peer-based reference groups on consumer decisions. J. Consum. Res. 19, 198–211. Clark, L., Bechara, A., Damasio, H., Aitken, M.R., Sahakian, B.J., Robbins, T.W., 2008. Differential effects of insular and ventromedial prefrontal cortex lesions on risky decision-making. Brain 131, 1311–1322. Craig, A.D., 2009. How do you feel — now? The anterior insula and human awareness. Nat. Rev. Neurosci. 10, 59–70. Critchley, H.D., Mathias, C.J., Dolan, R.J., 2001. Neural activity in the human brain relating to uncertainty and arousal during anticipation. Neuron 29, 537–545. Cunningham, W.A., Raye, C.L., Johnson, M.K., 2004. Implicit and explicit evaluation: fMRI correlates of valence, emotional intensity, and control in the processing of attitudes. J. Cogn. Neurosci. 16, 1717–1729. Eisenberger, N.I., Lieberman, M.D., Williams, K.D., 2003. Does rejection hurt? An fMRI study of social exclusion. Science 302, 290–292. Elster, J., 1989. Social norms and economic-theory. J. Econ. Perspect. 3, 99–117. Erk, S., Spitzer, M., Wunderlich, A.P., Galley, L., Walter, H., 2002. Cultural objects modulate reward circuitry. Neuroreport 13, 2499–2503. Fliessbach, K., Weber, B., Trautner, P., Dohmen, T., Sunde, U., Elger, C.E., Falk, A., 2007. Social comparison affects reward-related brain activity in the human ventral striatum. Science 318, 1305–1308. Grill-Spector, K., Knouf, N., Kanwisher, N., 2004. The fusiform face area subserves face perception, not generic within-category identification. Nat. Neurosci. 7, 555–562. Harenski, C.L., Hamann, S., 2006. Neural correlates of regulating negative emotions related to moral violations. Neuroimage 30, 313–324. Haugtvedt, Curtis P., P.M.H., Kardes, Frank R., 2008. Handbook of Consumer Psychology. Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E., Gintis, H., McElreath, R., 2001. In search of Homo economicus: behavioral experiments in 15 small-scale societies. Am. Econ. Rev. 91, 73–78. Herwig, U., Bruhl, A.B., Viebke, M.C., Scholz, R.W., Knoch, D., Siegrist, M., 2011. Neural correlates of evaluating hazards of high risk. Brain Res. 1400, 78–86. Hesselmann, G., Kell, C.A., Eger, E., Kleinschmidt, A., 2008. Spontaneous local variations in ongoing neural activity bias perceptual decisions. Proc. Natl. Acad. Sci. U. S. A. 105, 10984–10989. Hsu, M., Anen, C., Quartz, S.R., 2008. The right and the good: distributive justice and neural encoding of equity and efficiency. Science 320, 1092–1095. Huettel, S.A., Stowe, C.J., Gordon, E.M., Warner, B.T., Platt, M.L., 2006. Neural signatures of economic preferences for risk and ambiguity. Neuron 49, 765–775.

Ishii, H., Ohara, S., Tobler, P.N., Tsutsui, K.I., Iijima, T., 2012. Inactivating anterior insular cortex reduces risk taking. J. Neurosci. 32, 16031–16039. Izuma, K., 2012. The social neuroscience of reputation. Neurosci. Res. 72, 283–288. Izuma, K., 2013. The neural basis of social influence and attitude change. Curr. Opin. Neurobiol. 23, 456–462. Izuma, K., Adolphs, R., 2013. Social manipulation of preference in the human brain. Neuron 78, 563–573. Izuma, K., Saito, D.N., Sadato, N., 2010. Processing of the incentive for social approval in the ventral striatum during charitable donation. J. Cogn. Neurosci. 22, 621–631. Jakab, A., Molnar, P.P., Bogner, P., Beres, M., Berenyi, E.L., 2012. Connectivity-based parcellation reveals interhemispheric differences in the insula. Brain Topogr. 25, 264–271. 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. Klein, J.T., Platt, M.L., 2013. Social information signaling by neurons in primate striatum. Curr. Biol. 23, 691–696. Knutson, B., Rick, S., Wirnmer, G.E., Prelec, D., Loewenstein, G., 2007. Neural predictors of purchases. Neuron 53, 147–156. Kuhnen, C.M., Knutson, B., 2005. The neural basis of financial risk taking. Neuron 47, 763–770. Kurth, F., Zilles, K., Fox, P.T., Laird, A.R., Eickhoff, S.B., 2010. A link between the systems: functional differentiation and integration within the human insula revealed by meta-analysis. Brain Struct. Funct. 214, 519–534. Lascu, D.-N., Zinkhan, G., 1999. Consumer conformity: review and applications for marketing theory and practice. J. Mark. Theory Pract. 1–12. Lieberman, M.D., 2007. Social cognitive neuroscience: a review of core processes. Annu. Rev. Psychol. 58, 259–289. Mano, Y., Sugiura, M., Tsukiura, T., Chiao, J.Y., Yomogida, Y., Jeong, H., Sekiguchi, A., Kawashima, R., 2011. The representation of social interaction in episodic memory: a functional MRI study. Neuroimage 57, 1234–1242. Mason, M.F., Dyer, R., Norton, M.I., 2009. Neural mechanisms of social influence. Organ. Behav. Hum. Decis. Process. 110, 152–159. Mohr, P.N.C., Biele, G., Heekeren, H.R., 2010. Neural processing of risk. J. Neurosci. 30, 6613–6619. Montague, P.R., Lohrenz, T., 2007. To detect and correct: norm violations and their enforcement. Neuron 56, 14–18. Morris, J.S., 2002. How do you feel? Trends Cogn. Sci. 6, 317–319. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97–113. Oppenheimer, S.M., Kedem, G., Martin, W.M., 1996. Left-insular cortex lesions perturb cardiac autonomic tone in humans. Clin. Auton. Res. 6, 131–140. 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. Peter, J.P., Olson, J.C., 1996. Consumer Behavior and Marketing Strategy. Irwin, Chicago, IL. Preuschoff, K., Bossaerts, P., Quartz, S.R., 2006. Neural differentiation of expected reward and risk in human subcortical structures. Neuron 51, 381–390. Rilling, J.K., Sanfey, A.G., 2011. The neuroscience of social decision-making. Annu. Rev. Psychol. 62 (62), 23–48. Robertson, T.S., Kassarjian, H.H., 1990. Handbook of Consumer Behavior. Rook, D.W., Fisher, R.J., 1995. Normative influences on impulsive buying behavior. J. Consum. Res. 22, 305–313. Royet, J.-P., Plailly, J., Delon-Martin, C., Kareken, D.A., Segebarth, C., 2003. fMRI of emotional responses to odors:-influence of hedonic valence and judgment, handedness, and gender. Neuroimage 20, 713–728. Sanfey, A.G., 2007. Social decision-making: insights from game theory and neuroscience. Science 318, 598–602. 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. Saxe, R., Kanwisher, N., 2003. People thinking about thinking people — the role of the temporo-parietal junction in “theory of mind”. Neuroimage 19, 1835–1842. Simmons, A., Matthews, S.C., Paulus, M.P., Stein, M.B., 2008. Intolerance of uncertainty correlates with insula activation during affective ambiguity. Neurosci. Lett. 430, 92–97. Singer, T., Seymour, B., O'Doherty, J., Kaube, H., Dolan, R.J., Frith, C.D., 2004. Empathy for pain involves the affective but not sensory components of pain. Science 303, 1157–1162. Small, D.M., Gregory, M.D., Mak, Y.E., Gitelman, D., Mesulam, M.M., Parrish, T., 2003. Dissociation of neural representation of intensity and affective valuation in human gustation. Neuron 39, 701. Spitzer, M., Fischbacher, U., Herrnberger, B., Gron, G., Fehr, E., 2007. The neural signature of social norm compliance. Neuron 56, 185–196. Sugiura, M., Sassa, Y., Jeong, H., Wakusawa, K., Horie, K., Sato, S., Kawashima, R., 2012. Self-face recognition in social context. Hum. Brain Mapp. 33, 1364–1374. Tricomi, E., Rangel, A., Camerer, C.F., O'Doherty, J.P., 2010. Neural evidence for inequalityaverse social preferences. Nature 463, 1089–1091. van den Bos, W., Talwar, A., McClure, S.M., 2013. Neural correlates of reinforcement learning and social preferences in competitive bidding. J. Neurosci. 33, 2137–2146. Wänke, M., 2009. Social Psychology of Consumer Behavior. Psychology Pr. Xiang, T., Lohrenz, T., Montague, P.R., 2013. Computational substrates of norms and their violations during social exchange. J. Neurosci. 33, 1099–1108. Xu, Y.J., Summers, T.A., Belleau, B.D., 2004. Who buys American alligator? Predicting purchase intention of a controversial product. J. Bus. Res. 57, 1189–1198.