Decreased insular and increased midbrain activations during decision-making under risk in adolescents with excess weight

Authors

  • Elena Delgado-Rico,

    1. Department of Personality, Evaluation and Psychological Treatment, University of Granada, Granada, Spain
    Search for more papers by this author
  • Carles Soriano-Mas,

    1. Department of Psychiatry, Bellvitge University Hospital-IDIBELL, Barcelona, Spain
    2. Carlos III Health Institute, Centro de Investigación Biomédica en Red de Salud Mental (CIBERSAM), Madrid, Spain
    Search for more papers by this author
  • Juan Verdejo-Román,

    1. Department of Personality, Evaluation and Psychological Treatment, University of Granada, Granada, Spain
    Search for more papers by this author
  • Jacqueline S. Río-Valle,

    1. Department of Nursing, University of Granada, Granada, Spain
    Search for more papers by this author
  • Antonio Verdejo-García

    Corresponding author
    1. Institute of Neurosciences Federico Olóriz, University of Granada, Granada, Spain
    2. School of Psychology, Psychiatry and Psychological Medicine, Monash University, Melbourne, Victoria, Australia
    • Department of Personality, Evaluation and Psychological Treatment, University of Granada, Granada, Spain
    Search for more papers by this author

  • Disclosures: The authors declared no conflicts of interest.

  • Funding agencies: This study has been funded by grants PI 0416/2008 (BRAINOBE) from the Andalusian Health Service (Consejería de Salud), PSI2010-17290 (INTEROBE) from the Ministry of Innovation and Science (MICINN), and P-10-HUM-6635 (NEUROECOBE). Dr. Soriano-Más is funded by a Miguel Servet contract from the Carlos III Health Institute (CP10/00604).

Correspondence: Antonio Verdejo-García (antonio.verdejo@monash.edu or averdejo@ugr.es)

Abstract

Objective

Functional magnetic resonance imaging (fMRI) was used to explore the brain substrates of decisions under risk in excess weight adolescents. Decreased activations of the brain regions signaling risk (orbitofrontal cortex [OFC], insula) were expected during anticipation of higher rewards and increased activations of the brain regions involved in reward processing (OFC, striatum) were expected after reward receipt in excess weight adolescents compared to normal weight controls.

Design and Methods

Fifty-two adolescents (age range 12-17), classified in three groups as a function of BMI: obese (n = 21), overweight (n = 15), or normal weight (n = 16) performed the Risky-Gains task as described by Paulus et al. in the fMRI scanner.

Results

Excess weight adolescents, compared to normal weight controls, showed decreased left insular and increased midbrain activations during anticipation of risky choices. In addition, excess weight adolescents showed increased activations of the inferior frontal gyrus, parahippocampus, thalamus, and posterior brain regions after reward receipt.

Conclusions

Adolescents with excess weight showed reduced activations in brain regions signaling risk and increased activations in regions signaling reward during anticipation of decisions involving risk and reward. In addition, post-decision reward outcomes produced increased activations of regions involved in emotional salience in excess weight adolescents versus controls.

Introduction

The prevalence of adolescent overweight and obesity has sharply increased over the last two decades, rapidly reaching epidemic levels [1, 2]. To account for this rise, recent theoretical models underscore the role of decision-making skills as a key asset to regulate caloric intake in modern environments, since these are characterized by unrestricted access to food and strong media-driven appeals to eat caloric products [3]. Decision-making skills are particularly relevant in the case of adolescents, in whom brain developmental transitions seem to be hardwired to maximize reward at the expense of risk [4]. Neuroimaging studies have demonstrated that adolescents have hypersensitive striatal response to reward prediction [5, 6] and heightened activation of brain regions involved in fostering risk-taking (orbitofrontal cortex [OFC]) during decision-making [7]. Obesity-induced neural adaptations may further rewire these brain systems; in adults, excessive BMI levels are associated with reduced striatal dopamine function and lower metabolism of the OFC [8, 9].

In agreement with these notions, burgeoning evidence from neuroscience studies indicates that adolescents with excess weight have increased activations of brain systems involved on sensory-emotional processing (frontal operculum, insula) [10] and a deficient activation of the neural network supporting cognitive control of choice (several regions of the prefrontal cortex) in response to highly appetizing food stimuli [11]. Behaviorally, excess weight adolescents have demonstrated decision-making deficits in cognitive tests measuring the ability to choose between safe and risky (superficially rewarding) choices [12]. Furthermore, obese adolescents have increased cognitive disinhibition correlated with lower gray matter volumes in the OFC [13]. However, no studies to date have explored the neural substrates of decision-making under conditions of risk and reward in excess weight adolescents.

In this study, we used functional magnetic resonance imaging to explore the brain substrates of decisions under risk in excess weight adolescents. Decision-making was challenged using the Risky-Gains task, which opposes a less rewarding safe choice with more rewarding risky choices. This task has shown to robustly activate the brain systems involved in choice-related risk-taking and reward receipt (including the OFC, insula, and striatum). We expected decreased activations of the brain regions signaling risk (OFC, insula) during anticipation of higher rewards (risky vs. safe choices) and increased activations of the brain regions involved in reward processing (OFC, striatum) after receipt of reward (vs. punishment) in excess weight adolescents compared to normal weight controls.

Methods and Procedures

Participants

Fifty-two adolescents (age range 12-17) participated in this study. They were classified in three groups (obese [n = 21], overweight [n = 15], or normal weight [n = 16]) according to their BMI, following the criteria of the International Obesity Task Force (IOTF) defined by Cole et al. [14] or, in normal weight adolescents, according to age- and sex-adjusted Spanish-specific norms [15]. The demographical data and BMI of participants are summarized in Table 1. Participants were recruited from the Endocrinology Service of the Hospital “Virgen de las Nieves” in Granada, Spain, and from schools located in the same geographical area. To be included, they had to meet the following criteria: (i) age range between 12 and 17 years old; (ii) BMI values falling within the intervals categorized as overweight or obesity according to the IOTF (excess weight adolescents), or normal weight values (normal weight adolescents); (iii) absence of past/current evidence of neurological or psychological disorders; (iv) absence of significant abnormalities on magnetic resonance imaging (MRI) or any contraindications to MRI scanning (including claustrophobia and implanted ferromagnetic objects); and (v) absence of history of brain injury involving loss of consciousness for longer than 5 min. All of them had normal or corrected-to-normal vision.

Table 1. Socio-demographic characteristics and BMI for each study group
 Obese (n = 21)Overweight (n = 15)Normal weight (n = 16)
Sexnnn
Male647
Female15119
 Mean (SD)Mean (SD)Mean (SD)
Age14.29 (1.31)14.07 (1.67)13.88 (1.36)
BMI31.33 (2.92)24.65 (1.26)20.19 (2.80)

This study was approved by the Ethical Committee for Research in Humans of the University of Granada; all procedures were conducted in accordance with the Declaration of Helsinki. All participants and their parents were debriefed about study aims and detailed procedures, and both signed an informed consent form agreeing participation.

Experimental task

We used Risky-Gains task described by Paulus et al. [16]. The task consisted of 96 trials (5 s/trial). In each trial, participants are presented with the numbers 20, 40, and 80 in a fixed order. The task requires the participant to acquire as many points as possible by choosing between safe (20 points) and risky (40, 80 points) options. Each number (20, 40, or 80) is presented on the screen for 1 s, and participants are instructed to press a button while the selected number in on the screen in order to win the corresponding amount of points. The first number in the sequence [20] is always a safe choice (the participant always receives 20 points). However, the two subsequent choices (40 and 80) can be rewarded (+40/+80) or punished (−40/−80), in the latter cases meaning that the trial ends and the participant loses 40 or 80 points, respectively. Thus, although the subject may gain more points per trial by waiting until the 40 or 80 choices appear on the screen, there is also a risk of losing 40 or 80 points. Points accumulate from trial to trial, and the stake is shown at the top of the screen, being continuously updated. Participants received feedback immediately after making a response, so they could adapt their behavior to the feedback received.

Fifty-four trials belonged to the non-punished trial type category, in which participants could get as much as 80 points, while 24 were −40 punished and 18 were −80 punished trial types. Relevantly, the final score did not depend on subject's choices, so there was no advantage in selecting safe or risky options.

Imaging data acquisition

A 3.0 T clinical MRI scanner, equipped with an eight-channel phased-array head coil, was used (InteraAchieva, Philips Medical Systems, Eindhoven, The Netherlands) to obtain a T2*-weighted echo-planar imaging (EPI) during task performance (repetition time [TR] = 2000 ms, echo time [TE] = 35 ms, field of view [FOV] = 230 × 230 mm2, 96 × 96 matrix, flip angle = 90°, 21 4-mm axial slices, 1 mm gap, 243 scans). A sagittal three-dimensional (3D) T1-weighted turbo-gradient-echo sequence (160 slices, TR = 8.3 ms, TE = 3.8 ms, flip angle = 8°, FOV = 240 × 240, 1 mm3 voxels) was obtained in the same experimental session for anatomical reference. Stimuli were presented through magnetic resonance-compatible liquid crystal display goggles (Resonance Technology Inc., Northridge, California), and responses were recorded through Evoke Response Pad System (Resonance Technology Inc., Northridge, California).

Imaging data processing and analysis

The functional images were analyzed using Statistical Parametric Mapping (SPM8) software (Wellcome Department of Cognitive Neurology, Institute of Neurology, Queen Square, London, UK), running under Matlab R2009 (MathWorks, Natick, Massachusetts). Preprocessing included slice timing correction, reslicing to the first image of the time series, normalization, using affine and smoothly nonlinear transformations, to an EPI template in the Montreal Neurological Institute (MNI) space, and spatial smoothing by convolution with a 3D Gaussian kernel (full width at half maximum = 8 mm).

Data analyses

We defined three conditions of interest: (i) safe response (20 points trials), (ii) risky response (40 and 80 point trials), and (iii) punishment feedback (−40 and −80 point trials). The first two conditions were modeled as the time elapsed from the beginning of the trial to the participants' response. The last condition was modeled as the time elapsed between feedback presentation and the end of the trial. Accordingly, two contrasts were defined to study brain activations: a risky versus safe choice contrast and a reward versus punishment feedback contrast.

We first conducted three one-sample t-tests within each group (normal weight, overweight, and obese) in order to obtain the patterns of intragroup activations in the two contrasts of interest. The statistical threshold was set at P < 0.05 false discovery rate whole-brain corrected. Next, we conducted between-group comparisons using a mixed model ANOVA and linear contrasts to determine whether brain activations were related to BMI indices: normal weight versus overweight versus obese. Finally, we conducted two-sample t-tests to deconstruct the differences in brain activations between each group-pair (i.e., normal weight vs. overweight; normal weight vs. obese; overweight vs. obese). In these analyses, significance threshold was set at P < 0.001 (uncorrected).

Results

Behavioral results

Nine participants were excluded from this study because they made less than four risky choices, thus invalidating contrast interpretation. Also, four scans were excluded because of excessive motion artifacts. As a result, imaging data from 14 obese (mean age [±SD] was of 14.07 [±1.33], 9 female and 5 male, and BMI [weight/height2, kg/m2] was 31.14 [±2.66]), 13 overweight (mean age = 14.15 [±1.77], 9 female and 4 male, and BMI = 24.73 [±1.34]), and 13 normal weight (mean age = 13.69 [±1.18], 8 female and 5 male, and BMI = 20.36 [±2.49]) adolescents were used in the analysis.

Percentage of safe and risky responses per each group

The percentage of safe and risky responses is presented in Table 2. Participants with overweight showed a higher percentage of safe responses in comparison to the other two groups, although the interaction between group and response type was non-significant [F(2.37) = 2.089, P = 0.14].

Table 2. Percentage of safe and risky responses in the three study groups
 Rate, (%)
 SafeRisky
 Mean (SD)Mean (SD)
Normal weight45.94 (16.99)54.06 (16.99)
Overweight58.89 (12.26)41.11 (12.26)
Obese49.32 (19.90)50.68 (19.90)

Neuroimaging results

Risky–safe contrast

Results for risky–safe contrast are presented in Table 3. Within-subject contrasts (whole-brain analyses) showed that each of the three groups activated the right inferior frontal gyrus/anterior insular region during risky (vs. safe) choices. In addition, a significant midbrain activation was also observed in overweight and obese groups (see Figure 1A). Linear contrasts showed that activation of the left inferior frontal/anterior insular region during risky choices was progressively smaller with increasing BMI (normal weight > overweight > obese). Likewise, midbrain activation during risky choices significantly increased with BMI (obese > overweight > normal weight; see Figure 1B). Pairwise comparisons between all groups are presented in Supplementary Table 1.

Table 3. Brain activations observed in risky versus safe choices in within-group (one-sample) and between-group (linear contrasts) whole-brain analyses
 BrodmannSideMNI coordinatesVolume, mm3t
 XYZ
Normal weight
Inferior frontal gyrus/insula47/13R362402,2885.52
Overweight
MidbrainL/R4−20−102,0567.88
Inferior frontal gyrus/insula47R3824−22804.78
Obese
Midbrain L/R6−16−69926.40
Inferior frontal gyrus47R3422−14484.64
Normal weight > overweight > obese
Insula13L−2622109204.26
Obese > overweight > normal weight
Midbrain R4−18−82643.72
Figure 1.

(A) Within-group activations observed for risky versus safe choices overlaid on selected slices of a normalized brain. Inferior frontal/anterior insular activations (upper row) progressively decreased with increasing BMI, while midbrain activations (bottom row) showed the opposite pattern. (B) Left: cluster of increasing activation during risky choices with decreasing BMI, located in the left inferior frontal/anterior insular region. Right: cluster of increasing activation during risky choices with increasing BMI, located in the superior midbrain region. X and Z denote coordinates in standard MNI space. Voxels with P < 0.001 (uncorrected) are displayed in all cases to provide a better description of the anatomical extension of findings. Color bar indicates t value. Right hemisphere is displayed on the right. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Reward–punishment contrast

Results for reward–punishment contrast are presented in Table 4. Whole-brain analysis for each groups showed a significant activation in the reward system (inferior frontal gyrus and nucleus accumbens), and the occipital lobe in all groups during reward (vs. punish) feedback. The linear contrast showed that activations observed during rewarded trials in the inferior frontal gyrus, the thalamus, the cerebellum, and the hippocampal and parahippocampal regions were significantly related to BMI (obese > overweight > normal weight; see Figure 2). No regions of greater activation in normal weight subjects were observed. Pairwise comparisons are presented in Supplementary Table 2.

Table 4. Brain activations observed in rewarded versus punished trials in within-group (one-sample) and between-group (linear contrasts) whole-brain analyses
 BrodmannSideMNI coordinatesVolume, mm3t
XYZ
Normal weight
Superior frontal gyrus8/9R2030522,7607.28
Caudate R22−8302,5285.54
Caudate L−1812203,4805.48
Middle occipital gyrus18/19L−18−98101,1285.05
Middle occipital gyrus R24−96101123.97
Angular40R46−66422484.12
Overweight
Cerebellum posterior lobe L/R−14−84−2427,2247.09
Occipital lobe18/19L/R0−84−48,8046.72
Inferior frontal gyrus47/11L−3440−186486.50
Inferior frontal gyrus11R3444−182083.21
Caudate L−1810163,6726.39
Caudate R1824−122,2485.89
Parahippocampal gyrus R22−26−301,0324.18
Middle frontal gyrus10R124861,9683.98
Middle frontal gyrus10L−365667923.44
Hippocampal/parahippocampal gyrus R26−26−62,1283.43
Temporal inferior37R50−56−244243.35
Fusiform20L−38−14−223763.07
Obese
Occipital superior/cuneus18L−12−102142,4406.07
Caudate L−200221,2964.88
Caudate R2012−161524.36
Inferior frontal gyrus47L−1822−141844.22
Hippocampus R26−28−21604.00
Obese > overweight > normal weight
Inferior frontal gyrus47/11L−4636−61,5204.81
Thalamus L−24−2602004.29
Superior occipital gyrus18/19R18−86282,1604.27
Cerebellum posterior lobe L−30−70−285764.12
Hippocampal/parahippocampal gyrus R26−28−43924.12
Figure 2.

Clusters of increasing activation during rewarded versus punished trials with increasing BMI. They were located in the left inferior frontal gyrus, thalamus, hippocampus gyrus, superior occipital gyrus (extending to medial cuneus), and cerebellum posterior lobe. X, Y, and Z denote coordinates in standard MNI space. Voxels with P < 0.001 (uncorrected) are displayed in all cases to provide a better description of the anatomical extension of findings. Color bar indicates t value. Right hemisphere is displayed on the right. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Discussion

The main findings from this study were that excess weight adolescents, compared to normal weight peers, have a decreased left insular activation and an increased midbrain activation during risk-based decision-making. In addition, excess weight adolescents have increased inferior frontal gyrus, thalamus, parahippocampal, and posterior activations in response to reward receipt. These brain activation differences emerged in the absence of significant between-group differences on behavioral choice, such that they genuinely reflect different approaches to decision-making as a function of weight status. Therefore, we conclude that excess weight adolescents have differential patterns of brain activation when they face the risk associated with reward-related decisions and when they process the rewards obtained by these decisions.

The first relevant finding refers to a significantly decreased left anterior insular activation during the period preceding actual choice in the excess weight groups. Conversely, excess weight adolescents showed an increased activation in the midbrain, a hub of ascending monoaminergic bundles. The anterior insula is importantly involved in signaling the probability of aversive outcomes, thus guiding risk prediction and choices oriented to minimize losses [17, 18]. Accordingly, patients with insula damage fail to adjust their bets by the odds of winning in a risk-taking gamble task; they bet similarly high amounts of money irrespective of outcome probability [19]. This is consistent with the finding that excess weight adolescents have increased preference for risky decks in the Iowa Gambling Task [12]. The insula is also strongly associated with interoceptive sensitivity (the ability to perceive bodily feedback to regulate internal state), which is decreased as a function of increased body weight in adolescents [20]. Individual differences in interoceptive sensitivity significantly shape cognitive-affective processes including decision-making; for example, individuals with good interoceptive sensitivity tend to select less risky choices in the Iowa task [21]. Overall, the evidence suggests that excess weight adolescents may have dysfunctional insular-mediated processing of interoceptive information relevant for decision-making. On the other hand, they seem to rely on midbrain dopamine regions to ponder risky versus safe options, probably biasing preference toward immediate reinforcement [5, 6]. In real-life decision-making, these deficits may promote excessive meal intake at the expense of negative consequences, either immediate (e.g., belly ache) or postponed (e.g., restriction of social activities).

In line with the results from the pre-decisional stage, the second finding referred to an increased inferior frontal gyrus activation in response to reward versus punishment feedback in the excess weight group. This region has been associated with the evaluation of reward outcome following the risk-based decisions among healthy adolescents [22]. In addition, this region, along with others identified in the same contrast (e.g., parahippocampus and cerebellum), has been linked to incentive motivation toward food stimuli in adolescents [23]. Therefore, our findings are suggestive of the notion that, in excess weight adolescents, the persistent stimulation of incentive motivational systems by seeking of high caloric food may hypersensitize the brain substrates of reward processing in this group. Since the evaluation of outcomes is thought to slot in the formation of preferences for subsequent decisions [24], decision-making in excess weight adolescents may be overridden by sensory-emotional aspects of reward to the expense of risk [25]. This notion is in line with the proposed parallels between obesity and addiction [8, 26], based on the array of reward-related neurobiological alterations associated with increased BMI, including reduced striatal dopamine function and lower metabolism of the OFC [8, 9]. In sum, excess weight adolescents show hyperprocessing of reward outcomes in the inferior frontal gyrus and other brain regions importantly involved in reward and arousal processing, which may sensitize their decisional balance toward the rewarding properties of food products.

The main conclusion of this study is that adolescents with excess weight have reduced activations in brain regions signaling risk and increased activations in regions signaling reward during anticipation of decisions involving risk and reward. In addition, post-decision reward processing produced increased activations of several regions involved in emotional salience as a function of increased weight. These results demonstrate for first time the differential patterns of brain activation during decisions involving risk and reward in adolescents with excess weight: these patterns are suggestive of reduced signaling of risk and increased reactivity to reward during decisions involving both factors. Nonetheless, it is important to note that these differential patterns of brain activations arise in the absence of significant behavioral differences in the task choices. Therefore, future studies are warranted to explore whether these patterns can contribute to explain overeating as a risky choice in more ecological food-choice paradigms or everyday measures of food habits.

Ancillary