Abstract

Converging behavioral evidence indicates that temporal discounting, measured by intertemporal choice tasks, is inversely related to intelligence. At the neural level, the parieto-frontal network is pivotal for complex, higher-order cognitive processes. Relatedly, underrecruitment of the pFC during a working memory task has been found to be associated with steeper temporal discounting. Furthermore, this network has also been shown to be related to the consistency of intertemporal choices. Here we report an fMRI study that directly investigated the association of neural correlates of intertemporal choice behavior with intelligence in an adolescent sample (n = 206; age 13.7–15.5 years). After identifying brain regions where the BOLD response during intertemporal choice was correlated with individual differences in intelligence, we further tested whether BOLD responses in these areas would mediate the associations between intelligence, the discounting rate, and choice consistency. We found positive correlations between BOLD response in a value-independent decision network (i.e., dorsolateral pFC, precuneus, and occipital areas) and intelligence. Furthermore, BOLD response in a value-dependent decision network (i.e., perigenual ACC, inferior frontal gyrus, ventromedial pFC, ventral striatum) was positively correlated with intelligence. The mediation analysis revealed that BOLD responses in the value-independent network mediated the association between intelligence and choice consistency, whereas BOLD responses in the value-dependent network mediated the association between intelligence and the discounting rate. In summary, our findings provide evidence for common neural correlates of intertemporal choice and intelligence, possibly linked by valuation as well as executive functions.

INTRODUCTION

Behavioral studies have shown that temporal discounting (i.e., the propensity to devalue future rewards), which is also commonly considered as an indicator of behavioral impulsivity (Ainslie, 1975), is negatively related to intelligence in adults (Shamosh et al., 2008; de Wit, Flory, Acheson, McCloskey, & Manuck, 2007) and in adolescents (Freeney & O'Connell, 2010; Olson, Hooper, Collins, & Luciana, 2007). These findings were further supported by the results of a meta-analysis showing small to moderate inverse correlations (mean effect size r = −.23) between the rate of temporal discounting and intelligence (Shamosh & Gray, 2008). In a first attempt to reveal common neural processes underlying intelligence and temporal discounting, Shamosh et al. (2008) showed that the negative correlation between measures of intelligence and temporal discounting rates was partly mediated by higher BOLD response in the anterior pFC during a working memory (WM) task. They thus proposed that WM is involved in temporal discounting as choosing between different future reward alternatives requires the maintenance and integration of diverse information (Shamosh et al., 2008). However, to the best of our knowledge no study has yet directly tested whether BOLD responses in brain networks that are associated with individual differences in intelligence would also correlate with BOLD responses during intertemporal choice. Investigating such relations may shed new light on potential overlaps between brain circuitries underlying temporal discounting and intelligence.

Steeper rates of discounting delayed rewards, measured as the proclivity for immediate rewards, have been considered as an important component in the development and persistence of addictive behavior (Bickel et al., 2007), such as excessive substance use (Reynolds, Patak, & Shroff, 2007; Bickel, Odum, & Madden, 1999) and gambling (Holt, Green, & Myerson, 2003). The integrity of brain networks underlying the inverse relation between intelligence and temporal discounting behavior might thus be a protective factor against the development of addictive behavior, such as substance abuse (Sjolund, Allebeck, & Hemmingsson, 2012). This is of particular relevance for adolescent development: Given the maturational gap of the prefrontal top–down control circuitry lagging behind the development of subcortical bottom–up network (Casey, Jones, & Hare, 2008; Casey, Tottenham, Liston, & Durston, 2005; Steinberg, 2005), adolescence may be a critical period regarding the development of addictive behavior. The yet-to-mature prefrontal regulatory functions are vulnerable to subcortical circuitry's heightened sensitivity to rewarding stimuli, such as alcohol, illicit drug, or risky activities, that elicit high motivational or affective valences (Crone & Dahl, 2012).

Brain Networks of Intertemporal Choice

Research on intertemporal choice, that is, choosing between a small immediate reward and a larger but delayed reward, suggests that such decisions consist of two neurocognitive processes, and both may be associated with intelligence. On one hand, there is the process of evaluating the subjective values of the choice options. BOLD responses in brain regions, such as the ventral striatum (VS), ventromedial pFC, and posterior cingulate cortex, have been found to correlate with the subjective values of delayed rewards (Peters & Buchel, 2009; Kable & Glimcher, 2007). In addition, value-dependent processes have also been found to be associated with frontal regions, such as the perigenual ACC and the inferior frontal gyrus (IFG; Ripke et al., 2012; Hare, Camerer, & Rangel, 2009; Kable & Glimcher, 2007). On the other hand, intertemporal choice also relies on the process of comparing alternative options (Hoffman et al., 2008; Monterosso et al., 2007; McClure, Laibson, Loewenstein, & Cohen, 2004). Activity in frontal brain regions, such as the ventro- and dorsolateral pFC, the dorsal ACC, as well as parietal brain regions, like the intraparietal sulcus and the posterior parietal cortex, have been shown to be involved in this process (e.g., Basten, Biele, Heekeren, & Fiebach, 2010). This comparison process is based on the valuation of different choice options but is not directly involved in subjectively evaluating the options per se (see Rushworth, Kolling, Sallet, & Mars, 2012, for a review). In this manner the comparison process may be characterized as value independent. In line with this, Liu, Feng, Wang, and Li (2012) showed a dissociation of the valuation and decision process.

Focusing more specifically on the rate of temporal discounting, it has been shown that steeper discounting in adults was related to hypoactivity of the VS when processing the value of delayed rewards (Ballard & Knutson, 2009). Recently, we replicated this association in adolescents as well as in adults (Ripke et al., 2012) and did not observe differences between both age groups. Thus, interindividual differences in value-dependent processing and temporal discounting may be more related to other factors of individual differences than chronological age. One possible factor might be intelligence. BOLD responses in cortical (e.g., perigenual ACC, IFG, and ventromedial pFC) and subcortical (e.g., VS) regions involved in value processing may mediate the association between intelligence and the discounting rate. The IFG might be involved not only in value-dependent processing but also in inhibition during intertemporal choices. It has been shown that the IFG as part of the pFC is linked to the inhibition of prepotent responses (Gan et al., in press; Aron, Robbins, & Poldrack, 2004). Moreover, it has been proposed that the IFG inhibits the prepotent immediate choice during intertemporal decisions to thoroughly evaluate different choice alternatives (Ballard & Knutson, 2009). Along the same line, greater BOLD responses in the IFG have been observed during decisions for delayed rewards (Luo, Ainslie, Pollini, Giragosian, & Monterosso, 2012) and in individuals who showed lower discounting rates (Liu et al., 2012). Similarly, the ventromedial pFC has been shown to integrate magnitude and delay information (i.e., value) and to be less activated in more impulsive participants in conditions with longer delays (Ballard & Knutson, 2009). Thus, the ventromedial pFC might be another region in the frontoparietal network that is important for the association between temporal discounting and intelligence.

Besides the discounting rate, the consistency of choices is a second behavioral parameter of intertemporal choice. Consistency in intertemporal choice refers to the degree to which participants consistently select the alternative with the higher subjective value over the time course of the task. Low consistency of choices might be because of attentional fluctuations or limited WM capacity. Recently, we demonstrated that high consistency of intertemporal choices is associated with higher BOLD responses in a frontoparietal brain network as well as with completed brain maturation after adolescence (Ripke et al., 2012). Furthermore, the consistency in behavioral responses of choice RT tasks increases during child development (Li et al., 2004) and is associated with the maturation of performance monitoring (Li, Hammerer, Muller, Hommel, & Lindenberger, 2009) as well as dopamine D2 receptor binding in ACC in adults (MacDonald, Cervenka, Farde, Nyberg, & Backman, 2009).

Neural Correlates of Intelligence and Their Relation to Intertemporal Choice

Proposing neural correlates of intelligence, the parieto-frontal integration theory (Jung & Haier, 2007) suggests that a network consisting of the inferior (BA 39, BA 40) and superior parietal lobes (BA 7) interacts with the frontal regions (i.e., dorsolateral pFC, BA 6, BA 9, BA 10, BA 45, BA 46, BA 47) to serve the comparison of various solutions during reasoning and problem solving. ACC (BA 32) then engages in facilitating the selected responses and inhibiting the nonselected responses. This theory was further supported by studies showing that higher scores in measures of intelligence were related to higher gray matter volume in these areas (Colom et al., 2009; Karama et al., 2009; Johnson, Jung, Colom, & Haier, 2008). Recently, higher cortical thickness of the superior frontal cortex has been proposed as common neural basis of intelligence and impulsivity (Schilling et al., 2013) indicating a potential overlap in brain networks underlying both behaviors.

Candidate processes that link intertemporal choice behavior and intelligence are executive functions (EFs). For example, WM capacity might contribute to more consistent decisions, because it is necessary to have a distinctive representation of the decision criterion through the active maintenance of goal-directed information during the task to decide consistently (Shamosh & Gray, 2008). Furthermore, WM load has been shown to be related to more impulsive choices (Hinson, Jameson, & Whitney, 2003). Besides WM, other EFs, such as inhibition, executive control, and updating, are also important processes that might link intelligence and discounting behavior. These processes have been shown to be related to intelligence (Diamond, 2013). Moreover, EFs have been shown to be related to neural activity in a frontoparietal network (e.g., problem solving or WM) using PET (Duncan et al., 2000) or fMRI (Waiter et al., 2009). Furthermore, lesions in the ventromedial pFC have been shown to lead to insensitivity to future consequences during decisions that were therefore guided by immediate prospects only (Bechara, Damasio, Damasio, & Anderson, 1994).

In summary, the frontoparietal brain network (i.e., lateral pFC, dorsal ACC, intraparietal sulcus, and posterior parietal cortex) has been shown to be functionally associated with intelligence, and EFs have also been shown to be involved in intertemporal choices (Hoffman et al., 2008; Monterosso et al., 2007; McClure et al., 2004). This is not surprising, because online maintenance and integration of information are essential to decide between choice alternatives. Moreover, higher BOLD responses in these regions have been shown to be related to higher consistency of intertemporal choices (Ripke et al., 2012). Thus, it is very likely that the relation between intelligence and behavioral measures of intertemporal choice arise from the value-dependent and value-independent decision processes of intertemporal choice, because they share overlapping brain networks with intelligence and EFs, such as WM, inhibition, executive control, and updating. However, to date, no study has yet directly tested whether individual differences in intelligence are related to BOLD responses in brain circuitries that are involved in processes of intertemporal decision-making.

Study Aim and Hypotheses

The aim of the current study was to explore the neural basis for the associations between intelligence and behavioral parameters of intertemporal choice. In addition to expecting a negative correlation between intelligence and the rate of temporal discounting, we further expected a positive correlation between intelligence and consistency of intertemporal choices. Central to the study, we investigated potential shared neural substrates underlying intelligence and intertemporal choice.

As to the processing of reward valuation, we expected BOLD differences in the VS, where brain activity has been shown to be inversely related to the discounting rate, to be associated with intelligence as well. Furthermore, we also hypothesized correlations between individual differences in intelligence and BOLD responses in prefrontal cortical brain regions of value-dependent processing. For example the IFG has been shown in previous studies to be involved in inhibiting prepotent responses. Thus, a higher value signal in the IFG might be associated with stronger inhibition of choosing the immediate reward.

In terms of neural correlates regarding the value-independent decision process, we hypothesized that BOLD responses in frontoparietal regions relevant for decision processes during intertemporal choice would be related to behavioral parameters of intertemporal choice as well as intelligence.

We investigated these hypotheses in three steps. First, we investigated whether intelligence is related to BOLD responses during the value-dependent and value-independent processing of intertemporal choice. Second, we tested whether the behavioral measures of intertemporal choice (i.e., consistency of choices and discounting rate) are related to BOLD responses in the brain areas, which are associated with individual differences in intelligence. Third, we tested whether brain activity in these areas might mediate the association between intelligence and the behavioral measures of intertemporal choice.

METHODS

Sample

The data acquisition was part of the project “The Adolescent Brain,” which is funded by the German Federal Ministry of Education and Research. The project aims to investigate structural and functional brain development in the context of environmental and genetic factors from a longitudinal perspective and is related to the IMAGEN project, funded by the European Commission (Ripke et al., 2012; Schumann et al., 2010).

Two hundred sixty adolescent participants and one of their legal guardians signed informed consent and were invited to take part in the study. Of these participants, 235 (122 boys, mean age = 14.6, SD = 0.3, min = 13.7, max = 15.5) participated in the two parts of the intertemporal choice task (training session and imaging session). They received monetary compensation for their participation in the study and additional payments depending on their choices during the task, which ranged from €5 to €35.

Participants with disorders, including substance use disorder, were excluded from the sample (n = 4) based on ratings from the Development and Well-Being Assessment (Goodman, Ford, Richards, Gatward, & Meltzer, 2000). All participants had normal or corrected-to-normal vision. Seven of the adolescent data sets were incomplete because of technical difficulties during the imaging session and the data of 10 further adolescents were excluded because of signal drop out. Additional eight participants were excluded because of lack of intelligence measures. After these exclusions, the final data analysis included 206 participants.

Intertemporal Choice Task

The intertemporal choice task used in our study consisted of two sessions. First, a training session was performed outside the MRI scanner. This session served to estimate individual discounting rate as well as to train the participant for the subsequent imaging session. To give the participants time to understand the task and response procedure, the first three trials in the training session were not included in the behavioral analysis. After these three trials, each participant underwent 50 trials. In each of the trials, the participants had to choose either a small immediate amount of money or a larger amount of money to be paid after a delay. Before the task started, we instructed the participants that the immediate amount would be €20 in every trial. At the beginning of each trial, participants saw the amount of money and the duration of the delay (10, 30, 60, 120, or 180 days) of the delayed reward only. After 2 sec, participants had to indicate their preference by pressing either the left button (for the later alternative) or the right button (for the immediate option). Directly after their response, participants received feedback on the amount and delay chosen (either of the immediate or later reward) to ensure that participants could monitor their decisions. There were 10 trials for each of the five delays. After these 10 trials, the delay changed for the next 10 decisions. The training session was adaptive, that is, the amounts of money displayed increased or decreased based on the participant's decision in the previous trial. For instance, if the immediate amount was chosen, for the next trial the delayed amount increased by 50% of the difference between immediate and delayed reward. The delayed reward decreased by the same amount when it was chosen in the trial before. On the basis of the choices in these 50 trials, we estimated the individual discount parameter k. First, we estimated the indifference amount for each of the five delays, that is, the mean of the maximum delayed amount rejected and the minimum delayed amount chosen. The indifference amount represents the quantity (A) in the hyperbolic function:
formula

Furthermore, V represents the subjective value of a reward (during our training session, it was €20) in this equation, D represents the delay, and k represents the individual discount parameter. Using ordinary least square estimation, parameter k was estimated to best fit the hyperbolic curve consisting of 6 points, that is, €20 for a delay of 0 (immediate reward) and the indifference amounts of the five delays. We used a hyperbolic function, because previous studies found that it best fits the data (Simpson & Vuchinich, 2000; Kirby & Marakovic, 1995; Mazur, 1987).

We adapted the delayed rewards presented during the imaging session of the task to the individual discounting rate (k) in a way that (i) participants ought to choose the delayed reward in 50% of trials, according to the procedure of Kable and Glimcher (2007); (ii) the mean subjective value (V) of all delayed rewards would be the same (€30) for each participant; and (iii) the maximal value of all rewards was 20-fold the minimum value. Amounts and delays were computed in advance and presented in random order during the imaging session of the experiment. To meet all of the above criteria, we computed the amounts based on an immediate reward of 1 and multiplied all amounts by an individualized factor such that the mean of all delayed rewards was 30 Euro. This factor, depending on the individual k, was used to compute the individual immediate reward for each participant. Therefore, the immediate reward differed from that of the training session (€20) but was the same in all trials of the scanning session for one participant. Between participants, it ranged between €5 and €20. For further information regarding the adaptation, formulas, and sample sets of delayed rewards, see Ripke et al. (2012). Our aim was to make sure that possible k-related BOLD differences between participants did not simply depend on individual subjective value ranges of delayed rewards. Because of the formula of the hyperbolic function, keeping the immediate reward constant between participants would lead to different subjective value ranges and vice versa. This was the reason to use individual immediate rewards.

Participants were informed about the amount of the immediate reward right before the imaging session of the experiment started. In each trial, the delayed amount and the respective delay were presented for 2 sec (Figure 1). After a further period of 6 sec during which a fixation cross was displayed, the participants were prompted to respond within a response timeframe. An exclamation mark on the left or right side of the screen indicated to the participant which button was mapped onto the delayed reward option (the immediate option was mapped to the alternative button). To avoid lateralization effects of response, in 50% of the trials the decision for the delayed reward was mapped to the right button and in 50% of the trials to the left one.

Figure 1. 

Time course of one trial.

Figure 1. 

Time course of one trial.

Each trial of the experiment ended with a display of the participants' decision (i.e., either the delayed amount and delay or the immediate reward) followed by an intertrial interval with duration of 7 sec on average (range = 6–8 sec; uniform distribution). The whole session (90 trials) lasted 25 min. The participants were told that one of their choices would be randomly selected and paid later (in 10, 30, 60, 120, or 180 days) via bank transfer or immediately after scanning. We integrated this procedure to ensure that participants made realistic choices and to increase task relevance (Zink, Pagnoni, Martin-Skurski, Chappelow, & Berns, 2004).

Intelligence Measure

For assessing intelligence, we administered four subtests of the German version of the Wechsler intelligence scale for children and adolescents (Daseking, Petermann, & Petermann, 2007). The subtests included vocabulary, similarities, block design, and matrix reasoning. For the analysis in the current study, we computed a mean unit-weighted score of general intelligence (g) based on the z scores of the four subtests.

fMRI Data Acquisition

Scanning was performed with a 3T whole-body MR tomograph (Magnetom TRIO, Siemens, Erlangen, Germany) equipped with a standard head coil. For functional imaging, a standard EPI sequence was used (repetition time = 2410 msec, echo time = 25 msec, flip angle = 80°). fMRI scans were obtained from 42 transversal slices, orientated 30° clockwise to the AC–PC line, with a thickness of 2 mm (1 mm gap), a field of view of 192 × 192 mm2, and an in-plane resolution of 64 × 64 pixels, resulting in a voxel size of 3 × 3 × 3 mm3. To exclude structural abnormalities, a 3-D T1-weighted MP-RAGE image data set was acquired (repetition time = 1900 msec, echo time = 2.26 msec, field of view = 256 × 256 mm2, 176 slices, 1 × 1 × 1 mm3 voxel size, flip angle = 9°). Images were presented via NNL goggles (Nordic Neurolab, Bergen, Norway). Task presentation and recording of the behavioral responses were performed using Presentation software (version 11.1, Neurobehavioral Systems, Inc., Albany, CA).

Behavioral Data Analysis

We estimated two different behavioral parameters, the discount parameter k and the consistency of choices. Consistency of choices here denotes the degree to which participants consistently chose the alternative with the higher subjective value. To compute the consistency parameter, we ran a receiver operating characteristics curve analysis with subjective values of the delayed reward as a predictor for the respective choice. For each participant, we computed the area under the curve (AUC) from the data of the scanning session, which was supposed to be higher for more consistent participants (i.e., always choosing the reward with the higher value results in an AUC of 1; complete randomness of choices would yield an AUC of 0.5). Statistical testing and determination of AUC was computed with SPSS 19. The discounting parameter k for the imaging session was computed applying the fitting procedure described for data of the training session using MATLAB 7.1. Because AUC and k were not normally distributed, we log-transformed k and rank-transformed the AUC. The significance level for all statistical testing was set at α = 5% (two-tailed).

In light of findings suggesting that intelligence during adolescence is, in part, related to schooling (Cliffordson & Gustafsson, 2008; Ceci, 1991), we also tested whether environmental factors that are associated with school performance, that is, participants' sociodemographics (Holmlund, Lindahl, & Plug, 2011), might be related to intelligence and behavioral measures of intertemporal choice. Specifically, we analyzed the association between parents' educational attainment (mean of mother's and father's educational attainment), g, discounting rate log k, and consistency of choices (rank AUC).

MRI Data Analysis

We analyzed functional MRI data using statistical parametric mapping (SPM 5, Wellcome Department of Neuroimaging, London, United Kingdom, www.fil.ion.ucl.ac.uk/spm). For preprocessing, data were corrected for temporal differences in scan time to minimize temporal differences in slice acquisition and interscan head motions over course of the session. The scans were then normalized to the standard EPI template (MNI) and finally smoothed using an isotropic Gaussian kernel (8 mm FWHM).

Our first level model consisted of five regressors for different events and parametrical modulations. The first regressor contains the onsets of all presentations of the delayed rewards, and each onset was parametrically modulated by the amount of the delayed reward (second regressor). The amount was chosen as value representation to preserve differences in reward-related processing related to subjective discounting for later analysis steps (mediation). Following our hypothesis that these processes run in parallel, the Regressors 1 and 2 capture the decision-related processing. The first regressor captured the value-independent processing and the second regressor the value-dependent (amount-related) processing of the decision. The subsequent motor responses/feedbacks (separated for responses with the left and right hand) were modeled as third and fourth regressor. Trials with implausible decisions, for a reward with a subjective value lower than half of the alternative reward, and trials without response (missing trial) were regarded as invalid. This was done to make sure that only trials in which participants decided properly were included into the analysis. The number of excluded trials was very low (median missings (Mdmissings) = 1, interquartile range (IQRmissings) = 2; Mdinvalid = 2, IQRinvalid = 3). Invalid trials were modeled as the fifth regressor. For trials with implausible choices, the subsequent motor response was modeled using the same regressor as in valid trials. All events (zero duration) were modeled using the canonical hemodynamic response function. To alleviate the effects of participants' movement, we integrated the six realignment parameters (three translation and three rotation parameters) as regressors of no interest.

As we hypothesized that associations between individual differences in intelligence and behavior during intertemporal choice would rely on shared brain activation, we expected the overlap in brain networks relevant for intertemporal choice. Therefore, the first step of our second level analyses focused on regions that showed main effects of BOLD responses with respect to value-independent and value-dependent decision processing. Specifically, for the second level analysis, we set up two regression models with the covariate g and BOLD responses capturing value-dependent processing and value-independent decision processing as dependent variables. This was done to identify brain networks in which the BOLD response was related to g during either value-dependent or independent processing. The ROIs for these analyses were the main effects of value-dependent (amount) and value-independent decision processing, respectively. For details regarding the mask of our ROIs, see Figure 2 (A: value-related BOLD and B: decision-related BOLD). All reported results regarding the regressions reached the threshold of p < .05, false discovery rate (FDR) corrected within at least 25 contiguous voxels.

Figure 2. 

(A) Value-dependent BOLD signal (main effect, threshold t = 3.0, p < .05, FDR-corrected in 25 contiguous voxels). (B) Value-independent BOLD signal (main effect, threshold t = 4.89, p < .05, FDR-corrected in 25 contiguous voxels). ACC = dorsal ACC; dlPFC = dorsolateral pFC; vmPFC = ventromedial pFC; PCC = posterior cingulate cortex; BA = Brodmann's area.

Figure 2. 

(A) Value-dependent BOLD signal (main effect, threshold t = 3.0, p < .05, FDR-corrected in 25 contiguous voxels). (B) Value-independent BOLD signal (main effect, threshold t = 4.89, p < .05, FDR-corrected in 25 contiguous voxels). ACC = dorsal ACC; dlPFC = dorsolateral pFC; vmPFC = ventromedial pFC; PCC = posterior cingulate cortex; BA = Brodmann's area.

The second step was to test whether the BOLD responses within the networks found to be associated with intelligence by the above-described regression analyses, would, in turn, also be associated with the temporal discounting rate (log k) and consistency of behavioral choices (rank AUC). To obtain this, we correlated the mean BOLD signal changes from functional ROIs, which were binary masks of the respective activation map (g-related BOLD; see Results section). For the correlation analyses, significance level was set at α = 5% (two-tailed).

Because we hypothesized that the association between g and the behavioral measures of intertemporal choice might be mediated by BOLD response in the respective ROIs, we ran path analyses using SPSS AMOS 17 (Analysis of Moment Structures; IBM Corporation, Somers, NY) to test our model. The path model (Figure 3C and D) comprised connections between g, the behavioral measures of intertemporal choice as manifest variables, and the BOLD response in the networks identified, which were correlated with the respective behavioral measures as latent variable. The BOLD responses of each network were used as one common latent variable, because within each network the BOLD responses were highly correlated whereas between the networks they were not. This fact was hinting at two different networks rather than seven single ROIs. In each of the two path models, g had direct connections to the respective behavioral measure (1. log k and 2. rank AUC) and to BOLD responses inside the respective network (1. value-dependent decision network, 2. value-independent decision network). Furthermore, g was indirectly connected to the behavioral measures through the respective BOLD response. Significance for all relationships was determined at p < .05. Model fits were assessed by χ2 goodness-of-fit tests and by root mean square error of approximations, which are two common indices for deviance of the model from the data.

Figure 3. 

(A) Value-dependent BOLD response, which is associated with general intelligence (threshold t = 2.18, p < .05, FDR corrected). (B) Value-independent BOLD response, which is associated with general intelligence (threshold t = 3.50, p < .05, FDR corrected). (C) Path model regarding the mediation of the association between g and the discount rate (log k) through BOLD in the g-related value-dependent network. (D) Path model regarding the mediation of the association between g and the choice consistency (rank AUC) through BOLD in the g-related value-independent network. Asterisks indicate significant effects. ACC = perigenual ACC; dlPFC = dorsolateral pFC; vmPFC = ventromedial pFC; log k = discount rate; rank AUC = choice consistency.

Figure 3. 

(A) Value-dependent BOLD response, which is associated with general intelligence (threshold t = 2.18, p < .05, FDR corrected). (B) Value-independent BOLD response, which is associated with general intelligence (threshold t = 3.50, p < .05, FDR corrected). (C) Path model regarding the mediation of the association between g and the discount rate (log k) through BOLD in the g-related value-dependent network. (D) Path model regarding the mediation of the association between g and the choice consistency (rank AUC) through BOLD in the g-related value-independent network. Asterisks indicate significant effects. ACC = perigenual ACC; dlPFC = dorsolateral pFC; vmPFC = ventromedial pFC; log k = discount rate; rank AUC = choice consistency.

RESULTS

Behavioral Results

The median of k was Mdk = .029 with an interquartile range of IQRk = .038 and the median of the AUC was MdAUC = .936 with an interquartile range of IQRAUC = .07. The discounting rate k and AUC were not correlated (r(204) = −.076, p = .278). We assessed intelligence (g) as unit-weighted score based on four z-transformed subtests of the WISC (SD = 2.90, range = −9.32 to 6.04). As expected, log k was negatively correlated with intelligence (g: r(204) = −.237, p = .001), whereas the rank AUC was positively correlated with intelligence (g: r(204) = .171, p = .014).

The correlation analysis of the associations between parents' educational attainment and the adolescents' g, log k, and rank AUC revealed positive correlations between g and parents' educational attainment (r(204) = .460, p < .001). The temporal discounting rate (log k) was negatively correlated with educational attainment (r(204) = −.230, p = .001), whereas consistency of choices (rank AUC) was not correlated to parents' education. After controlling for g, the correlation between parental education and the discounting rate was attenuated but still significant (rs(204) = −.167, p = .019). Statistical testing between the zero-order correlation and the partial correlation was based on the comparison of the Fisher Z-transformed correlation coefficients using z statistics as proposed by Olkin and Finn (1995) and revealed no significant differences (z = −1.23, p = .109).

Imaging Results

Neural Correlates of Intelligence during Value-dependent Decision Processing

The regression analysis revealed a frontal network consisting of the perigenual ACC, the IFG, and the medial pFC where higher BOLD responses were related to higher g. Furthermore, a positive correlation between the BOLD response and g was also detected in the VS. These four brain regions were used for further analyses as g-related value-dependent decision network (Figure 3A).

Neural Correlates of Intelligence during Value-independent Decision Processing

In line with previous findings, the regression analyses testing whether BOLD responses in the value-independent decision network would be related to g revealed positive correlations between g and the BOLD response in the precuneus (BA 7) as well as in the dorsolateral pFC (BA 9) and the occipital lobe (BA 18, BA 19, BA 31). These areas were used as functional ROIs for further correlational analyses as g-related value-independent decision network (Figure 3B).

Correlations of BOLD Responses in Functional ROIs Related to Intelligence with Discounting Rate and Consistency

The correlations between the behavioral parameters of intertemporal choice and BOLD response were tested in different functional ROIs. These ROIs were exactly the regions where BOLD responses were related to g. Functional ROIs of the value-dependent decision network included the perigenual ACC, the IFG, the ventromedial pFC, and the VS as separate functional ROIs. For the value-independent decision network, these regions were the precuneus and adjacent parietal region of the inferior parietal cortex (BA 7), the dorsolateral pFC (BA 9) and the occipital areas of BA 18 and BA 19.

The analysis revealed that value-related BOLD responses in the perigenual ACC, the IFG, the ventromedial pFC, and the right VS was inversely correlated to log k. There was no association between value-related BOLD responses in the perigenual ACC, the IFG, the ventromedial pFC, and the VS and choice consistency (rank AUC; Table 1). Consistency of choices (rank AUC) was positively correlated with BOLD responses in the dorsolateral pFC, the precuneus, and the occipital lobe during decisions. There was no association between log k and the decision-related BOLD response (Table 2).

Table 1. 

Associations between Behavioral Parameters of Intertemporal Choice and Value-dependent BOLD Response inside Intelligence g-related ROIs

ROIPeak CoordinatesT maxBehavioral MeasureCorrelations
xyz
Perigenual ACC (BA 32) 18 30 3.22 log k r = −.173 (p = .013) 
rank AUC r = .049 (p = .486) 
IFG −39 21 −12 3.69 log k r = −.199 (p = .004) 
rank AUC r = .074 (p = .288) 
Ventromedial pFC −9 33 −3 3.73 log k r = −.189 (p = .007) 
rank AUC r = .043 (p = .541) 
VS 
 Right 18 3.01 log k r = −.246 (p < .001) 
rank AUC r = .082 (p = .242) 
 Left −15 12 −6 3.17 log k r = −.262 (p < .001) 
rank AUC r = .134 (p = .055) 
ROIPeak CoordinatesT maxBehavioral MeasureCorrelations
xyz
Perigenual ACC (BA 32) 18 30 3.22 log k r = −.173 (p = .013) 
rank AUC r = .049 (p = .486) 
IFG −39 21 −12 3.69 log k r = −.199 (p = .004) 
rank AUC r = .074 (p = .288) 
Ventromedial pFC −9 33 −3 3.73 log k r = −.189 (p = .007) 
rank AUC r = .043 (p = .541) 
VS 
 Right 18 3.01 log k r = −.246 (p < .001) 
rank AUC r = .082 (p = .242) 
 Left −15 12 −6 3.17 log k r = −.262 (p < .001) 
rank AUC r = .134 (p = .055) 

log k = discounting rate; rank AUC = choice consistency.

Table 2. 

Associations between Behavioral Parameters of Intertemporal Choice and Value-independent BOLD Response inside Intelligence g-Related ROIs

ROIPeak CoordinatesT maxBehavioral MeasureCorrelations
xyz
Precuneus (BA 7) −66 54 5.67 log k r = −.020 (p = .780) 
rank AUC r = .231 (p = .001) 
Dorsolateral pFC (BA 9) 
 Right 39 51 27 4.35 log k r = −.044 (p = .528) 
rank AUC r = .138 (p = .048) 
 Left −39 48 24 3.58 log k r = −.052 (p = .457) 
rank AUC r = .134 (p = .056) 
Occipital lobe (BA 18/BA 19) −72 5.74 log k r = −.036 (p = .612) 
rank AUC r = .208 (p = .003) 
ROIPeak CoordinatesT maxBehavioral MeasureCorrelations
xyz
Precuneus (BA 7) −66 54 5.67 log k r = −.020 (p = .780) 
rank AUC r = .231 (p = .001) 
Dorsolateral pFC (BA 9) 
 Right 39 51 27 4.35 log k r = −.044 (p = .528) 
rank AUC r = .138 (p = .048) 
 Left −39 48 24 3.58 log k r = −.052 (p = .457) 
rank AUC r = .134 (p = .056) 
Occipital lobe (BA 18/BA 19) −72 5.74 log k r = −.036 (p = .612) 
rank AUC r = .208 (p = .003) 

log k = discounting rate; rank AUC = choice consistency.

Path Analysis to Test the Mediation of the Effect of g on Intertemporal Choice Behavior (log k and rank AUC) through BOLD Responses in the Value-dependent and -independent Decision Network

Resulting from the correlation analyses, we identified two g-related networks relevant for the behavior during intertemporal choices: (1) the value-dependent network consisting of the perigenual ACC, the IFG, the ventromedial pFC, and the VS and (2) the value-independent network consisting of the dorsolateral pFC, the precuneus, and the occipital lobe. The two path models that tested the mediation used the BOLD inside the respective network as latent variable.

The path analysis revealed that the association between g and k was partially mediated by the BOLD response in the network of ACC, the VS, the ventromedial pFC, and the IFG. Both direct and indirect effects of g on k were significant (Figure 3C), indicating that at least a portion of the association between g and k is mediated by the activity in these brain regions. The χ2 goodness-of-fit test and the root mean square error of approximation (RMSEA), two common indices for deviance of the model from the data, indicated that our data fit the model well (χ2 = 3.882; df = 8; p = .868; RMSEA = 0.000). For the association between g and the AUC, the path analysis revealed that BOLD responses in the dorsolateral pFC, the precuneus, as well as the occipital lobe fully mediated the association between g and AUC. The direct effect of g on AUC was no longer significant, whereas the indirect effect was (Figure 3D). Again, the model fit indices (χ2 = 1.604, df = 4, p = .808, RMSEA = 0.000) indicated that our data fit the model very well.

DISCUSSION

In this study, we found direct associations between the neural response during intertemporal choice and intelligence. Higher intelligence was associated with higher activation during value-dependent processing in a frontal network consisting of the perigenual ACC, the IFG, and the ventromedial pFC as well as the VS (value-dependent network). Importantly, BOLD responses in these regions were negatively correlated with the temporal discounting rate. Additionally, we found that higher intelligence was also related to higher BOLD in the dorsolateral pFC, the precuneus, and the occipital lobe (value-independent network) during the value-independent decision processing, which in turn was positively correlated with the consistency of choices. Path analyses testing direct and indirect effects of g on the behavioral measures of intertemporal choice indicated that the associations between intelligence and temporal discounting or consistency of choices are at least partly mediated by activation in the respective brain network.

The first process essential for intertemporal choice is the value-dependent decision processing (Peters & Buchel, 2010; Kable & Glimcher, 2007). Our results revealed that a higher BOLD response in a network consisting of the perigenual ACC, the IFG, the ventromedial pFC, and the VS, while processing the value of delayed rewards, was positively related to general intelligence. Moreover, higher BOLD signals in this value-dependent network were also negatively related to the discounting rate. This implicates a possible neural mechanism mediating the link between intelligence and discounting. The association between higher IFG activation and lower discounting has been shown before by Liu and Feng (2012). This effect and earlier findings of higher BOLD in the IFG being related to the inhibition of prepotent behavior (Swick, Ashley, & Turken, 2011; Aron et al., 2004) and during intertemporal choice (Ballard & Knutson, 2009) strengthen the hypothesis that higher BOLD signal in the IFG might be a possible link between higher intelligence scores and lower discounting of delayed rewards. Negative correlations between intelligence and discounting were observed in several behavioral studies (Freeney & O'Connell, 2010; Shamosh & Gray, 2008; Shamosh et al., 2008; de Wit et al., 2007). Regarding the ventromedial pFC, its role for the integration of magnitude and delay information has been proposed by Ballard and Knutson (2009). Furthermore, decisions involving the evaluation of future consequences have been shown to be impaired in patients with ventromedial pFC lesions (Bechara et al., 1994). Thus, inhibition processes in the IFG and better integration of future consequences in the ventromedial pFC might be possible processes that make delayed reward options more appealing to more intelligent participants. Also related to value-based decision-making, it has been shown that the VS not only represents value information, but also updating information about prediction errors and thus guides decisions (Rolls, McCabe, & Redoute, 2008). Additionally, it has been shown that the prediction error-related BOLD signal in the VS was positively correlated with fluid intelligence (Schlagenhauf et al., 2013). In line with these findings regarding the VS, our data revealed that BOLD responses in the VS, which represent the subjective value of delayed rewards during intertemporal choice (Kable & Glimcher, 2007), was positively correlated with intelligence and negatively with discounting of delayed rewards. The notion that these neural processes might be the link between intelligence and behavior was further supported by a path analysis of our data. This analysis showed that the BOLD signal in ACC, IFG, ventromedial pFC, and VS mediated, at least partly, the association between intelligence and the discounting rate.

The second process investigated in the current study is the value-independent decision processing. The positive correlation between the decision-related BOLD response in the dorsolateral pFC (BA 9/BA 10) and the precuneus and intelligence is consistent with studies that found greater brain activity in these areas during reasoning or WM tasks (Waiter et al., 2009; Duncan et al., 2000) and a positive association between WM and intelligence (Diamond, 2013). Recently, it has been shown that decision-making becomes less strategic when WM capacity is restricted by a demanding secondary task (Otto, Gershman, Markman, & Daw, 2013). Thus, WM abilities might be the link between choice behavior and intelligence. For example, to make consistent intertemporal choices the evaluation algorithm has to be kept in WM during the duration of the task. Indeed, in our study intelligence was positively related to the consistency of choices and to BOLD response in the frontoparietal network consisting of the dorsolateral pFC (BA 9/BA 10), the precuneus (BA 7), and areas in the occipital lobe (BA 18/BA 19). Occipital areas have been shown to be related to visual WM (Xu & Chun, 2006). Together with the frontoparietal network, the occipital lobe has been implicated in visual memory, especially during multitasking (Deprez et al., 2013), higher task demands, and attention (Nelissen, Stokes, Nobre, & Rushworth, 2013). Thus, the occipital areas, identified in our data might contribute to higher consistency of choices. The path that tested whether the association between intelligence and consistency is mediated by neural activity in the value-independent decision network revealed a full mediation. Thus, the higher BOLD response during intertemporal choices mediates the behavioral correlation between intelligence and decision consistency.

The cluster in the dorsolateral pFC included BA 10, a region that was also identified by Shamosh et al. (2008). In line with the interpretation of Shamosh et al., the BA 10 is part of the value-independent decision network and the BOLD might reflect integration of information for the subsequent decision. Our findings show that higher choice consistency is associated with higher BOLD response in this area. However, we did not find a relation between BOLD activity in BA 10 and the discounting rate. This difference might be related to the analysis of our data, because we separated value-independent decision processing, which was related to consistency of choices and value-related decision processing, which was related to the discounting rate.

Intellectual abilities in adolescents are related to schooling (van Tuijl & Lesernan, 2007; Ceci & Williams, 1997; Ceci, 1991) and the parents' educational attainment (Bulut, 2013). The results of our study might hint at a possible mechanism how education and further contextual factors in the adolescent's developmental context influence temporal discounting possibly mediated by general intelligence. In our data, parents' educational level was positively correlated to their children's general intelligence and negatively to the discounting rate, but not to consistency of choices. After controlling for general intelligence the effect of parental education on the adolescents' discounting rates was attenuated but still remained significant. Further indirect support for this line of reasoning comes from several studies that showed associations between education in school and delay discounting in different populations (Bauer & Chytilova, 2010; Reimers, Maylor, Stewart, & Chater, 2009; Harrison, Lau, & Williams, 2002; Kirby et al., 2002). Because substance use has been shown to be related to steeper discounting (Bickel et al., 2007), one might speculate that better education at school and at home might be related to less impulsive behavior and hence to lower risk for substance use. Speculating about the mediating role of intelligence, higher education, and a supportive environment might be protecting factors, which both have an effect on the discounting rate and subsequently on substance use behavior.

Lastly, a few caveats of our study should be discussed. First, our sample only consists of adolescents. But, as mentioned in the outset, it is of special interest to understand the associations between intelligence, temporal discounting, and neural processing in adolescence. In light of the results of our former study indicating differences between adults and adolescents only during value-independent decision processing, one might expect that the association between individual differences in intelligence and reward processing could also be found in adults. Thus, the proposed protective role of intelligence regarding impulsive behavior or substance abuse may also be important for later periods in life. Future studies with adult samples are necessary to further investigate these associations. A second limitation of our study is that the correlations between intelligence and the behavioral measures represent, though significant, only small to moderate effects, although the effect sizes of our findings are in line with previous findings (e.g., Shamosh & Gray, 2008, showed a mean effect of r = −.23 in their meta-analysis).

In conclusion, this study provides direct evidence for common brain networks underlying individual differences in temporal discounting rate, consistency of choices, and intelligence. Stronger BOLD responses in a value-dependent decision network consisting of ACC, IFG, ventromedial pFC, and VS as well as in a value-independent decision network of the dorsolateral pFC, the precuneus, and the occipital lobe may mediate the positive relations between a higher level of intelligence, lower discounting rates, and more consistent choices. Furthermore, the integrity of the common brain networks identified here and their interactions with other environmental factors (e.g., parental educational resources) may play important roles in impulsive behavior.

Acknowledgments

The authors thank the radiographers Romy Bänsch, Silke Geißler, Kerstin Raum, and Veronika Ziesch and our student workers for assistance with data collection. The study was supported by the Bundesministerium für Bildung und Forschung (BMBF, grant 01EV0711), the Deutsche Forschungsgemeinschaft (DFG, grants SM 80/5-2, SM 80/7-1, and SFB 940/1), and the European Sixth Framework Programme (LSHM-CT-2007-037286).

Reprint requests should be sent to Michael N. Smolka or Stephan Ripke, Section of Systems Neuroscience, Department of Psychiatry and Psychotherapy, Technische Universität Dresden, Würzburger Str. 35, 01187 Dresden, Germany, or via e-mail: michael.smolka@tu-dresden.de, stephan.ripke@tu-dresden.de.

REFERENCES

Ainslie
,
G.
(
1975
).
Specious reward: A behavioral theory of impulsiveness and impulse control.
Psychological Bulletin
,
82
,
463
496
.
Aron
,
A. R.
,
Robbins
,
T. W.
, &
Poldrack
,
R. A.
(
2004
).
Inhibition and the right inferior frontal cortex.
Trends in Cognitive Sciences
,
8
,
170
177
.
Ballard
,
K.
, &
Knutson
,
B.
(
2009
).
Dissociable neural representations of future reward magnitude and delay during temporal discounting.
Neuroimage
,
45
,
143
150
.
Basten
,
U.
,
Biele
,
G.
,
Heekeren
,
H. R.
, &
Fiebach
,
C. J.
(
2010
).
How the brain integrates costs and benefits during decision-making.
Proceedings of the National Academy of Sciences, U.S.A.
,
107
,
21767
21772
.
Bauer
,
M.
, &
Chytilova
,
J.
(
2010
).
The impact of education on subjective discount rate in Ugandan villages.
Economic Development and Cultural Change
,
58
,
643
669
.
Bechara
,
A.
,
Damasio
,
A. R.
,
Damasio
,
H.
, &
Anderson
,
S. W.
(
1994
).
Insensitivity to future consequences following damage to human prefrontal cortex.
Cognition
,
50
,
7
15
.
Bickel
,
W. K.
,
Miller
,
M. L.
,
Yi
,
R.
,
Kowal
,
B. P.
,
Lindquist
,
D. M.
, &
Pitcock
,
J. A.
(
2007
).
Behavioral and neuroeconomics of drug addiction: Competing neural systems and temporal discounting processes.
Drug and Alcohol Dependence
,
90
,
S85
S91
.
Bickel
,
W. K.
,
Odum
,
A. L.
, &
Madden
,
G. J.
(
1999
).
Impulsivity and cigarette smoking: Delay discounting in current, never, and ex-smokers.
Psychopharmacology
,
146
,
447
454
.
Bulut
,
S.
(
2013
).
Intelligence development of socio-economically disadvantaged pre-school children.
Anales de Psicologia
,
29
,
855
864
.
Casey
,
B. J.
,
Jones
,
R. M.
, &
Hare
,
T. A.
(
2008
).
The adolescent brain.
Annals of the New York Academy of Sciences
,
1124
,
111
126
.
Casey
,
B. J.
,
Tottenham
,
N.
,
Liston
,
C.
, &
Durston
,
S.
(
2005
).
Imaging the developing brain: What have we learned about cognitive development?
Trends in Cognitive Sciences
,
9
,
104
110
.
Ceci
,
S. J.
(
1991
).
How much does schooling influence general intelligence and its cognitive components? A reassessment of the evidence.
Developmental Psychology
,
27
,
703
722
.
Ceci
,
S. J.
, &
Williams
,
W. M.
(
1997
).
Schooling, intelligence, and income.
American Psychologist
,
52
,
1051
1058
.
Cliffordson
,
C.
, &
Gustafsson
,
J. E.
(
2008
).
Effects of age and schooling on intellectual performance: Estimates obtained from analysis of continuous variation in age and length of schooling.
Intelligence
,
36
,
143
152
.
Colom
,
R.
,
Haier
,
R. J.
,
Head
,
K.
,
Alvarez-Linera
,
J.
,
Quiroga
,
M. A.
,
Shih
,
P. C.
,
et al
(
2009
).
Gray matter correlates of fluid, crystallized, and spatial intelligence: Testing the P-FIT model.
Intelligence
,
37
,
124
135
.
Crone
,
E. A.
, &
Dahl
,
R. E.
(
2012
).
Understanding adolescence as a period of social-affective engagement and goal flexibility.
Nature Reviews Neuroscience
,
13
,
636
650
.
Daseking
,
M.
,
Petermann
,
U.
, &
Petermann
,
F.
(
2007
).
Assessment of intelligence with the HAWIK-IV.
Kindheit und Entwicklung
,
16
,
250
259
.
de Wit
,
H.
,
Flory
,
J. D.
,
Acheson
,
A.
,
McCloskey
,
M.
, &
Manuck
,
S. B.
(
2007
).
IQ and nonplanning impulsivity are independently associated with delay discounting in middle-aged adults.
Personality and Individual Differences
,
42
,
111
121
.
Deprez
,
S.
,
Vandenbulcke
,
M.
,
Peeters
,
R.
,
Emsell
,
L.
,
Amant
,
F.
, &
Sunaert
,
S.
(
2013
).
The functional neuroanatomy of multitasking: Combining dual tasking with a short term memory task.
Neuropsychologia
,
51
,
2251
2260
.
Diamond
,
A.
(
2013
).
Executive functions.
Annual Review of Psychology
,
64
,
135
168
.
Duncan
,
J.
,
Seitz
,
R. J.
,
Kolodny
,
J.
,
Bor
,
D.
,
Herzog
,
H.
,
Ahmed
,
A.
,
et al
(
2000
).
A neural basis for general intelligence.
Science
,
289
,
457
460
.
Freeney
,
Y.
, &
O'Connell
,
M.
(
2010
).
Wait for it: Delay-discounting and academic performance among an Irish adolescent sample.
Learning and Individual Differences
,
20
,
231
236
.
Gan
,
G.
,
Guevara
,
A.
,
Marxen
,
M.
,
Schwarz
,
D.
,
Neumann
,
M.
,
Jünger
,
E.
,
et al
(
in press
).
Alcohol-induced impairment of inhibitory control is linked to attenuated brain responses in right fronto-temporal cortex.
Biological Psychiatry
.
Goodman
,
R.
,
Ford
,
T.
,
Richards
,
H.
,
Gatward
,
R.
, &
Meltzer
,
H.
(
2000
).
The development and well-being assessment: Description and initial validation of an integrated assessment of child and adolescent psychopathology.
Journal of Child Psychology and Psychiatry and Allied Disciplines
,
41
,
645
655
.
Hare
,
T. A.
,
Camerer
,
C. F.
, &
Rangel
,
A.
(
2009
).
Self-control in decision-making involves modulation of the vmPFC valuation system.
Science
,
324
,
646
648
.
Harrison
,
G. W.
,
Lau
,
M. I.
, &
Williams
,
M. B.
(
2002
).
Estimating individual discount rates in Denmark: A field experiment.
American Economic Review
,
92
,
1606
1617
.
Hinson
,
J. M.
,
Jameson
,
T. L.
, &
Whitney
,
P.
(
2003
).
Impulsive decision making and working memory.
Journal of Experimental Psychology: Learning Memory and Cognition
,
29
,
298
306
.
Hoffman
,
W. F.
,
Schwartz
,
D. L.
,
Huckans
,
M. S.
,
McFarland
,
B. H.
,
Meiri
,
G.
,
Stevens
,
A. A.
,
et al
(
2008
).
Cortical activation during delay discounting in abstinent methamphetamine dependent individuals.
Psychopharmacology
,
201
,
183
193
.
Holmlund
,
H.
,
Lindahl
,
M.
, &
Plug
,
E.
(
2011
).
The causal effect of parents' schooling on children's schooling: A comparison of estimation methods.
Journal of Economic Literature
,
49
,
615
651
.
Holt
,
D. D.
,
Green
,
L.
, &
Myerson
,
J.
(
2003
).
Is discounting impulsive?. Evidence from temporal and probability discounting in gambling and non-gambling college students.
Behavioural Processes
,
64
,
355
367
.
Johnson
,
W.
,
Jung
,
R. E.
,
Colom
,
R.
, &
Haier
,
R. J.
(
2008
).
Cognitive abilities independent of IQ correlate with regional brain structure.
Intelligence
,
36
,
18
28
.
Jung
,
R. E.
, &
Haier
,
R. J.
(
2007
).
The parieto-frontal integration theory (P-FIT) of intelligence: Converging neuroimaging evidence.
Behavioral and Brain Sciences
,
30
,
135
154
.
Kable
,
J. W.
, &
Glimcher
,
P. W.
(
2007
).
The neural correlates of subjective value during intertemporal choice.
Nature Neuroscience
,
10
,
1625
1633
.
Karama
,
S.
,
Ad-Dab'bagh
,
Y.
,
Haier
,
R. J.
,
Deary
,
I. J.
,
Lyttelton
,
O. C.
,
Lepage
,
C.
,
et al
(
2009
).
Positive association between cognitive ability and cortical thickness in a representative US sample of healthy 6 to 18 year-olds.
Intelligence
,
37
,
145
155
.
Kirby
,
K. N.
,
Godoy
,
R.
,
Reyes-Garcia
,
V.
,
Byron
,
E.
,
Apaza
,
L.
,
Leonard
,
W.
,
et al
(
2002
).
Correlates of delay-discount rates: Evidence from Tsimane' Amerindians of the Bolivian rain forest.
Journal of Economic Psychology
,
23
,
291
316
.
Kirby
,
K. N.
, &
Marakovic
,
N. N.
(
1995
).
Modeling myopic decisions—Evidence for hyperbolic delay-discounting within-subjects and amounts.
Organizational Behavior and Human Decision Processes
,
64
,
22
30
.
Li
,
S. C.
,
Hammerer
,
D.
,
Muller
,
V.
,
Hommel
,
B.
, &
Lindenberger
,
U.
(
2009
).
Lifespan development of stimulus-response conflict cost: Similarities and differences between maturation and senescence.
Psychological Research-Psychologische Forschung
,
73
,
777
785
.
Li
,
S. C.
,
Lindenberger
,
U.
,
Hommel
,
B.
,
Aschersleben
,
G.
,
Prinz
,
W.
, &
Baltes
,
P. B.
(
2004
).
Transformations in the couplings among intellectual abilities and constituent cognitive processes across the life span.
Psychological Science
,
15
,
155
163
.
Liu
,
L.
, &
Feng
,
T. Y.
(
2012
).
The neural predictors of choice preference in intertemporal choice.
Brain Research
,
1436
,
92
100
.
Liu
,
L.
,
Feng
,
T. Y.
,
Wang
,
J. F.
, &
Li
,
H.
(
2012
).
The neural dissociation of subjective valuation from choice processes in intertemporal choice.
Behavioural Brain Research
,
231
,
40
47
.
Luo
,
S.
,
Ainslie
,
G.
,
Pollini
,
D.
,
Giragosian
,
L.
, &
Monterosso
,
J. R.
(
2012
).
Moderators of the association between brain activation and farsighted choice.
Neuroimage
,
59
,
1469
1477
.
MacDonald
,
S. W. S.
,
Cervenka
,
S.
,
Farde
,
L.
,
Nyberg
,
L.
, &
Backman
,
L.
(
2009
).
Extrastriatal dopamine D2 receptor binding modulates intraindividual variability in episodic recognition and executive functioning.
Neuropsychologia
,
47
,
2299
2304
.
Mazur
,
J. E.
(
1987
).
An adjusting procedure for studying delayed reinforcement.
In
M. L.
Commons
,
J. L.
Mazur
,
J. A.
Nevin
, &
H.
Rachlin
(Eds.),
Quantitative analyses of behavior: The effect of delay of intervening events on reinforcement value
(
Vol. 5
, pp.
55
73
).
Hillsdale, NJ
:
Erlbaum
.
McClure
,
S. M.
,
Laibson
,
D. I.
,
Loewenstein
,
G.
, &
Cohen
,
J. D.
(
2004
).
Separate neural systems value immediate and delayed monetary rewards.
Science
,
306
,
503
507
.
Monterosso
,
J. R.
,
Ainslie
,
G.
,
Xu
,
J.
,
Cordova
,
X.
,
Domier
,
C. P.
, &
London
,
E. D.
(
2007
).
Frontoparietal cortical activity of methamphetamine-dependent and comparison subjects performing a delay discounting task.
Human Brain Mapping
,
28
,
383
393
.
Nelissen
,
N.
,
Stokes
,
M.
,
Nobre
,
A. C.
, &
Rushworth
,
M. F. S.
(
2013
).
Frontal and parietal cortical interactions with distributed visual representations during selective attention and action selection.
The Journal of Neuroscience
,
33
,
16443
16458
.
Olkin
,
I.
, &
Finn
,
J. D.
(
1995
).
Correlations redux.
Psychological Bulletin
,
118
,
155
164
.
Olson
,
E. A.
,
Hooper
,
C. J.
,
Collins
,
P.
, &
Luciana
,
M.
(
2007
).
Adolescents' performance on delay and probability discounting tasks: Contributions of age, intelligence, executive functioning, and self-reported externalizing behavior.
Personality and Individual Differences
,
43
,
1886
1897
.
Otto
,
A. R.
,
Gershman
,
S. J.
,
Markman
,
A. B.
, &
Daw
,
N. D.
(
2013
).
The curse of planning: Dissecting multiple reinforcement-learning systems by taxing the central executive.
Psychological Science
,
24
,
751
761
.
Peters
,
J.
, &
Buchel
,
C.
(
2009
).
Overlapping and distinct neural systems code for subjective value during intertemporal and risky decision making.
The Journal of Neuroscience
,
29
,
15727
15734
.
Peters
,
J.
, &
Buchel
,
C.
(
2010
).
Neural representations of subjective reward value.
Behavioral Brain Research
,
213
,
135
141
.
Reimers
,
S.
,
Maylor
,
E. A.
,
Stewart
,
N.
, &
Chater
,
N.
(
2009
).
Associations between a one-shot delay discounting measure and age, income, education and real-world impulsive behavior.
Personality and Individual Differences
,
47
,
973
978
.
Reynolds
,
B.
,
Patak
,
M.
, &
Shroff
,
P.
(
2007
).
Adolescent smokers rate delayed rewards as less certain than adolescent nonsmokers.
Drug and Alcohol Dependence
,
90
,
301
303
.
Ripke
,
S.
,
Hubner
,
T.
,
Mennigen
,
E.
,
Muller
,
K. U.
,
Rodehacke
,
S.
,
Schmidt
,
D.
,
et al
(
2012
).
Reward processing and intertemporal decision making in adults and adolescents: The role of impulsivity and decision consistency.
Brain Research
,
1478
,
36
47
.
Rolls
,
E. T.
,
McCabe
,
C.
, &
Redoute
,
J.
(
2008
).
Expected value, reward outcome, and temporal difference error representations in a probabilistic decision task.
Cerebral Cortex
,
18
,
652
663
.
Rushworth
,
M. F. S.
,
Kolling
,
N.
,
Sallet
,
J.
, &
Mars
,
R. B.
(
2012
).
Valuation and decision-making in frontal cortex: One or many serial or parallel systems?
Current Opinion in Neurobiology
,
22
,
946
955
.
Schilling
,
C.
,
Kuhn
,
S.
,
Paus
,
T.
,
Romanowski
,
A.
,
Banaschewski
,
T.
,
Barbot
,
A.
,
et al
(
2013
).
Cortical thickness of superior frontal cortex predicts impulsiveness and perceptual reasoning in adolescence.
Molecular Psychiatry
,
18
,
624
630
.
Schlagenhauf
,
F.
,
Rapp
,
M. A.
,
Huys
,
Q. J. M.
,
Beck
,
A.
,
Wustenberg
,
T.
,
Deserno
,
L.
,
et al
(
2013
).
Ventral striatal prediction error signaling is associated with dopamine synthesis capacity and fluid intelligence.
Human Brain Mapping
,
34
,
1490
1499
.
Schumann
,
G.
,
Loth
,
E.
,
Banaschewski
,
T.
,
Barbot
,
A.
,
Barker
,
G.
,
Buchel
,
C.
,
et al
(
2010
).
The IMAGEN study: Reinforcement-related behaviour in normal brain function and psychopathology.
Molecular Psychiatry
,
15
,
1128
1139
.
Shamosh
,
N. A.
,
DeYoung
,
C. G.
,
Green
,
A. E.
,
Reis
,
D. L.
,
Johnson
,
M. R.
,
Conway
,
A. R. A.
,
et al
(
2008
).
Individual differences in delay discounting relation to intelligence, working memory, and anterior prefrontal cortex.
Psychological Science
,
19
,
904
911
.
Shamosh
,
N. A.
, &
Gray
,
J. R.
(
2008
).
Delay discounting and intelligence: A meta-analysis.
Intelligence
,
36
,
289
305
.
Simpson
,
C. A.
, &
Vuchinich
,
R. E.
(
2000
).
Reliability of a measure of temporal discounting.
Psychological Record
,
50
,
3
16
.
Sjolund
,
S.
,
Allebeck
,
P.
, &
Hemmingsson
,
T.
(
2012
).
Intelligence quotient (IQ) in adolescence and later risk of alcohol-related hospital admissions and deaths—37-year follow-up of Swedish conscripts.
Addiction
,
107
,
89
97
.
Steinberg
,
L.
(
2005
).
Cognitive and affective development in adolescence.
Trends in Cognitive Sciences
,
9
,
69
74
.
Swick
,
D.
,
Ashley
,
V.
, &
Turken
,
U.
(
2011
).
Are the neural correlates of stopping and not going identical? Quantitative meta-analysis of two response inhibition tasks.
Neuroimage
,
56
,
1655
1665
.
van Tuijl
,
C.
, &
Lesernan
,
P. P. M.
(
2007
).
Increases in the verbal and fluid cognitive abilities of disadvantaged children attending preschool in the Netherlands.
Early Childhood Research Quarterly
,
22
,
188
203
.
Waiter
,
G. D.
,
Deary
,
I. J.
,
Staff
,
R. T.
,
Murray
,
A. D.
,
Fox
,
H. C.
,
Starr
,
J. M.
,
et al
(
2009
).
Exploring possible neural mechanisms of intelligence differences using processing speed and working memory tasks: An fMRI study.
Intelligence
,
37
,
199
206
.
Xu
,
Y. D.
, &
Chun
,
M. M.
(
2006
).
Dissociable neural mechanisms supporting visual short-term memory for objects.
Nature
,
440
,
91
95
.
Zink
,
C. F.
,
Pagnoni
,
G.
,
Martin-Skurski
,
M. E.
,
Chappelow
,
J. C.
, &
Berns
,
G. S.
(
2004
).
Human striatal responses to monetary reward depend on saliency.
Neuron
,
42
,
509
517
.