Edited by: Leonhard Schilbach, University Hospital Cologne, Germany
Reviewed by: Bernd Weber, Rheinische-Friedrich-Wilhelms Universität, Germany; Bojana Kuzmanovic, Research Center Juelich, Germany
*Correspondence: Sung-il Kim, Department of Education, Brain and Motivation Research Institute (bMRI), Korea University, Anam-Dong 5 Ga, Seongbuk-Gu, Seoul, 136-701, South Korea e-mail:
This article was submitted to the journal Frontiers in Human Neuroscience.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This study investigates differential neural activation patterns in response to reward-related feedback depending on various reward contingencies. Three types of reward contingencies were compared: a “gain” contingency (a monetary reward for correct answer/no monetary penalty for incorrect answer); a “lose” contingency (no monetary reward for correct answer/a monetary penalty for incorrect answer); and a “combined” contingency (a monetary reward for correct answer/a monetary penalty for incorrect answer). Sixteen undergraduate students were exposed to the three reward contingencies while performing a series of perceptual judgment tasks. The fMRI results revealed that only the “gain” contingency recruited the ventral striatum, a region associated with positive affect and motivation, during overall feedback processing. Specifically, the ventral striatum was more activated under the “gain” contingency than under the other two contingencies when participants received positive feedback. In contrast, when participants received negative feedback, the ventral striatum was less deactivated under the “gain” and “lose” contingencies than under the “combined” contingency. Meanwhile, the negative feedback elicited significantly stronger activity in the dorsal amygdala, a region tracking the intensity and motivational salience of stimuli, under the “gain” and “lose” contingencies. These findings suggest the important role of contextual factor, such as reward contingency, in feedback processing. Based on the current findings, we recommend implementing the “gain” contingency to maintain individuals’ optimal motivation.
Motivation, a major determinant of behavior, helps individuals to engage in goal-directed behaviors (McClelland,
A wealth of neuroimaging studies have identified several key brain regions associated with reward, motivation, and emotion, including the ventral striatum, the orbitofrontal cortex (OFC), and the amygdala (see Davis and Whalen,
Evidence also revealed that reward-sensitive brain regions are highly context-dependent (Holroyd et al.,
In our opinion, an identical reward-related feedback may be perceived differently depending on whether the context is promotion-focused (gain context) or prevention-focused (lose context), resulting in differential emotional and motivational responses. For example, in a behavioral study comparing the effects of three reward contingencies on performance satisfaction of middle school students, Bong and Kim (
In the present study, we aimed to investigate how reward contexts with different focuses (e.g., promotion or prevention) would influence the neural responses during feedback processing. We manipulated three different types of reward contingencies (the “gain”, “lose”, and “combined” contingencies) and compared the brain activities in various regions of interest (ROI) (ventral striatum, OFC, and amygdala) during feedback processing. We expected different brain activation patterns in response to feedback depending upon the type of reward contingency.
The study was approved by Korea University’s Institutional Review Board for human participants. A total of sixteen healthy, right-handed undergraduate students from several private universities in Seoul (seven males, mean age = 22.4,
We used a mixed blocked/event-related design, with three independent runs (one each for the “combined,” “gain,” and “lose” contingencies) during scanning. Each contingency consisted of 40 trials, yielding 120 trials in total. Each trial lasted 11 s, with each contingency lasting 7 min and 20 s. The entire scan took 22 min. For each trial, the task stimulus was presented for 2 s, followed by a random fixation. The durations of the fixation were determined by a sequence of random numbers generated by Excel, ranging from 1 to 4 s with an interval of 0.5 s. The average duration of fixation was 2 s. After the random fixation, the participants received facial feedback for 1 s, notifying them of whether they had succeeded or failed. After another random fixation, participants received monetary feedback that lasted 2 s. This was followed by a random inter-trial interval also ranging from 1 to 4 s (average of 2 s; see Figure
The experimental design was programmed and presented using E-PRIME v. 1.1. software. Before scanning, participants performed practice trials to familiarize themselves with the task and the response procedure. During the scanning, participants performed the task in the fMRI scanner through the rear-projection monitor. They were asked to judge whether the number of the stimuli on the screen met the criterion by clicking either the left button (if they believed that the number met the criteria) or the right button (if they believed that the number did not meet the criteria) on a keyboard connected to the fMRI machine. To prevent the practice effect as well as fatigue effect from repetitive performance of the same tasks, we incorporated slightly different tasks into each contingency. Specifically, during the “combined” contingency, participants looked at a computer screen filled with 14 to 16 figures for 2 s. The task was to determine whether the number of figures on the screen was 15 or not. Similarly, participants were asked to detect whether or not the number of letters was 16 during the “gain” contingency and to determine whether or not the number of digits was 16 during the “lose” contingency. The pilot test showed that there was no difference among these tasks in terms of task difficulty and task interest.
Three patterns of feedback were used in accordance with the three types of reward contingencies. Under the “combined” contingency, participants started with KRW 20,000. For a correct answer, an image of a smiley face was accompanied and followed by monetary feedback indicating that the participant received KRW 1000. For an incorrect answer, on the other hand, an image of a sad face was accompanied and followed by monetary feedback indicating that the participant lost KRW 1000. Under the “gain” contingency, participants started with KRW 0. For a correct answer, a smiley face was presented and followed by monetary feedback indicating that the participant received KRW 1000. For an incorrect answer, a sad face was presented and followed by monetary feedback indicating no monetary change. Under the “lose” contingency, participants started with KRW 40,000. For a correct answer, a smiley face was presented and followed by monetary feedback indicating no monetary change. For an incorrect answer, a sad face was presented and followed by monetary feedback indicating that the participant lost KRW 1000. Figure
Unbeknownst to the participants, the feedback was predetermined regardless of their actual performance and all of the participants received identical feedback throughout the whole experiment. We used this method to prevent potential large differences in performance among the participants so that they could end up with the same amount of rewards within each contingency. The success rate was set as 50% for all three contingency conditions, and the sequence of positive and negative feedback was randomized within each contingency. Because the task stimulus was presented for a relatively short period of time (2 s), the high level of task difficulty made the bogus performance feedback credible to the participants. In their post-scanning interviews, all participants reported that they believed the feedback was based on their actual performance. After they had finished the whole experiment, participants were fully debriefed.
For the run sequence, the “combined” contingency was always presented first because it creates a reward context with no specific focus (neither promotion nor prevention). Therefore, it was used as a baseline condition. For the purpose of counterbalancing the order effect of the “gain” and “lose” contingencies, half of the participants were scanned in a “combined-gain-lose” sequence and the other half were scanned in a “combined-lose-gain” sequence.
The experiment was conducted at Ewha Womans University Mokdong Hospital. Images were acquired using a Philips Intera Achieva 3T MRI scanner (Philips Medical Systems, Andover, MA, USA). Functional data were obtained using a single-shot gradient echo-planar imaging (EPI) sequence (TR = 2000 ms, TE = 30 ms, flip angle = 90°, field of view (FOV) = 240 mm, ascending, 36 3-mm-thick slices, with no gap). After the first run, high-resolution T1-weighted three-dimensional volumes were acquired for anatomical localization (TR = 9.8 ms, TE = 4.6 ms, 160 slices, voxel size = 1 × 1 × 1 mm).
Imaging data were preprocessed and analyzed by Statistical Parametric Mapping (SPM 5, Department of Cognitive Neuroscience, London, U.K.) in the Matlab (Mathworks Inc., USA) environment. During preprocessing, functional images were first realigned to the first volume to compensate for subtle head motions. All participants’ head motions were less than 3 mm in any translation within each run. Data were then corrected for differences in timing of slice acquisition, normalized to EPI templates implemented in the SPM, and spatially smoothed using an 8 mm full width at half maximum isotropic (FWHM) Gaussian kernel.
After the preprocessing, statistical analyses were performed on each participant’s data using a general linear model (GLM) in SPM. The analyses were performed by modeling facial feedback (success and failure events for each contingency), monetary feedback (success and failure events for each contingency), and task stimuli as regressors. Participants’ response times during task phases and realignment parameters were also included as regressors in the statistical model. Changes in the Blood-Oxygen-Level-Dependent (BOLD) signal were assessed by linear combinations of the estimated GLM parameters (beta values). Because participants were informed of the contingency condition before they started each run, the present study particularly focused on comparing the facial feedback phase. In other words, once the participants received facial feedback, they were automatically aware that they would receive reward, lose money, or obtain no monetary reward/penalty.
Individual contrast images were estimated by contrasting the beta value of the positive and negative facial feedback against the implicit baseline within each contingency. Thus six types of contrast images were estimated. All individual contrast images were then collected to further examine the statistical significance of the evoked hemodynamic response in a second level random effects analysis. We first conducted a whole-brain 2 × 3 factorial ANOVA with feedback (positive or negative feedback) and contingency (“combined”, “gain”, or “lose” contingency) as factors to test the main effects of the feedback and the contingency as well as the potential interaction of the two factors on brain activation. We also conducted two separate one-way ANOVAs with each feedback valence as factors to explicitly test how positive and negative feedback may recruit distinct patterns of brain activation under various contingencies. The statistical criterion was set at
We then conducted functional ROI analyses and a series of
Participants’ average response times were 1326.88 ms (
First, the 2 × 3 factorial ANOVA revealed a significant main effect of feedback on brain activation in the bilateral ventral striatum, the OFC, the anterior cingulate cortex, and the inferior parietal lobule (
There was also a significant main effect of reward contingency on brain activation in
Brain regions | BA | R/L | Cluster | MNI Coordinates | |||
---|---|---|---|---|---|---|---|
Ventral striatum | L | 292 | −12 | 8 | −12 | 5.90 | |
R | 319 | 12 | 8 | −12 | 5.59 | ||
Anterior cingulate cortex | 32 | R | 81 | 12 | 42 | 6 | 4.22 |
32 | R | 4 | 46 | 2 | 3.78 | ||
32 | R | 12 | 38 | 16 | 3.58 | ||
OFC | 11 | L | 17 | −34 | 42 | −16 | 4.14 |
Inferior parietal lobule | 40 | L | 17 | −44 | −56 | 42 | 3.90 |
Ventral striatum | L | 69 | −14 | 14 | −2 | 4.07 | |
R | 12 | 12 | 14 | −8 | 3.89 | ||
R | 27 | 16 | 12 | −20 | 4.60 | ||
Amygdala | L | 45 | −30 | −8 | −8 | 4.44 | |
R | 16 | 30 | −8 | −18 | 4.41 | ||
Uncus | 34 | L | 126 | −16 | 4 | −22 | 5.22 |
Putamen | L | 69 | −20 | −2 | 0 | 4.39 | |
Superior temporal gyrus | 22 | L | 63 | −42 | 6 | −24 | 4.38 |
Cerebellum | 30 | 0 | −52 | −8 | 4.36 | ||
sgACC | 25 | L | 12 | −10 | 24 | −12 | 4.34 |
Inferior temporal gyrus | 19 | L | 14 | −52 | −62 | 0 | 4.34 |
Precentral gyrus | 4 | R | 13 | 40 | −26 | 68 | 3.99 |
Inferior frontal gyrus | 44 | R | 14 | 50 | 16 | 12 | 3.88 |
We further conducted two separate one-way ANOVAs with different feedback valences to test how positive and negative feedback may differently elicit neural activities in
Brain regions | BA | R/L | Cluster | MNI Coordinates | |||
---|---|---|---|---|---|---|---|
Ventral striatum | R | 13 | 14 | 12 | −18 | 3.23 | |
Ventral striatum | L | 24 | −16 | 8 | −4 | 3.46 | |
Amygdala | L | 24 | −30 | −6 | −8 | 3.09 |
We examined the effects of different types of reward contingencies on emotional and motivational responses during feedback processing by comparing the brain activation in several reward-sensitive regions. We found differential pattern of neural activities in the ventral striatum and the amygdala depending upon the type of reward contingency.
First, significant difference in the ventral striatum activation was observed across the three contingencies during overall feedback processing. Functional ROI results indicate that the ventral striatum showed positive activation only under the “gain” contingency and was deactivated under the “lose” and the “combined” contingencies. Moreover, two separate one-way ANOVAs with different feedback valences revealed that when participants received positive feedback, the ventral striatum showed significantly stronger activation under the “gain” contingency than under the other two contingencies. On the other hand, the ventral striatum was less deactivated under the “gain” contingency than under the “combined” contingency in response to negative feedback. The ventral striatum is known as the main reward area responsible for hedonic experience and its activation in response to a variety of reward has been reported (e.g., Delgado et al.,
Second, differential activation patterns of the amygdala were witnessed across the three reward contingencies. Overall feedback produced significantly higher activation in the bilateral amygdala under the “gain” and the “lose” contingencies than under the “combined” contingency. It is well documented that the amygdala plays an important role in processing negative and unpleasant emotions, such as fear and disgust (see Calder et al.,
In addition, we also found that the sgACC showed different activation patterns across the three contingencies during overall feedback processing. It positively activated only under the “lose” contingency but deactivated under the “gain” and the “lose” contingencies. Interestingly, although the sgACC activation was witnessed from the main effect of contingency (regardless of the feedback valence), we found significant activation in the corresponding sgACC region (MNI coordinates:
There was also a significant main effect of feedback on the lateral OFC in the 2 × 3 ANOVA analysis. Specifically, positive feedback has produced significantly stronger activation in the lateral OFC than negative feedback. The OFC is the critical brain region for value judgment (Grabenhorst and Rolls,
Taken together, the findings of this study have practical implications for designing reward contexts that could beget positive affect and enhance individuals’ motivation. Depending upon the reward contingency, individuals could perceive an identical feedback differently and in turn experience different emotions and motivations. Among the three types of contingencies, we recommend implementing the “gain” contingency, in which a reward is given for success and no punishment is given for failure, because it shows the most adaptive pattern of emotional and motivational responses to both positive and negative feedback. Our interview data support this argument as most of the participants felt more satisfied with the “gain” contingency. Among the sixteen participants, nine rated the “gain” contingency, four rated the “combined” contingency, and only three rated the “lose” contingency as most satisfactory.
Several limitations of the present study as well as suggestions for future research need to be addressed. First, we used bogus feedback regardless of participants’ actual performance. Although all the participants believed that the feedback was based on their actual performance, it would be ideal for future research to use real feedback based on participants’ actual performance. Second, it would be worthwhile to further investigate if verbal feedback without a monetary reward would have a similar effect because verbal praise and punishment are more frequently used in educational settings.
The present study investigated individuals’ emotional and motivational responses to three different types of reward contingencies during feedback processing. It contributes to the existing literature by demonstrating that contextual effect of reward could elicit distinct neural activities during feedback processing. In particular, the results indicate that the “gain” contingency is more likely to produce positive affect and maintain individuals’ motivation. Therefore, we suggest implementing reward/punishment systems based on the “gain” contingency to maintain individuals’ motivation during task performances.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This research was supported by the WCU (World Class University) Program funded by the Korean Ministry of Education, Science and Technology, consigned to the Korea Science and Engineering Foundation (Grant no. R32-2008-000-20023-0).