Conceived and designed the experiments: MdG CT GN. Performed the experiments: MdG. Analyzed the data: BE UP GN. Wrote the paper: BE MdG CT GN. Technical support: CT. Establishing an appropriate scanning sequence: CT.
The authors have declared that no competing interests exist.
The attribution of personal relevance, i.e. relating internal and external stimuli to establish a sense of belonging, is a common phenomenon in daily life. Although previous research demonstrated a relationship between reward and personal relevance, their exact neuronal relationship including the impact of personality traits remains unclear.
Using functional magnetic resonance imaging, we applied an experimental paradigm that allowed us to explore the neural response evoked by reward and the attribution of personal relevance separately. We observed different brain regions previously reported to be active during reward and personal relevance, including the bilateral caudate nucleus and the pregenual anterior cingulate cortex (PACC). Additional analysis revealed activations in the right and left insula specific for the attribution of personal relevance. Furthermore, our results demonstrate a negative correlation between signal changes in both the PACC and the left anterior insula during the attribution of low personal relevance and the personality dimension novelty seeking.
While a set of subcortical and cortical regions including the PACC is commonly involved in reward and personal relevance, other regions like the bilateral anterior insula were recruited specifically during personal relevance. Based on our correlation between novelty seeking and signal changes in both regions during personal relevance, we assume that the neuronal response to personally relevant stimuli is dependent on the personality trait novelty seeking.
Various imaging studies tried to clarify and to uncover the neuronal basis of our self, indicating an increased interest in this mysterious topic. In this context, a variety of different aspects and concepts of the self were investigated by neuroscientists, for instance self-recognition
In this study, we focussed on a clearly distinguishable concept of the self, the attribution of personal relevance to everyday stimuli
The above mentioned neuroimaging studies consistently showed the involvement of a set of brain regions in different aspects and concepts of the self. These studies were able to detect various subcortical and cortical regions like the medial orbitofrontal cortex, the ventromedial prefrontal cortex (VMPFC) or the pregenual anterior cingulate cortex (PACC), the dorsomedial prefrontal, the ventrolateral prefrontal cortex, the anterior insula, the amygdala and the ventral and dorsal striatum
The neuronal networks underlying personal relevance and reward show a strong overlap. Recent studies
Personality, or more specifically temperament, makes a major contribution to human behaviour. Various brain imaging studies tried to disentangle the complex relationship between personality, its neurobiological foundations and human behaviour
The general aim of our study was to investigate the behavioural and neuronal relationship between the attribution of personal relevance and reward.
First, we identified brain regions involved in both, reward processing and the attribution of personal relevance. Second, we identified specific brain regions for personal relevance. Finally, we correlated our obtained imaging data with the dimensions of Cloninger's Temperament and Character Inventory. Relying on previous research
It should be noted, that this study was based on an experimental paradigm used in our previous study
As mentioned above, we had an a priori hypothesis concerning the involvement of the anterior insula and the pregenual anterior cingulate cortex in the attribution of personal relevance and therefore concentrated our correlation analysis on these two (anatomical) defined regions, whereas our whole brain analysis for overlapping and non-overlapping regions between reward and personal relevance is rather exploratory.
The presented study was approved by the institutional review board of the Department of Neurology, University of Magdeburg, Germany and by the ethical committee of the Medical Faculty, University of Magdeburg, Germany. After a detailed explanation of the study, all subjects gave their written informed consent.
We investigated 19 right-handed healthy subjects (12 men and 7 women, mean age 30.7 years, SD 7.1, range 23 to 50 years) without any neurological or psychiatric illness.
We applied different psychological tests for the behavioural characterizations of our subjects, including Beck Depression Inventory (BDI;
We applied a well-established paradigm
The experiment contained three different types of tasks. During reward trials subjects had to perform a gambling task, where they could either win or lose. During personal relevance-evaluation trials subjects indicated whether a stimulus was of high or low relevance to them. The third task was a control task in which subjects had to assess the orientation of a presented stimulus. The sequence of all trial types was designed to be as similar as possible to allow for comparison.
The whole experiment consists of eight runs (four reward runs, two personal relvance runs and two control runs) presented in a pseudo-randomized order.
All trials began with the presentation of a
During the decision phase of reward trials subjects were instructed to press either the left or the right button in order to gamble about amounts of their reimbursement. In the feedback phase they were informed whether they had won or lost, symbolised by a plus or a minus sign. The state bar reflected the subject's new total after the previous win or loss. Subjects were made believe that their luck during the gambling trials had direct influence on their performance however, the proportion of wins and losses was predefined and almost identical for all subjects.
During the decision phase of personal relevance evaluation trials, subjects had to evaluate the presented picture and determine whether it was of high or low personal relevance. In the feedback phase of these trials an equality sign was presented when the button press was delivered in time. In contrast to both of the other tasks the minus sign was only presented when no response occurred. We decided to present an equality sign instead of the plus sign to make sure that this task had no rewarding component. The state bar was presented in these trials as well for consistency reasons. Subjects were instructed that it had no meaning and the actual value fluctuated around the midline.
In the decision phase of the control trials it was the subject's task to identify the alignment of the presented picture. All stimuli had the shape of a rectangle, half of the stimuli were horizontally aligned and half of them vertically. In the feedback phase, a plus or minus sign was presented for correct and incorrect trials respectively. As in the personal relevance-evaluation task, the feedback display contained the fluctuating state bar that was irrelevant in these trials.
After every 8 trials a baseline event occurred, in which only the four empty location frames were presented.
Each task included the presentation of all three different types of stimuli (gambling, alcohol and food stimuli) taken from the International Affective Picture System and the Normative Affective Picture System. The stimuli were chosen to maximize our ability to investigate the specific relationship between reward and personal relevance. Based on previous imaging experiments we selected stimuli that show a strong reward value such as natural reinforcers i.e. food
Functional data was collected using a 3-Tesla whole body MRI system (Siemens Trio, Erlangen, Germany) equipped with an 8-channel head coil. 32 T2*-weighted echo-planar images (EPI) per volume with blood oxygenation level-dependent (BOLD) contrast were obtained (matrix 64×64, field-of-view 224×224 mm, spatial resolution: 3,5×3,5×4 mm, TE = 30 ms, TR = 2000 ms, flip angle 80°). The slices were acquired parallel to the AC-PC plane in an odd-even interleaved acquisition order. Subjects had to complete eight scanning runs with 210 volumes per run. The first four volumes of each run were discarded.
The functional data was preprocessed and statistically analysed using the SPM2 software package (Wellcome Department of Cognitive Neuroscience, University College London, UK;
Briefly, all functional images were slice time corrected with reference to the first slice acquired, corrected for motion artifacts by realignment to a mean functional image and spatial normalized to a standard T1-weighted template provided by SPM2. The normalization was generated by warping the subject's anatomical T1-weighted scan on the T1-template and applying these parameters to all functional images
All relevant periods (i.e. the decision phase, the feedback phase and the baseline phase) and all four conditions (reward win, reward lose, high personal and low personal relevant trials) were included in the SPM model. In each reward run five conditions are modelled (decision lose, feedback lose, decision win, feedback win and baseline), each personal relevance run also contains five conditions (decision low personal relevance, feedback low personal relevance, decision high personal relevance, feedback high personal relevance and baseline), whereas the intertrial interval was not modelled separately. For our contrasts of interest we used the feedback phase to provide as much coherence as possible to our time course analysis. A statistical model for each subject was computed by convolving a canonical response function
For specification of regions only active during evaluation of personal relevance, we conducted a masking analysis implemented in SPM2 for the contrast “high personal relevance > low personal relevance” exclusively masked with “reward win > reward lose”. We thresholded the images for p<0.05 [uncorrected] for the mask and p<0.05 [FDR] for the main contrast for at least 10 contiguous voxels.
Determination of common regions for “reward win” and “high personal relevance” was calculated by a conjunction analysis implemented in SPM2 for the contrasts “reward win > reward lose” and “high personal relevance > low personal relevance”. The threshold of the resulting statistical map was p<0.001 [uncorrected] for at least 10 contiguous voxels
In a second step we performed a detailed analysis based on functional and anatomical regions of interest. First, we extracted the fMRI raw data using the Marseille Region of Interest Toolbox software package MarsBaR 1.86 (
Reaction times during reward (mean 852.6 ms, SD 154.7) were significantly faster than during the attribution of personal relevance (mean 991.9 ms, SD 105.7) (t(18) = −4.579, p<0.001) (
Furthermore, we analysed the relationship between personal relevance and novelty seeking using repeated measurements analysis of variance (ANOVA). We observed a significant main effect of “task” (high vs. low personal relevance) (F(1,10) = 25.961, p<0.001) and a significant interaction between “task” (high vs. low personal relevance) and “group” (high vs. low novelty seekers) (F(1,10) = 8.746, p = 0.014).
For a more detailed analysis, we divided our study sample in three groups (low, medium and high novelty seeking) according to the lower and upper 33th percentile of the novelty seeking score and compared the reaction time between low and high novelty seeking individuals. This analysis revealed a significantly faster response during the attribution of high personal relevance in high novelty seekers compared to low novelty seekers (t(10) = 2.836, p = 0.009, independent samples
The mean value of novelty seeking (NS) as measured by Cloninger's Temperament and Character Inventory was 19.3 (SD 6.3). For the temperament dimension novelty seeking we found a positive correlation with the character dimension self-transcendence (r[Pearson] = .506, p = 0.027), whereas we were not able to detect significant correlations between novelty seeking and the other temperament and character dimensions (see
For determination of regions active during reward and the processing of high personal relevant pictures, we performed a conjunction analysis between the contrasts “high personal relevance >low personal relevance” and “win>lose”. This approach revealed activations in the right pregenual cingulate cortex (PACC), the right and left caudate nucleus and the right ventrolateral prefrontal cortex (VLPFC) (coordinates according to the MNI stereotactical space: PACC (2, 40, 16), right caudate (10, 10, 10), left caudate (−12, 10, 12) and right VLPFC (46, 42, 8),
The second level group statistic for the above mentioned contrast revealed activations in the right pregenual cingulate cortex (PACC), the right ventrolateral prefrontal cortex (VLPFC) and in the right and left caudate nucleus adjacent to the ventral striatum. The images on the far left show the statistical maps calculated with SPM2. The two diagrams in each line show the mean normalized fMRI signal changes (y-axis) for the conditions win and lose as high and low personal relevance (high pr, low pr) with t = 0 for the start of the feedback phase in healthy subjects. (
ROI name | Contrast | coordinates [MNI] | p [FDR] | t-value | z-value |
left caudate | −12, 10, 12 | 0,073 | 4,32 | 3,85 | |
right caudate | 10, 10, 10 | 0,086 | 3,55 | 3,27 | |
right putamen | 30, −6, −4 | 0,067 | 5,22 | 4,47 | |
right VLPFC | 46, 42, 8 | 0,078 | 4,20 | 3,76 | |
right PACC (BA32) | 2, 40, 16 | 0,073 | 4,31 | 3,85 | |
right insula |
22, 24, −6 | 0,069 | 4,52 | 4,00 | |
right dorsomedial PFC | 8, 44, 48 | 0,071 | 4,46 | 3,95 | |
right anterior insula | 28, 20, 8 | 0,037 | 3,77 | 3,44 | |
left anterior insula | −34, 26, 2 | 0,028 | 4,23 | 3,78 | |
right premotor cortex | 6, 8, 60 | 0,024 | 5,07 | 4,37 | |
left Insula | −34, 10, 0 | 0,029 | 4,14 | 3,72 | |
right supragenual ACC | 4, 24, 24 | 0,035 | 3,87 | 3,51 |
conjunction of “high personal relevance > low personal relevance” with “win > lose”.
“high personal relevance > low personal relevance” exclusively masked with “win > lose”.
extending to the basal ganglia.
VLPFC: ventrolateral prefrontal cortex, PACC: pregenual anterior cingulate cortex, PFC: prefrontal cortex, ACC: anterior cingulate cortex.
We supposed that there are also regions specific for the processing of personal relevance. For confirmation of this hypothesis, we calculated an exclusive masking analysis between the contrasts “high personal relevance > low personal relevance” and “win > lose” (
The second level group statistic for the above mentioned contrast revealed activations in the right and left anterior insula and the right premotor cortex. The images on the far left show the statistical maps calculated with SPM2. The two diagrams in each line show the mean normalized fMRI signal changes (y-axis) for the conditions win and lose as high and low personal relevance (high pr, low pr) with t = 0 for the start of the feedback phase in healthy subjects. (error bar: standard deviation).
Our independent anatomically-based PACC region of interest (ROI) confirmed our above mentioned results by showing the same neuronal distinction between reward and personal relevance. In the PACC we observed a significant differentiation between “win” and “lose” (t(18) = 6.093; p<0.001), as well as between “high personal relevance” and “low personal relevance” (t(18) = 2.158; p = 0.045) (
The independent anatomical ROI encompassing the left anterior insula also confirmed our functional imaging results. We observed a significant differentiation concerning our conditions “high personal relevance” and “low personal relevance” (t(18) = 2.482; p = 0.023). Between “win” and “lose” no significant distinction was observable (t(18) = 1.47; p = 0.159) (see
Furthermore, our anatomical ROI encompassing the right anterior insula also showed a differentiation between high and low personal relevance (t(18) = 2.047; p = 0.056, statistical trend) and not between win and lose (t(18) = 0.791; p = 0.439) (
To confirm our findings from both the conjunction and the exclusive masking analysis, we also calculated the contrast “high personal relevance > low personal relevance” in SPM. As expected, we observed activations in the right and left ventral striatum (VS), the right and left anterior insula, the right VLPFC and the PACC. It is important to note that some of these regions are specific for personal relevance (e.g. left and right anterior insula), whereas other regions differentiate between reward and personal relevance (PACC, bilateral striatum). This observation underlines our results derived from the conjunction and masking analysis mentioned above. (For more details, see
To disentangle the complex relationship between personal relevance, reward and personality, we correlated the temperament dimension novelty seeking and its subscales (NS1: “exploratory excitability vs. stoic rigidity”, NS2: “impulsiveness vs. reflection”, NS3: “extravagance vs. reserve” and NS4: “disorderliness vs. regimentation”) with the functional imaging data obtained from our anatomy-based insula and PACC regions of interest.
In the case of the PACC, we found a significant negative correlation between the mean normalized fMRI signal for the condition “low personal relevance” and novelty seeking (r[Pearson] = −.551; p = 0.014) (
Region | NS | HA | RD | P | SD | C | ST |
Right PACC | r = −.551* |
r = .19 |
r = .227 |
r = −.158 |
r = .417(*) |
r = .511* |
r = −.302 |
Left anterior insula | r = −.483* |
r = .256 |
r = .133 |
r = .238 |
r = −.192 |
r = .13 |
r = .023 |
PACC: pregenual anterior cingulate cortex, NS: novelty seeking, HA: harm avoidance, RD: reward dependence, P: persistence, SD: self-directedness, C: cooperativeness, ST: self-transcendence.
Pearson correlation coefficients [r], significant correlations are labelled (**
For the left anterior insula, correlation analysis revealed a significant negative correlation between the mean normalized fMRI signal for the condition “low personal relevance” and novelty seeking (r[Pearson] = −.483, p = 0.036) (
We here investigated the behavioural and neuronal relationship between the attribution of personal relevance, reward and personality. Our data indicate neuronal overlap between personal relevance and reward in the PACC, the bilateral caudate nucleus bordering to the ventral striatum and the right VLPFC. Neural activity in the left anterior insula was recruited specifically during personal relevance as distinguished from reward, whereas we observed a statistical trend concerning the differentiation between high and low personal relevance in the right anterior insula. Signal changes in both left anterior insula and PACC during low personal relevance correlated negatively with novelty seeking as measured by Cloninger's Temperament and Character Inventory (TCI). Taken together, these data suggest both neuronal overlap and distinction between reward and personal relevance. In addition, the neuronal response during the attribution of personal relevance is modulated by the temperament dimension of novelty seeking.
The ACC can be separated according to its main function into more a cognitive dorsal part and a rostral-ventral affective division (ACCad)
This is very well compatible especially with personal reference since any personally relevant stimulus is strongly affectively coloured be it positively or negatively
A possible explanation for our correlation results arises from the fact, that activity in the ACC is more pronounced when external information, i.e. low relevant pictures, requires additional processing with conflicting internal states
More indirect support comes from studies with psychiatric patients. Manic patient exhibit in elevated states an increase in novelty seeking scores
Although there is evidence that emotion and personal relevance can be distinguished on neuronal and behavioural level
Our main finding concerned the specific involvement of the bilateral anterior insula in personal relevance. The insula has been involved in interoceptive awareness, emotion processing and consciousness
This is well in accordance with a recent study
Our findings extend these results by showing that the insula is specifically involved in the attribution of personal relevance rather than in reward processing although both processes contain a relevant input from the brain's emotional system
The observed negative correlation between low personal relevance and novelty seeking was mainly based on the subscale NS2: “impulsiveness vs. reflection”. Individuals high scoring on NS2 are described as excitable, dramatic and impulsive, whereas individuals low scoring on NS2 are described as thoughtful, analytic and focussed. This is well in accordance with our behavioural results. People with high NS scores showed faster reaction times and (following the correlation) higher degrees of deactivation, so both behavioural and neuronal measurements indicate strong reagibility to external stimuli. In contrast, people with low NS scores show slower reaction times and a more pronounced activation in the left anterior insula. Since our data were mainly based on the differences in the NS2, such differential reagibility may be related to the above described psychological profiles of high and low NS2.
Although our masking analysis revealed the bilateral anterior as specific for personal relevance, it should be noted that our results concerning the right anterior insula are ambiguous. For instance, we were only able to observe a statistical trend for the differentiation between high and low personal relevance in the right anterior insula.
Relying on a conjunction analysis, we observed recruitment of the PACC, the bilateral caudate, the right VLPFC, the right putamen and the right DMPFC during reward and personal relevance. This is consistent with studies reporting involvement of these regions in both reward
Especially subcortical regions like the bilateral caudate and the putamen seem to play an important role in the overlap between reward and personal relevance. Being part of the reward system
These regions are part of the so-called “valuation system”
4. Schilbach L, Eickhoff S, Rotarska-Jagiela A, Fink G, Vogeley K (2008) Minds at rest? Social cognition as the default mode of cognizing and its putative relationship to the “default system” of the brain. Conscious Cogn 17: 457-467.
Contrast “(high personal reference) > (low personal reference)” Activations and fMRI signal changes in regions derived from the contrast “high personal relevance > low personal relevance”. The images on the far left show the t-contrast calculated with SPM2. The two diagrams in each line show the mean normalized fMRI signal changes (y-axis) for the conditions win and lose as high and low personal relevance (high pr, low pr) with t = 0 for the start of the feedback phase in healthy subjects. (error bar: standard deviation) The second level group statistic for the above mentioned contrast revealed activations in the right (10, 8, 2; z = 3.70; p[FDR]<0.01; k>20) and left (−8, 8, 4; z = 4.05; p[FDR]<0.01; k>20) ventral striatum (VS), the right (36, 28, 4; z = 4.57; p[FDR]<0.01; k>20) and left (−38, 16, −2; z = 5.23; p[FDR]<0.01; k>20) anterior insula, the right ventrolateral prefrontal cortex (VLPFC) (44, 36, 10; z = 4.16; p[FDR]<0.01; k>20) and the pregenual cingulate cortex (PACC) (0, 40, 16; z = 3.94; p[FDR]<0.01; k>20). As expected, we observe activations in regions specific for the differentiation between high and low personal relevance like e.g. the bilateral anterior insula. Moreover, the bilateral VS, the VLPFC and the PACC show a differentiation in both domains, reward and personal relevance. This supports our proposed model for neuronal integration and differentiation between reward and personal relevance.
(1.31 MB TIF)
Reaction time during personal relevance in high and low novelty seeking individuals Reaction time in high and low novelty seeking individuals After division of our study population in three groups (high NS (n = 6): mean 25.3 (SD 3.7), medium NS (n = 7): mean 20.0 (SD 1.7), low NS (n = 6): mean 12.3 (SD 3.0)), we compared the average reaction time for reward (win and lose) and personal relevance (high and low personal relevance). Concerning the reward task, we observed no significant difference between high and low novelty seekers (t(10) = 0.611; p = 0.277), whereas in the personal relevance task high novelty seekers responded faster than low novelty seekers (t(10) = 1.413; p = 0.094, statistical trend). t-test for independent variables, 1-sided Error bar: standard deviation
(0.03 MB TIF)
Schematic illustration of the paradigm used in this study.
(0.39 MB TIF)
fMRI results for the right anterior insula a. fMRI signal changes in the right anterior insula. The two diagrams show the mean normalized fMRI signal changes (y-axis) for the conditions win and lose and high and low personal relevance (high pr, low pr) with t = 0 for the start of the feedback phase in healthy subjects. (error bar: standard deviation) b. mean normalized fMRI signal for the timepoints 6 to 8 sec after the beginning of the feedback phase. The mean normalized fMRI signal indicates a statistical trend for the differentiation between the conditions “high personal relevance” and “low personal relevance” (t(18) = 2.047; p = 0.056) in the right anterior insula, whereas we were not able to observe a significant differentiation between “win” and “lose” (t(18) = 0.791; p = 0.439). t-test for paired variables, 2-sided Error bar: standard deviation
(0.46 MB TIF)
Characteristics of the study population. mean (Standard deviation) Abbreviations: MWT-A (german: Mehrfachwortschatzintelligenztest
(0.04 MB DOC)
Correlation between the dimensions of Cloninger's Temperament and Character Inventory (n = 19). Pearson correlation coefficients [r], significant correlations are labelled (**p<0.01, *p<0.05, (*)p<0.1), two-sided Abbreviations: NS: novelty seeking, HA: harm avoidance, RD: reward dependence, P: persistence, SD: self-directedness, C: cooperativeness, ST: self-transcendence
(0.03 MB DOC)
Correlation between the subscales of the temperament dimensions novelty seeking and the mean fMRI signal (6 to 8 seconds) obtained for the condition low personal relevance. Pearson correlation coefficients [r], significant correlations are labelled (**p<0.01, *p<0.05, (*)p<0.1), two-sided Abbreviations: NS1: novelty seeking subscale 1 - exploratory excitability vs. stoic rigidity, NS2: novelty seeking subscale 2 - impulsiveness vs. reflection, NS3: novelty seeking subscale 3 - extravagance vs. reserve, NS4: novelty seeking subscale 4 - disorderliness vs. regimentation PACC: pregenual anterior cingulate cortex
(0.03 MB DOC)
MNI coordinates of activations for the contrast. Abbreviations: VLPFC: ventrolateral prefrontal cortex, IFG: inferior frontal gyrus, ACC: anterior cingulated cortex, DMPFC: dorsomedial prefrontal cortex, SMA: supplementary motor area, BA32: Brodman Area 32
(0.04 MB DOC)
Correlation between the different temperament dimensions and the mean fMRI signal (6 to 8 seconds) obtained for the conditions high personal relevance, low personal relevance, win and lose. Pearson correlation coefficients [r], significant correlations are labelled (**p<0.01, *p<0.05, (*)p<0.1), two-sided Abbreviations: NS: novelty seeking, HA: harm avoidance, RD: reward dependence, P: persistence, PACC: pregenual anterior cingulate cortex
(0.04 MB DOC)
The authors thank Sascha Moerth and Michael Rotte for their comments on conception and design and Diana Moritz, Ulrike Bruer, Rene Thiemann and Rabea Paus for assistance in data collection and analysis. The authors also thank the staff members of the Department of Neurology for their support and collaboration.