A human subcortical network underlying social avoidance revealed by risky economic choices

5 figures and 2 additional files

Figures

Task.

(A) Trial structure with human partner. Participants chose between a risky and a safe option. Risky options were monetary gambles, in the form of a virtual game of dice played against a partner presented as an image, with two known potential outcomes (0 or 3 Euros). In the fMRI task, these images were replaced with the name of the partner, learned before the scan. Choosing the safe option led to the receipt of a certain monetary amount, which was varied across trials (0 to 3 Euros, 0.5 Euro steps, equal probability, random order). Choosing the risky option resulted in the display of the outcome of the gamble (win, 3 Euro or no win, 0 Euro) and the display of the partner’s response. At the end of the game, the amount gained in one randomly selected trial was paid out in real currency. (B) Trial structure with computer partner, identical to trial with human partner except for the fact that the partner was a computer. (C) The six human partners with neutral facial expression. (D) One of the partners’ faces displaying no emotion, admiration and condescension (left to right); the emotional expressions were presented as videos.

https://doi.org/10.7554/eLife.45249.003
Behavioural results showing an association between the value of social engagement and social anxiety level.

(A) Probability of choosing the safe (no gamble) option instead of the gamble as a function of the amount offered in the safe option, with fitted cumulative Gaussian functions (red and blue lines). The value of engaging in each kind of gamble was estimated as the certainty equivalent of the gamble (CE50), which is the amount offered in the safe option associated with 50% safe choices. Example data are shown from participants with anxiety score (LSAS) below (A) and above the median (B): less anxious participants tended to choose the safe option against the computer partner at lower amounts than against the human partner, indicating a tendency to seek the social situation; more anxious participants showed the opposite tendency, indicating social avoidance. Each marker is the mean of 10 trials. (C) The value of social engagement (CE50 Human – CE50 Computer) decreased with social anxiety score (LSAS) (Study 1; n = 68). (D) Replication of the findings in Study 2 (n = 47); same display as in B. P(safe)=probability of choosing the ‘safe’ option [0–1]. Grey-shaded areas represent the 95% confidence intervals around the slope of the regression line.

https://doi.org/10.7554/eLife.45249.004
Amygdala and ventral striatum activation during the task.

(A) Decision (safe or risky) could be decoded with 56.4% accuracy from amygdala activation. Decoding accuracy map resulting from an MVPA searchlight analysis of the BOLD signal obtained at the time of the decision, thresholded at p<0.05 FWE-corrected across anatomically-defined amygdalae. (B) Amygdala response to game outcome: response to feedback from the human partner (facial expression) minus the response to feedback from the human partner (abstract symbol). Threshold was p<0.05 FWE-corrected for multiple comparisons at the voxel level across the whole brain. (C) Clusters in bilateral nucleus accumbens showed significant activation increases in response to wins compared to no wins. Threshold was p<0.05 FWE-corrected for multiple comparisons at the voxel level within the anatomically-defined ventral striatum. R: right; y = MNI coordinate.

https://doi.org/10.7554/eLife.45249.005
Variation of the neural activation as a function of choice, partner, social anxiety level (LSAS score) and the value of social engagement (CE50 human – CE50 computer) during the decision.

(A–C) and the outcome (D–F) phases of the experiment. (A) During the decision phase of the trial (no face was shown), the left amygdala response was higher when participants chose the risky option against the human rather than the computer; this difference was not found when participants chose the safe option. (B) Similar effects were observed in the right amygdala. (C) In the left amygdala, the response difference between risky choices with human vs. computer partners and safe choices with human vs. computer partners (interaction: risky - safe choices X human - computer) decreased with the measured value of social engagement. (D) During the outcome stage of the game, the left amygdala response to outcomes of gambles against a human (=emotional expressions) compared to a computer increased with LSAS and (E) decreased with the measured value of social engagement. (F) In the right nucleus accumbens, the response difference between winning against a human partner rather than a computer partner decreased with LSAS. A.u.: arbitrary units. In panels A and B, central tendency is the mean and indicated by bar height, error bars indicate standard error of the mean. Grey-shaded areas represent the 95% confidence intervals around the slope of the regression line.

https://doi.org/10.7554/eLife.45249.006
Functional connectivity associated with the decision to engage in the social situation.

(A) Connectivity between left NAcc and right amygdala increased with the level of social anxiety. (B) Connectivity between right NAcc and right amygdala showed the same effect. (C) Cluster of voxels in the perigenual anterior cingulate cortex (pACC, BA24a/b; circle) showing functional connectivity modulation with right nucleus accumbens (arrow) as a function of the partner and social anxiety. Threshold was p<0.05 FWE-corrected for multiple comparisons across the whole brain at the cluster level (based on p<0.001 uncorrected threshold, minimum cluster size = 100 voxels). (D) Connectivity between right NAcc and pACC decreased proportionally to the level of social anxiety. Abbreviations as in Figures 3 and 4; pACC: perigenual anterior cingulate.

https://doi.org/10.7554/eLife.45249.007

Additional files

Supplementary file 1

Supplementary behavioural results.

Four tables displaying additional results: Supplementary file 1A and 1B show descriptive statistics for questionnaire data from Study one and Study two respectively, Supplementary file 1C and 1D show all steps of a backward stepwise regression analysis for the experiment data of Study one and Study 2, respectively.

https://doi.org/10.7554/eLife.45249.008
Transparent reporting form
https://doi.org/10.7554/eLife.45249.009

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Johannes Schultz
  2. Tom Willems
  3. Maria Gädeke
  4. Ghada Chakkour
  5. Alexander Franke
  6. Bernd Weber
  7. Rene Hurlemann
(2019)
A human subcortical network underlying social avoidance revealed by risky economic choices
eLife 8:e45249.
https://doi.org/10.7554/eLife.45249