DATA_for_JASP_Anxiety.xlsx (55.3 kB)
Download file

Risk and Loss aversion in pathologically anxious and matched control individuals

Download (55.3 kB)
This dataset contains data collected on 25 pathologically anxious individuals (meeting criteria for GAD) and 23 matched healthy controls on a task measuring gambling behavior, its modulation by incidental emotional cues, and working memory for these emotional cues.
Computational modelling derived from Prospect Theory was employed to formally estimate risk and loss aversion parameters from the gambling data, revealing that relative to healthy controls, anxious individuals exhibit enhanced levels of risk aversion, but no difference in loss aversion.
There was no effect of emotional cues on gambling or working memory (and no group differences on these variables).
The dataset shows parameter estimates from the Prospect Theory model: Mu (inverse temperature, free parameter), Lambda (loss aversion), and Rho (risk preference). The parameters were log-transformed because of their distribution (Log Mu, Log Loss Aversion, Log Risk Aversion). A Maximum A Posteriori (MAP) estimation procedure was also run to corroborate the main results, leading to another set of parameter estimates labelled "MAP". The next columns of the data set shows group (anxious vs control), MDD group (whether the anxious individuals also met criteria for Major Depression), Trait anxiety and BDI scores. Log-transformed parameter values are then shown for each emotion condition separately. Then come gamble propensity data, calculated across all trials, separately for each gamble type, and then for each emotion condition; followed by working memory performance and reaction time data.
For more details please refer to the published version of the paper:
Charpentier CJ, Aylward J, Roiser JP, Robinson OJ (2017) Enhanced Risk Aversion, But Not Loss Aversion, in Unmedicated Pathological Anxiety. Biological Psychiatry. in press


UCL Grand Challenge award, MRC Career Development Award (Grand No. MR/K024280/1)