Article Figures & Data
Figures
- Figure 1.
The gambling task and the probability weighting function of prospect theory. A, Each trial consisted of a self-paced choice between a sure option (on the left) and a risky gamble (on the right), followed by visual confirmation of the choice (a frame around the chosen option) and fixation. The sure amount in the next trial was adjusted based on the choice (increased if gamble was chosen, decreased if the sure option was chosen), with the gamble being fixed. After six choices, the sure amount that was reached provided an indifference point between the two options, defined as the certainty equivalent of the gamble. A new series of choices involving a new gamble was then started (in total, 10 gambles in the gain domain and 10 gambles in the loss domain). No feedback was provided on the outcome of the choices. B, The solid black line represents a typical probability weighting function, with overweighting of low probabilities and underweighting of moderate to high probabilities. The dashed diagonal line represents neutrality with regard to sensitivity to probabilities.
- Figure 2.
Dopaminergic modulation of probability distortion. A, Boxplot illustrating the drug effect (sulpiride-placebo) on the distortion parameter γ of the probability weighting function in the gain domain, across all participants. Box height represents the interquartile range (IQR), black line represents the median, and whiskers represent the largest and smallest values no further than 1.5*IQR. Single data points are values located outside the whiskers. B, Within-subject paired observations of γ estimates in the placebo and sulpiride conditions for both experimental groups (different illustration of the result presented in Fig. 2A). C, Fitted probability weighting function, based on the median estimates of δ (elevation) and γ (distortion) parameters across all participants. The shaded areas illustrate the variance of γ across participants, with the boundaries corresponding to the probability weighting function plotted with median δ, and 25th and 75th percentile γ.
- Figure 3.
Fitted probability weighting function based on group median estimates of δ (elevation) and γ (distortion). Across groups, sulpiride decreased probability distortion in the gain domain compared with placebo (A, B). When examining the placebo condition alone, pathological gamblers showed a decreased elevation of their probability weighting function in the loss domain compared with healthy controls (C, D).
Tables
Variable Healthy Controls (n = 21) Pathological Gamblers (n = 16) p Range M SD Range M SD Age 18–52 32.1 11.4 21–50 35.8 8.8 0.29 Net income (€) 0–3570 1691 1123 750–3250 1750 949 0.87 Body mass index 18.3–30.9 23.1 3.2 20.8–26.9 23.9 2.0 0.38 SOGS 0–2 0.2 0.5 6–18 12.4 3.9 <0.001 FTND 0–5 0.6 1.4 0–8 2.5 2.9 0.014 Number of current smokers – 10 – – 10 – 0.37 AUDIT 0–14 6.2 3.8 0–15 7.7 4.6 0.27 HADS anxiety 0–12 2.7 2.8 1–12 4.9 3.4 0.035 HADS depression 0–10 1.6 2.3 0–15 4.9 4.4 0.006 Verbal IQ 85–128 106 9.5 77–123 103 12.3 0.43 M, Mean; SD, Standard Deviation; FTND, Fagerström test for nicotine dependence (Heatherton et al., 1991); AUDIT, alcohol use disorders identification test (Saunders et al., 1993); HADS, hospital anxiety and depression scale (Zigmond and Snaith, 1983).
Gamble index i Variable 1 2 3 4 5 6 7 8 9 10 x 1200 1200 600 1200 600 1000 1200 1200 1200 1200 p 1/6 2/6 2/6 2/6 2/6 2/6 2/6 3/6 4/6 5/6 y 0 0 0 600 300 400 900 0 0 0 x is the larger amount of money in the gamble that could be won or lost with probability p; y is the smaller amount of money in the gamble that could be won or lost with probability 1 − p. x and y are in €. For losses, the amounts of money were the same but negative.
Parameter Controls Gamblers Placebo Sulpiride Placebo Sulpiride Mdn IQR Mdn IQR Mdn IQR Mdn IQR αgains 0.74 0.66 0.80 0.42 0.80 0.37 1.16 0.81 αlosses 1.02 1.14 1.28 0.64 1.92 1.46 1.36 0.95 δgains 0.99 0.86 0.93 0.85 0.90 0.51 0.68 0.61 δlosses 1.08 0.80 0.83 0.61 0.42 0.31 0.78 0.99 γgains 0.55 0.50 0.66 0.54 0.64 0.35 0.89 0.49 γlosses 0.97 0.61 1.06 0.60 0.88 0.78 0.72 0.72 Mdn, median; IQR, interquartile range.
Extended Data 1
Raw data and MATLAB code for the prospect theory parameter estimation procedure implemented in the study. The code includes scripts for the main parameter estimation steps, parameter recovery, and checking the quality of the estimation. Download Extended Data 1, ZIP file.
In this issue
