Skip to main content
. 2016 Jun 1;41(11):2658–2667. doi: 10.1038/npp.2016.68

Figure 3.

Figure 3

(a) Mean gambling percentage as a function of drug (L-DOPA vs placebo) and body weight (high vs low body weight). Larger average gambling percentage under L-DOPA compared with placebo was observed in low-weight (t(15)=2.2, p=0.044) but not high-weight participants (t(15)=0.5, p=0.625). (b) Relationship between the effect of L-DOPA (relative to placebo) on average gambling percentage and body weight, where blue circles represent females (n=16) and green circles represent males (n=16). We observed a significant inverse partial correlation between average gambling percentage under L-DOPA and body weight, controlling for average gambling percentage under placebo (b; r(29)=−0.378, p=0.043). A significant inverse partial correlation was found within males (r(13)=−0.516, p=0.049) but not within females (r(13)=−0.078, p=0.782), possibly because of a ceiling effect, as females had substantial lower weight than males. This suggests that weight and not gender mediated an effect of drug on average gambling percentage. (c) Difference in gambling percentage between L-DOPA (white bars) and placebo (grey bars) in different groups of subjects and different conditions. Participants are grouped according to two dimensions: (i) weight, where data for high and low body weight groups, created based on a median split, are, respectively, displayed in the first and second row of panels; (ii) gambling slope (ie, the beta weight associated with EV in a logistic regression model of gambling choice), where data for participants with negative and positive gambling slope in the first session are displayed in the first and second column of panels, respectively. For each group of participants, mean gambling percentage (on the y axis) is reported for each of four standardized bins of increasing EV separated for each context (LC: low-value context; HC: high-value context). Considering low body weight participants alone, this figure shows an increased mean gambling percentage with L-DOPA compared with placebo in all EV bins and contexts and for both positive and negative gambling slope participants (bin 3 in high value context for positive gambling slope participants is an exception but not statistically significant). This is confirmed by a lack of an interaction between the different bins and the drug/placebo manipulation.