40 Temporal discounting and male depression

Abstract

Male depression is an attempt to extend the definition of depression to address an atypical cluster of symptoms that have been associated with suicide completion in community-dwelling populations. These include externalizing behaviour, avoidance, emotional suppression, irritability, substance misuse, and risk-seeking behaviour. Temporal discounting is a measure of economic impulsivity and a possible transdiagnostic neurocognitive domain that plays an essential role in everyday decision-making. Previous studies have found mixed results relating economic impulsivity to depression. We hypothesized that this divergence could be attributed to the heterogeneity of depression and contrasted temporal discounting between general ICD-10 depression and the specific, and possibly more impulsive male depression.We administered a temporal discounting task to 170 in-ward psychiatric patients and 176 healthy controls (HC). All patients fulfilled the ICD-10 criteria for moderate to severe depression. The patients were separated into a male-depressed group (MD) and a general major depression group (MDD) using median split of the Male Depression Risk Score (MDRS-22). HC had subthreshold scores for depression and had no neuropsychiatric history.We used the Bayesian information criterion (BIC) to compare 3 models of delay discounting. All models were based on the hyperbolic discounting model Q(R,D,k)=R/(1+k*D) where R =reward, k=discounting rate, and D=delay, comparing the subjective value of the immediate option Qi=Q(Ri,0,k) against the delayed option Qd=Q(Rd,D,k). Models differed in their treatment of noise in generating the probability piof making the impulsive choice of the immediate option (withσ(z) = 1/(1 + exp(−z))):(1) preference-temperature model (k, β): pi=σ(β(Qi– Qd))(2) preference-uncertainty model (μ, σ): draw log ks~N(μ, σ):if Qi> Q(Rd,D, ks), pi=1; otherwise pi=0(3) trembling-hand model (k, β, lapse):λ = (3/56)* σ(lapse)+1e-4 pi=(1–2λ(σ(β(Qi– Qd))+λThe preference-temperature model yielded the best fit (BIC(1)=3.817, BIC(2)=3.893, BIC(3)=7.001). There was no significant group difference in the model-specific parameters: discount ratekand inverse temperatureβ. There was a correlation between choosing the immediate reward and MDRS-22 (r=0.234, p<0.001), as well as with Beck’s Depression Inventory (BDI-II) (r=0.191, p<0.001), but no significant group differences (ANOVA: F=0.454, p=0.501).We could not find any model-based differences between the three groups. However, model- agnostic results suggested that male depressiveness, but also ICD-10 depressiveness, were associated with more impulsive choices. A noteworthy limitation was a strong correlation between BDI-II and MDRS-22 in our sample.