Evidence accumulation and associated error-related brain activity as computationally-informed prospective predictors of substance use in emerging adulthood

Abstract

Rationale: Substance use peaks during the developmental period known as emerging adulthood (ages 18-25), but not every individual who uses substances during this period engages in frequent or problematic use. Although individual differences in neurocognition appear to predict use severity, mechanistic neurocognitive risk factors with clear links to both behavior and neural circuitry have yet to be identified. Here, we aim to do so with an approach rooted in computational psychiatry, an emerging field in which formal models are used to identify candidate biobehavioral dimensions that confer risk for psychopathology.

Objectives: We test whether lower efficiency of evidence accumulation (EEA), a computationally characterized individual difference variable that drives performance on the go/no-go and other neurocognitive tasks, is a risk factor for substance use in emerging adults.

Methods and results: In an fMRI substudy within a sociobehavioral longitudinal study (n = 106), we find that lower EEA and reductions in a robust neural-level correlate of EEA (error-related activations in salience network structures) measured at ages 18-21 are both prospectively related to greater substance use during ages 22-26, even after adjusting for other well-known risk factors. Results from Bayesian model comparisons corroborated inferences from conventional hypothesis testing and provided evidence that both EEA and its neuroimaging correlates contain unique predictive information about substance use involvement.

Conclusions: These findings highlight EEA as a computationally characterized neurocognitive risk factor for substance use during a critical developmental period, with clear links to both neuroimaging measures and well-established formal theories of brain function.

Keywords: Computational psychiatry; Diffusion model; Drift rate; Evidence accumulation; Salience network; Substance use.

Fig. 1

a Schematic of the go/no-go task. Participants are presented with a string of letters and are instructed to respond on trials with any letter other than “X” (“go” trials) but to withhold from responding on trials where an “X” is presented (“no-go” trials). b Simplified schematic of the diffusion decision model (DDM) description of the go/no-go task in cases with relatively high (top) vs. low (bottom) efficiency of evidence accumulation (EEA). The DDM assumes that responses on choice response time (RT) and go/no-go tasks are the result of a decision variable that drifts over time, on the basis of noisy evidence gathered from the stimulus, until it reaches one of two boundaries which represent each possible choice (e.g., to respond vs. withhold). When decision processes on individual “go” and “no-go” trials, which are represented by the light green and light red traces, respectively, terminate at one of the boundaries, the corresponding choice is made. The boundaries are set at 0 and parameter a, and the decision variable begins at a starting value z. A non-decision time (Ter) parameter accounts for time taken up by processes peripheral to the decision (e.g., the motor response). The drift rate parameter (v) determines the average rate at which the decision variable drifts toward the boundary for the correct choice (v.go for “go” trials, v.nogo for “no-go trials) and can be used as a measure of EEA in individual differences analyses. Lines above the “respond” boundaries represent RT distributions for responses on correct “go” trials (green) and failed “no-go” trials (red). The case with lower EEA exhibits fewer correct “go” responses, more incorrect “no-go” responses, and more variable RTs (a greater proportion of long RTs in the skewed right tail) due to lower v

Fig. 2

Spheres (8-mm radius) centered about coordinates (see Table 3 in “Results” section) of our 8 regions of interest (ROIs; white numbers in lower right indicate z-coordinates)

Fig. 3

Scatterplots and simple regression lines of the association between our summary measure of error-related activation (PC1) and individuals’ drift rates from the go/no-go task, including drift rate on “go” trials (v.go), drift rate on “no-go” trials (v.nogo), and the composite drift rate measure from across all trials (v.avg). The association between PC1 and false alarm (FA) rates on “no-go” trials, the primary index of inhibition on the task, is also displayed for comparison. Red points and lines represent these relationships for the subsample of individuals with substance use outcome data (n = 106) and pink points and lines represent the same relationships when all individuals in the full sample (N = 143) are included

Fig. 4

Scatterplots of associations in which drift rates (v.go, v.nogo, v.avg) and our summary measure of error-related activation (PC1) predict individuals’ values of the age 22–26 substance use composite (SC). Simple regression lines are displayed in black