41 Adaptive Implementation of Cognitive Control in School-Aged Children with ADHD: A Diffusion-Model Analysis

Abstract

Objective:Because cognitive resources are limited, models of cognitive control predict that additional control is engaged only if it improves task performance. Increased response caution, which occurs when individuals increase the threshold of information needed before making a decision, is one example of cognitive control adaptation. While previous studies have measured increased response caution via increased reaction time, the diffusion model can be used to derive a boundary separation parameter that directly indexes response caution and eliminates capturing alternative influences on reaction time. This study aims to determine if school-aged children, either with or without ADHD, show adaptive changes in response caution during a set-shifting task. These groups have demonstrated mixed results when analyzing reaction time, so this study utilizes diffusion modeling to measure response caution more directly. The set-shifting task presents switches in a random order such that they cannot be predicted; therefore, increasing response caution is only adaptive following errors, called post-error slowing (PES), but not following switch trials. It is predicted that children will show increased response caution only when adaptive. If child with ADHD adapt their response caution fundamentally differently, then there will be individual differences in change in boundary separation.Participants and Methods:Children ages 8-12 with (n=193) and without (n=70) ADHD completed the Navon set-shifting task. Participants saw one of four global shapes made up of local shapes and were asked to identify one or the other based upon the background color. Of the 144 trials, 70 presented a switch between global and local. Trials were presented in the same randomized order for all participants, self-paced, and followed by feedback on correctness. The diffusion model parameters boundary separation (a), drift rate (v), and nondecision time (Ter) were estimated by condition, including a) post-error versus after correct and b) post-switch versus post-same. For PES analyses, only participants with a sufficient number of errors for modeling were included (ADHD n=113, control n=19).Results:Participants were slower on trials immediately following errors (F(1, 130)=119.76, p<.001, n2=.48) and switches (F(1, 261)=154.93, p<.001, n2=.37). In PES, slowing was attributable to increased boundary separation, F(1, 130)=16.11, p<.001, n2=.11, as well as slower drift rate and longer nondecision time (both p<.01, n2 >.05). However, as predicted, post-switch slowing was only attributable slower drift rate and longer nondecision time (both p<.001, n2 >.10), not increased boundary separation, F(1, 261)=0.77, p=.38, n2<.01. Overall, children with ADHD had slower drift rates (F(1, 261)=4.63, p<.001, n2=.10) and narrower boundary separation (F(1, 261)=10.56, p=.001, n2=.04). However, there were no ADHD x trial-type interactions for PES or post-switch (both p>.33, n2<.01).Conclusions:School-aged children demonstrated increased response caution following errors, but not following switches. This demonstrates an adaptive use of cognitive control. The diffusion model was crucial in determining this, as reaction time slowed following switches for reasons unrelated to cognitive control. Additionally, although children with ADHD demonstrated slower drift rates and narrower boundary separation overall, they showed no differences when adapting response caution.