A Low-Level Perceptual Correlate of Behavioral and Clinical Deficits in ADHD

Abstract

In many studies of attention-deficit hyperactivity disorder (ADHD), stimulus encoding and processing (perceptual function) and response selection (executive function) have been intertwined. To dissociate deficits in these functions, we introduced a task that parametrically varied low-level stimulus features (orientation and color) for fine-grained analysis of perceptual function. It also required participants to switch their attention between feature dimensions on a trial-by-trial basis, thus taxing executive processes. Furthermore, we used a response paradigm that captured task-irrelevant motor output (TIMO), reflecting failures to use the correct stimulus-response rule. ADHD participants had substantially higher perceptual variability than controls, especially for orientation, as well as higher TIMO. In both ADHD and controls, TIMO was strongly affected by the switch manipulation. Across participants, the perceptual variability parameter was correlated with TIMO, suggesting that perceptual deficits are associated with executive function deficits. Based on perceptual variability alone, we were able to classify participants into ADHD and controls with a mean accuracy of about 77%. Participants' self-reported General Executive Composite score correlated not only with TIMO but also with the perceptual variability parameter. Our results highlight the role of perceptual deficits in ADHD and the usefulness of computational modeling of behavior in dissociating perceptual from executive processes.

Keywords: ADHD; executive function; psychophysics; task-switching; variability; visual perception.

Figure 1.. Task design. A) Trial sequence example. A feature dimension cue indicated whether orientation (cross)–depicted here–or color (colored circle) was relevant, while a simultaneous endogenous spatial cue (line segment) indicated which side (left or right) was relevant. Thus, the participant received one of four possible cue screens. We always chose the spatial cue randomly. The participant had to respond whether the orientation of the ellipse on the relevant side was clockwise or counterclockwise with respect to vertical or whether its color was more yellow or more blue, with the associated set of keys (left or right). The color and orientation continua are shown above the stimulus screen, with the dashed line at vertical and respectively mid-level green. To respond, the participant could press any one of eight keys, but only two were task-relevant on a given trial; the other six keys being considered task-irrelevant motor output. The participant received correctness feedback. B) Left: Cue–relevant stimulus–relevant response buttons pairings for the four types of trials as they arise from the four feature and spatial cue combinations (2 × 2). Relevant is marked with pink for visualization only. Pressing any other button would result in task-irrelevant motor output. Right: During Ori and Col blocks, only two types of trials are possible, while during Switch blocks, all four trial types are possible.

Figure 2.. Dissociation of perceptual and executive processes. Schematic of the early perceptual encoding and late stimulus–response rule selection (executive) processes that may play a role in this task, and the corresponding task metrics.

Figure 3.. ADHD participants had higher task-irrelevant motor output and longer and more variable reaction times. A) Proportion of TIMO across conditions. Here and elsewhere, values represent medians across participants and error bars the bootstrapped 95% confidence intervals. B) Empirical cumulative density functions of reaction times, collapsed across all conditions. Thin lines: individual participants. Thick lines: median for the RT distribution collapsed across all participants in a group. C) Reaction time median by condition and group. Throughout the article, we use RT median because reaction time distributions are not Gaussian. D) Reaction time variability metric, the τ parameter from ex-Gaussian distribution fits, by condition and group.

Figure 4.. Fitted psychometric curves and parameters; attention-deficit hyperactivity disorder participants had higher perceptual variability. A) Psychometric curve fits across all conditions. Here and elsewhere, n.u. stands for normalized units. Thin lines: individual participants. Thick lines: medians for each group. For fits overlaid on top of data, see Mihali et al. (, Appendix, Figure A8). B) Perceptual variability parameter values, medians, and bootstrapped 95% confidence intervals. Top inset plot: black psychometric curve has low noise, while the gray has higher noise. C) Lapse rate. Top inset plot: black psychometric curve has low lapse, while the gray has higher lapse.

Figure 5.. Logistic regression based on task metrics can classify participants into attention-deficit hyperactivity disorder and controls with accuracies larger than 70%. A) Dots: combinations of log TIMO and log perceptual variability (σ) across participants. Dashed lines: logistic regression classifiers trained on log σ only (olive), TIMO only (old rose), and both (black). B) Full receiver operating characteristic curves obtained by varying the diagnosis threshold for the three classifiers in A, as well as for one based on all five behavioral metrics (purple). C) Full ROC curves, this time with stratified 10-fold cross-validation, for the same classifiers as in B.