Dissociable effects of valence and arousal in adaptive executive control

Abstract

Background: Based on introspectionist, semantic, and psychophysiological experimental frameworks, it has long been assumed that all affective states derive from two independent basic dimensions, valence and arousal. However, until now, no study has investigated whether valence and arousal are also dissociable at the level of affect-related changes in cognitive processing.

Methodology/principal findings: We examined how changes in both valence (negative vs. positive) and arousal (low vs. high) influence performance in tasks requiring executive control because recent research indicates that two dissociable cognitive components are involved in the regulation of task performance: amount of current control (i.e., strength of filtering goal-irrelevant signals) and control adaptation (i.e., strength of maintaining current goals over time). Using a visual pop-out distractor task, we found that control is exclusively modulated by arousal because interference by goal-irrelevant signals was largest in high arousal states, independently of valence. By contrast, control adaptation is exclusively modulated by valence because the increase in control after trials in which goal-irrelevant signals were present was largest in negative states, independent of arousal. A Monte Carlo simulation revealed that differential effects of two experimental factors on control and control adaptation can be dissociated if there is no correlation between empirical interference and conflict-driven modulation of interference, which was the case in the present data. Consequently, the observed effects of valence and arousal on adaptive executive control are indeed dissociable.

Conclusions/significance: These findings indicate that affective influences on cognitive processes can be driven by independent effects of variations in valence and arousal, which may resolve several heterogeneous findings observed in previous studies on affect-cognition interactions.

Figure 1. Experimental materials and results.

(A) Examples of search displays in the control condition (left panel) and the pop-out distractor condition (right panel). Participants were instructed to search for the tilted pop-out target and to ignore an occasionally occurring luminance pop-out distractor. (B) Left panel: Mean interference effects (reaction time on pop-out distractor trials minus reaction time on control trials) as a function of valence state (negative, positive) and arousal state (high, low). Right panel: Mean conflict-adaptation effects (interference on trials following pop-out distractor trials minus interference on trials following control trials) as a function of valence state (negative, positive) and arousal state (high, low). Error bars indicate standard errors.

Figure 2. Results of the Monte Carlo simulation of Botvinick et al’s (30) model of control.

(A) Effect of the parameters β, λ, and α on mean control. (B) Effect of β, λ, and α on mean control adaptation. (C) Correlation between mean control and mean control adaptation for fixed β, λ with variable α. (D) Correlation between mean control and mean control adaptation for a fixed level of α.