Dissociable influences of reward motivation and positive emotion on cognitive control

Abstract

It is becoming increasingly appreciated that affective and/or motivational influences contribute strongly to goal-oriented cognition and behavior. An unresolved question is whether emotional manipulations (i.e., direct induction of affectively valenced subjective experience) and motivational manipulations (e.g., delivery of performance-contingent rewards and punishments) have similar or distinct effects on cognitive control. Prior work has suggested that reward motivation can reliably enhance a proactive mode of cognitive control, whereas other evidence is suggestive that positive emotion improves cognitive flexibility, but reduces proactive control. However, a limitation of the prior research is that reward motivation and positive emotion have largely been studied independently. Here, we directly compared the effects of positive emotion and reward motivation on cognitive control with a tightly matched, within-subjects design, using the AX-continuous performance task paradigm, which allows for relative measurement of proactive versus reactive cognitive control. High-resolution pupillometry was employed as a secondary measure of cognitive dynamics during task performance. Robust increases in behavioral and pupillometric indices of proactive control were observed with reward motivation. The effects of positive emotion were much weaker, but if anything, also reflected enhancement of proactive control, a pattern that diverges from some prior findings. These results indicate that reward motivation has robust influences on cognitive control, while also highlighting the complexity and heterogeneity of positive-emotion effects. The findings are discussed in terms of potential neurobiological mechanisms.

Fig. 1

General schematic of the study structure. Participants came for two experimental sessions: the emotion session (AX-CPT under neutral and positive emotion conditions) and the reward session (AX-CPT under baseline and reward conditions). A Self-Assessment Manikin (SAM) was administered at intervals throughout the sessions, and passive viewing of IAPS images was completed following the emotion AX-CPT task runs

Fig. 2

AX-CPT trial structure. In the neutral block, only neutral IAPS images were presented. In the positive block, neutral and positive IAPS images were intermixed and presented randomly on each trial. In baseline and reward trials, only two neutral IAPS images were presented per participant (chosen via random counterbalance): Participants were told in the baseline block that these images had no meaning, and in the reward block they were explicitly informed which one signified incentive and which did not

Fig. 3

Task performance measures in the reward and emotion sessions. The top panels show trial-related incentive effects (Rew–Neut vs. Rew–Rew trials) and block-related incentive effects (Baseline–Rew vs. Rew–Neut trials) in (a) error rates and (b) RTs; the bottom panels show trial-related emotion effects (Pos–Neut vs. Pos–Pos trials) and block-related emotion effects (Baseline–Emot vs. Pos–Neut trials) in (c) error rates and (d) RTs. Significant contrasts (p < .05) are marked with asterisks

Fig. 4

Proactive index measures in each condition for the reward session (Baseline–Rew, Rew–Neut, Rew–Rew) and the emotion session (Baseline–Emot, Pos–Neut, Pos–Pos): (a) calculated with error rates; (b) calculated with RTs. Significant contrasts (p < .05) are marked with asterisks

Fig. 5

Pupil time courses for the block-based incentive contrast in (a) the reward session (Baseline–Rew vs. Rew–Neut) and (b) the emotion session (Baseline–Emot vs. Pos–Neut). The pretrial period (–200 to 0 ms), during which pupil magnitudes were analyzed for this contrast, is shaded

Fig. 6

Block-based incentive (reward session data) and emotion (emotion session data) effects, as averaged pupil magnitudes in the pretrial period (–200 to 0 ms). Significant contrasts (p < .05) are marked with asterisks

Fig. 7

Pupil time courses for the trial-based incentive contrast in (a) the reward session (Rew–Neut vs. Rew–Rew) and (b) the emotion session (Pos–Neut vs. Pos–Pos). The cue maintenance period (2,550–2,800 ms), during which pupil magnitudes were analyzed for this contrast, is shaded

Fig. 8

Trial-based incentive (reward session data) and emotion (emotion session data) effects as averaged pupil magnitudes in the cue maintenance period (2,550–2,800 ms). Significant contrasts (p < .05) are marked with asterisks

Fig. 9

Scatterplots showing significant and trend-level pupil–behavior correlations: (a) Baseline–Rew RT proactive index versus pupil dilation for B trials (p = .04); (b) Rew–Neut error proactive index versus pupil dilation for B cues (p = .024); (c) Rew–Rew minus Rew–Neut proactive indices (from errors) plotted against pupil dilation for B cues (p = .085); (d) Pos–Pos minus Pos–Neut proactive indices (from errors) plotted against pupil dilation for B cues (p = .053)