Reward, punishment, cognitive control and context in decision making
|Zusammenfassung der Sitzung|
Reward and punishment processing have tremendous impact on our behavior and decisions. Yet, contexts and cognitive control may alter their impact on behavior. In this symposium, we bring together studies investigating cognitive control and the impact of reward and punishment processing on behavioral and electrocortical outcomes in different contexts.
The first study investigated electrocortical responses of the receiver in an ultimatum game to social cues of successful, costly punishment. The fairness of the offer was considered as well as the reward of getting an adequate social reaction to costly punishment. Further exploring the interrelation of reward and punishment, the second study used two three-armed bandit tasks to investigate feedback-locked frontal midline theta power in reward gain versus punishment avoidance learning. Additionally, personality traits were considered. Adding the context of a second chance, the third study focused on the impact of a second stage as a receiver and relevant personality traits in the ultimatum game on behavior, fairness related reward processing (FRN) and cognitive control related EEG-signals (midfrontal theta). Centralizing cognitive control, the fourth study investigated the influence of cognitive effort investment in a flanker task with varying demands and payoffs. Behavioral reactions as well as midfrontal theta band activations revealed interactions of cognitive effort investment with reward and demand. Focusing on the meta-cognition of control, the fifth study investigated the impact of need for cognition on metacontrol (switching to accurate but cognitively effortful strategies), in a sequential decision-making task. A computational reinforcement-learning model was used to explore this relation.
Neural correlates of social cues that indicate successful costly punishment
1Translational Social Neuroscience Unit, Center of Mental Health, Department of Psychiatry, Psychosomatic and Psychotherapy, University of Würzburg, Würzburg, Germany; 2Department of Psychology I, Institute of Psychology, University of Würzburg, Würzburg, Germany; 3Division Personality Psychology and Psychological Assessment, Freie Universität Berlin, Berlin, Germany
Costly punishment is the attempt to punish interaction partners for deviant behavior, such as violating fairness rules, at personal costs. We examined interaction processes in which a counterpart showed a socio-emotional response to punishment, which indicates whether the punishment was successful or not. In a modified ultimatum game, emotional facial expressions of the proposer in response to the receiver's decision served as feedback stimuli. We focused on the neural correlates of this response and its influence on subsequent decision making. Results from a series of experiments are consistent with the concept of costly punishment as an intentional action following norm-violating behavior, in our case unfair offers by the proposer. Socio-emotional stimuli were found to have a significant influence on perception and behavior in economic negotiations. Especially, smiling facial expressions in response to an accepted offer consistently led to an increase in acceptance rates. Moreover, both rewarding reactions following the acceptance of an offer (smiling compared to neutral facial expression) as well as reactions indicating successful punishment (sad compared to neutral facial expression) elicited a reward positivity in the time frame of the Feedback-related Negativity, indicating that punishment was the intended outcome. Individual differences in depression showed that the reward positivity for smiling facial expressions as feedback stimuli decreased with higher depression scores.
Frontal theta oscillations reflect positive prediction error processing which is amplified in threat avoidance learning and modulated by trait neuroticism/anxiety in monetary reward learning
1University of Marburg, Germany; 2Goldsmiths, University of London, UK
Frontal midline theta (FMθ) has been associated with prediction error processing and trait anxiety and thus has been hypothesized to reflect unspecific cognitive control processing, such as in threat scenarios that elicit state fear and negative affect. However, most studies in reinforcement learning on FMθ and the influence of negative affective personality traits have used paradigms involving reward- rather than threat-related reinforcer. Accordingly, the role of FMθ in threat-related reinforcement learning and its associations with personality traits from fear and anxiety domains is widely unknown. Here n = 105 participants underwent one reward-related and one punishment-related three-armed bandit task. Feedback in the reward task signalled monetary reward (+10 Cent) versus nonreward (+0 Cent) and feedback in the punishment task signalled nonpunishment (no noise burst) vs. punishment (noise burst titrated to match aversiveness of monetary nonreward). While FMθ was amplified following negative vs. positive valent feedback in both tasks, there were task-dependent differences in the processing of positive prediction errors (PE+) which indicated the initiation of subsequent behavioural exploitation. Single-trial regression analyses demonstrated that FMθ scaled with a computationally derived PE+ that was calculated using a state-action value function. This effect was significantly stronger in the punishment vs. reward task but only modulated by trait neuroticism/anxiety in the reward but not punishment task. In line with the adaptive control hypothesis, our results suggest that FMθ as an index of PE+ processing was particularly sensitive for threat avoidance learning and linked to trait neuroticism/anxiety in non-threatening reward learning scenarios.
Play it again Sam! The influence of a second offer in the ultimatum game on decision making of the receiver and their offer related EEG responses.
1Julius-Maximilians Universität Würzburg, Deutschland; 2Universitätsklinikum Würzburg, Deutschland; 3Freie Universität Berlin, Deutschland
In economic neuroscience, the ultimatum game is often used to investigate bargaining behavior. We used an ultimatum game and a two-stage ultimatum game with 92 participants to determine the influence of an additional offer after a previous rejection on the acceptance rates and the electro-cortical responses of the receivers to fair and unfair offers. Additionally, the influence of traits including greed and anxiety on the decisions were investigated.
The results led to the conclusion that higher offers lead to more acceptance in general, but a second stage in the ultimatum game influences the behavioral responses with higher rejection rates if a second offer is available.
Concerning the electrophysiological measurements, the (single-trial) reward positivity as an indicator of fairness evaluation was higher if the offer was more generous. If a second offer was available, this effect was not present. In the final stage of the two-stage ultimatum game, the effect was present again. Midfrontal theta as an indicator of cognitive effort was lower for higher offers in the ultimatum game, while it was higher for more generous offers, if a second stage was still available in the two-stage ultimatum game. There, it could be used as a predictor for the behavioral responses.
Interestingly, the investigated traits only moderated the behavioral responses of the two-stage ultimatum game.
The study shows differences in trait relation and electro-cortical correlates of behavior when a second bargaining stage is added to the ultimatum game. It provides insights about electro-cortical correlates and trait moderations of bargaining responses.
The relationship of cognitive effort investment and reward processing under varying demand
TU Dresden, Deutschland
Traits related to dispositional differences in cognitive effort investment (CEI) like need for cognition (NFC) and self-control have been shown to be related to effort-based decision making. Based on previous findings, we assumed that CEI, as an integrative measure of the willingness and tendency to exert effortful control, would be related to behavioral and psychophysiological indices of effort investment in a typical cognitive control task under varying demand and payoff. Specifically, performance and midfrontal theta power (FMθ) in the electroencephalogram are sensitive to cognitively demanding tasks and were shown to exhibit different patterns for individuals with high vs. low NFC as one core aspect of CEI: performance decreased more strongly for individuals with low vs. high NFC, and high NFC individuals showed a demand-congruent increase in FMθ, pointing to a more efficient allocation of cognitive resources. In the present study (N ~ 145), we examined the relationship of CEI and behavioural measures, i.e., reaction time (RT), error rate (ER) and FMθ, during a flanker task with varying demand and payoff. Whereas the analysis of the behavioral data revealed significant effects for demand (RT, ER), demand x payoff (RT, ER) and payoff x CEI (RT), significant effects regarding FMθ emerged for demand and demand x CEI. Taken together, we mapped CEI onto objective markers of the willingness to exert cognitive effort. Our result may further our understanding of person x situation interactions with regard to effort investment in goal-directed behavior. Open data and reproducible analysis scripts will be available at OSF.
Need for cognition does not account for individual differences in metacontrol of decision making
1Faculty of Psychology, Technische Universität Dresden, Germany; 2Max Planck Institute for Human Development, Germany; 3Cluster of Excellence “Science of Intelligence”, Technische Universität Berlin, Germany; 4Department of Psychology, Concordia University, Canada; 5PERFORM centre, Concordia University, Canada
Humans show metacontrol of decision-making towards different reward magnitudes. Specifically, when higher rewards are at stake, individuals increase their reliance on a more accurate but cognitively effortful, model-based reinforcement-learning strategy. We investigated whether the personality trait Need for Cognition (NFC) explains individual differences in metacontrol. NFC reflects an individual’s intrinsic motivation for cognitively demanding activities and previous studies showed that individuals low in NFC are more strongly affected by different reward magnitudes in how much they engage in cognitively effortful activities. Based on these findings, we expected more reward-based metacontrol in individuals low in NFC. In two independent studies (N = 126 and N = 205), we assessed reliance on model-free and model-based reinforcement-learning strategies by means of a decision making task. Across trials, the magnitude of available rewards was manipulated. We found that participants showed metacontrol, i.e. they relied more strongly on a model-based strategy when rewards were amplified. In contrast to our expectations, NFC did not account for individual differences in metacontrol of decision making in both studies. In fact, a Bayesian analysis provided moderate to strong evidence against a relationship between NFC and metacontrol. Beyond this, NFC was also not related to the overall reliance on model-based reinforcement learning. Our findings show that while individuals differed in their intrinsic motivation to exert cognitive effort, extrinsic rewards modulated their engagement in an effortful decision-making strategy in a similar way. This suggests a differential role of NFC for the regulation of cognitive effort in decision making and executive functions.