Human oscillatory activity associated to reward processing in a gambling task

https://doi.org/10.1016/j.neuropsychologia.2007.07.016Get rights and content

Abstract

Previous event-related brain potential (ERP) studies have identified a medial frontal negativity (MFN) in response to negative feedback or monetary losses. In contrast, no EEG correlates have been identified related to the processing of monetary gains or positive feedback. This result is puzzling considering the large number of brain regions involved in the processing of rewards. In the present study we used a gambling task to investigate this issue with trial-by-trial wavelet-based time–frequency analysis of the electroencephalographic signal recorded non-invasively in healthy humans. Using this analysis a mediofrontal oscillatory component in the beta range was identified which was associated to monetary gains. In addition, standard time–domain ERP analysis showed an MFN for losses that was associated with an increase in theta power in the time–frequency analysis. We propose that the reward-related beta oscillatory activity signifies the functional coupling of distributed brain regions involved in reward processing.

Introduction

In order to successfully navigate through a busy day, we need to constantly assess the values and uncertainties attached to different options and to adapt our behavior according to the outcome of an action which might or might not match our predictions and hopes. The function of positive (rewards) and negative feedback signals (punishments) in this scenario is to guide behavior and to mediate learning (Schultz, 2006). The brain network activated in reward processing comprises the orbitofrontal cortex, amygdala, ventral striatum/nucleus accumbens, prefrontal cortex and anterior cingulate cortex (Delgado, Nystrom, Fissell, Noll, & Fiez, 2000; Gottfried, O’Doherty, & Dolan, 2003; Knutson, Fong, Bennett, Adams, & Hommer, 2003; Knutson, Westdorp, Kaiser, & Hommer, 2000; O’Doherty, Kringelbach, Rolls, Hornak, & Andrews, 2001). However, although the neural circuit involved in reward processing is quite well defined, the specific roles of each region and the integration of information in this circuit are not well understood.

Several authors have proposed that in order to integrate a disparate number of different rewards the brain uses a common network that converges in a final pathway that informs about the nature of the reward (comparison process) and about the possible courses of action in the future (Montague & Berns, 2002; Shizgal, 1997). Neurophysiological studies in animals revealed dopaminergic neurons in the midbrain projecting, among other regions, to the ventral striatum responding selectively to unpredicted events: (i) they are mostly responsive to appetetive events that are better than predicted, (ii) they do not respond to well-predicted rewards, and (iii) a negative signal (i.e. decreased activity) is elicited when an appetetive event is worse than predicted (Mirenowicz & Schultz, 1994; Schultz, Dayan, & Montague, 1997; Tremblay & Schultz, 2000; for similar results in humans: Berns, McClure, Pagnoni, & Montague, 2001; O’Doherty et al., 2001). The pattern observed in these neurons contrasts to that observed in prefrontal, anterior, and posterior cingulate cortex neurons which exhibit only a unidirectional “error signal” (Ito, Stuphorn, Brown, & Schall, 2003; McCoy, Crowley, Haghighian, Dean, & Platt, 2003; Watanabe, 1989, however see Kim, Shimojo, & O’Doherty, 2006; Tom, Fox, Trepel, & Poldrack, 2007).

In non-invasive electrophysiological studies in humans using event-related brain potentials (ERPs), a similar “error signal” has been observed emanating from the medial prefrontal cortex (Carter et al., 1998, Ullsperger and von Cramon, 2001). Frontocentral negativity, called error-related negativity (ERN, Falkenstein, Hohnsbein, Hoormann, & Blanke, 1990; Gehring, Goss, Coles, Meyer, & Donchin, 1993) has been recorded after performance errors. In addition, a similar medial frontal negativity (MFN) has also been observed after feedback informing that a response had been incorrect (Holroyd & Coles, 2002; Muller, Moller, Rodriguez-Fornells, & Munte, 2005) or after feedback informing about the amount of money lost in a gambling task (Gehring & Willoughby, 2002). Gehring and Willoughby (2004) have used a Morlet wavelet-based time–frequency analysis of the average ERN and MFN components to analyze oscillatory activity underlying these ERP responses. A theta oscillatory response (4–7 Hz) with a maximum at midline electrodes was found in relation to both ERP components. This result corroborated previous time–frequency studies of the ERN component (Luu, Tucker, & Makeig, 2004; Yordanova, Falkenstein, Hohnsbein, & Kolev, 2004). As the scalp distribution of the theta oscillatory activity was slightly lateralized to the right and more anterior than for the MFN compared to the ERN component, their neural generators may be partially distinct (Nieuwenhuis, Slagter, von Geusau, Heslenfeld, & Holroyd, 2005; Muller et al., 2005). Luu and Tucker (2001) have suggested that the midline theta oscillatory process underlying the ERN component may reflect the broad coordination of several brain regions (which include the anterior cingulate cortex and several subcortical regions) that participate in the action regulation system. This system might be involved in learning the appropriateness of a behavior in a specific context, monitoring the outcome of actions and switching to a different behavior when unexpected outcomes occur.

One important limitation of the Gehring and Willoughby (2004) study is that the time–frequency analysis was applied only to averaged ERP activity, which only accounts for the electrical activity reflected by the evoked potential. In contrast, single trial time–frequency analysis provides information that cannot be gleaned from averaged ERPs or time–frequency analysis of average data (see, for example, Makeig et al., 2002; Tallon-Baudry, Bertrand, Delpuech, & Pernier, 1997): (i) there is no information loss due to the averaging process in the time or spectral domains; (ii) it is possible to study systematic variations between single trials, and (iii) changes of power of non-phase-locked activity in a given frequency band can be assessed (Tallon-Baudry et al., 1997), which is especially important for higher frequencies. Thus, fine-grained wavelet-based time–frequency analysis could help to solve the question as to why no specific electrophysiological effects have been recorded for affirmative feedback information, monetary gains or correct responses using standard time–domain ERP analysis or time–frequency analysis applied to averaged ERP data. In the present study we therefore employed trial-by-trial wavelet-based time–frequency analysis in addition to standard ERP analysis in a simplified version of the gambling task developed by Gehring and Willoughby (2002).

Section snippets

Participants

Twenty-five right-handed healthy undergraduate psychology students of the University of Barcelona participated in the experiment (seven men, mean age 23.7 ± 5.4 (S.D.)) for monetary compensation. Written consent was obtained prior to the experiment. The experiment was approved by the local ethical committee.

Design

We used a simplified version of the gambling task (Gehring & Willoughby, 2002) in which valence (reward/monetary gain or punishment/monetary loss) and correctness (correct or incorrect choice)

Results

The mean gain of the participants after performing the 768 trials was 7.4 ± 467.3 points. Ten participants ended the experiment gaining some points, 14 participants lost points and one participant ended the experiment with 0 points.

Discussion

Rewards and punishments are major forces in the modification of behavior. Using event-related brain potentials in the human, a negative component (MFN) has been observed for feedback information signaling negative outcomes (e.g. monetary loss; Gehring & Willoughby, 2002; Holroyd & Coles, 2002; Muller et al., 2005, Nieuwenhuis et al., 2005). By implementing trial-by-trial time–frequency analysis we were able to study the relationship between oscillatory activity and the processing of reward and

Conclusion

Time–frequency analysis of feedback related activity revealed two different oscillatory responses. First, a frontocentral theta component was seen in response to negative feedback, which underlies the previously described medial frontal ERP negativity. By contrast, a high-frequency frontocentral beta response was associated with positive feedback (monetary gains), whose function might be to functionally couple different brain areas during reward processing.

Acknowledgements

This research was supported by research grants of Spanish Government (SEJ2005-06067/PSIC to ARF and SEC2001-3821C0501 to AAP), the Ramon y Cajal research program and the Volkswagenstiftung to ARF and TFM. TFM is also supported by the DFG and the BMBF. Special thanks to Estela Camara, Anna Mestres-Missé and Lluis Fuentemilla for their technical help and comments at various stages of the project.

References (56)

  • P. Luu et al.

    Frontal midline theta and the error-related negativity: Neurophysiological mechanisms of action regulation

    Clinical Neurophysiology

    (2004)
  • A.N. McCoy et al.

    Saccade reward signals in posterior cingulate cortex

    Neuron

    (2003)
  • P.R. Montague et al.

    Neural economics and the biological substrates of valuation

    Neuron

    (2002)
  • S.V. Muller et al.

    Brain potentials related to self-generated and external information used for performance monitoring

    Clinical Neurophysiology

    (2005)
  • D.J. Schutter et al.

    Electrophysiological ratio markers for the balance between reward and punishment

    Brain Research Cognitive Brain Research

    (2005)
  • P. Shizgal

    Neural basis of utility estimation

    Current Opinion in Neurology

    (1997)
  • M. Steriade

    Grouping of brain rhythms in corticothalamic systems

    Neuroscience

    (2006)
  • L.T. Trujillo et al.

    Theta EEG dynamics of the error-related negativity

    Clinical Neurophysiology

    (2007)
  • M. Ullsperger et al.

    Subprocesses of performance monitoring: A dissociation of error processing and response competition revealed by event-related fMRI and ERPs

    Neuroimage

    (2001)
  • M. Watanabe

    The appropriateness of behavioral responses coded in post-trial activity of primate prefrontal units

    Neuroscience Letter

    (1989)
  • J. Yordanova et al.

    Parallel systems of error processing in the brain

    Neuroimage

    (2004)
  • G.S. Berns et al.

    Predictability modulates human brain response to reward

    Journal of Neuroscience

    (2001)
  • M.M. Botvinick et al.

    Conflict monitoring and cognitive control

    Psychological Review

    (2001)
  • S.L. Bressler et al.

    Episodic multiregional cortical coherence at multiple frequencies during visual task performance

    Nature

    (1993)
  • C.S. Carter et al.

    Anterior cingulate cortex, error detection, and the online monitoring of performance

    Science

    (1998)
  • R. Courtemanche et al.

    Synchronous, focally modulated beta-band oscillations characterize local field potential activity in the striatum of awake behaving monkeys

    Journal of Neuroscience

    (2003)
  • M.R. Delgado et al.

    Tracking the hemodynamic responses to reward and punishment in the striatum

    Journal of Neurophysiology

    (2000)
  • M. Falkenstein et al.

    Effects of errors in choice reaction task on the ERP under focused and divided attention

  • Cited by (211)

    • Alpha Event-Related Desynchronization During Reward Processing in Schizophrenia

      2023, Biological Psychiatry: Cognitive Neuroscience and Neuroimaging
    View all citing articles on Scopus
    View full text