Higher-order comparative reward processing is affected by noninvasive stimulation of the ventromedial prefrontal cortex
Abstract
Introduction. A crucial skill, especially in rapidly changing environments, is to be able to learn efficiently from prior rewards or losses and apply this acquired knowledge in upcoming situations. Often, we must weigh the risks of different options and decide whether an option is worth the risk or whether we should choose a safer option. The ventromedial prefrontal cortex (vmPFC) is suggested as a major hub for such basic but also higher-order reward anticipation and processing. Dysfunction in this region has been linked to cognitive risk factors for depression and behavioral addictions, including reduced optimism and feedback learning. Here, we test whether modulations of vmPFC excitability via noninvasive transcranial direct current stimulation (tDCS) can alter reward anticipation and reward processing.
Methods. In a financial gambling task, participants chose between a higher and a lower monetary risk option and eventually received feedback whether they won or lost. Simultaneously feedback on the unchosen option was presented as well. Behavioral and magnetoencephalographic correlates of reward processing were evaluated in direct succession of either excitatory or inhibitory tDCS of the vmPFC.
Results. We were able to show modulated reward approach behavior (expectancy of greater reward magnitudes) as well as altered reevaluation of received feedback by vmPFC-tDCS as indicated by modified choice behavior following the feedback. Thereby, tDCS not only influenced early, rather basic reward processing, but it also modulated higher-order comparative feedback evaluation of gains and losses relative to alternative outcomes. The neural results underline this idea, as stimulation driven modulations of the basic reward related effect occurred at rather early time intervals and were followed by stimulation effects related to comparative reward processing. Importantly, behavioral ratings were correlated with neural activity in left frontal areas.
Discussion. Our results imply a dual function of the vmPFC consisting of approaching reward (as indicated by more risky choices) and elaborately evaluating outcomes. In addition, our data suggest that vmPFC activity is associated with adaptive decision-making in the future via modulated behavioral adaptation or reinforcement learning.
Data
Behavioral and neural data (MEG - MN5) of a study comparing excitatory and inhibitory stimulation and their influence on behavioral and neural correlates of gambling. The neural data can be read with the free emegs software (Peyk, De Cesarei & Junghöfer, 2011). The data is in the so-called source space, which means that the underlying sources of the MEG signal were estimated. For this purpose the L2-Minimum-Norm estimation was used (Hämäläinen and Ilmoniemi, 1994).
We provided two datasets: The higher order reward processing data set contains the independent variables alternative outcome
+50 = ...51.100.MN5,
+30 = ...31&34.100.MN5,
+20 = ...22&23.100.MN5,
+10 = ...11.100.MN5,
-10 = ...12.100.MN5,
-20 = ...21&24.100.MN5,
-30 = ...33&32.100.MN5,
-50 = ...52.100.MN5
and stimulation
sham = ...100.MN5
excitatory = ...200.MN5.
The all data set contains the independent variables
11 = Gain5 miss Loss5
12 = Loss5 miss Gain5
21 = Gain5 miss Gain25 (gain and incorrect)
22 = Gain25 miss Gain5 (gain and correct)
23 = Loss5 miss Loss25 (loss and correct)
24 = Loss25 miss Loss5 (loss and incorrect)
31 = Gain5 miss Loss25
32 = Loss25 miss Gain5
33 = Loss5 miss Gain25
34 = Gain25 miss Loss5
51 = Gain 25 miss Loss25
52 = Loss25 miss Gain25
and stimulation
sham = ...100.MN5
excitatory = ...200.MN5.
The data can be organized with the provided Matlab Scripts: Find...Files.
The behavioral data can be analyzed with the provided R-script. The R-script will recode all the independent variables into a meaningful form.
