Towards a Functional Neuromarker of Impulsivity: Feedback-Related Brain Potential during Risky Decision-Making Associated with Self-Reported Impulsivity in a Non-Clinical Sample

Abstract

Risk-taking is part of the multidimensional nature of impulsivity, consisting of an active engagement in behaviors or choices with potentially undesirable results, with probability as the cost for an expected reward. In order to understand the neurophysiological activity during risky behavior and its relationship with other dimensions of impulsivity, we have acquired event-related-potential (ERP) data and self-reported impulsivity scores from 17 non-clinical volunteers. They underwent high-resolution electroencephalography (HR-EEG) combined with an adapted version of the Balloon Analogue Risk Task (BART), and completed the Barratt Impulsiveness Scale (BIS-10) and the Urgency, Premeditation, Perseverance, Sensation Seeking, Impulsive Behavior Scale (UPPS). The ERP components were sensitive to valence (FRN, P300) and risk/reward magnitude (SPN, RewP). Our main finding evidenced a positive correlation between the amplitude of the P300 component following positive feedback and both the global UPPS score and the (lack of) perseverance UPPS subscale, significant for several adjacent electrodes. This finding might suggest an adaptive form of impulsive behavior, which could be associated to the reduction on the difference of the P300 amplitude following negative and positive feedback. However, further investigation with both larger clinical and non-clinical samples is required.

Keywords: decision-making; electroencephalography (EEG); event-related potentials (ERPs); feedback processing; impulsivity; risk-taking.

Figure 1

Schematic diagram for the BART adapted to EEG (80 balloon trials split into 4 blocks). At the beginning of each trial, the balloon is orange and cannot explode; the only choice the subjects have is to pump the balloon. Each inflation increases the score in the trial by one point, as displayed in the center of the balloon. At a random point in time, the balloon turns blue, and a fixation cross appears in its center, indicating an imminent outcome. There are two randomly assigned outcome possibilities: (i) balloon explosion—the so far accumulated trial score is lost and a new one automatically starts—or (ii) doubling of the trial score—the new (doubled) score is displayed and in the sequence the word ‘Cagnotte’ (cash-out) followed by a question mark appears on the screen. In the latter, the subjects have the choice to either cash-out the trial gains, ending the current trial and starting a new one, or continue pumping the balloon from where it was (the balloon becomes once again orange), until it turns blue with a fixation cross and a new outcome arrives. During a trial, the probability of a balloon explosion (risk) increases with the number of times the balloon has turned blue.

Figure 2

Mean number of adjusted blue balloons per block on the adapted version of the BART. A significant difference was found between the first block and the three consecutive blocks. Error bars represent standard error.

Figure 3

Negative correlation between risk-taking behavior at the BART and the (lack of) premeditation subscale of the UPPS (r = −0.67, p < 0.01).

Figure 4

Grand average ERP waveforms after the onset of a loss and a win feedback on electrodes Fz and Cz. The FRN component appears between 275 and 330 ms, while the P300 ERP occurs between 375 and 575 ms after feedback.

Figure 5

Correlations with the P300 response (μV) after positive feedback on the FPz electrode. (a) Negative correlation with the mean number of adjusted blue balloons (r = −0.49, p < 0.05). (b) Significant positive correlation with the (lack of) perseverance subscale of the UPPS (r = 0.79, p < 0.001).

Figure 6

Influence of reward magnitude on the SPN and RewP components on the electrode Cz. (a) Neural responses during the anticipation of smaller and larger rewards; a significantly more negative SPN is present for large rewards in the time window of −200 and 0 ms before the feedback appears. (b) Neural response during reward processing; a significantly more positive RewP is observed following larger rewards, between 150 and 350 ms after positive feedback.

Figure 7

Localization of neural responses relative to a reward in the time window of the P300. (a) Topographic representation of neural responses. (b) Source localization.