A comparison of reward processing during Becker-DeGroot-Marschak and Vickrey auctions: An ERP study

Abstract

Vickrey auctions (VA) and Becker-DeGroot-Marschak auctions (BDM) are strategically equivalent demand-revealing mechanisms, differentiated only by a human opponent in the VA, and a random-number-generator opponent in the BDM. Game parameters are such that players are incentivized to reveal their private subjective values (SV) and behavior should be identical in both tasks. However, this has been repeatedly shown not to be the case. In this study, the neural correlates of outcome feedback processing during VA and BDM were directly compared using electroencephalography. Twenty-eight healthy participants bid for household products which were then divided into high- and low-SV categories. The VA included a human opponent deception to induce a social environment, while in reality a random-number-generator was used in both tasks. A P3 component peaking at 336 ms over midline parietal sites showed more positive amplitudes for high bid values, and for win outcomes in the VA but not the BDM. Both auctions also elicited a Reward Positivity potential, maximal at 275 ms along the central midline electrodes, that was not modulated by auction task or SV. Further, an exploratory N170 potential in the right occipitotemporal electrodes and a vertex positive potential component were stronger in the VA relative to the BDM. Results point to an enhanced cortical response to bid outcomes during VA task in a potential component associated with emotional control, and to the occurrence of face-sensitive potentials in VA but not in BDM auction. These findings suggest modulation of bid outcome processing by the social-competitive aspect of auction tasks. Directly comparing two prominent auction paradigms affords the opportunity to isolate the impact of social environment on competitive, risky decision-making. Findings suggest that feedback processing as early as 176 ms is facilitated by the presence of a human competitor, and later processing is modulated by social context and subjective value.

Keywords: EEG; N170; P3; RewP; reward; subjective value; willingness-to-pay.

FIGURE 1

(a) A no‐win trial in the BDM task. b) A win trial for the VA task. For both tasks, each trial began with a fixation cross for 1 s, followed by the auction item for 2 s, which is then joined by a sliding scale from £0 to £12 in increments of 25p on which to select their bid. Participants were instructed to select their bid on the scale, and once they were happy with their decision, to submit the bid by clicking on the button in the bottom right‐hand corner. In the VA, in two‐thirds of the trials, this was followed by the phrase “wait for opponent” and a loading GIF indicating that the other player has yet to submit their bid, which lasted for either 1–2 or 5–6 s. This was then followed by a blank screen for 1 s. In the BDM, the screen was blank for 1 s before presenting the outcome of the trial. The outcome of the trial was then presented for 2 s. EEG triggers were synced to the onset of auction feedback.

FIGURE 2

Butterfly plots of grand average ERPs in response to outcome presentation for (a) VA task and (b) BDM task. Epochs for distinct ERP components, N170, P2, and P3, are highlighted with gray bars, and the corresponding averaged topographies across the selected epochs are shown above.

FIGURE 3

Win–no‐win contrasts (a, b) and high vs. low bid values contrasts (c, d) in BDM and VA tasks for the N170 and the VPP. (a) Whole scalp topographic maps displaying grand average ERPs for each outcome condition at time point 182 ms. Electrodes used in statistical analysis (129, 55, 99, 100, 101) are highlighted in white. (b) Grand average ERP waveforms across all participants and subjective value conditions comparing the four outcome conditions: BDM win (light blue), BDM no‐win (navy), VA win (light orange), and VA no‐win (dark orange). Epoch of interest (172–192 ms post‐feedback‐onset) highlighted in gray. (c) Whole scalp topographic maps displaying grand average ERPs for each value condition at time point 182 ms. (d) Grand average ERP waveform across all participants and outcome conditions comparing the four value conditions: BDM low‐value (dark green), BDM high‐value (light green), VA low‐value (red), and VA high‐value (pink).

FIGURE 4

The Win–No‐win contrast in BDM and VA tasks in the RewP component. Left: Whole scalp topographic maps displaying differences in grand average ERPs at time point 275 ms. Three electrodes used in statistical analysis numbered 6, 129, and 55, are highlighted in white. Right: Grand average win–minus–no‐win ERP difference waveform across all subjects and product value conditions comparing BDM (blue) and VA (orange) win–minus–no‐win difference waveforms. Epoch of interest, 260–290 ms post‐feedback‐onset, highlighted in gray.

FIGURE 5

The win–no–win contrast (a, b) and the high vs low‐value bids contrast (c, d) in BDM and VA tasks in the P3 component. (a) Whole scalp topographic maps displaying grand average ERPs for each outcome condition at time point 331 ms. Three electrodes used in statistical analysis numbered 6, 129, and 55, are highlighted in white. (b) Grand average ERP waveforms across all subjects and product value conditions comparing the four outcome conditions: BDM win (light blue), BDM no‐win (navy), VA win (light orange), and VA no‐win (dark orange). Epoch of interest (316–346 ms post‐feedback‐onset) highlighted in gray. (c) Whole scalp topographic maps displaying grand average ERPs for each value condition at time point 331 ms. (d) Grand average ERP waveform across all subjects and outcome conditions comparing the four value conditions: BDM low‐value (dark green), BDM high‐value (light green), VA low‐value (red), and VA high‐value (pink).