The ERP, frequency, and time–frequency correlates of feedback processing: Insights from a large sample study

Abstract

Human learning, at least in part, appears to be dependent on the evaluation of how outcomes of our actions align with our expectations. Over the past 23 years, electroencephalography (EEG) has been used to probe the neural signatures of feedback processing. Seminal work demonstrated a difference in the human event‐related potential (ERP) dependent on whether people were processing correct or incorrect feedback. Since then, these feedback evoked ERPs have been associated with reinforcement learning and conflict monitoring, tied to subsequent behavioral adaptations, and shown to be sensitive to a wide range of factors (e.g., Parkinson's disease). Recently, research has turned to frequency decomposition techniques to examine how changes in the EEG power spectra are related to underlying learning mechanisms. Although the literature on the neural correlates of feedback processing is vast, there are still methodological discrepancies and differences in results across studies. Here, we provide reference results and an investigation of methodological considerations for the ERP (reward positivity) and frequency (delta and theta power) correlates of feedback evaluation with a large sample size. Specifically, participants (n = 500) performed a two‐armed bandit task while we recorded EEG. Our findings provide key information about the data characteristics and relationships that exist between the neural signatures of feedback evaluation. Additionally, we conclude with selected methodological recommendations for standardization of future research. All data and scripts are freely provided to facilitate open science.

A Window into the Rational Mind: The Neural Underpinnings of Human Reasoning

“…One participant was removed due to excessive noise in their EEG data and two others were removed as outliers determined via an evaluation of Cook's distances during analyses (see below). Thus, here we 19 , respectively, to achieve 95% power -confirming our sample size which exceeds these calculations. All participants were recruited through an online sign-up system wherein they earned course credit within psychology courses, were native English speakers, and had normal or corrected-to-normal vision.…”

Single-trial modeling separates multiple overlapping prediction errors during reward processing in human EEG

“…FRN peak latencies were modulated by RPE value, and similar to losses and wins, FRN peak latency was earlier for neutral feedback with negative than positive RPEs. However, Williams et al found an opposite FRN latency shift with wins occurring earlier than losses in a large ( n = 500) gambling dataset using only visual feedback 54 . The latency of FRN and RewP ERPs in our data are also earlier than reported in recent meta-analyses 38 , likely because feedback in our task includes auditory components, which generate faster FRN latencies 33 .…”

“…This interpretation is complicated by more recent observations of a P3-like positivity called the Reward Positivity (RewP) that tracks RPE magnitude specifically on positive outcomes (see ref. 46 for review) 47 – 52 that overlaps in space and delta frequencies with other non-valenced P3 components 53 , 54 . Importantly, this suggests losses and wins generate distinct FRN and RewP ERPs with opposite polarities that interact to some degree depending on their overlap in time and space 48 , 55 .…”

“…However, individual ERP peaks are variable and may not correspond to unique neural sources 56 – 58 , rendering their use as reference measures questionable. For example, the P2 shows confounding effects of surprising positive reward 54 , 59 . Difference waves are commonly used to isolate target variables such as valence by subtracting ERPs across conditions matched for confounding variables (e.g., magnitude or probability).…”

“…Difference waves are commonly used to isolate target variables such as valence by subtracting ERPs across conditions matched for confounding variables (e.g., magnitude or probability). Indeed, win-loss difference waves are commonly used as the operational definition of the FRN and RewP 38 , 54 , 60 , 61 . However, this subtraction logic can determine neither which ERP nor which condition was modulated and thus is not well suited to unraveling the dual multiplicity of ERPs and learning signals, particularly since the FRN and RewP may have distinct neural mechanisms supporting different computational roles 52 , 62 , 63 .…”