Decoding the Source of Reward Positivity: Insights from Multi-Modal EEG-fMRI Analysis

Authors: Malte Rudo Güth, Yang Yang, Clay Holroyd, Travis E. Baker

Presentation type: Poster at SNUFA 2024 online workshop (5-6 Nov 2024)

Abstract

Over two decades of EEG and neuroimaging research have consistently identified Reward Positivity (RewP) as a positive-going deflection in the event-related potential (ERP) elicited by reward feedback but not by negative feedback. While this component is widely believed to originate from or near the anterior cingulate cortex (ACC), its precise neural source remains contentious. A significant complication in resolving this issue arises from the modality-dependent limitations inherent in current neuroimaging techniques. For instance, EEG/MEG source localization is inherently hampered by the inverse problem, which limits precise spatial resolution. On the other hand, conventional fMRI analyses can reveal activation across multiple brain areas but lack the temporal precision necessary to link these activations to specific ERP components like the RewP. Additionally, mass-univariate approaches for combining EEG and fMRI data often fall short in analyzing the multidimensional nature of the information, which is crucial for understanding the dynamics of neural activity. Twenty-eight right-handed undergraduate students (MAge = 23.5 years, SDAge = 4.24 years; 16 female) were recruited from Rutgers University-Newark and the New Jersey Institute of Technology. Simultaneous EEG-fMRI data was recorded while participants performed a T-maze virtual task. In the current study, we propose that these limitations can be mitigated by applying neural decoding through multivariate pattern analysis (MVPA). This technique leverages brain activity recordings to make causal predictions using classification algorithms. Specifically, we suggest that decoding simultaneous EEG-fMRI data during tasks that elicit robust RewPs will provide critical insights into the neural origins of RewP or other ERP components. In essence, if the pattern of EEG activity allows for the prediction or decoding of a reward stimulus, it implies that the stimulus-related information is represented within the fMRI data, precisely where the pattern was identified.