نهمین کنگره نقشه برداری مغز

نمایش مقاله

Title:

A Model-Based Framework for Brain Mapping: From Hidden Variables to EEG Signatures

Subject:

Artificial Intelligence and Brain Mapping

Presenter Author:
Name:	Fatemeh Shakibafard

Author:
Name 1:	Fatemeh Shakibafard - Corresponding - Presenter
Affiliation 1:	Department of Cognitive Modelling & Brain and Computation (CM&BC), Institute for Cognitive Science Study, Tehran, Iran.

Name 2:	Saeed Setayeshi
Affiliation 2:	Department of Physics and Energy Engineering, Amirkabir University of Technology, Tehran, Iran.

Name 3:	Mehdi Tehrani-Doost
Affiliation 3:	Research Center for Cognitive and Behavioral Sciences, Tehran University of Medical Sciences, Tehran, Iran.

Name 4:	Mohammad Nasehi
Affiliation 4:	Cognitive and Neuroscience Research Center (CNRC), Tehran Medical Sciences Branch, Islamic Azad University, Tehran, Iran.

Name 5:
Affiliation 5:

Name 6:
Affiliation 6:

Name 7:
Affiliation 7:

Name 8:
Affiliation 8:

Name 9:
Affiliation 9:

Name 10:
Affiliation 10:

Abstract

Original Research

introduction :

Computational modeling has become an essential approach in neuroscience for making latent neural variables accessible—internal processes that are not directly observable in behavior or recorded brain signals, yet fundamentally shape learning, decision-making, and response inhibition. The cortico-basal ganglia-thalamo-cortical loop (CBGTC loop) occupies a central position in this framework, as dopaminergic signaling within this circuit encodes reward prediction errors in line with temporal-difference learning and supports the acquisition of action values and policies. Despite this well-established theoretical link, relating such latent computational variables to brain-mapping data, particularly EEG, remains a major challenge.

Methods:

In this study, we developed a basal ganglia–inspired actor–critic model for a Go/No-Go task. The critic component was designed to capture action-value evaluation and was conceptually aligned with ventral striatal function, whereas the actor governed action-selection policy and was associated with dorsal striatal circuitry. Learning emerged through reward prediction error signals with asymmetric dopaminergic modulation for positive and negative outcomes. This formulation allowed key latent variables—such as action value, response inhibition tendency, and action conflict—to be explicitly derived from the model’s behavior. Model validity was first assessed using human behavioral data by comparing group-level hit and false-alarm rates. To examine the relevance of the framework for brain mapping, EEG data recorded during No-Go trials were subsequently analyzed. Event-related potential components linked to response inhibition, particularly N2 and P3, were extracted from central and parietal regions and related to both behavioral measures and model-derived latent variables.

Results:

The model successfully reproduced group-level behavioral patterns observed in human participants. EEG analyses further showed that group differences in behavioral performance were accompanied by systematic changes in N2 and P3 amplitudes. Importantly, these neural differences could be meaningfully interpreted within the computational framework, reflecting variations in latent variables such as action conflict and inhibitory policy.

Conclusion:

Together, these findings suggest that inspired computational models can function as interpretive bridges between EEG signals and underlying neural computations. By integrating computational modeling with EEG, this work illustrates a model-based approach to brain mapping that moves beyond signal description toward the reconstruction and interpretation of latent neural variables. Without making clinical claims, this framework offers a promising foundation for future computationally informed neuroimaging research. (This work is based upon research funded by Iran National Science Foundation (INSF) under project No 4042222.)

Keywords:

Brain Mapping Computational Modeling Basal Ganglia Circuits Electroencephalography (EEG) Response Inhibition