Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Functional brain organization is stable within individuals across years.

bioRxiv : the preprint server for biology·2026

Same author

Efficient Deep Learning Models for Predicting Individualized Task Activation From Resting-State Functional Connectivity.

Human brain mapping·2026

Same author

Refining RDoC Using Individual-Level Task fMRI Factor Models Reveals Reproducible and Clinically Relevant Brain-Wide Motifs.

bioRxiv : the preprint server for biology·2026

Same author

Building an open ecosystem for molecular neuroimaging: standards and tools from the OpenNeuroPET initiative.

bioRxiv : the preprint server for biology·2026

Same author

Quality assessment and control of unprocessed anatomical, functional and diffusion MRI of the human brain using MRIQC.

Nature protocols·2026

Same author

Investigating the analytical robustness of the social and behavioural sciences.

Nature·2026

Same journal

Investigating the Neural Origins of Ear-EEG: A Correlation Study Using Scalp EEG Source Reconstruction.

NeuroImage·2026

Same journal

Hysteresis effects in visual and auditory perception and the comparison of underlying neural mechanisms - an EEG study.

NeuroImage·2026

Same journal

Short-term audio-tactile training affects cortical auditory speech-envelope tracking for incongruent but not congruent stimuli.

NeuroImage·2026

Same journal

Dissociable Neurocognitive Mechanisms of State and Trait Anxiety in Working Memory: Threat-Induced Alterations in Decision Dynamics and Attenuation of Large-Scale Network Reconfiguration.

NeuroImage·2026

Same journal

Neuro-Ocular Amyloid Characterization in Alzheimer's Disease via Cross-Site PET-MRI and Hierarchical Cross-Attention Driven Multimodal Representation Learning.

NeuroImage·2026

Same journal

Whole-brain network dynamics underlying intolerance of uncertainty.

NeuroImage·2026

See all related articles

Search research articles

Related Experiment Video

Updated: Aug 2, 2025

P300-Based Brain-Computer Interface Speller Performance Estimation with Classifier-Based Latency Estimation

P300-Based Brain-Computer Interface Speller Performance Estimation with Classifier-Based Latency Estimation

Published on: September 8, 2023

Benchmarking explanation methods for mental state decoding with deep learning models.

Armin W Thomas¹, Christopher Ré², Russell A Poldrack³

¹Stanford Data Science, Stanford University, 450 Serra Mall, 94305, Stanford, USA.

|April 14, 2023

Summary

This summary is machine-generated.

Deep learning models decode mental states from brain activity. Explanation methods show a trade-off between accurately reflecting the model and aligning with existing neuroscience, guiding researchers in choosing the best approach.

Keywords:

Benchmark Deep learning Explainable AI Mental state decoding Neuroimaging

More Related Videos

A Novel Experimental and Analytical Approach to the Multimodal Neural Decoding of Intent During Social Interaction in Freely-behaving Human Infants

A Novel Experimental and Analytical Approach to the Multimodal Neural Decoding of Intent During Social Interaction in Freely-behaving Human Infants

Published on: October 4, 2015

Simultaneous Scalp Electroencephalography EEG, Electromyography EMG, and Whole-body Segmental Inertial Recording for Multi-modal Neural Decoding

Simultaneous Scalp Electroencephalography EEG, Electromyography EMG, and Whole-body Segmental Inertial Recording for Multi-modal Neural Decoding

Published on: July 26, 2013

Related Experiment Videos

Last Updated: Aug 2, 2025

P300-Based Brain-Computer Interface Speller Performance Estimation with Classifier-Based Latency Estimation

P300-Based Brain-Computer Interface Speller Performance Estimation with Classifier-Based Latency Estimation

Published on: September 8, 2023

A Novel Experimental and Analytical Approach to the Multimodal Neural Decoding of Intent During Social Interaction in Freely-behaving Human Infants

A Novel Experimental and Analytical Approach to the Multimodal Neural Decoding of Intent During Social Interaction in Freely-behaving Human Infants

Published on: October 4, 2015

Simultaneous Scalp Electroencephalography EEG, Electromyography EMG, and Whole-body Segmental Inertial Recording for Multi-modal Neural Decoding

Simultaneous Scalp Electroencephalography EEG, Electromyography EMG, and Whole-body Segmental Inertial Recording for Multi-modal Neural Decoding

Published on: July 26, 2013

Area of Science:

Neuroscience
Artificial Intelligence
Cognitive Science

Background:

Deep learning (DL) models are increasingly used for mental state decoding from neuroimaging data.
Explainable artificial intelligence (XAI) methods are crucial for understanding DL model decisions in neuroscience.
Functional Magnetic Resonance Imaging (fMRI) is a key neuroimaging technique for studying brain activity.

Purpose of the Study:

To benchmark prominent XAI methods in the context of mental state decoding using fMRI data.
To evaluate the trade-off between explanation faithfulness and alignment with empirical neuroscience evidence.
To provide guidance for neuroimaging researchers on selecting appropriate XAI methods for DL models.

Main Methods:

Benchmarking of various XAI methods on multiple fMRI datasets.
Analysis of mental state decoding performance using DL models.
Quantitative assessment of explanation faithfulness and empirical alignment.

Main Results:

A gradient exists between explanation faithfulness and alignment with existing neuroscience knowledge.
Highly faithful explanation methods may align less well with established empirical evidence.
Less faithful methods might offer better alignment with prior neuroscience findings.

Conclusions:

The choice of XAI method in mental state decoding involves a trade-off between model fidelity and interpretability based on prior knowledge.
Neuroimaging researchers should carefully consider this trade-off when selecting XAI techniques.
Guidance is provided to aid researchers in making informed decisions for understanding DL-based mental state decoding.