Related Experiment Video
Updated: Aug 31, 2025
11:33
Transcranial Magnetic Stimulation for Investigating Causal Brain-behavioral Relationships and their Time Course
Published on: July 18, 2014
43.1K
The multiple indicator multiple cause model for cognitive neuroscience: An analytic tool which emphasizes the
Adon F G Rosen1, Emma Auger1, Nicholas Woodruff1
1Department of Psychology, University of Oklahoma, Norman, OK, United States.
Frontiers in Psychology
|August 22, 2022
Summary
This study reveals that traditional methods for analyzing cognitive data can obscure important brain-behavior relationships. The Multiple Indicator Multiple Cause (MIMIC) model, however, effectively identifies and removes problematic items, improving the sensitivity of neuroimaging analyses.
Area of Science:
- Cognitive Neuroscience
- Psychometrics
- Neuroimaging Data Analysis
Background:
- Cognitive neuroscience relies on neuroimaging data, often summarized using sum-scores or item response theory (IRT).
- These traditional methods may overlook item dimensionality and quality, potentially including poorly performing items.
- This can lead to unexplored parameter bias and differential item functioning (DIF) in cognitive neuroscience research.
Purpose of the Study:
- To demonstrate how two-stage approaches can introduce parameter bias in cognitive neuroscience data.
- To highlight the manifestation of differential item functioning (DIF) in cognitive neuroscience.
- To showcase the Multiple Indicator Multiple Cause (MIMIC) model's ability to identify and remove DIF items for more sensitive brain-behavior relationship modeling.
Main Methods:
- A simulation study was conducted to explore parameter bias in sum-scores and IRT compared to a MIMIC model.
- An empirical study involved participants performing an emotional identification task with concurrent electroencephalogram (EEG) data acquisition.
- The MIMIC model was used to identify DIF in emotion identification items based on P200 event-related potential (ERP) amplitude and latency, and its sensitivity was compared to two-stage approaches.
Main Results:
- The simulation demonstrated parameter bias in sum-scores and IRT compared to the MIMIC model.
- Instances of DIF were identified in all four emotion categories during the empirical study and subsequently removed.
- Only the MIMIC model successfully identified significant brain-behavior relationships, outperforming traditional two-stage methods.
Conclusions:
- Item performance in cognitive tasks can be effectively assessed using subject-specific biomarkers.
- The Multiple Indicator Multiple Cause (MIMIC) model offers enhanced sensitivity for uncovering complex item-level brain-behavior relationships.
- The MIMIC model is a valuable tool for improving the rigor and interpretability of cognitive neuroscience research.
