Colin Galvin

BWH Job Title:

Research Assistant

Academic Rank:

N/A

Department/Division/Lab:

Neurosurgery

Golby Lab

Authors:

Colin Galvin, Ben Phan, Laura Rigolo, Jeffrey Girard, Einat Liebenthal, Alexandra J Golby, Yanmei Tie

Movies as Naturalistic Stimuli for Comprehensive Brain Mapping

Abstract

Background:
There is a growing need for a repository of movie clips that can be used in naturalistic mapping of language and other high-order cognitive functions using fMRI and other brain mapping techniques. The development of a diverse movie database, validated by reliable emotional elicitation, is a crucial first step in developing a naturalistic fMRI paradigm with movie-watching condition1 for comprehensive brain mapping1,2. A universal movie dataset will promote adoption of movie fMRI for brain mapping with an easier paradigm than traditional task-based fMRI, especially for patients with neurological deficits.
This study aims to develop a database of movie clips that contain segments with and without highly comprehensible spoken dialogue, that elicit patterns of emotional reactivity, and are therefore suitable for naturalistic stimulation of language and emotion networks during fMRI brain mapping.
Methods:
Twenty-two movie clips containing spoken dialogue with actors in camera-facing orientation were chosen from free and publicly available websites. Films of different genres were selected to fit one of 5 general valence and arousal categories: negative high arousal, negative low arousal, neutral, positive low arousal, and positive high arousal. The clips ranged between 2-7 mins with a naturally evolving narrative structure and time-varying emotional content.
Healthy subjects without neurologic or mental health conditions were recruited and screened leading to 97 subjects (58 females, age range: 18-60 years old) being enrolled into this study. Subjects were asked to participate in 1 or 2 movie watching sessions (each 90 mins), conducted over Zoom, where they were asked to provide dynamic ratings of valence and arousal of the clips. Subject dynamic ratings were collected using CARMA (https://carma.jmgirard.com), a media annotation program that collects continuous ratings while displaying audio and video files3. Subjects rated emotional valence on a continuous scale from -4 (very negative) to +4 (very positive). After each clip, participants answered questions about their ability to see and hear the movie, familiarity with the movie, 3 comprehension questions about the content, and holistic ratings of its emotional salience and affective dimensions using the short version of the Positive and Negative Affect Scale short form (PANAS)4.
Results:
The clips evoked extensive emotional valence and specific emotions which were confirmed by analyses of the distribution of dynamic and holistic ratings. For all clips, the dynamic ratings were normally distributed. The inter-rater reliability in dynamic ratings, calculated as the average-measures consistency intraclass correlation (ICC), was above 0.90. A variance decomposition analysis indicated that the largest sources of variance in dynamic valence ratings were the movie clip (38.7%), clip by rater interaction (19.3%), and clip segment (13.4%), whereas rater on its own accounted for only 5.1% of the variance. Comprehension questions yielded an average of 95.77% accuracy across all clips.
Conclusion:
This study developed a diverse movie database with consistent emotional elicitation across subjects which is needed for developing a movie fMRI paradigm for comprehensive brain mapping. Future work will examine how genre, familiarity and relative demographics influence the level of language comprehension and emotional reactivity.

References:
1 Tie Y, Rigolo L, Ovalioglu AO, Olubiyi O, Doolin KL, Mukundan S, et al. A new paradigm for individual subject language mapping: Movie-watching fMRI. J Neuroimaging 2015;25:710–20. https://doi.org/10.1111/jon.12251.
2 Yao S, Rigolo L, Yang F, Vangel MG, Wang H, Golby AJ, Liebenthal E, Tie Y. Movie-watching fMRI for presurgical language mapping in patients with brain tumour. J Neurol Neurosurg Psychiatry. 2022 Feb;93(2):220-221. doi: 10.1136/jnnp-2020-325738. Epub 2021 Jun 25. PMID: 34172562; PMCID: PMC8709882.
3 Girard JM. CARMA: Software for Continuous Affect Rating and Media Annotation. Journal of Open Research Software 2014;2:e5. https://doi.org/10.5334/jors.ar.
4 Mackinnon A, Jorm AF, Christensen H, Korten AE, Jacomb PA, Rodgers B. A short form of the Positive and Negative Affect Schedule: evaluation of factorial validity and invariance across demographic variables in a community sample. Personality and Individual Differences 1999;27:405–16. https://doi.org/10.1016/S0191-8869(98)00251-7.