Abstract
While univariate functional magnetic resonance imaging (fMRI) data analysis methods have been utilized successfully to map brain areas associated with cognitive and emotional functions during viewing of naturalistic stimuli such as movies, multivariate methods might provide the means to study how brain structures act in concert as networks during free viewing of movie clips. Here, to achieve this, we generalized the partial least squares (PLS) analysis, based on correlations between voxels, experimental conditions, and behavioral measures, to identify large-scale neuronal networks activated during the first time and repeated watching of three ∼5-min comedy clips. We identified networks that were similarly activated across subjects during free viewing of the movies, including the ones associated with self-rated experienced humorousness that were composed of the frontal, parietal, and temporal areas acting in concert. In conclusion, the PLS method seems to be well suited for the joint analysis of multi-subject neuroimaging and behavioral data to quantify a functionally relevant brain network activity without the need for explicit temporal models.
Highlights
Converging evidence suggests that complex naturalistic scenes and stimuli elicit neuronal responses more reliably than simplified stimuli in conventional laboratory experiments (Mechler et al, 1998; Yao et al, 2007; Belitski et al, 2008)
The subjects were presented with three movie clips (4 33, 5 17, and 5 08 in duration) twice, in an order that was randomized across subjects, during 3T (Tim Trio, Siemens, Erlangen, Germany) functional magnetic resonance imaging (fMRI) [echo-planar imaging (EPI); repetition time/echo time (TR/TE) = 2,000/30 ms, field of view (FOV) = 220 × 220 mm, matrix = 64 × 64, slice thickness = 4 mm, flip angle = 90◦]
At a lower threshold, we found that the correlation between fMRI time series and degree of humorousness in the first viewing was larger at the bi-hemisphere temporal–parietal junction (TPJ) than that in the second viewing, while the third latent variable (LV) in the behavioral partial least squares (PLS) analysis suggested that bi-hemispheric superior temporal sulcus (STS) and dorsolateral prefrontal cortex (DLPFC) show the difference in the behavior–brain correlation between repeated viewings
Summary
Converging evidence suggests that complex naturalistic scenes and stimuli elicit neuronal responses more reliably than simplified stimuli in conventional laboratory experiments (Mechler et al, 1998; Yao et al, 2007; Belitski et al, 2008). PLS results were compared with the conventional univariate analysis results to show that the proposed multivariate analysis is more sensitive in identifying neural correlates to the processing of complex naturalistic stimuli This multivariate network analysis approach may be applied to analyze multi-subject neuroimaging data without explicit temporal models
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
