The Inter-Subject Correlation of EEG in Response to Naturalistic Stimuli

Inter-subject correlation is a measure of the similarity of the brain activity of a group of people as they respond to the same naturalistic stimulus, typically a story or video, meant to simulate a real world experience. This thesis tests the hypothesis that the correlation of the brain responses of a group of people is indicative of stimulus engagement. The rationale is that the content of the stimulus drives brain activity in a consistent manner, while internal thoughts are divergent and result in uncorrelated activity. The inter-subject correlation (ISC) of neural responses have previously been assessed with fMRI, EEG, and MEG. Here, EEG will assess ISC, thereby examining the correlation of the early responses to a stimulus. Engagement has been examined previously with self-report assessments of interest. These ratings are noisy, subject to bias, and do not measure how engagement evolves over time. In this thesis, engagement is defined as a commitment to devote a scarce resource, such as attention or time, to a stimulus. In the experiment presented here, subjects were allowed limited time with the stimuli, thus forcing them to engage with the content they determined to be most compelling. This behavioral metric strongly correlated with ISC of the EEG, thus validating it as a measure of neural engagement. Interestingly, higher ISC was also indicative of a shared perception of the passage of time across subjects. This suggests that when people are engaged with a stimulus, their perception of time is also driven by that stimulus, rather than by an internal sense of time. If people are more engaged at the time of encoding, it is likely that they will better remember their experiences. Memory was therefore assessed three weeks after subjects heard salient emotional narratives. Individuals whose EEG responses during the stories correlated more strongly with their peers had stronger memories of the events in the stories. ISC was also tested as a predictor of retention in the context of online educational videos. Again, the similarity between each subject’s brain activity and that of his or her peers corresponded with memory for factual information in a subsequent test. It is possible that people with different backgrounds do not engage with the world in similar ways, and their neural responses will therefore correlate more strongly with people who are most similar to them. To address this notion, ISC was compared across the dimensions of age and gender. In a population with ages ranging from 5 - 44 years old, ISC weakens with age and is stronger in males than it is in females. This result is consistent with the idea that age and experience are marked by an increase in the repertoire of neural representations. Adults may therefore have more variable interpretations that mediate their sensory responses to stimuli. Alternatively, if ISC is truly assessing engagement in this context, the result may demonstrate that adults are less susceptible to the influence of outside stimuli since they have more powerful internal voices that distract them. Whichever the ultimate reason for this change, the gender disparity may also be related to a developmental difference because the deviation between males and females in ISC is strongest in young ages, a period when anatomical findings show that young males are less neurally mature than young females. Although ISC is implicated in fundamental processes such as engagement, memory, and development, the neural underpinnings of this signal are unclear. The spatial distribution of the EEG signal that drives ISC appears similar on the surface of the scalp across stimuli with different narrative content, and between different stimulus modalities. The similarity of the topography of correlated activity across sensory modalities may indicate that this activity is supramodal and is therefore generated by a region that is impervious to the stimulus modality. To assess ISC’s dependence on stimulus modality and stimulus type, the modulation of ISC was compared with the fMRI BOLD responses to the same stimuli. This analysis revealed that ISC is mostly modulated by sensory regions, and that the extent of the regions involved depends on the content of the stimulus. These areas, which are largely driven by immediate processing of the stimulus at a fast timescale, are therefore implicated in higher-level behaviors such as engagement and memory