Poster #10 - Infant functional brain networks associated with temperamental fear and reactivity

• In Event: Poster Session 17

Abstract

While several infant functional magnetic resonance imaging (fMRI) studies have examined infant temperament (Rogers et al, 2017, Thomas et al, 2018; Graham et al, 2016), these studies have relatively small sample sizes (ns<45) and rely on parent-report of temperament. Parent reports and observed measures of temperament offer unique lenses, but it remains unclear how to combine them in fMRI research. This study identifies key functional brain networks associated with both parent-reports and direct observations of temperament.

We integrated data from the Baby Connectome Project (BCP; n=63) and Origins of Infant Temperament (OIT; n=38). Both studies obtained fMRI data from sleeping 4-to-6-month-olds and concurrent temperament assessments. For the BCP, temperament was assessed using the Infant Behavior Questionnaire (IBQ), a parent report. For OIT, temperament was evaluated using both the IBQ and an observational assessment of reactivity. Across both studies, we focused on the IBQ-fear composite because it has previously been shown to correlate with observed measures of reactivity (Filippi et al, 2020). Observational assessments of reactivity were rated in terms of negative affect, positive affect, and motor arousal. Resting state functional connectivity (rsfc) was computed between 200 regions of interest. Brain-behavior associations were tested using enrichment. Enrichment is an approach that identifies brain network pairs that have a greater density of strong brain-behavior correlations. False positive rates were determined via permutation. All analyses controlled for age and sex.

Harmonized data from BCP and OIT (n=101) was utilized to examine the association between rsfc and IBQ-fear. Connectivity within the visual, dorsal attention (DAN), and control networks was associated with parent reported fear. Additionally, findings manifest between many other networks, including the salience, somatomotor, default mode (DMN), and limbic networks (See Table 1 for specific network pair details). Next, we examined the association between rsfc and observed reactivity (OITn=38). Connectivity within the DAN, DMN, and control networks related to observed negative reactivity, with both consistency and inconsistency with IBQ data. While connectivity between multiple networks related to reactivity, the exact network pairs identified differed in terms of hemisphere and brain regions. A follow-up conjunction analysis demonstrated that greater connectivity between right hemisphere Control- left hemisphere somatomotor networks was associated with both parent-reported fear and reactivity (p<.04; See Figure 1).

Together, these data illustrate that the brain networks implicated in early fear and reactivity are broadly consistent. However, there were differences in the specific network connections implicated. Future work should determine whether identifying brain-behavior associations that converge across different temperament measurements are more stable predictors of later behavior. Additional independent datasets are critical to testing such hypotheses. Overall, these data illustrate how measurement of temperament impacts brain-behavior associations.

Authors

• Courtney Filippi, New York University School of Medicine
  Presenting Author

• Anderson Winkler

• Daniel S. Pine, National Institutes of Health

• Nathan A Fox, University of Maryland - College Park