Research. In my Social Neuroscience Lab we study the brain basis of non-verbal social cognition – particularly how consciously or unconsciously perceived information helps us assess others’ intentions, goals and mental states. My research integrates functional and structural magnetic resonance imaging (MRI), infrared eye tracking, high-density electroencephalography (EEG), magnetoencephalography (MEG) and electrocorticography (ECoG). By investigating sbrain tructural and functional connectivity patterns as subjects engage in evaluating the social behaviors of others, we aim to develop insights into how our brains processes the continuous daily barrage of dynamic and fleeting incoming social information.

Methods training. Science is important, but must be based on a solid scaffold of valid scientific method. To that end, I have co-written a textbook on MEG-EEG methods [Riitta Hari & Aina Puce (2017) MEG-EEG Primer, New York, Oxford University Press, https://global.oup.com/academic/product/meg-eeg-primer-9780190497774?cc=us&lang=en&]. I am active in an international effort devoted to encouraging best practices in neuroimaging, open science and data analysis/sharing [see Pernet et al., (2018) https://osf.io/a8dhx/].

Talk Title: 

Re-using MEEG data and maximizing its value: considerations of statistical power & white matter connectivity

Abstract:

Human neurophysiological data are a rich source of spatio-temporal information. They deepen our understanding of brain-behavior relationships, when combined with other data types [e.g. white matter tractography]. I will present two studies, where MEG data and ECoG data have been re-used to answer specific questions related to experimental design/data analysis or to science itself.

1. MEG and statistical power as a function of number of subjects and trials.

Simulated ‘experiments’ were performed on the MEG resting-state data of 89 subjects from the Human Connectome Project [HCP]. Each subject’s data were divided into two ‘conditions’. In a ‘signal condition’ a dipolar source was injected at a known anatomical location, but not in a ‘noise condition’. First, we detected significant differences at sensor level with classical paired t-tests across subjects, and group-level detectability of simulated effects varied greatly with anatomical origin. Second, we examined which spatial properties of the sources most affected detectability. Not surprisingly, the most detectable effects originate from source positions closest to the sensors with tangential orientation with respect to the sensor array. Also, cross-subject variability in orientation affected group-level effect detectability, aiding detection in regions with small variability and hindering detection where it was large. A considerable covariation of source position, orientation, and cross-subject variability in native brain space was seen, making it difficult to assess effects of these variables independently of one another. Therefore, a second set of similar simulations was performed. This time, spatial properties were controlled independently of individual anatomy, allowing the effects of position, orientation, and cross-subject variability of on detectability to be examined. Results confirmed the strong impact of distance and orientation on source detectability and showed that orientation variability across subjects affects detectability, whereas position variability does not. Study bottom line? [A] Strict unequivocal recommendations as to ideal number of trials and subjects for experiments cannot be realistically provided for MEG studies. [B] Spatial constraints underlying expected sources of activity while planning experiments must be considered at the experimental design stage. This is imperative for network level analyses in cognitive neuroscience studies, where activity in nodes in a network may not be equally well detected. [See: Chaumon M, Puce A, George N. (2020) bioRxiv doi: https://doi.org/10.1101/852202]

2. ECoG, white matter tractography and information flow in face processing pathways.

Intracerebral EEG from 11 epileptic patients were analyzed while they had viewed initially presented neutral faces that turned fearful or happy with, or without, an associated lateral gaze change [amygdala data published as Huijgen et al (2015) doi: 10.1093/scan/nsv048]. N200 field potential peak amplitudes were used for identifying the loci for face processing, and latency served as an indicator of transmission of information within the system. Results indicated that face onset processing begins in inferior occipital cortex and moves, in parallel, anteroventrally to fusiform and inferior temporal cortices. In contrast, the superior temporal sulcus [STS] responded selectively to gaze changes with augmented field potential amplitudes for averted versus direct gaze, and large effect sizes relative to other regions of the network. An overlap analysis of active intracerebral electrodes from the 11 patients with posterior white matter tract endpoints [from 1066 healthy brains, Human Connectome Project] was then performed. Study bottom line? From the combination of ECoG and white matter tract endpoint data it appears that inferior occipital and temporal sulci likely broadcast their information - the former dorsally to intraparietal sulcus, and the latter to both fusiform and superior temporal cortex. These data indicate that: [A] Inferior temporal cortex should be included in future face processing models. [B] Superior temporal cortex is critical for dynamic gaze processing. [See: Babo-Rebelo M, Puce A et al., (2020) submitted*]

*I should be able to provide a weblink to a preprint soon...