Multivariate neural signatures for health neuroscience: Assessing spontaneous regulation during food choice

Abstract

Establishing links between neural systems and health can be challenging since there isn't a one-to-one mapping between brain regions and psychological states. Building sensitive and specific predictive models of health-relevant constructs using multivariate activation patterns of brain activation is a promising new direction. We illustrate the potential of this approach by building two "neural signatures" of food craving regulation using multivariate machine learning and, for comparison, a univariate contrast. We applied the signatures to two large validation samples of overweight adults who completed tasks measuring craving regulation ability and valuation during food choice. Across these samples, the machine learning signature was more reliable. This signature decoded craving regulation from food viewing and higher signature expression was associated with less craving. During food choice, expression of the regulation signature was stronger for unhealthy foods and inversely related to subjective value, indicating that participants engaged in craving regulation despite never being instructed to control their cravings. Neural signatures thus have the potential to measure spontaneous engagement of mental processes in the absence of explicit instruction, affording greater ecological validity. We close by discussing the opportunities and challenges of this approach, emphasizing what machine learning tools bring to the field of health neuroscience. neural signature | multivariate fMRI | health neuroscience | craving regulation | food valuation

Neural indicators of food cue reactivity, regulation, and valuation and their associations with body composition and daily eating behavior

“…Therefore, for each psychological process, we included two meta-analytic maps corresponding to the association and uniformity tests on NeuroSynth. We also included a single unthresholded, whole-brain neural signature of food craving regulation (Cosme et al, 2020), as a secondary measure of cognitive control specific to food craving. We assessed the degree to which participants expressed the patterns of interest by treating each multivariate pattern and each participant-level condition contrast as a vector of weights and taking the dot product of each combination (Cosme et al, 2020;Doré et al, 2017) using the 3ddot function in AFNI.…”

“…We also included a single unthresholded, whole-brain neural signature of food craving regulation (Cosme et al, 2020), as a secondary measure of cognitive control specific to food craving. We assessed the degree to which participants expressed the patterns of interest by treating each multivariate pattern and each participant-level condition contrast as a vector of weights and taking the dot product of each combination (Cosme et al, 2020;Doré et al, 2017) using the 3ddot function in AFNI. This process yields one scalar "pattern expression value" (PEV) for each condition and multivariate pattern, and this can generally be interpreted as relative evidence for the target/referent psychological process being present or engaged for that participant.…”

“…Whereas most research to date has utilized controlled tasks to assess neural activation during self-control, recent research has administered tasks and stimuli with fewer constraints and little to no instructions, enabling naturally-occurring psychological processes to be captured (e.g., Reggente et al, 2018;Vanderwal et al, 2019). Indeed, this approach has been applied by measuring brain activation associated with food cue reactivity and self-regulation in response to naturalistic food cues (i.e., food commercials; Rapuano et al, 2017) and during food choice (Cosme et al, 2020).…”

“…While these studies utilized patterns derived from meta-analysis, others have utilized machine learning to develop whole-brain patterns indexing target psychological processes, also known as neural signatures (Chang et al, 2015;Wager et al, 2013;Woo et al, 2017). In the eating domain, Cosme and colleagues (2020) showed that patterns of activation indexing food craving regulation derived using machine learning exhibited greater out-of-sample accuracy than patterns derived from univariate contrasts when classifying whether people were regulating their food cravings or not (Cosme et al, 2020).…”

“…Here, we selected several a priori multivariate patterns representing cognitive processes related to food cue reactivity, regulation, and valuation selected using the NeuroSynth database (Yarkoni et al, 2011), as well as the aforementioned neural signature of food craving regulation (Cosme et al, 2020). For each participant and for each multivariate pattern, we computed pattern expression values, which can be interpreted as an index of how strongly the target cognitive process was engaged during exposure to appetizing food cues.…”