Pattern classification differentiates decision of intertemporal choices using multi-voxel pattern analysis.

In daily life, individuals frequently make trade-offs between the small-but-immediate benefits and large-but-delayed profits. This type of decision is known as intertemporal choice. Previous studies have uncovered the neurobiological mechanism of the intertemporal choice, but it still remains unclear how the patterns of brain activity predict the decisions of intertemporal choices. To fill this gap, we used functional magnetic resonance imaging (fMRI), in conjunction with the machine learning technique of multi-voxel pattern analysis (MVPA), to ascertain the predictive capability of the neuronal pattern for classifying individuals' intertemporal decisions across two independent samples. To further probe how this neuronal pattern worked in predicting individual intertemporal decision, we drew on the Power Atlas to examine the accuracies of classification within each regional mask as well. Classification findings showed that the pattern of neuronal activity over the whole-brain can correctly classify the accuracies of individual decisions up to 84.3%. Encouragingly, further analysis shows that the neuronal information encoded in three brain functional networks can predict individuals' decisions with significant discriminative power in cross-samples, namely the valuation network (e.g., striatum), the cognitive control network (e.g., dorsolateral prefrontal cortex) and the episodic prospection network (e.g., amygdala, parahippocampus gyrus, insula). Collectively, these findings advance our comprehension of the neuronal mechanism of human intertemporal decisions, and substantially reshape our understanding for this cardinal behaviour from behavioural-brain scheme to brain-behavioural configuration.