Since noxious stimulation usually leads to the perception of pain, pain has traditionally been considered sensory nociception. But its variability and sensitivity to a broad array of cognitive and motivational factors have meant it is commonly viewed as inherently imprecise and intangibly subjective. However, the core function of pain is motivational-to direct both short-and long-term behavior away from harm. Here, we illustrate that a reinforcement learning model of pain offers a mechanistic understanding of how the brain supports this, illustrating the underlying computational architecture of the pain system. Importantly, it explains why pain is tuned by multiple factors and necessarily supported by a distributed network of brain regions, recasting pain as a precise and objectifiable control signal.Despite the advent of brain imaging, a clear picture of how pain is processed in the brain has been much harder to unravel than anticipated, being beset by three problems. First, pain is associated with robust responses in multiple and diverse brain regions, most of which are not specific to pain (at least on a macroscopic scale), and so it has been hard to ''pin down'' the pain system to any specific brain region. Second, pain is an inherently private percept, but an individual's self-reports of pain can vary widely from moment to moment, and it has remained unclear whether this fluctuation represents irreducible noise and subjectivity or a precise tuning of pain based on hidden factors. Third, pain is exquisitely sensitive to a broad range of emotional, environmental, and cognitive factors-a phenomenon called endogenous modulation. Although this has led to an appreciation that pain is more than a simple readout of nociceptive input, it has not led to any satisfactory unified explanation as to what pain really is. This has left the view that pain is simply a highly variable and malleable representation of assumed actual or potential tissue damage.In this review, we propose a model of pain that centralizes its role as a learning and control signal and argue that this can solve these problems. We begin with a perspective of how theories of pain have evolved over recent decades, and how insights have emerged that have moved thinking beyond purely sensory accounts of pain. We then argue that current accounts still don't fully capture how pain controls behavior to minimize harm, which is its primary function. Importantly, although this is often achieved by immediate nocifensive responses, a substantial part of this comes from learning-allowing an animal to mitigate or avoid predictable harm long into the future. The foundations of a learning account of pain are rooted in psychological models of animal learning, and we describe how these can be developed in computational terms to provide a mechanistic model of the architecture of the pain system. Critically, we argue that this requires pain to be shaped by a set of factors to optimize its role as a learning and control signal and review evidence that suggests that a great...

The limited efficacy of available antidepressant therapies may be due to how they affect the underlying brain network. The purpose of this study was to develop a melancholic MDD biomarker to identify critically important functional connections (FCs), and explore their association to treatments. Resting state fMRI data of 130 individuals (65 melancholic major depressive disorder (MDD) patients, 65 healthy controls) were included to build a melancholic MDD classifier, and 10 FCs were selected by our sparse machine learning algorithm. This biomarker generalized to a drug-free independent cohort of melancholic MDD, and did not generalize to other MDD subtypes or other psychiatric disorders. Moreover, we found that antidepressants had a heterogeneous effect on the identified FCs of 25 melancholic MDDs. In particular, it did impact the FC between left dorsolateral prefrontal cortex (DLPFC)/inferior frontal gyrus (IFG) and posterior cingulate cortex (PCC)/precuneus, ranked as the second ‘most important’ FC based on the biomarker weights, whilst other eight FCs were normalized. Given that left DLPFC has been proposed as an explicit target of depression treatments, this suggest that the limited efficacy of antidepressants might be compensated by combining therapies with targeted treatment as an optimized approach in the future.

Machine learning classifiers for psychiatric disorders using resting-state functional magnetic resonance imaging (rs-fMRI) have recently attracted attention as a method for directly examining relationships between neural circuits and psychiatric disorders. To develop accurate and generalizable classifiers, we compiled a large-scale, multi-site, multi-disorder neuroimaging database. The database comprises resting-state fMRI and structural images of the brain from 993 patients and 1,421 healthy individuals, as well as demographic information such as age, sex, and clinical rating scales. To harmonize the multi-site data, nine healthy participants (“traveling subjects”) visited the sites from which the above datasets were obtained and underwent neuroimaging with 12 scanners. All participants consented to having their data shared and analyzed at multiple medical and research institutions participating in the project, and 706 patients and 1,122 healthy individuals consented to having their data disclosed. Finally, we have published four datasets: 1) the SRPBS Multi-disorder Connectivity Dataset 2), the SRPBS Multi-disorder MRI Dataset (restricted), 3) the SRPBS Multi-disorder MRI Dataset (unrestricted), and 4) the SRPBS Traveling Subject MRI Dataset.