About this Research Topic
While the influence of neuroscience on the study of Artificial Intelligence - most obviously in the use of artificial neural networks - is widely recognised, it is important to acknowledge the reciprocity of this relationship. Advances in probabilistic inference and machine learning have had a profound impact upon theoretical neurobiology, and upon approaches used to study the brain empirically. This impact ranges from theoretical constructs, like ‘the Bayesian brain’, through to detailed accounts of neuronal architectures inspired by the message passing schemes used for efficient probabilistic inference. Recently, these ideas have been extended to a broader social and anthropological context, giving rise to the emerging field of ‘variational neuroethology’. The probabilistic underpinning of these provides a common language that enables communication between diverse fields within (and beyond) neuroscience.
These approaches, unified by their appeal to recent developments in abductive inference, offer exciting new perspectives on the function of the healthy brain. Crucially, they also provide accounts of brain dysfunction that have been exploited through computational approaches to psychiatry and neurology. This is important for understanding the computational deficits that underwrite behavioural syndromes and to motivate the development of novel therapies. It is also important for phenotyping of patients who fall within broad, heterogeneous, syndrome categories to facilitate a precision medicine approach; in which existing therapies may be targeted towards specific computational pathologies. The use of abductive reasoning (and specifically Bayesian inference) in addressing these problems is especially important, in that this relies upon forward (generative) models of how data are generated. The advantage of this is that it mandates the development of mechanistic (computational) accounts of brain function.
In this Article Collection, we aim to bring together concepts of brain function and dysfunction that draw from techniques in Artificial Intelligence and Machine Learning to inform our understanding of biological cognition.
We welcome submissions that deal with healthy or pathological neuronal computation from an abductive, probabilistic perspective. This includes computational nosology (classification of psychiatric or neurological disease), accounts of healthy or pathological inference in the brain, or the use of new techniques from machine learning to understand empirical neurobiological or behavioural findings.
We encourage submission of original research, review papers, and perspective papers that fall within the following (broad) categories:
- Use of Artificial Intelligence and Machine Learning to evaluate generative models of brain or behavioural data;
- Probabilistic inference in the brain (e.g. as a way of understanding aspects of cognition, emotion, action, and perception);
- Computational psychiatry and neurology (e.g. accounts of pathological inference, methods for quantitative phenotyping of patients).
Keywords: Computational psychiatry, bayesian brain, abductive inference, computational nosology, artificial intelligence, neuroscience, generative model
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.