Brains and Machines

Why Brains and Machines? Brains are machines. A brain is an organ, it has a purpose. And it’s constrained by the laws of physics, the laws of probability – the laws of the known universe. It needs to fit within these laws to perform its function. So, it is subject to the same kinds of trade-offs, constraints and costs that ‘machines’ in the engineering sense are subject.

It is now possible to measure and manipulate signals in the brain at single-cell resolution, over extended periods and with minimal unwanted impact on nervous system function. This has resulted in tremendous progress in neuroscience in the last decade and a flurry of new research technologies, clinical interventions and diagnostic tools. At the same time, this progress has resulted in a number of era-defining challenges:

analysing, interpreting and managing a deluge of new data
designing biocompatible sensors, hardware and algorithms to interface with living nervous systems
developing theoretical principles for understanding how brains process information
anticipating societal challenges and disruption due to increased human-brain-machine interaction

Conversely, many of the recent advances in information engineering and automation, particularly AI and Machine Learning, have been heavily inspired by neural architectures. This suggests an approach that marries traditional engineering disciplines with all domains of neuroscience. The Brains and Machines theme thus aims to exploit progress in neurally-inspired engineering and data science with major open challenges in neuroscience, biotechnology and biomedicine. This provides opportunities that move far beyond the traditional basic/translational divide in neuroscience to address emerging societal challenges that will define this century and the next.

Brains and Machines encapsulates the development of artificial intelligence approaches with applications in neuroscience and mental health, including computational neuroscience approaches and artificial networks to advance: a) our understanding of the workings of the brain, b) the design of neuro-inspired artificial systems, c) diagnosis of disease and prediction of treatment outcomes, paving the way to a personalised approach to mental health and brain disorder.

There are four major areas of research covered by the theme:

Machines interfaced with brains

This area covers research on recording or stimulation devices and pipelines used in research or clinical settings, as well as closed-loop brain machine interfaces and prosthetic devices. Research in this area thus spans electrophysiology, non-invasive imaging (EEG, fMRI), semiconductor and material science, optics, biocompatible materials, signal processing, control engineering and robotics.

Brains as machines: neural computation

Research that aims to understand the computational principles of the nervous system falls into this category, as well as research on brain-inspired computing. This includes artificial intelligence, neuromorphic computing and autonomous (intelligent) robotics.

Machines understanding brains

The task of analysing and interpreting neural data is a formidable challenge and neural data science is emerging as its own discipline. This is happening at a time when AI and Machine Learning methods are increasingly central to finding patterns in data and generating predictions. Thus, there is a deep synergy between brain-inspired computation and traditional statistical modelling and inference. This trend is reflected in flagship conferences such as NeurIPS, which attract research on neuroscience, machine learning and AI interchangeably.

Philosophical, social and political aspects of brain-machine interactions

It is impossible to ignore societal implications of advances in the three areas above and this demands new research in: the Ethics of machine intelligence and bioengineering brain tissue; how living with machines affects our brain and mind and our social interactions; how our knowledge of human biases should influence the machines we build; how personal data should be handled; ethics of commercialisation and innovation.

No single research area sits within any existing University Department or School. Cross-disciplinary training, collaboration and organisation is therefore essential for this theme to progress in Cambridge and for Cambridge to become competitive in this theme internationally. These challenges were identified as being both intrinsic to the nature of the theme as well as resulting from a number of institutional challenges. This theme brings together a broad range of researchers from Cambridge-based institutes such as the MRC Laboratory of Molecular Biology, as well as many different University departments including Engineering, Computer Science & Technology, Clinical Neurosciences, MRC Cognition and Brain Sciences Unit, Physiology, Development and Neuroscience, Psychiatry, Psychology, Chemical Engineering & Biotechnology and Applied Mathematics & Theoretical Physics.