Multistability and metastability: understanding dynamic coordination in the brain

J. A. Scott Kelso 0 Intelligent Systems Research Centre, University of Ulster , Derry , UK 1 Human Brain and Behavior Laboratory, Center for Complex Systems and Brain Sciences, Florida Atlantic University , Boca Raton, FL 33435 , USA Articles on similar topics can be found in the following collections behaviour (554 articles) cognition (377 articles) neuroscience (542 articles) theoretical biology (71 articles) Receive free email alerts when new articles cite this article - sign up in the box at the top right-hand corner of the article or click here - Subject collections Email alerting service Multistability and metastability: understanding dynamic coordination in the brain J. A. Scott Kelso1,2,* Multistable coordination dynamics exists at many levels, from multifunctional neural circuits in vertebrates and invertebrates to large-scale neural circuitry in humans. Moreover, multistability spans (at least) the domains of action and perception, and has been found to place constraints upon, even dictating the nature of, intentional change and the skill-learning process. This paper reviews some of the key evidence for multistability in the aforementioned areas, and illustrates how it has been measured, modelled and theoretically understood. It then suggests how multistability when combined with essential aspects of coordination dynamics such as instability, transitions and (especially) metastabilityprovides a platform for understanding coupling and the creative dynamics of complex goal-directed systems, including the brain and the brain behaviour relation. 1. INTRODUCTION In addressing the subject of multistability, we are advised to follow Socrates: first ask what? Then ask why? So, what is multistability and how do we understand it? In previous work, my colleagues and I have considered multistability and its cognate aspects (fluctuations, instability, transitions, metastability, hysteresis, adaptation, etc.) in vision, speech, language, motor and neural dynamics ([1]; see also earlier studies [2 4] for related approaches). In the last 15 years, many empirical generalizations and modelling developments have taken place on the subject, at least in part owing to the advent of structural and functional neuroimaging and increasingly sophisticated analysis and computational modelling tools. Here, however, after a few general words on what multistability is, I wish to focus on the question of why multistability occurs in the first place. For that one has to go beyond traditional disciplinary boundaries and fully embrace the science of complex systems tailored to the goal-directed and functional aspects of living thingscoordination dynamicswhere multistability is not some freak phenomenon [5], but rather is close to the very core of the way things are. The hope is that asking the why question may reshape our perspective by providing not only a language for understanding multistability but also a rationale for its occurrence, indeed its ubiquity. One contribution of 10 to a Theme Issue Multistability in perception: binding sensory modalities. 2. WHAT IS MULTISTABILITY? Multistability is a universal, essentially nonlinear aspect of matter and its organizationfrom molecular arrangements and chemical reactions to multistability in the meaning of words and actions and beyond [6]. When we talk of multistability, we are usually talking about stable states or attractors; the stability of a state depends on how quickly the system returns to a state following a perturbation.1 In the brain, attractors correspond to stable patterns of reverberating activity in neural populations that support and sustain themselves. The image is one of a changing dynamic landscape where population activity shifts from one attractive state to another. Fluctuations in the nervous system occur on many levels, from channel noise in synaptic transmission to neural populations and neural networks ([7] for review), destabilizing self-sustaining patterns and causing switching from one attractor state to another. A specific example comes from recent biophysical modelling [8] in which multistability and scale-invariant fluctuations arise in resting state cortical activity as a result of noisy input into thalamic neurons modulated by cortical feedback. In short, if a system has multiple coexisting attractors and noise is sufficiently strong to cause switching among stable states, it may be said to be multistable. Understanding multistability may proceed in (at least) two ways. In the first, one may seek specific mechanisms for a given instance of multistable phenomena. The search for specific mechanism proceeds by identifying the physico-chemico-bio-psycho- basis of a chosen exemplar of interest. A second way is to seek dynamic laws and principles of multistability. In this approach, the same dynamic laws of multistability are expected to exist across multiple scales of observation regardless of their specific mechanistic realization. Such a law-based perspective leads us to notice parallels between different expressions of multistability across a wide range of systems, functions and timescales, seeing the connections between them and trying to uncover the functional roles that multistability may play [9,10]. As we shall see, multistability appears in unexpected guises, playing a central role where we might not even expect it to such as in skill learning and intentional change. Universal dynamics springs from inferred principles, specific mechanisms from close examination of particular cases. Dynamical laws and specific mechanisms can thus be seen as complementary [11]: both are needed for a comprehensive understanding of multistability and cognate phenomena. This apparent tension between dynamics and mechanism may be seen as an advantage: universal dynamics allows us to see the connections across manifold expressions of multistability in a level- and mechanism-independent way. At the same time, we need particular realizations if we are ever to extract general laws and principles [12], thereby focusing our attention on mechanisms. It should be noted that many physicists take the word mechanism to mean that dynamical laws at one level are connected to, if not to be eventually replaced by, dynamical laws at lower levels (as in the case of quantum mechanics). Whether one accepts it or not, such a perspective does not deny the possibility of eventually identifying universal (dynamical) mechanisms for multistability and related phenomena across scales and levels of observation. The present contribution may be seen as a move in that direction. 3. WHY MULTISTABILITY? Why are systems, particularly complex biological systems (including, but not necessarily restricted to the brain) composed of very many interacting parts and processes and capable of producing a large repertoire of coordinated patterns of behaviour, multistable in the first place? Any answer to the question why multistability? is likely to be (...truncated)