Lecture 6: Brain Functioning and its Challenges Jordi Soriano Fradera Dept. Física de la Matèria Condensada, Universitat de Barcelona UB Institute of Complex Systems September 2016
1. The brain: a true complex system The study of the brain is the study of ourselves, and on how the dynamic interaction of 10 11 neurons with 10 14 connections shapes information processing, decision making and cognition. The brain works. Nonlinear neurons, nonlinearly coupled in a noisy environment shape exquisitely regular macroscopic observables such as brain waves, precisely controlled physical movements, and efficient multi-tasking.
1. The brain: a true complex system The excellent performance of the brain indicates that robust mechanisms area at play, and at a spatiotemporal scales that go beyond spikes and biochemistry. We have seen how nonlinear dynamics can give rise to patterned behavior in an autonomous manner, i.e. without external control. Physics is continuously pouring new concepts and tools to understand brain performance. Attractor dynamics, stochastic facilitation and criticality aspects are prominent examples However, putting aside physics and nonlinear dynamics temporary, how do we face the analysis of the brain? Which aspects of brain functioning are more challenging?
2. Brain research: a daunting (and expensive) endeavor Brain research is nowadays highly interdisciplinary (neurobiology, genetics, medicine, psychology, physics, mathematics, computer science, instrumentation & engineering, ) and international cooperation has become central. Research initiatives have budgets comparable to high-energy physics and space exploration. And strong impact in the media.
-BRAIN Initiative. - Brain activity map. Human Brain Project
- A meticulous virtual copy of the human brain Henry Markram
The Human Brain Project fosters data sharing and communication among different disciplines (physiology and medicine, computer science, physics, ) Bottom-up approach. (molecular) (regions) (whole organ)
Fosters technological developments. Top-down approach. Neuroscience Techniques Observational data
Technology to measure activity Technology to physically map the brain
Zebrafish Light-sheet microscopy
Prevedel et al., Nat. Methods (2014) Shipley et al., Front. Neur. Circ. (2014) C. Elegans Calcium imaging during free movement
C. Elegans. Optogenetics (neuronal excitation through light)
3. How do we explore the brain? Conceptually, one must get information from both structure and neuronal activity, and link the physical map of connections with function and cognition.
3. How do we explore the brain? Factually, the study of the brain involves several spatial and temporal scales. From a network perspective, nodes can be single neurons, assemblies of them or parcellations containing millions of cells (voxels). The brain is embedded in a 3D environment that combines 2D and 3D functional modules, with strong wiring restrictions, evolution-inherited units, redundant circuits and enormous energy consumption. Multi-scale, multi-temporal analysis is challenging given the inherent difficulty (at the moment impossibility) to access all neurons simultaneously, monitor their activity, and translate that information into actual behavior and cognition.
3. How do we explore the brain? Practically, the comprehension of brain functioning involves the combination of several experimental techniques. Like a jigsaw puzzle, they all contribute to get the whole picture.
4. How do we get data? For functional connectivity, the most important techniques are fmri, PET, MEG, and EEG. They typically measure activity in small brain regions or parcellations, either at rest or during the development of specific tasks, both in health and disease. The intensity of the signal along time provides the time series to be analyzed, and functional interactions inferred. 1 n --- C(1,2) = 0.55 C(1,n) = 0.72 C(2,n) = 0.34 fmri
4. How do we get data? For structural connectivity, a recent technique has become very important: diffusion tensor imaging, which measures the anisotropy in water diffusion across the brain. Diffusion is faster parallel to nerve fibers. Data is represented using tractography, a computational technique for representing the fibers.
5. Challenge I. Detailed structural connectivity Structural connections represent anatomical, actual paths. They are obtained from noninvasive techniques such as diffusion imaging and tractography. The human brain exhibits modularity and hierarchical organization, with the existence of hubs and a rich-club core (interconnected high-degree nodes). Hubs are nodes that have a central role linking modules and other higher structures together.
5. Challenge I. Detailed structural connectivity The rich club is shaped by interconnected hubs, and plays an important role in providing short paths among distance brain regions and supporting integration of information. The challenge is to get an accurate network representation of the brain, and in healthy and diseased scenarios. In disease, hubs and rich club are affected, compromising brain s ability to combine integration and segregation. Rich club core
6. Challenge II. Functional connectivity Functional connections represent statistical relations between brain areas. Most of the data represents interrelations during cognitive or physical tasks. But an important fraction of data is centered on quantifying brain s cross-taking when no particular tasks are being conducted: the resting state network. Challenges: 1) Improve the algorithms for data treatment and establish unified analysis tools. 2) Develop tools to obtain a directed and weighted description of brain regions interaction, i.e. build effective networks.
6. Challenge II. Functional connectivity 3) Understand the interplay between the structural and functional networks, for instance to assess highly vulnerable regions and the flow of information. Structural Functional
7. Challenge III. Brain damage and neurological disorders A wealth of research explores the changes in the structural and functional networks upon damage (e.g. stroke, lesion) or neurodegeneration (e.g. Alzheimer's, Parkinson s ), and their mutual relationship. structural damage Healthy controls Functional damage AD patients
7. Challenge III. Brain damage and neurological disorders A first important challenge is the understanding and characterization of those network measures that are central in describing brain deterioration, such as hubs or rich club cores. This sets an urgency for fostering research in network theory. (Alzheimer's Disease, AD) AD targets cortical hubs
7. Challenge III. Brain damage and neurological disorders A second important challenge is the development of tools for early prognosis of brain alterations and neurodegeneration, and the development of new/better techniques for therapeutic treatment. Exploratory experiments for inducing external noise Parkinson treatment Experiments in vitro could provide significant insight in this direction.
8. Challenge IV. Resting state The resting state is background spontaneous activity in the brain. It can be well characterized in measurements in which the subject is not performing any specific tasks. The resting state is both puzzling and challenging: - Rich spatiotemporal patterns across brain regions. - Origin unknown, but seems associated to anatomical connectivity and noise. - The resting state patterns significantly change when a task is being conducted. Counterintuitively, activity may abruptly decrease in the region associated with the task.
8. Challenge IV. Resting state The resting state seems a strategy of the brain for continuously probing the network while waiting for concrete action. For optimum performance, this would imply the existence of long range spatiotemporal correlations among brain regions. Critical behavior! Data suggests that resting state (and the critical behavior) is spontaneously driven, and possible linked to attractor dynamics. Several neurodegenerative disorders cause severe physical damage to neurons and connections, leading to malfunctional nonlinear units. This suffices to significantly alter the properties of bifurcation points, limit cycles and attractors.
8. Challenge IV. Resting state Attractor model for the resting state
9. Final remark The different challenges for understanding the brain can be explored in small living neuronal networks and computer models. The following lectures will show how neuronal cultures, small living neuronal networks in vitro, can help comprehending many open challenges introduced here.
End of lecture 6
TAKE HOME MESSAGE: - The brain is a major complex system. The understanding of its functioning is proving elusive despite knowing well the structure and operation of its building blocks, the neurons. - Brain is studied at diverse spatial and temporal scales, and current research implies the use of several different tools across different disciplines. - The major challenges in brain research include the characterization of the structural and functional connectivity maps, in health and disease, and the understanding of the resting state. Questions and discussion aspects: - Do you think the Human Brain Project will provide comprehensive data? How do you tune all the parameters and initial conditions in such a simulation? - What other challenges do you think I have left in the drawer?
References R. A. Poldrack and M. J. Farah, Progress and challenges in probing the human brain, Nature (2015). M. Rubinov and O. Sporns, Complex network measures of brain connectivity: Uses and interpretations, NeuroImage (2010). R. Frackowiak and H. Markram, The future of human cerebral cartography: a novel approach, Philosophical Transacations B (2016). D. S. Bassett and E.T. Bullmore, Human Brain Networks in Health and Disease, Curr Opin Neurol. (2009). O. Sporns, The Human Connectome: A Structural Description of the Human Brain, PLoS Comput Biol (2005). P. Hagmann et al., Mapping the Structural Core of Human Cerebral Cortex, PLoS Biol (2008). M de la Iglesia-Vaya et al., Brain Connections Resting State fmri Functional Connectivity, Novel Frontiers of Advanced Neuroimaging (2013). G. Deco et al., Emerging concepts for the dynamical organization of resting-state activity in the brain, Nat Rev Neurosci (2011). D. Plenz, The Critical Brain, Physics (2013). G. Deco et al., Resting brains never rest: computational insights into potential cognitive architectures, Trends Neurosci (2013). A. Haimovici, Brain Organization into Resting State Networks Emerges at Criticality on a Model of the Human Connectome, Phys Rev Lett (2013).