Dietmar Plenz - US grants

1. Neuronal avalanches are increasingly recognized to be important for cortex function. My Section took the lead in organizing the first conference on Criticality in Neural Systems in collaboration with Ernst Niebur, Johns Hopkins University. In April 2012, the 2-day conference took place on the NIH campus in Bethesda at the Natcher Conference center with about 100 attendees and featured 19 international and national speakers and posters. Since then, a book with about 22 chapters and international authors, most of who presented at the conference, has been assembled and was published in Spring 2014 (Plenz, D. & Niebur, E. Criticality in Neural Systems, Wiley-VCH, Berlin (2014). The book covers all major aspects of criticality in the brain and is on track to become a standard text book for a rapidly increasing field of critical phenomena in the brain. Besides being the main editor, my Section has contributed 4 chapters covering our major accomplishments demonstrating criticality in the brain from in vitro preparations to the awake animals and normal human subjects. 2. Neuronal avalanches identify critical brain dynamics at which several aspects of information processing are optimized as demonstrated in our previous work. Several classes of critical systems have been identified based on the precise critical exponents that control a systems performance at criticality. One of the main issues though was the proper identification of power laws in critical cortical dynamics. We clarified several misconceptions in the literature regarding the proper identification of cut-offs in scale-invariant critical dynamics. The proper identification of the valid range of power law scaling is a prerequisite for careful evalution of critical brain dynamics (Yu et al. 2014 PLoS One). Abstract: Scale-Invariant Neuronal Avalanche Dynamics and the Cut-off in Size Distributions Identification of cortical dynamics strongly benefits from the simultaneous recording of as many neurons as possible. Yet current technologies provide only incomplete access to the mammalian cortex from which adequate conclusions about dynamics need to be derived. Here, we identify constraints introduced by sub-sampling with a limited number of electrodes, i.e. spatial windowing, for well-characterized critical dynamics ― neuronal avalanches. The local field potential (LFP) was recorded from premotor and prefrontal cortices in two awake macaque monkeys during rest using chronically implanted 96-microelectrode arrays. Negative deflections in the LFP (nLFP) were identified on the full as well as compact sub-regions of the array quantified by the number of electrodes N (10 95), i.e., the window size. Spatiotemporal nLFP clusters organized as neuronal avalanches, i.e., the probability in cluster size, p(s), invariably followed a power law with exponent 1.5 up to N, beyond which p(s) declined more steeply producing a cut-off that varied with N and the LFP filter parameters. Clusters of size s ≤ N consisted mainly of nLFPs from unique, non-repeated cortical sites, emerged from local propagation between nearby sites, and carried spatial information about cluster organization. In contrast, clusters of size s > N were dominated by repeated site activations and carried little spatial information reflecting greatly distorted sampling conditions. Our findings were confirmed in a neuron-electrode network model. Thus, avalanche analysis needs to be constrained to the size of the observation window to reveal the underlying scale-invariant organization produced by locally unfolding, predominantly feed-forward neuronal cascades. 3. To facilitate the use of neuronal avalanche metrics, we published a software package custom-designed in python that allows for relatively easy power law fits to acquired data (Alstott et al., 2014, PLoS One). Abstract: powerlaw: a Python package for analysis of heavy-tailed distributions Power laws are theoretically interesting probability distributions that are also frequently used to describe empirical data. In recent years selective statistical methods for fitting power laws have been developed, but appropriate use of these techniques requires significant programming and statistical insight. In order to greatly decrease the barriers to using good statistical methods for fitting power law distributions, we developed the powerlaw Python package. This software package provides easy commands for basic fitting and statistical analysis of distributions. Notably, it also seeks to support a variety of user needs by being exhaustive in the options available to the user. The source code is publicly available and easily extensible. 4. We filed for a patent that allows quantification of the behaviorally detrimental effects of sleep deprivation using neuronal avalanche based metrics (Plenz et al. 2013). For this project, our group did not perform primary data collection, but performed secondary analysis of clinical data provided by Dr. Giulio Tononi at the University of Wisconsin, USA.

? DESCRIPTION (provided by applicant): The cortex is a laminated structure that is thought to underlie sequential information processing. Sensory input enters layer 4 (L4) from which activity quickly spreads to superficial layers 2/3 (L2/3) and deep layers 5/6 (L5/6) and other cortical areas eventually leading to appropriate motor responses. Sensory responses themselves depend on ongoing, i.e. spontaneous cortical activity, usually in the form of reverberating activit from within or distant cortical regions, as well as the state and behavioral context of the animal. Receptive field properties of neurons can rapidly and adaptively be reshaped when an animal is engaged in a behavioral task, indicating that encoding of stimuli is dependent on task- or context-dependent state. Responses also depend on ongoing cortical dynamics in a lamina-dependent fashion and differ between the awake and anesthetized state. The intricate neuronal interplay between behavioral context, ongoing activity, and sensory stimulus underlying cortical representations is unknown. Specifically, we do not know how neuronal circuits shape these emergent dynamics within and between laminae, and we do not know which neurons encode which aspect of a sensory stimulus. One shortcoming of all prior studies of sensory processing is that only a few neurons are sampled, and thus information about the interactions between neurons, and between neuron and global brain state is lacking. Here we address these challenges by developing new in vivo 2-photon imaging technology that allows rapid imaging and stimulation in multiple focal planes and new computational and information theoretic techniques to extract network dynamics at the single neuron and population level. These measures go beyond paired measures and take synergistic interactions between neurons into account. We use these new techniques to investigate the 3D single cell and population activity patterns in the auditory cortex in mice. We investigate the influence of single neurons relative to the synergistic influence of specific groups of neurons (the crowd) on network dynamics and ultimately behavior of the animal.

Sensory representations in the brain and an animal?s perception change in many ways over many time scales. Over tens or hundreds of milliseconds, ongoing neuronal activity contributes substantially to response variability in primary sensory areas. Effort and arousal typically vary over seconds or minutes, and are also associated with major changes in sensory representations and behavioral performance. Perceptual learning occurs over hours and days, imparting new perceptual capabilities. Working with all these forms of variability in stimulus processing, the brain maintains (in the case of ongoing activity) and improves (in the case of arousal and learning) behavioral outcomes. It is commonly assumed that maintenance and improvements in behavioral outcome depend primarily on changes in the corresponding sensory representation, yet this is far from certain. New methods of two-photon stimulation are ideal for probing how much the contributions of different cortical neurons change across behavioral states or as animals learn new perceptual tasks. The proposed experiments take advantage of the experimental accessibility of stimulus-response associations in primary sensory cortices to identify mechanisms and principles in neuronal circuits that maintain and improve behavioral outcome in the context of brain state changes over many timescales. These studies will test whether the high spatiotemporal variability of ongoing activity reflects higher-order statistics of neuronal population activity that ensure the most informative stimulus processing and best behavioral outcomes. The impact of effort and arousal will be addressed at cellular resolution by identifying changes in population representations and readout in primary sensory cortex between different behavioral states and during saccadic suppression. Perceptual learning experiments will probe the contributions to behavioral performance of individual neurons in the olfactory bulb, primary auditory cortex, and primary visual cortex, and determine how those contributions change and can be manipulated over the course of perceptual learning. Collectively, these experiments will provide a far more precise and granular view of how sensory representations vary over different time scales, and new information on how the decoding of those representations can change over time.

Project Summary To survive, organisms must both accurately represent stimuli in the outside world, and use that representation to generate beneficial behavioral actions. Historically, these two processes ? the mapping from stimuli to neural responses, and the mapping from neural activity to behavior ? have largely been treated separately. Of the two, the former has received the most attention. Often referred to as the ?neural coding problem,? its goal is to determine which features of neural activity carry information about external stimuli. This approach has led to many empirical and theoretical proposals about the spatial and temporal features of neural population activity, or ?neural codes,? that represent sensory information. However, there is still no consensus about the neural code for most sensory stimuli in most areas of the nervous system. The lack of consensus arises in part because, while it is established that certain features of neural population responses carry information about specific stimuli, it is unclear whether the brain uses (?reads?) the information in these features to form sensory perceptions. We have developed a theoretical framework, based on the intersection of coding and readout, to approach this problem. Experimentally informing this framework requires manipulating patterns of neuronal activity based on, and at the same spatiotemporal scale as, their natural firing patterns during sensory perception. This work must be done in behaving animals because it is essential to know which neural codes guide behavioral decisions. In the first phase of this project (funded by the BRAIN Initiative), we developed the technology necessary for realizing this goal. In the present proposal, we will extend our patterned neuronal stimulation technology and apply it to answer long-standing questions about neural coding and readout in the visual, olfactory, and auditory systems. We will pioneer the capacity to determine which neurons within a network are encoding behaviorally relevant information, and also to determine the extent to which temporal patterns of those neurons? activity are being used to guide behavior. Finally, we will study these neural coding principles across changes in behavioral state and during learning to determine how internal context and past experience shape coding and readout. The contributions of the proposed work will be three-fold. First, we will provide the neuroscience community with the tools needed to test theories of how neural populations encode and decode information throughout the brain. Second, we will reveal fundamental principles of spatiotemporal neural coding and readout in the visual, olfactory, and auditory systems of behaving animals. And third, our unifying theoretical framework for cracking neural codes will allow the broader neuroscience community to resolve ongoing debates regarding neural coding that have been previously stalemated by considering only half of the coding/readout problem.