The ability of network theories to predict functional output from ensembles of individual noisy and unreliable elements is (most often) demonstrated numerically

or (rarely) derived analytically under severely limiting simplifications. However, the simplified network units used in these theoretical studies are remote from the richness of biological neuronal entities, the complexity of their networks, and their interactions with the environment. Thus most neuronal network concepts are yet to be tested experimentally in physiological settings. Inhibitors,research,lifescience,medical This state-of-the-art points towards an acute need for controlled multi-level experimental access to large networks of real neurons over the wide range of relevant time and length scales (milliseconds to weeks; micrometers to millimeters). An ideal experimental system should serve both as a source for fresh insights as well as a natural test bed for verification or modification of existing theories. What are the requirements from an experimental network system? The system should Inhibitors,research,lifescience,medical allow for references simultaneous stimulation and recordings from many individual neurons and individual synapses; long-term monitoring Inhibitors,research,lifescience,medical and manipulation of both activity and structure

over the wide range of relevant time and length scales; enforcement of developmental constraints at both the structural and functional levels; access to chemical modulation; and controllability of connection between elements as well as between ensembles. Such omni-potentiality is practically impossible at the level of behaving organisms or in

preparations where preformed structures are examined in vitro (e.g. brain slices). Here we review the use of multi-site interaction with large cortical networks Inhibitors,research,lifescience,medical developing ex vivo, in a culture dish, to study basic biophysical aspects of synchronization, adaptation, learning, and representation Inhibitors,research,lifescience,medical in neuronal assemblies. Out of various alternatives, large, random, cultured networks of cortical neurons developing ex vivo are most appropriate experimental model systems for studying the general questions of learning and memory at the population level. An extensive survey of the properties of large, random, cortical networks developing ex vivo may be found in recent reviews.2,3 These Entinostat networks are relatively free of predefined constraints and intervening variables, yet the electrophysiological, biochemical, and pharmacological properties of their neurons are by and large identical to neurons in vivo.4–9 The proportions of different cell types are practically identical to those found in vivo.10–12 Unlike slice preparations, the ex vivo developing networks are not cut out of a larger system to which their structures are particularly fitted, and in the absence of which they might function aberrantly. Indeed, alternative models, such as acute cortical selleck inhibitor slices and cultured slices, allow one to explore “what is there”, but not “how it got to be there”.