However, even in these relatively simple systems, specific elements of a functional repertoire may be distributed across multiple cell types—a circuit is certainly more than the sum of its parts. 3 MA As one moves from peripheral circuits that carry out relatively fixed routines to CNS circuits that mediate increasingly complex behaviors, the relationships between the number of cell types and function are less obvious. It is not immediately apparent how different structures utilize discrete cell types

in order to mediate distinct but related forms of neural computation. For example, why do the entorhinal cortex and hippocampus organize at least several scores of distinct cell types into nested maps comprised of grid and place cells in order to mediate spatial learning (Parra et al., 1998 and Thompson et al., 2008), whereas check details the cerebellar cortex can execute its complex procedural learning tasks with only a dozen or so discrete cell types (Llinás and Welsh, 1993 and Gao et al., 2012)? We also lack an adequate explanation for the hundreds of distinct cell types thought to be present in the cerebral cortex, even considering its lamination, variations in local architectonic

structure, and exceedingly complex functional properties. One feature of nervous systems that may explain some of the cell-type diversity evident in complex systems is the ability of circuit activity to be modulated remotely by neuropeptides and other small mediators (Bargmann, 2012). Given the very specific expression patterns observed for a large number of neuropeptide and G protein-coupled receptors in the mammalian brain, segregation of these modulatory pathways of into distinct circuit elements offers opportunities for simultaneous customized control of multiple circuits by the release of a wide variety of peptides, lipids, and other small molecules. Examples of this type of global modulation in response to internal states in mammals include the regulation of emotion by serotonin (Meneses and Liy-Salmeron, 2012) and neuropeptides (Love, 2013), the induction of “sickness behaviors”

in response to prostaglandins (Pecchi et al., 2009), and the modulation of feeding behavior by peripherally produced peptides (Friedman, 2009). Given that the cell-surface receptors mediating these complex behavioral states converge onto a small number of intracellular effector pathways, their segregation into different cell types may be required in order to optimize their effects. Consider the actions of serotonin in the cerebral cortex. Several serotonin receptors are expressed in the cortex, each with a different distribution across cortical cell types. Htr3a receptors, for example, are ionotropic and expressed in a range of interneuron classes that include neurogliaform cells that are thought to function for volume transmission of GABA (Oláh et al., 2009) and bipolar VIP-expressing populations that function selectively in disinhibition (Dávid et al., 2007).