Neural circuits process sensory information, generate motor output, and create spontaneous activity
A neuron never works alone. Even in the most primitive nervous systems, all neurons participate in synaptically interconnected networks called circuits. In some hydrozoans (small jellyfish), the major neurons lack specialization and are multifunctional. They serve simultaneously as photodetectors, pattern generators for swimming rhythms, and motor neurons. Groups of these cells are repetitively interconnected by two-way electrical synapses into simple ring-like arrangements, and these networks coordinate the rhythmic contraction of the animal's muscles during swimming. This simple neural network also has the flexibility to command defensive changes in swimming patterns when a shadow passes over the animal. Thus, neuronal circuits have profound advantages over unconnected neurons.
In more complex animals, each neuron within a circuit may have very specialized properties. By the interconnection of various specialized neurons, even a simple neuronal circuit may accomplish astonishingly intricate functions. Some neural circuits may be primarily sensory (e.g., the retina) or motor (e.g., the ventral horns of the spinal cord). Many circuits combine features of both, with some neurons dedicated to providing and processing sensory input, others to commanding motor output, and many neurons (perhaps most) doing both. Neural circuits may also generate their own intrinsic signals, with no need for any sensory or central input to activate them. The brain does more than just respond reflexively to sensory input, as a moment's introspection will amply demonstrate. Some neural functions—such as walking, running, breathing, chewing, talking, and piano playing—require precise timing, with coordination of rhythmic temporal patterns across hundreds of outputs. These basic rhythms may be generated by neurons and neural circuits called pacemakers because of their clock-like capabilities. The patterns and rhythms generated by a pacemaking circuit can always be modulated—stopped, started, or altered—by input from sensory or central pathways. Neuronal circuits that produce rhythmic motor output are sometimes called central pattern generators; we discuss these in a section below.
This chapter introduces the basic principles of neural circuits in the mammalian central nervous system (CNS). We describe a few examples of specific systems in detail to illuminate general principles as well as the diversity of neural solutions to life's complex problems. However, this topic is enormous, and we have necessarily been selective and somewhat arbitrary in our presentation.
Nervous systems have several levels of organization
The function of a nervous system is to generate adaptive behaviors. Because different species face unique problems, we expect brains to differ in their organization and mechanisms. Nevertheless, certain principles apply to most nervous systems. It is useful to define various levels of organization. 
N16-1 We can analyze a complex behavior—reading the words on this page—in a simple way, with progressively finer detail, down to the level of ion channels, receptors, messengers, and the genes that control them. At the highest level, we recognize neural subsystems and pathways (see Chapter 10), which in this case include the sensory input from the retina (see Chapter 15) leading to the visual cortex, the central processing regions that make sense of the visual information and the motor systems that coordinate movement of the eyes and head. Many of these systems can be recognized in the gross anatomy of the brain. Each specific brain region is extensively interconnected with other regions that serve different primary functions. These regions tend to have profuse connections that send information in both directions along most sensory/central motor pathways. The advantages of this complexity are obvious; while you are interpreting visual information, for example, it can be very useful simultaneously to analyze sound and to know where your eyes are pointing and how your body is oriented.
N16-1
Levels of Organization of the Nervous System
Contributed by Barry Connors
EFIGURE 16-1 (Data from Shepherd GM: Neurobiology, 3rd ed. New York, Oxford University Press, 1994.)
The systems of the brain can be more deeply understood by studying their organization at the cellular level. Within a local brain region, the arrangement of neurons and their synaptic connections is called a local circuit. A local circuit typically includes the set of inputs, outputs, and all the interconnected neurons that are essential to functions of the local brain region. Many regions of the brain are composed of a large number of stereotyped local circuits, almost modular in their interchangeability, that are themselves interconnected. Within the local circuits are finer arrangements of neurons and synapses sometimes called microcircuits. Microcircuits may be repeated numerous times within a local circuit, and they determine the transformations of information that occur within small areas of dendrites and the collection of synapses impinging on them. At even finer resolution, neural systems can be understood by the properties of their individual neurons (see Chapter 12), synapses, membranes, molecules (e.g., neurotransmitters and neuromodulators), and ions as well as the genes that encode and control the system's molecular biology.
Most local circuits have three elements: input axons, interneurons, and projection (output) neurons
One of the most fascinating things about the nervous system is the wide array of different local circuits that have evolved for different behavioral functions. Despite this diversity, we can define a few general components of local circuits, which we illustrate with two examples from very different parts of the CNS: the ventral horn of the spinal cord and the cerebral neocortex. Some of the functions of these circuits are described in subsequent sections; here, we examine their cellular anatomy.
All local circuits have some form of input, which is usually a set of axons that originate elsewhere and terminate in synapses within the local circuit. A major input to the spinal cord (Fig. 16-1) is the afferent sensory axons in the dorsal roots. These axons carry information from somatic sensory receptors in the skin, connective tissue, and muscles (see pp. 383–389). However, local circuits in the spinal cord also have many other sources of input, including descending input from the brain and input from the spinal cord itself, both from the contralateral side and from spinal segments above and below. Input to the local circuits of the neocortex (Fig. 16-2) is also easily identified; relay neurons of the thalamus send axons into particular layers of the cortex to bring a range of information about sensation, motor systems, and the body's internal state. By far, the most numerous type of input to the local circuits of the neocortex comes from the neocortex itself—from adjacent local circuits, distant areas of cortex, and the contralateral hemisphere. These two systems illustrate a basic principle: local circuits receive multiple types of input.
FIGURE 16-1 Local circuits in the spinal cord. A basic local circuit in the spinal cord consists of inputs (e.g., sensory axons of the dorsal roots), interneurons (both excitatory and inhibitory), and output neurons (e.g., α motor neurons that send their axons through the ventral roots).
FIGURE 16-2 Local circuits in the neocortex. A basic local circuit in the neocortex consists of inputs (e.g., afferent axons from the thalamus), excitatory and inhibitory interneurons, and output neurons (e.g., pyramidal cells).
Output is usually achieved with a subset of cells known as projection neurons, or principal neurons, which send axons to one or more targets. The most obvious spinal output comes from the α motor neurons, which send their axons out through the ventral roots to innervate skeletal muscle fibers. Output axons from the neocortex come mainly from large pyramidal neurons in layer V, which innervate many targets in the brainstem, spinal cord, and other structures, as well as from neurons in layer VI, which make their synapses back onto the cells of the thalamus. However, as was true with inputs, most local circuits have multiple types of outputs. Thus, spinal neurons innervate other regions of the spinal cord and the brain, whereas neocortical circuits make most of their connections to other neocortical circuits.
Rare, indeed, is the neural circuit that has only input and output cells. Local processing is achieved by additional neurons whose axonal connections remain within the local circuit. These neurons are usually called interneurons or intrinsic neurons. Interneurons vary widely in structure and function, and a single local circuit may have many different types. Both the spinal cord and neocortex have excitatory and inhibitory interneurons, interneurons that make very specific or widely divergent connections, and interneurons that either receive direct contact from input axons or process only information from other interneurons. In many parts of the brain, interneurons vastly outnumber output neurons. To take an extreme example, the cerebellum has ~1011 granule cells—a type of excitatory interneuron—which is more than the total number of all other types of neurons in the entire brain!
The “principles” of local circuits outlined here have many variations. For example, a projection cell may have some of the characteristics of an interneuron, as when a branch of its output axon stays within the local circuit and makes synaptic connections. This branching is the case for the projection cells of both the neocortex (pyramidal cells) and the spinal cord (α motor neurons). On the other hand, some interneurons may entirely lack an axon and instead make their local synaptic connections through very short neurites or even dendrites. In some rare cases, the source of the input to a local circuit may not be purely synaptic but chemical (as with CO2-sensitive neurons in the medulla; see p. 714) or physical (as with temperature-sensitive neurons in the hypothalamus; see p. 1199). Although the main neurons within a generic local circuit are wired in series (see Figs. 16-1 and 16-2), local circuits, often in massive numbers, operate in parallel with one another. Furthermore, these circuits usually demonstrate a tremendous amount of crosstalk; information from each circuit is shared mutually, and each circuit continually influences neighboring circuits. Indeed, one of the things that makes analysis of local neural circuits so exceptionally difficult is that they operate in highly interactive, simultaneously interdependent, and expansive networks.