Auditory information is critical for vocal imitation and other elements of social life in song birds. In zebra finches, neural centers that are necessary for the acquisition and production of learned vocalizations are known, and they all respond to acoustic stimulation. However, the circuits by which conspecific auditory signals are perceived, processed, and stored in long-term memory have not been well documented. In particular, no evidence exists of direct connections between auditory and vocal motor pathways, and two newly identified centers for auditory processing, caudomedial neostriatum (Ncm) and caudomedial hyperstriatum ventrale (cmHV), have no documented place among known auditory circuits. Our goal was to describe anatomically the auditory pathways in adult zebra finch males and, specifically, to show the projections by which Ncm and vocal motor centers may receive auditory input. By using injections of different kinds of neuroanatomical tracers (biotinylated dextran amines, rhodamine-linked dextran amines, biocytin, fluorogold, and rhodamine-linked latex beads), we have shown that, as in other avian groups, the neostriatal field L complex in caudal telencephalon is the primary forebrain relay for pathways originating in the auditory thalamus, i.e., the nucleus ovoidalis complex (Ov). In addition, Ncm and cmHV also receive input from the Ov complex. Ov has been broken down into two parts, the Ov "core" and "shell," which project in parallel to different targets in the caudal telencephalon. Parts of the field L complex are connected among themselves and to Ncm, cmHV, and caudolateral Hv (clHV) through a complex web of largely reciprocal pathways. In addition, clHV and parts of the field L complex project strongly to the "shelf" of neostriatum underneath the song control nucleus high vocal center (HVC) and to the "cup" of archistriatum rostrodorsal to another song-control nucleus, the robust nucleus of the archistriatum (RA). We have documented two points at which the vocal motor pathway may pick up auditory signals: the HVC-shelf interface and a projection from clHV to the nucleus interfacialis (NIf), which projects to HVC. These data represent the most complete survey to date of auditory pathways in the adult male zebra finch brain, and of their projections to motor stations of the song system.

Odors evoke complex responses in locust antennal lobe projection neurons (PNs)-the mitral cell analogs. These patterns evolve over hundreds of milliseconds and contain information about odor identity and concentration. In nature, animals often encounter many odorants in short temporal succession. We explored the effects of such conditions by presenting two different odors with variable intervening delays. PN ensemble representations tracked stimulus changes and, in some delay conditions, reached states that corresponded neither to the representation of either odor alone nor to the static mixture of the two. We then recorded from Kenyon cells (KCs), the PNs' targets. Their responses were consistent with the PN population's behavior: in some conditions, KCs were recruited that did not fire during single-odor or mixture stimuli. Thus, PN population dynamics are history dependent, and responses of individual KCs are consistent with piecewise temporal decoding of PN output over large sections of the PN population.

Olfactory processing in the insect antennal lobe is a highly dynamic process, yet it has been studied primarily with static step stimuli. To approximate the rapid odor fluctuations encountered in nature, we presented flickering "white-noise" odor stimuli to the antenna of the locust and recorded spike trains from antennal lobe projection neurons (PNs). The responses varied greatly across PNs and across odors for the same PN. Surprisingly, this diversity across the population was highly constrained, and most responses were captured by a quantitative model with just 3 parameters. Individual PNs were found to communicate odor information at rates up to approximately 4 bits/s. A small group of PNs was sufficient to provide an accurate representation of the dynamic odor time course, whose quality was maximal for fluctuations of frequency approximately 0.8 Hz. We develop a simple model for the encoding of dynamic odor stimuli that accounts for many prior observations on the population response.