1. In most natural environments, sound waves from a single source will reach a listener through both direct and reflected paths. Sound traveling the direct path arrives first, and determines the perceived location of the source despite the presence of reflections from many different locations. This phenomenon is called the "law of the first wavefront" or "precedence effect." The time at which the reflection is first perceived as a separately localizable sound defines the end of the precedence window and is called "echo threshold." The precedence effect represents an important property of the auditory system, the neural basis for which has only recently begun to be examined. Here we report the responses of single neurons in the inferior colliculus (IC) and superior olivary complex (SOC) of the unanesthetized rabbit to a sound and its simulated reflection. 2. Stimuli were pairs of monaural or binaural clicks delivered through earphones. The leading click, or conditioner, simulated a direct sound, and the lagging click, or probe, simulated a reflection. Interaural time differences (ITDs) were introduced in the binaural conditioners and probes to adjust their simulated locations. The probe was always set at the neuron's best ITD, whereas the conditioner was set at the neuron's best ITD or its worst ITD. To measure the time course of the effects of the conditioner on the probe, we examined the response to the probe as a function of the conditioner-probe interval (CPI). 3. When IC neurons were tested with conditioners and probes set at the neuron's best ITD, the response to the probe as a function of CPI had one of two forms: early-low or early-high. In early-low neurons the response to the probe was initially suppressed but recovered monotonically at longer CPIs. Early-high neurons showed a nonmonotonic recovery pattern. In these neurons the maximal suppression did not occur at the shortest CPIs, but rather after a period of less suppression. Beyond this point, recovery was similar to that of early-low neurons. The presence of early-high neurons meant that the overall population was never entirely suppressed, even at short CPIs. Taken as a whole. CPIs for 50% recovery of the response to the probe among neurons ranged from 1 to 64 ms with a median of approximately 6 ms. 4. The above results are consistent with the time course of the precedence effect for the following reasons. 1) The lack of complete suppression at any CPI is compatible with behavioral results that show the presence of a probe can be detected even at short CPIs when it is not separately localizable. 2) At a CPI corresponding to echo threshold for human listeners (approximately 4 ms CPI) there was a considerable response to the probe, consistent with it being heard as a separately localizable sound at this CPI. 3) Full recovery for all neurons required a period much longer than that associated with the precedence effect. This is consistent with the relatively long time required for conditioners and probes to be heard with equal loudness. 5. Conditioners with either the best ITD or worst ITD were used to determine the effect of ITD on the response to the probe. The relative amounts of suppression caused by the two ITDs varied among neurons. Some neurons were suppressed about equally by both types of conditioners, others were suppressed more by a conditioner with the best ITD, and still others by a conditioner with the worst ITD. Because the best ITD and worst ITD presumably activate different pathways, these results suggest that different neurons receive a different balance of inhibition from different sources. 6. The recovery functions of neurons not sensitive to ITDs were similar to those of ITD-sensitive, neurons. This suggests that the time course of suppression may be common among different IC populations. 7. We also studied neurons in the SOC. Although many showed binaural interactions, none were sensitive to ITDs. Thus the response of this population may not be
