(A) Tuning to interaural time difference (ITD) emerges from the convergence of inputs selective for the same ITD but different frequencies (F1–F3) (Takahashi and Konishi, 1986; Mazer, 1998; Pena and Konishi, 2000; best ITD indicated by the dashed line). While ITD-selective neurons respond to a broad range of frequencies (here the black bold curve represents the neuron's combined response for frequencies F1, F2 and F3), their inputs are narrowly tuned to frequency (each input only responding to F1, F2 or F3). Because the inputs are narrowly tuned to frequency, the responses at each input vary with the phase difference between the left and right ears (IPD) of their preferred frequency, as shown by the sinusoidal curves. (B) A sound emitted by a single source (B, left) in front of the owl is filtered by the head and decomposed in narrow frequency channels by the cochlea. The localization cue corresponds to an IPD in each frequency channel (F1–F3). In a different context (B, right), the target frontal sound (yellow) is emitted concurrently with another sound source from a different location (blue). The blue source interferes with the yellow target and shifts the resultant IPDs in each frequency channel (shown in green). Black dotted lines indicate IPD responses for the target frontal source alone, for comparison. While IPD shifts greatly for some frequencies (F1 and F3), in others (F2) IPD is more robust to the presence of another source. Thus in this example, F2 carries the most reliable IPD cue. To provide a clearer visualization that IPD is encoded at different frequencies, F1–F3 inputs remain plotted as a function of ITD in (B).