1. We investigated the influence of sound direction on the frequency-tuning characteristics of neurons in the frog inferior colliculus, the torus semicircularis. For each neuron, we used tone bursts to determine the frequency-threshold curves (FTCs) for three to seven loudspeaker azimuths. The loudspeaker was mounted on a rotatable arc and could be swung through the frontal field between positions opposite the ear that was contralateral to the recording site (C90-degrees) and the ear that was ipsilateral to the recording site (I90-degrees). 2. Frequency-tuning data from 83 units showed that the characteristic frequency (CF) shift through a 180-degrees change in loudspeaker azimuth was typically small, i.e., 85% of neurons showed maximum absolute changes in CF that were < 0.4 octaves. Paired comparisons of CFs for each neuron when the loudspeaker was located at C90-degrees, I90-degrees, and the frontal midline position (0-degrees) revealed no significant differences (P > 0.2) between azimuths. The magnitude of CF shift between different sound directions showed no systematic pattern. 3. In contrast to the CF, midbrain neurons showed distinct changes in the minimum threshold (MT) at CF across 180-degrees of azimuth. The maximum absolute change in MT ranged from 0 to 38 dB, with a mean of 10.9 dB. A pair-wise comparison of MTs, for each neuron, derived with the speakers at C90-degrees, 0-degrees, and I90-degrees, revealed that the MT typically increased when the loudspeaker was rotated toward I90-degrees (P < 0.0001). 4. The most prominent effect of rotating the loudspeaker from C90-degrees to I90-degrees was a narrowing of the FTC. Sharpness of tuning in simple V-shaped FTCs was most directly shown by changes in the Q factors at 10, 20, and 30 dB above MT as a function of sound direction. A pair-wise comparison for individual neurons showed that all Q factors were significantly larger (sharper tuning) for I90-degrees compared with C90-degrees (P < 0.02). The Q20dB Values were also significantly larger for 190-degrees compared with 0-degrees (P < 0.05). For the majority of units, the maximum Q10dB and Q20dB values were displayed when the loudspeaker was positioned at I90-degrees; however, the maximum Q30dB was distributed nearly equally for the three azimuths. These results suggest that a change in sound direction has the most robust effect on tuning properties of the FTCs nearest the MT. 5. Similar to neurons with simple V-shaped FTCs, neurons with complex FTCs showed a narrowing of the response areas when the loudspeaker was rotated from C90-degrees to I90-degrees. However, with this shift in sound direction, the shape of the FTC could change from V-shaped to W-shaped, a broad W-shaped to a narrow V-shaped, or transform into separate response regions. 6. To examine how the shape of the FTC changes with sound direction, we derived threshold difference curves (TDCs) by computing the differences between the FTCs for one sound direction and the FTC for C90-degrees. Evaluation of TDCs revealed that changes in the shape of the FTCs were most often associated with shifts in either the low-frequency flank alone or the low- and high-frequency flanks together. 7. The results of this study showed that midbrain neurons exhibited increased sharpening of frequency tuning as a function of the direction of free-field sounds. The mechanisms underlying this sharpening of frequency selectivity are unclear but likely involve an increase in binaural inhibition. Binaural interactions, by sharpening frequency tuning, may serve to increase an animal's ability to analyze complex sound spectra.
