1. We studied the spatial distributions of amplitude tuning (monotonicity of rate-level functions) and response threshold of single neurons along the dorsoventral extent of cat primary auditory cortex (AI). To pool data across animals, we used the multiple-unit map of monotonicity as a frame of reference. Amplitude selectivity of multiple units is known to vary systematically along isofrequency contours, which run roughly in the dorsoventral direction. Clusters sharply tuned for intensity (i.e., ''nonmonotonic'' clusters) are located near the center of the contour. A second nonmonotonic region con be found several millimeters dorsal to the center. We used the locations of these two nonmonotonic regions as reference points to normalize data across animals. Additionally. to compare this study to sharpness of frequency tuning results, we also used multiple-unit bandwidth (BW) maps as references to pool data. 2. The multiple-unit amplitude-related topographies recorded in previous studies were closely approximated the previously reported individual case maps when the multiple-unit monotonicity or the map of bandwidth (in octaves) of pure tones to which a cell responds 40 dB above minimum threshold were used as the pooling reference. When the map of bandwidth ( in octaves) of pure tones to which a cell responds 10 dB above minimum threshold map was used as part of the measure, the pooled spatial pattern of multiple-unit activity was degraded. 3. Single neurons exhibited nonmonotonic rate-level functions more frequently than multiple units. Although common in single-neuron recordings (28%). strongly nonmonotonic recordings ( firing rates reduced by >50% at high intensities) were uncommon (8%) in multiple-unit recordings. Intermediately nonmonotonic neurons (firing rates reduced between 20% and 50% at high intensities) occurred with nearly equal probability in single-neuron( 28%) and multiple unit (26%) recordings. The remaining recordings for multiple units (66%) and single units (44%) were monotonic ( firing rates within 20% of the maximum at the highest tested intensity). 4. In ventral AI (AIv). the topography of monotonicity for single units was qualitatively similar to multiple units, although single units were on average more intensity selective. In dorsal AI (AId) we consistently found a spatial gradient for sharpness of intensity tuning for multiple units: however, for pooled single units in AId there was no clear topographic gradient. 5. Response (intensity) thresholds of single neurons were not uniformly distributed across the dorsoventral extent of AI. The most sensitive neurons were consistently located in the nonmonotonic regions. The scatter of single neuron intensity threshold was smallest at these locations and increased gradually toward more dorsal and ventral locations. 6. The existence of a specialized region for near-threshold stimuli along the AIv/AId border is revealed in these experiments. Neuronal recordings in this region are sharply tuned for frequency and amplitude. have low intensity thresholds, have low scatter in characteristic frequency and threshold, and selectively respond to narrowband stimuli within 40 dB of the cortical intensity threshold. Nonmonotonic neurons have been shown to shift their spike count-versus-level functions linearly in response to a continuous noise masker. Neurons in the ventral nonmonotonic region thus might serve as fine spectral/amplitude filters that only respond to frequency-banded components with intensities just above the cat's threshold in the presence of background noise. 7. The results of this study support the parceling of AI into at least two physiologically distinct subdivisions. The ventral subdivision (AIV) has a complete single-unit topographic representation of stimulus intensity. Low-intensity signals elicit maximal response at a signal detection region, located at the dorsal extreme of AIv at the AIV-AId border. Neurons respond better to higher-intensity signals progressively ventrally until the AII border is approached. Aid is well suited for differential frequency analysis and contains a single-unit topograph for stimulus bandwidth, as previously reported.
