Intensity discrimination: a severe departure from Weber's law.

These experiments were designed to assess the importance of different types of information which might be used in detecting intensity changes for pure tones. Thresholds for detecting an intensity change, expressed as 10 log (delta I/I), were measured over a wide range of frequencies and levels under conditions where one or more sources of information was either present or was removed. Spread of excitation was restricted by using bandstop noise centered at the signal frequency. Information conveyed by dynamic responses to signal onsets and offsets was eliminated by masking onsets and offsets with bursts of bandpass noise. Phase-locking information was eliminated by using high-frequency signals (above 5 kHz). Dynamic responses to signal onsets and offsets appear to play little role in intensity discrimination. Phase locking does appear to be important since Weber's law or a near-miss to it was observed at low frequencies, whereas at high frequencies performance deteriorated at moderate sound levels, and improved again at high levels. A preliminary experiment, using 225-ms stimuli revealed only a small midlevel deterioration at high frequencies. However, when 30-ms stimuli were used a large deterioration was observed, performance being worse when bandstop noise was presented with the tone. Hence at short durations and high frequencies spread of excitation seems to be important: When it is restricted by bandstop noise values of 10 log (delta I/I) observed at moderate levels it can be as large as 14 dB. The results of the experiments are consistent with a bimodal distribution of thresholds in primary auditory neurons; at intermediate levels neither population will operate effectively.(ABSTRACT TRUNCATED AT 250 WORDS)