210 



AUDITORY SIGNALS 



needed because of the restricted nature of the 

 method. 



TurnbuU (35) has shown how duration 

 affects frequency discrimination. His tech- 

 nique provided a standard duration tone 

 followed after a silent interval by a tone of 

 shorter and variable duration. Discrimina- 

 tion became rapidly worse at durations of 

 less than 100 milliseconds, and was almost 

 impossible below about 20 milliseconds. 



Experiments in progress by the author 

 indicate that frequency discrimination as a 

 function of duration depends on whether a 

 short tone is compared to a long one, or 

 whether two short tones are compared. 

 Particularly at the very short durations, 

 discrimination is better when both tones 

 have the same duration. Doughty and Gar- 

 ner (2) have shown that most tones in the 

 middle and higher range of audio frequencies 

 must last for about 10 milliseconds for them 

 to sound like a tone at all. Later experi- 

 ments by these same authors (3) also showed 

 that tones tend to sound lower in pitch when 

 they become very short, but that this pitch 

 loss is relatively slight and occurs only at 

 very short durations (under 50 milliseconds) . 



Experiments by Harris (16) indicate that 

 frequency discrimination is considerably 

 poorer when a masking noise is great enough 

 almost to mask the tones. When tones are 

 about 10 to 15 db above the masked thresh- 

 old, however, the presence of the masking 

 noise has little effect. 



Factors which affect intensity discrimina- 

 tion undoubtedly also affect frequency dis- 

 crimination, and research is needed to deter- 

 mine just how much of an effect is present, 

 and where the effects occur. 



Complex Discrimination 



An area where a great deal of research is 

 needed is in the discrimination of complex 

 tones of various sorts, and in the complex 

 discrimination process. The literature in 

 this area is very weak. 



Complex Tones 



We know practically nothing about simple 

 intensity and frequency discrimination of 

 complex tones. Any research in this area 

 would be useful, but particularly we need to 

 know how discrimination of complex tones 

 compares with discrimination of pure tones. 

 Is frequency discrimination better when a 

 tone has a high harmonic content, as in 

 square waves and pulse wave forms? It 

 may be that discrimination can be improved 

 considerably by using such tones. Miller 

 (23) has studied intensity discrimination 

 with noise, and found discrimination to be 

 about the same as for tones. 



Discrimination of Spectrum Changes 



Another kind of problem about which we 

 know practically nothing is the discrimina- 

 tion of spectrum changes. The recent re- 

 search on speech seems to indicate that we 

 understand speech by discriminating the 

 changes in spectrum through time. We 

 need to know discrimination limits for har- 

 monic content. There is little question that 

 humans can tell the difference between a 

 pure tone and one of high haraionic content, 

 but we do not know how precise that dis- 

 crimination can be. A possible step in this 

 direction has been taken by Karlin (21), who 

 presented data on the discrimination of the 

 pitch and loudness of noise spectra. 



Glide Tones 



Pepinsky (26) reports an experiment on 

 the perception of frequency-modulated 

 tones. His tones changed frequency in one 

 direction over a short period of time, and he 

 reports that the ability to detect the change 

 in pitch depends both on the amount of the 

 frequency change and on the duration of 

 the change, a greater frequency change being 

 required for shorter durations. His experi- 

 ments, however, were done at only one fre- 

 quency, and do not provide sufficient data 

 for all purposes. More research is needed 

 on the perception of this tj'^pe of tone, and 



