Correlations, but ... (Roy Patterson )

Subject: Correlations, but ...
From:    Roy Patterson  <roy.patterson(at)MRC-APU.CAM.AC.UK>
Date:    Tue, 16 Jun 1998 10:42:36 +0100

Chuck Watson wrote - "I would be curious to know ... whether there is a correlation between, say, critical ratios, band-narrowing critical bands, and notched-noise CB's. Sounds like a 70's-era dissertation." Work on this was actually done in the 70's and 80's. In our paper Patterson, R.D., Nimmo-Smith, I., Weber, D.L., and Milroy, R. (1982). The deterioration of hearing with age: Frequency selectivity, the critical ratio, the audiogram, and speech threshold. J. Acoust. Soc. Am., 72, 1788-1803. there is a large correlation matrix (p1794) including the ERB parameter, p, and the critical ratio, Ps(0). It also includes the efficiency constant K. The table shows that p correlates highly with a number of parameters, but not with Ps(0) and K. the high correlations associated with p arise from the low standard deviation of the measure, which in turn arises from the sensitivity of the notched-noise method. The low correlations with Ps(0) and K arise because frequency selectivity and efficiency are confounded in the Ps(0) measure, as we explain in the introduction to the paper. Conclusion: The Critical Ratio is a particularly bad measure of auditory filter bandwidth! With regard to Fletcher's band narrowing experiment, its disadvantages are discussed at length in Patterson, R.D. and Henning, G.B. (1977). Stimulus variability and auditory filter shape. J. Acoust. Soc. Am., 62, 649-664. Briefly, although the power-spectrum math for the band narrowing method and the notched-noise method are very similar, it is important to understand that the band narrowing version has two very serious weaknesses: 1) it confounds stimulus variability with filter shape, so you need full detection theory stats for the band narrowing derivation and no one ever uses them (Section III.C, Fig. 4). 2) the band narrowing version of the experiment is particularly insensitive (Section IV), whereas the notched-noise method is highly sensitive. Cheers Roy __________________________________________________ Roy D. Patterson Centre for the Neural Basis of Hearing Physiology Department, University of Cambridge Downing Street, Cambridge, CB2 3EG phone 44 (1223) 333819 office phone 44 (1223) 333837 lab fax 44 (1223) 333840 department email roy.patterson(at) Email to AUDITORY should now be sent to AUDITORY(at) LISTSERV commands should be sent to listserv(at) Information is available on the WEB at

This message came from the mail archive
maintained by:
DAn Ellis <>
Electrical Engineering Dept., Columbia University