Subject: From: Diana Deutsch <ddeutsch(at)UCSD.EDU> Date: Tue, 29 Apr 1997 12:30:01 -0800I have some comments on Dick Warren's comment on the 'high-low' illusion. First, verbal tranformation effects in general are indeed not new, as I write in the booklet accompanying my CD: 'Verbal transformations have also been produced in different ways, for example by the psychologist Richard Warren' (p5). Second, Warren did not 'discover' verbal transformation effects. In a paper published by Warren in the Journal of the Audio Engineering Society, 1983, vol 31, pp 623, he describes such effects as having existed since the early part of the century, quoting, among others, Titchener, 1915, and Skinner, 1936. Warren's description of Skinner's work reads (p626) '..a phonographic record of a series of faint and indistinct vowels, such as ''ee'', ''oo'', ''ah'', ''uh'', repeated over and over. After several repetitions of these meaningless patterns, meaning seemed to ''summate'', and listeners were convinced that they heard words and phrases related to some personal aspect of their lives'..... Third, Warren implies that he discovered or 'reported' the 'high-low illusion', but this is categorogally untrue. More generally, Warren writes that verbal transformations occur best when the words are 'clearly enunciated' (as in his recent note to this list) - which is the last way one would describe the 'high-low' pattern. Warren's closest work is one in which a word such as 'vrine' is repeated dichotocally, so that the word is delivered to the two ears asynchronously. Warren concluded that the right and left ears behave functionally equivalently in this situation. To quote from the same paper: 'there were no such differences, and that changes were functionally equivalent on the two sides. (p628)' There is a description of this particular work in this paper, together with sound demonstrations accompanying the paper (on soft LP), in the JAES issue (which I guest-edited). Fourth, the factors producing verbal transformations in the 'high-low' pattern are different from those reported previously by others, in that they depend on the perceptual fusion of sounds coming from the left and the right, and in an asymmetrical way that is analogous to the octave illusion. Indeed, as is evident in my booklet, I generated the 'high-low' illusion in an attempt to obtain a version of the octave illusion using verbal stimuli. To this end, I chose the words 'high' and low', not only for their meaning, but also because their vowel spectra are promising ones for inducing an effect such as the octave illusion. For the same reason, I also shaped the time-varying amplitude envelopes for the words 'high' and 'low', so that they would be reasonably similar to each other. And indeed, when listening to this pattern, the sounds that are heard as coming to the right ear do indeed differ from those that are heard as coming to the left. This effect is not as strong as the octave illusion, though its strength varies across subjects. It is certainly not the same as the effects reported by Warren, since for the 'high-low' pattern many people report clear differences in the sounds that are perceived as coming from the right earphone or loudspeaker as opposed to the left one, and Warren has written that the sound patterns he has worked with are not associated with a left-right asymmetries. So in sum, the 'high-low' demo on the CD is particularly conducive to obtaining verbal transformations (effects that have been around for the best part of a century), though I did not originally generate it for this purpose. I believe that the alternating spectra employed here, and generated in this particular fashion, are particularly conducive to perceptual fusion, (just as this type of alternation gives rise to the perceptual fusion that occurs with the octave illusion), and that this fusion induces ambiguities that in turn cause listeners to perceive transformations. And in contrast to the effects studied by Warren, the fusion of the sounds coming from the left and right sides of space here gives rise to the perception of speech sounds which are quite different from those that are heard as coming from either channel when presented alone. Finally,since these sound demonstrations are readily available, I invite readers to listen to Warren's demos, published in the JAES issue referenced above, and to compare these with the 'high-low' demo on my CD. I hope this clears up any misconceptions that may have been generated by Warren's note. Diana Deutsch -------------------------------- Diana Deutsch Department of Psychology University of California, San Diego La Jolla, CA 92093, USA tel: 619-453-1558 fax: 619-453-4763 e-mail: ddeutsch(at)ucsd.edu