Perceptual distance between consonant tokens is highly predictable from internal-model responses.

Evidence

Figure 3 shows scatter plots of the perceptual distance versus the modeled distances obtained using P (panel a) and Pmod (panel b). It can be observed that the perceptual distances were remarkably well-predicted using P as well as Pmod , with a Pearson’s r of 0.96 and 0.98, respectively. — Dijk, Pim van, Physiology, Psychoacoustics and…, 5609