Re: maximum 'tatum' speed perception

[Erick Gallun <gallun@xxxxxxxxxxxxxxxxxxxxx>]

At 12:39 PM 4/8/02, Jeff Bilmes wrote:

> Actually, I would surmise that 8-10Hz is a fairly large underestimate
> of the maximum rate at which tatum perception can occur. Certainly Art
> Tatum, for whom the term tatum was named, had a tatum rate that far
> exceeded 10Hz (it was probably around 13-16Hz, or possibly higher).
This made me think of the distinction raised in the Pastore, Harris and
Kaplan (1982) paper "Temporal order identification: some parameter
dependencies" in JASA, 71(2), pp. 430-436.  In their experiments, subjects
were asked to make judgements about tone-pairs with onset
asynchronies.  Either the high tone or the low tone came on first. In a
two-alternative forced-choice task, onset asynchrony thresholds with brief
(10 ms overlap) signals, were measured at roughly 5 ms.  When the same
subjects were then asked to "label" the same stimuli as high-low or
low-high, threshold performance was only achieved at 9-10 ms.  As the
overlap duration was increased (and consequently the duration of the entire
stimulus) up to 300 ms of overlap, a doubling of thresholds was
reported.  This suggests that while 8-10 Hz (roughly 10 ms) is a reasonable
estimate for an identification task with standard listeners, the human ear
could in fact perform the task at rates of up to 20 Hz.

Erick Gallun
Hafter Auditory Lab
UC Berkeley