Re: Effect of duration on pitch perception

[Christian Kaernbach <auditory@xxxxxxxxxxxx>]

Dear Martin,

The short duration to recognize the pitch of a complex as compared for 
that of a pure tone is indeed a most interesting finding. I would, 
however, not conclude, that this rules out a place theory of pitch. It 
is a step in this direction, but the argument is not yet complete.

Firstly, it is not sufficient to compare the minimum time for pitch 
perception for a pure tone at 100 Hz with a complex tone featuring 100, 
200, 300, 400, 500, and 600 Hz. Imagine that a pure tone of 500 Hz is 
perceived with stimuli as short as 15 ms, and a pure tone of 600 Hz 
within, say, 12 ms. A place theory would then conclude that after 15 ms 
there is enough information available to go for a pitch of 100 Hz. So 
the next step would have to be to repeat the experiment with pure tones 
at all frequencies of the components of the complex tone.

Secondly, any leaky integrator theory of information processing would 
predict that a place mechanism getting more info (at six locations) 
would be faster than if it is fed with less info (at one location). I am 
not sure how to deal with this argument, but I would guess that it 
should be possible to compare pitch perception performance with the 
output of optimal leaky integrator models of place perception and test 
whether this is a possible explanation. Input data would probably be: 
duration/performance curves (not only "thresholds") for pure tones at 
all six locations of the components of a complex tone, plus the same 
diagram for the complex tone.

I think that temporal coding plays an enormous role in hearing, and that 
this role is not yet appreciated by everybody. However, one should not 
be too fast in ruling out place theories. One point for place theories 
that I can hardly think of how to deal with is the perception of pitch 
of pure tones with very high frequencies, say above 4000 Hz. These 
pitches are admittedly imprecise, but they are perceived as pitches. 
Another such point is the pitch of Zwicker's aftertones. My guess is 
that (as so very often) the truth is at neither of the extremes (but not 
precisely in the middle, wherever that would be), and that full pitch 
perception is established if both temporal and place cues are available.

Best regards,
Christian

> Patterson et al. (1983) reported that the minimum stimulus duration for the
> perception of pitch of a complex tone with f0 =3D 100 Hz was < 20 ms. For the
> perception of pitch of a pure-tone with f =3D100 Hz, however, the minimum
> stimulus duration was > 80 ms.
...
> The most obvious conclusion that has to be drawn from the results is that
> they ruled out the place-pattern hypothesis of pitch. If there were a
> "harmonic grid" (Erik) in the brain that needed a stimulus duration of > 80
> ms to excite the 100 Hz position, the same grid could NOT be excited at its
> positions of 200, 300, 400, 500, and 600 Hz in a stimulus duration of < 20
> ms.