Re: missing fundamental references (Peter Cariani )


Subject: Re: missing fundamental references
From:    Peter Cariani  <peter(at)epl.meei.harvard.edu>
Date:    Fri, 8 Dec 2000 15:51:10 -0500

It is an interesting question. That they see periodicities in the FFR's that are consistent with the pitches of inharmonic complex tones (first effect of pitch shift, de Boer's rule) I think is strong evidence for a population-based temporal code for pitch at the level of the midbrain. One of the main doubts that I have about time-to place conversions in the midbrain (via units with bandpass MTF's is that the behavior of MTF-tuned responses would not be expected to retain the autocorrelation-like character of these pitch shifts. The other is that the MTF tunings appear to broaden at higher SPL's, which is not at all consistent with how the pitch percept behaves. A priori I think it is highly unlikely that the auditory system "pays attention to" one subpopulation of neurons at low levels and another at high levels, but it is difficult to rule out such hypotheses. So yes, I think Greenberg's FFR observations are entirely consistent with a temporal mechanism for pitch that operates at least as far up as the IC. Why these periodicities are not seen with latencies for other brainstem s= tructures, I don't know (certainly one would see them in averaged cochlear potentials and certainly all the major cell populations in the cochlear nucleus are driven, some better than others, by the fine time structure of the stimulus. I have seen the interval correlates of the pitch shifts of inharmonic AM tones in primarylike, cho= pper, and pauser responders, although these patterns are weaker for pausers. Maybe it has to do with their electrode placement and/or cancellation technique -- I would have to go back and look more closely at the papers. As for temporal representations for pitch above the IC, we should remember that the FFR is giving us information about the synchronized component of the population response and that timing information can be present in single = units with a range of latencies such that it would not be seen in a scalp averaged gross potential (for example, my guess is that the upper limit of scalp cortical AEP's that is seen is probably between 100-200 Hz, but Steinschneider's group has seen periodicities up to 300 Hz in current source density analysis of local fi= eld potentials, and there are occasional reports of phase locking in cortical neurons to still higher periodicities (Semple's group at Neurosciences reported in their poster a cortical unit that phase-locked to 1 kHz). So, I believe that this question is still unresolved (will let you know if we find anything). It could be the case that (not in any order of plausibility) 1) the temporal information is converted into a form that we do not yet u= nderstand (e.g. something like Hopfield's relative latency code) 2) the temporal information is in fact transformed into MTF-tuned activation profiles in a manner we don't fully understand (such that the representati= on takes care of de Boer's rule and doesn't degrade at high levels) 3) the timing information is put onto more and more neurons in a sparser and sparser way and multiplexed with other information (e.g. re: location) as= one proceeds up the pathway (a million neurons each with one interval at the f= undamental) 4) that the computations that make use of timing information involve top-down/bottom up interactions at the level of IC and/or thalamus that we don't understand (some sort of time-time comparison that requires reverberating circuits) 5) it's all a covert rate-place code, and we have been looking at the wrong neurons Take your pick. Peter Martin Braun wrote: > > Dear Peter, > > encouraging to see that we both (warmly) recommended Greenberg, S., et = al. > (1987) to Ani. > > I also added three papers by Galbraith & Co (see below), who replicated= and > extended Greenberg's results. I didn't send them to the list but to him > personally (11-29 that was), because Ani had announced he would send hi= s > harvest to the list anyway. > > But the fact that you mailed yours via the list gives me the chance to = a > possibly interesting question. > > I am a bit surprised that you put a special emphasis on these findings.= The > FFR to f0 is only seen with a latency corresponding to wave V of the AB= R, > never with a latency corresponding to the earlier waves. Wave V has its > origin in the upper brainstem, most likely the IC. > > Question: > Are these relations compatible with your hypotheses on extraction of > f0-pitch ? > > Martin > > Galbraith, G.C., 1994. Two-channel brain-stem frequency-following respo= nses > to pure tone and missing fundamental stimuli. Electroenceph. clin. > Neurophysiol. 92, 321-330. > > Galbraith, G.C., Doan, B.Q., 1995. Brainstem frequency-following and > behavioral responses during selective attention to pure tone and missin= g > fundamental stimuli. Int. J. Psychophysiol. 19, 203-214. > > Galbraith, G.C., Bhuta, S.M., Choate, A.K., Kitahara, J.M., Mullen, T.A= ., > 1998. Brain stem frequency-following responses to dichotic vowels durin= g > attention. Neuroreport 9, 1889-1893. > > Martin Braun > Neuroscience of Music > Gansbyn 14 > S-671 95 Kl=E4ssbol > Sweden > nombraun(at)post.netlink.se


This message came from the mail archive
http://www.auditory.org/postings/2000/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University