Re: Informational Masking (Peter Lennox )


Subject: Re: Informational Masking
From:    Peter Lennox  <peter(at)LENNOX01.FREESERVE.CO.UK>
Date:    Fri, 2 Mar 2001 15:18:12 -0000

Right. I said I was probably mixing several phenomena in together. Unsurprisingly, I underestimated how many! Taking speech as an example, which is a very peculiar 'special case' kind of listening (happy to expand/expound if necessary), I can think of many examples where I'm able to repeat, after a period, a very reasonable representation of the sound of the sentence which had been spoken to me, even though so doing produces a sentence which is complete nonsense. On re-hearing the original, one can hear the similarities, and in fact if my long-suffering wife (who is admittedly annoyed by such antics!) doesn't exactly repeat the sentence, but paraphrases in some way, I find it fairly easy to arrive at the original, using 'meaning' to enrich the 'stored approximation' of the *sound* of the sentence. Note that i find I can't do this if I understood the original at the time of reception. Note also that, in the 'gobbledegook' reconstruction of the sentence, it's usually the consonants that I've got wrong (.....I think). The vowel sounds seem fairly right, and the rhythm overall seems right, even though the division into words has often gone awry. Now this may be a slightly different thing to the experience of seeking out something which has made a noise, then stopped. Here, (and I'm talking about 'real world' examples rather than psychophysics-lab examples) I find that a sort of 'mental reiteration' of the sound may yield additional spatial information (I don't particularly mean inter-aural difference information, as this works even as one moves about), which may be used to 'home in' on the likely position-and-source. But both these examples may well relate to a finite storage capacity (even if that capacity is able to re-iterate to itself in the absence of further input), because new input seems to obliterate the phenomenon. The point is that this sort of memory store may well be useful in holding 'undigested' chunks (sorry about the unappetising metaphor!) pending apprehension of further, useful information which may well be of utility in processing it. And to that extent, it does seem to relate to speech processing (I think Jackendoff wrote something about this). Now, this may be somewhat different to the way a sentence heard a few moments ago suddenly makes sense in the light of a new sentence; here, we are talking about a sentence which has been processed, but doesn't quite fit into an unerstood scheme of things until new facts are revealed. But, for me, the difficulty I've always had with Gibson's approach comes up because this type of 'unprocessed-sensation storage' doesn't seem as necessary in vision as in hearing. In insisting that all information is external to the percipient, and allthough he stressed the notion of an ambulant percipient, I never really got the sense of the role of various types of memory in perception. Sorry if I've oversimplified this a bit. But the idea that we perceive the 'now' only in the light of already processed contexts from the past, a sort of unidirectional flow of information-with-time, simply will not do (to quote Gibson!). The alternative is to say that memory (in various forms of sophistication of processing) is a (the?) central part of 'real-world' perception, and in which case doesn't this imply that some sort of internal scheme of things is required, to map things onto? That this scheme-of-things might be highly metaphorical seems fine (in other words we remember the 'meaning' of things and events rather than the actuality of their physical characteristics), but doesn't this absolutely seem to require a buffer storage for 'un-metaphoricised' sense-data? regards, ppl ----- Original Message ----- From: "Christian Kaernbach" <chris(at)PSYCHOLOGIE.UNI-LEIPZIG.DE> To: <AUDITORY(at)LISTS.MCGILL.CA> Sent: 02 March 2001 07:16 Subject: Re: Informational Masking > Peter Lennox wrote: > > I've also noticed what a complex phenomenon this information-masking > > is, compared to signal-masking. If my wife speaks to me when I'm > > trying to fix this blessed computer, it can take me 15 minutes to > > respond! (if at all). I'm not just being facetious here; I seem to > > observe that it is possible to 'store' apparently (fairly) > > unprocessed 'sound chunks', and 're-listen' to them a short time > > later, and *only then* understand them. > > I guess you are storing not unprocessed sound chunks but auditorily > well-processed items on a priority-based schedule. Proof: tell your wife > to intersperse once a week sentences like "I'm going to switch off the > main fuse, so all electric devices will go down" in the same voice as > she would say "I'm back in ten minutes but I just have to get this silk > scarf." She should say both types of sentences in self-chosen wordings, > and let action follow. I bet you will process the first type of sentence > at quite a different speed. > > ;-) Christian Kaernbach >


This message came from the mail archive
http://www.auditory.org/postings/2001/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University