From: Patrick Wilken (patrickw@cs.monash.edu.au)
Date: Fri Sep 03 1999 - 21:49:06 MDT
>This is interesting. How would it explain the large variance across people in
>performance on memory tasks - lists of random numbers, for instance. Is buffer
>size variable, or do some people pay more attention to each number. (whatever
>that would mean.)
My interpretation is that for visual tasks (I have no idea what happens
when you learn a list of words) that the more items you attend to the worse
the signal for each *individual* item. It just turns out that the trade-off
between holding more items and increasing noise balances out and the
effective capacity remains constant.
This limit does appear to be attentional and not memory related. So long as
local motion cues are masked this capacity limit is present. So changes can
be made to an item in a scene without someone noticing so long as there
were visual transients elsewhere attracting attention. Likewise visual
tracking tasks strongly suggest we can only track 4 items.
Ron Rensink (one of the key researchers in this area) told me the story of
the Great Mephisto (?) from the 19th C who was able to produce a goat out
of thin air. Mephisto would have a helper who came on before him and worked
the audience up saying Mephisto was delayed, some emergency had come up
etc. Suddenly the great magician appears off the side waving a huge red
banner and shouting out to the audience as he strides out to the middle of
the stage. As he reaches the center of the stage he casts the banner aside
and suddenly in its place seemingly out of thin air is a goat. Alas our
little brains just aren't able to process everything at once and so never
noticed that the goat was under his arm the whole time.
ciao, patrick
----------------------------------------------------------------------------
Patrick Wilken
Editor: PSYCHE: An International Journal of Research on Consciousness
Board Member: The Association for the Scientific Study of Consciousness
http://psyche.cs.monash.edu.au/ http://www.assc.caltech.edu/
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:05:01 MST