Re: Miller: Magical Number 7 +/- 2

From: Harnad, Stevan (harnad@cogsci.soton.ac.uk)
Date: Mon Nov 20 1995 - 20:58:33 GMT


> From: "Parker, Chris" <C.C.Parker@soton.ac.uk>
> Date: Thu, 16 Nov 1995 08:53:58 +0000 (GMT)
>
> GM> If we are very ignorant, then when we make an observation, it
> GM> gives us a lot of information. [and vice versa]
>
cp> Isolated tribespersons in the Amazon region sees a tv will
cp> they get more information than me? If we only thought of affordances,
cp> they would see something to sit on? I would see a more affordances.

Not quite what we have in mind. More like: If there are numbers
from 1 to 100, and I know one of them is right, and my lunch depends on
it, but I don't know which, then if you tell me it's over 50, you've
doubled my chances of getting lunch. But if I know that it's either 48,
58, 68, or 78, then you've only increased my chances by 25%

Information is whatever resolves uncertainty among alternatives (on
which something that matters depends, presumably). The more alternatives
information eliminated, the more informative it is. The less informed
you are, the easier it is to reduce the alternatives. As you know more,
only more and more specific information will help you.

> GM> This aymptotic value we take to be the channel capacity..
>
cp> I thought if you gave someone too much information they saturate
cp> and the capacity suddenly drops or is that multidimensional info??

Persumably if you exceed channel capacity, it will have bad effects,
whether you exceed it in one channel or many. But this is not the kind
of thing Miller is discussing here (fatigue, overload, etc.).

> GM> Everday experiences teaches us that we can identify accurately
> GM> any one of several hundred faces..
>
cp> Yes but not just any faces. Don't all eskimos look the same?

Unless you're an eskimo; and that's presumably not for genetic, but for
experiential reasons.



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:56 GMT