Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!usc!orion.oac.uci.edu!unogate!xing!rick
From: rick@xing (Richard Ottolini)
Newsgroups: comp.ai.philosophy
Subject: Re: Direct awareness
Message-ID: <1991Jun21.152443.27543@unocal.com>
Date: 21 Jun 91 15:24:43 GMT
References: <25348@samsung.samsung.com> <8570@awdprime.UUCP> <603@ckgp.UUCP>
Sender: news@unocal.com (USENET News)
Organization: Unocal Corporation
Lines: 31

Awareness is hard to define.  Look at the thrashing in this newsgroup.

I suggest considering the more measurable quality of "attention".
At the sensation and perception level, the parameters of attention have
been measured for time & space resolution, persistence, multipleness, etc.
Each sense and level of mental activity has a little different range of
attention.  For example, hearing has the highest time resolution on the
order of milliseconds between stimuli.  The persistence or "nowness" of
a sound sensation is a good fraction of a second.  We can focus on that
single conversation at a cocktail party or listen to several, but not a large
number of, canversations at a time.  Vision, touch, smell, muscular coordination,
etc. have different parameters.

Of interest is "meta-attention", that of non-perceptual mental acitivities.
Like for example, memory studies have determined that we can keep about seven
abstractions in short term memory at time.  There is the sense of "I-ness"
when we direct attention to one main stream of tasks.  Then there is dissociation
where one part of the brain stands back and watches what is happening elsewhere.
Then you can see the cacophony of Marvin's mental agents at work.  Meditation
practices and sometimes mind-altering drugs seek to junp between sharp attentive
focus and disassociation to strengthen control over the mind.

As to whether human-mental models of attention is a good model for computer
attention or awareness, or vice-versa, is another question.  The "blackboard"
types of expert systems have some focus of attention, that is neither single
threaded nor massively diverse. Marvin's society of agents is a relevant model
too.  Have there been any interesting recent computer experiments along this
line?

-Rick Ottolini
(Hiding in industry after 18 years of lurking around Stanford and MIT.)