home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!elroy.jpl.nasa.gov!sdd.hp.com!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!destroyer!gumby!yale!yale.edu!jvnc.net!darwin.sura.net!paladin.american.edu!auvm!BEN.DCIEM.DND.CA!MMT
- Message-ID: <9208181555.AA20169@chroma.dciem.dnd.ca>
- Newsgroups: bit.listserv.csg-l
- Date: Tue, 18 Aug 1992 11:55:42 EDT
- Sender: "Control Systems Group Network (CSGnet)" <CSG-L@UIUCVMD.BITNET>
- From: mmt@BEN.DCIEM.DND.CA
- Subject: Re: separating action and perception
- Lines: 28
-
- [Martin Taylor 920818 11:50]
- (Penny Sibun 920817)
-
- > there's just tons of literature on
- >perception that suggest that perc. is a *lot* of work; in fact it
- >involves action--perhaps to the degree that p. and a. are inseparable.
- >(``active vision'' is a current buzzphrase.)
-
- This was cast as a criticism of PCT, but in fact it seems to be a reasonable
- description of how PCT says perception works. An ECS that is trying to
- achieve a percept can go only by its error signal. It doesn't know what
- it is doing when it amplifies that error signal into an output that contributes
- to the reference signals of unknown lower-level ECSs. All it knows is that
- the connections are such that producing this output normally leads to a
- reduction in its error (i.e. a percept more like the percept it is trying to
- achieve). From here, I can continue your quote to describe the sort of thing
- that might occur: " if i'm driving into the
- sun, i'll have to squint to see anything. if it's sleeting, i'll have
- to peer around the bouncing balls of ice and the flailing wipers. at
- *any* time, my eyes are having to track the road, since the car is
- constantly bouncing me around. "
-
- A good description of the actions of a PCT hierarchy.
-
- Martin
-
- "Statistics are easy to understand. Unfortunately they are even easier to
- misunderstand." (Me)
-