From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!decwrl!mcnc!aurs01!throop Mon May 25 14:06:21 EDT 1992
Article 5756 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!decwrl!mcnc!aurs01!throop
>From: throop@aurs01.UUCP (Wayne Throop)
Newsgroups: comp.ai.philosophy
Subject: Re: Grounding: Real vs. Virtual
Message-ID: <60713@aurs01.UUCP>
Date: 19 May 92 21:22:00 GMT
References: <60703@aurs01.UUCP> <78417@netnews.upenn.edu>
Sender: news@aurs01.UUCP
Lines: 52

> peter@sysnext.library.upenn.edu (Peter C. Gorman)
>> throop@aurs01.UUCP (Wayne Throop)
>> Consider a robot interacting and demonstrating competence against a
>> virtual world, and another robot interacting and demonstrating
>> competence against the real world.  The two robots will (by hypothesis)
>> end up in identical physical states, yet one "has semantics" and the
>> other doesn't.
> It seems to me that a 'grounded' robot is necessarily more than the
> virtual-world robot with some transducers strapped on.

No, that's backwards.  The "virtual", "ungrounded" robot is
a real-world robot with a virtual-reality headset over its optics,
virtual-reality servo-driven pressurepads over its tactile sensors,
and so on.  Both robots are made in the same factory, in the same
batch of robots, made from the exact same materials.  But one is
placed in the "real world", and the other in a digitally simulated
virtual reality.  The simulation is at a granularity beyond the 
ability of the robot's sensors to detect.

As I understand Harnad's notion of semantic grounding, one of these
robots would (potentially) have semantics, genuine understanding,
the ability to think, and so on, while the other robot, since it
wasn't grounded, would not.

Yet if the two were ever mixed up, neither their mothers nor anybody
else could ever tell them apart.  They will both say they like playing
with blocks.  They will both say their favorite block is the nice
heavy red one that makes the pleasing thud when they drop it.  And
so on and on.  But one is just uttering screeches and squalks,
while the other is having an intelligent conversation.

Now I'll agree that this "grounding" distinction is quite
self-consistent.  It just seems.... peculiar to take such care to label
one as "squalking" and the other "talking" based on history alone.

I could be misunderstanding something, of course.

But let's also throw in another scenario.  Two identical birds, one placed
in a simulator at hatching, the other raised normally in the real world.
The simulator-bound bird has a virtual reality headset, pressure pads on
its wings, etc, etc.  It is placed in a virtual reality and grows up, and
learns to pseudo-fly.  It real-world counterpart learns to really-fly.

Now (granting the accuracy of the simulation), we remove the
virtual reality suit from the simulator-bound bird.

The question now is: "can the bird fly".  NOT, mind you, HAS it
ever flown, but CAN it fly?  Any ability it MIGHT have is ungrounded
in reality, yet I'd certainly take long odds that such a bird
can fly (again, assuming the accuracy of the simulator).

Wayne Throop       ...!mcnc!aurgate!throop


