X-Message-Number: 8230
Subject: testing hypotheses
Date: Wed, 21 May 1997 14:00:52 -0400
From: "Perry E. Metzger" <>

> From: Will Dye <>
> 
> > Science depends on constructing TESTABLE hypotheses, you see. If you
> > can't test it, its religion, not science. I can hypothesize that there
> > is an invisible undetectable six foot tall bunny following Robert
> > Ettinger at all times, and no one could prove me wrong -- but since no
> > one can conduct a test of any sort to prove me wrong, we can ignore
> > the question as religious, not scientific.
> 
> To my reading, these statements imply that science consists only of 
> statements that are falsifiable at the time they are made,

No, it is fine to propose a hypothesis that cannot be immediately
tested -- I can happily propose that the planet Pluto has a core made
out of a certain substance, a hypothesis we can't test for many
decades to come. However, this is different from a hypothesis that is
not falsafiable. I can conceive of the experiment to falsify my Pluto
hypothesis even if I can't actually conduct it -- at least in theory,
I can fly to Pluto and check what the core is made out of. Indeed,
even if I can't conceive of an experiment immediately, any hypothesis
that bifurcates the world into "the world that is consistant with the
hypothesis" and "the world that isn't" eventually lends itself to
testing.

The distinction we make is with a hypothesis that can NEVER be tested
-- such a hypothesis is not science.

For example, take the notion of "there is an *undetectable* bunny
following you". Well, obviously, that can't *ever* be tested, can it?
I mean, the bunny is *undetectable*. It doesn't move objects, it has
no weight, etc. The bunny doesn't change the course of observable
events in any way. The existance or absense of the bunny creates no
distinctions in worldly phenomena, so the hypothesis has no
conceivable test. This is thus "Religion".

Similarly, we are told by fools like Searle that even if a computer
(or, more likely, a robot) acts just like a human, converses, laughs,
chatters away about politics, it can't be "conscious" -- that
consciousness is not detectable through external behavior of the
"thinking system".

Okay, then. If consciousness HAS NO EXTERNAL CONSEQUENCES AT ALL -- if
it creates no distinction in the observable behavior of a system --
then what is the test we can *EVER* conduct to show it is there or not
there?

If consciousness is a purely unobservable phenomenon how can we ever
conceive of a way to test the "consciousness hypothesis"?

> I'm not against the idea that science needs a falsifiable question.  
> Popper came up with a great heuristic, which, like Occam's Razor, 
> helps us quickly identify the paths which will more quickly lead to 
> satisfying results.  But I argue that it's only a heuristic, not a 
> reliable definition that draws a clear line between "science" and 
> "religion".  I do not believe that such a definition exists.  

Science is the only mechanism we have for reasoning about
phenomenology that works. The scientific method requires *tests* of
any hypothesis.  If you have a "consciousness hypothesis" that has no
phenomenological consequences -- if it cannot be observed by any means
-- how can you possibly conduct scientific reasoning about it?

Understand -- the undetectable bunny may be following Ettinger, but if
I cannot ever know, why should I even consider the question?

> From:  (Thomas Donaldson)
> 
> As an example, consider cosmology. There seem to be several problems:
> apparently lots of missing mass, no sign that protons decay so far, and 
> others. Does that mean that cosmology is not a science because we cannot
> decide NOW between several alternatives?

No. The alternatives have phenomenological consequences that we can
conceivably eventually test. Maybe it takes millenia, but the notions
are ones that have consequences.

The "consciousness hypothesis" as advanced by Searle, or Ettinger, has
no phenomenological consequences almost by definition. It *can never
be tested*. It is the undetectable bunny.

> Furthermore, although I may have misunderstood you, I understood that you
> were proposing that Ettinger lived in a simulated world, and Ettinger
> was arguing that he did not. If it's YOUR hypothesis that he lives in 
> a simulated world, can you provide means to disprove it?

You misunderstood.

Ettinger said "no one could live out their life in a simulated world".

I proposed a test. My remark was a way of saing "Mr. Ettinger, provide
us with the experiment by which someone inside a simulated world can
determine if they are in the 'real universe' or in a
simulation". Saying "I have attached you to a simulation" is just
shorthand for that.

> Here is an idea of how we might prove or disprove some ideas about 
> consciousness. We assume, not as a matter of religion but because we
> know that we are conscious, that normal people are conscious.

You have already started with a stream of logical fallacies.

1) You are saying that people have a property that you have not
   defined. No one in this argument has been able to provide any sort
   of reasonable definition of consciousness.
2) You are asking me to assume something for which I have no evidence
   at all. You say "well, I am conscious, so I assume others
   are". Well, I know that I am looking at the clock in my
   office. Does that mean you are? Of course not. Is this a trivial
   comment? No. For all I know, the "consciousness phenomenon" might
   be because I was hit upside the head properly when I was two years
   old, and you and the other zombies walking around just *say* you
   are conscious. The "assumption" that you must be conscious since I
   am conscious is like the "assumption" that you must have a glowing
   happy viewpoint about the world because I do, or that you must like
   chocolate cake because I do, or any similar notion.
3) You are starting FROM THE BEGINNING with the assumption that you
   are trying to prove. Not good, eh?

> So how does this work? Well, we take person X, and see what parts of
> his brain are active when he is appears conscious to us, and what
> parts are not.

How do you define "consciousness" well enough to define the
"appearance of consciousness"?

> Clearly if X is UNconscious we cannot show him that these areas are NOT
> active. However we could, for instance, make a movie of these areas,
> and show it to him later. If X agrees that there is a very close       
> match between activity in these brain areas and HIS OWN FEELING of 
> consciousness, then we've shown TO X that these areas relate. Show 
> the same to Y,Z,S,T etc and see if they too agree. We know that our
> brain anatomies at that high level coincide closely, from which we
> can conclude that we ALL are conscious when these areas are active.

How, though, does this permit us to answer the question "is this
computer conscious"?

This definition ASSUMES an answer. It assumes that consciousness is
the product of particular brain areas lighting up. Does this mean that
an alien that landed that lacked a brain anatomy anything like ours
must not experience consciousness? Or that a computer cannot
experience consciousness?

Perry

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=8230