Re: testability and hypotheses (long) [and getting longer]

[rowe@lepomis.psych.upenn.edu (Mickey Rowe)]

Wow, what a can of worms I seem to have opened...

Ken Carpenter (Crpntr@ix.netcom.com) says while addressing my
concerns:

> But as I got older, did more research on my own, I realized that
> there is a lot in paleontology that can not be tested without a time
> machine.

As an outsider to paleontology, I probably have a very different
perspective on such issues.  It seems to me that in your field it is
extremely easy to generate ideas that cannot currently (or perhaps
ever) be tested, and even those ideas that can be tested probably
can't be tested as thoroughly as, for example, the hypothesis that
electrostatic forces vary in proportion to 1/r^2.  However, it would
concern *me* if it didn't concern *you* while you were reviewing
papers.

Somewhat tying into the thread on dinosaur toys and their effects on
children, I think the thing that concerned me the most about "Jurassic
Park" (the movie) was the descriptions of Tyrannosaurus' hunting
strategy.  In the book, Alan Grant formulated the hypothesis that
Tyrannosaurus couldn't see stationary objects *AFTER* it didn't eat
him as he stood motionless against its nose.  In the movie, Alan Grant
claimed that T. rex vision was "based on motion" while he was still at
the dig site.  Last fall, a 6 year old stopped by the exhibit at which
I was volunteering at ANSP, and his proud parent nudged him into
telling me what he "knows" about Tyrannosaurs, that their vision was
"based on movement".  I wanted to prod him a little bit by asking how
we could know that, but I think I chickened out (not wanting to
embarrass the parent or ask too much of the kid -- of course that's
basically saying that I let the bad old pop-culture creators win that
battle without a fight).  In case it's not obvious, I'm saying I think
that people really ought to be made more aware of the nature of
science and the importance of methodically testing ideas.  I'm
concerned to the extent that we might not be in sync on this issue.

> Indeed, [Mike Williams] discussed the topic at some length in his
> 1994 paper "Catastrophic versus noncatastrophic extinction of the
> dinosaurs: testing, falsification, and the burden of proof.  Journal
> of Paleontology 68:183-190."

I intend to take a look at that paper as soon as I get back over to
the library.  In the mean time, let me mention that I covered my own
posterior quarter by using the word "testability" (as have others)
rather than "falsifiability".  I'm somewhat sensitive to issues of
falsifiability because it frequently comes up in talk.origins where
creationists are told that creationism isn't science because it isn't
falsifiable.  Of course, I don't want to discuss creationism here (if
anybody else does, please write to me personally) but I think that the
position of many creationists must be kept in mind: it's all in the
past, and it can't be repeated, so mainstream paleontology, geology,
etc. are no better than creationism on that score.

What I would argue is that the difference is (or should be) that
scientific ideas are those which are formulated based upon
observation, and can be rejected based upon observation.  In short,
the ideas are testable by comparing their consequences to physical
evidence.  In practical situations, scient*ists* have their own
strengths, and some people are better at generating theories while
others are better at collecting data.  At the former extreme I'd use
Robert Bakker as a canonical example, and he sometimes irks me (in
writing--I've never met the guy) because he seems more attached to his
ideas than the data.  That's an extreme I think we should always work
hard to avoid regardless of what particular field we're exploring
(mind you I've been heavily influenced by the writings attributed to
Richard Feynman).

>    To focus on one area as an example, I would argue that almost all
> that we infer about dinosaur behavior is not testable. [...]  We
> make inferences, draw conclusions from living analogies [...] But
> again, we are drawing inferences that may not be valid.  That is why
> paleobehavioral papers are full of terms like "might have,"
> "probable," etc. i.e. conditional terms (cover your ass phrases).

I agree 100%.  While working at the exhibit mentioned above, I tried
hard to get visitors to appreciate that perspective as I answered
questions about the lifestyles of the big and Cretaceous (sorry, best
I could do on short notice :-).  However, if I personally were
reviewing a paper, I'd be hesitant to recommend acceptance if the
author(s) couldn't even suggest a method for testing the main premise
(not having seen the push-me-pull-you-saurus paper, I'm not saying
you've done something I wouldn't have; I'm just making sure you
understand the depth of my concern over your previous remark).

In a forum like this, or when you're sitting around a campfire at your
dig site, or when you're mulling things over in a journal club,
untestable scenarios are fine, and perhaps when electronic publishing
takes over from printing presses the "literature" may be more open to
such things.  In a current peer-reviewed paper, though...  To go along
with your herding dinosaur comments, I think it's perfectly fine to
suggest that certain species moved in herds because of, for example,
trackway evidence and apparent death assemblages.  The idea is still
testable because it may be possible to show that the tracks or
carcasses were not actually laid down at the same time.  Or (Peter
Dodson's statistics notwithstanding) we could suddenly find hundreds
of dinosaur remains associated only with contemporaneous individuals
of different species.  Such findings would only be suggestive, of
course, and would make you want to buy stock in companies selling ass
coverers, but they would still provide a test of the hypothesis that
certain dinosaurs travelled in herds.  As a reviewer, it may not be
your job to suggest tests.  As an author it should be.  And as a
reviewer it should probably be your job to recognize when an author
hasn't done their job.  But what the hell do I know; I haven't even
defended my thesis yet.

-- 
Mickey Rowe     (rowe@lepomis.psych.upenn.edu)