[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
RE: "running" elephants - locomotary analoges
Date: Mon, 7 Apr 2003 22:36:37 EDT
Hutchinson made an interesting comment that reveals a flaw with his
methodology. It is the digital versus the real world problem.
"I'm not a big fan of using particular extant animals as "models", analogs,
or what have you for dinosaurs. I prefer understanding the principles and
mechanisms that make living animals work, and using multiple lines of
evidence to see how those principles might apply to extinct animals. I don't
find analogs/models as very testable (even indirectly) or insightful. They
mostly head toward dead ends, not in new or fertile research directions."
H appears to be pretty much abandoning what is far and away the best real
world method for investigating animal locomotion, comparative anatomy and
function. in favor of speculative digital simulations. Rather shocking
Sorry, but this is a blatant mischaracterization of what I think.
For example, this sentence is ignored:
"I prefer understanding the principles and mechanisms that make living
animals work, and using multiple lines of evidence to see how those
principles might apply to extinct animals." Where is "digital" in this
sentence and how is this not "real world"? Where did I say comp anat and
function are abandoned? I merely avoid use of analogies, but I still use
homologies and other approaches.
It might be a good idea to ask me what I think rather than assume for
rhetorical purposes. I love comp anat etc. and have published several
papers on it, right? This in itself falsifies this statement about "my
digital methodology" without even having to ask. As for how then I come to
different conclusions about tyrannosaur running, etc., it is in how the
anatomy is interpreted, which will be dealt with in forthcoming
papers. Trackways, anatomy, scaling, computer modeling, etc. are all
important to me-- this is what I consider to be an integrative approach,
combining multiple lines of evidence, as is said above.
I do feel, however, that biomechanics is ultimately the best way to test
most functional hypotheses, but not all, and not always using a computer
model. When combined with the aforementioned lines of evidence, many of
which can be included in a biomechanical model anyway, biomechanics shows
how animals work in a quantitative fashion. The power scaling Greg talks
about in his post is derived from decades of experiments with real animals
and, gasp, even the use of computers to analyze the data! Not from saying
"chickens are like humans in some ways so their power requirements must be
similar", which would be an analogy-based approach. When analogies with
extant animals are used, in my opinion they head down dead-ends. Once
someone concludes an ornithomimid worked like an ostrich then there is
often little else they can say, and too often the differences are swept
under the table in favor of similarity.
And so on, I have many more misgivings with the method that will be
explained in later papers. Perhaps this seems like a straw man, indeed
there are many shades of gray in what an analogy is and what it tells you,
but I do see the approach used in an erroneous way frequently. I am not
saying that anyone uses only this approach, but it is used as a primary
tool in some research rather than deeper mechanistic approaches such as
As for "digital vs. real world," one need look no further than my work to
see that this dichotomy is false. Unlike most other paleontologists who
talk about animal function, I have worked firsthand in experiments with
living birds, crocodiles, and elephants, with more experiments to
come. Hence the sentence "I prefer understanding the principles
and mechanisms that make living animals work". Where is this hands-on
understanding in the published research of others? Citations of Hatari and
nature documentaries replace direct observation and experimentation in many
accounts. How is that real world? A VCR and television were used, does
that make that approach "digital"? (OK, analog...) Many such people I
have talked to also do not even do dissections of living animals, instead
relying solely on pictures. Dissection was the core of my research work
and is fundamental to any biomechanics I do, with or without computers.
I rest my case; my work speaks for itself. I have not only published
digital studies and never will rely only on that. As for my dislike of
analogies without a deeper understanding of mechanisms, one can argue with
that. I argue that the link between form and function is complex and hence
function is often hard to predict from form. Biomechanics allows one to
quantify that linkage and tease it apart, within bounds of
confidence. Analogy seldom does; it is a qualitative approach (x works
like y because form in x is similar to y).
Hutchinson's claim that his methodology is more testable than anatomical
comparisons is spurious, since there is no way to observe if they are
correctly reconstructing the actual performance of fossil organisms that have
no directly comparable living examples.
I would argue that it is a better test because it gets at underlying
mechanisms, how animals work, in order to explain how an extinct animal
worked. An analogy with a rhino or elephant does not, at least very
well. That a computer is sometimes used is irrelevant. The mechanisms are
complex and hence sometimes it is impossible or inefficient to understand a
mechanism in depth without a computer to help. I'm no math genius;
computers are a crutch to make research go faster or even make it possible.
In fact, the actual power output of
leg muscles needed to run at a given speed in an animal of given size
directly tests Hutchinson's methodology for estimating the same, and
convincingly falsifies it.
It does not if one realizes that hands-on experimental work has also shown
that power output of leg muscles is often not the primary determinant of
running speed. This is in the literature. Sorry, but again people used a
computer rather than intuition or paper to do their simple math, so maybe
this real world approach is distasteful to Greg.
Hutchinson also suggests that the bouncing gait of elephants "costs more
energy per step because of the more flexed limbs." Again this is digital s
imulation being stated as though it were reality.
I did not use any digital simulations.... This follows the same logic and
methods as the standard literature in physiology and biomechanics, some of
which Greg implicitly cites in his post.
The real world cost of
locomotion in elephants plots a little below the standard line so if anything
they are energy efficent.
Slow walking is very different from running. Again, explained in the
The supposed extra cost of walking with flexed legs
is based on experiments in which humans move with abnormal gaits, and
abnormal gaits are always energy inefficient compared to the norm. If
ostrichs could be compelled to walk with their leg joints held straight they
would probably be less energy efficient than with their normally flexed legs.
Although it is again counterintuitive, it has never been demonstrated that
leg posture has a significant and consistent influence on power requirments,
since many flexed legged animals are more energy efficient than straight
kneed humans and elephants.
I won't spend energy explaining why this is patently false; it's in the
literature and we've rehashed this before.
...there are questions we will probably never
be able to answer no matter what methodlogy is employed. For example whether
some or all were sprinters of distance runners is strongly dependent upon
muscle fiber composition (lots of white fiber means short range, lots of red
long range), which as far as I know is not preserved directly or indirectly
in the fossil record.
Of course, this goes without saying. The "you need a time machine to test
your hypothesis" applies to anything done in historical science, and is not
a deeply intellectual criticism. We all know it but also know that various
indirect methods are useful as tests.
John R Hutchinson
NSF Postdoctoral Research Fellow
Biomechanical Engineering Division
Durand 209, BME
Stanford, CA 94305-4038
(650) 736-0804 lab
(415) 871-6437 cell
(650) 725-1587 fax