[Taxacom] dinosaurs and wolves

Kirk Fitzhugh kfitzhugh at nhm.org
Wed Oct 3 18:40:14 CDT 2012


Hi Pierre,

Brief replies, so as not to try the collective taxacom patience.

Yes, in actuality why-questions dictate the limits of current methods, 
and it is the fact that the nature of those questions have not been 
addressed in the systematics literature that we have a plethora of 
algorithms and contradictory answers. Evidential relevance cannot be 
dependent on the nature of inference, because relevance has to be 
decided before one can engage in rational inference (cf. Carl Hempel's 
work). When speaking of 'how best to infer,' that only makes sense in 
the context of abduction, and then we have to acknowledge that our 
available 'models' [sic] do not offer anything beyond profound 
vagueness; look at any cladogram for the lack of causal specifics. I'm 
not too worried about what theories are, or are not, used for the 
abduction of phylogenetic hypotheses. We almost never proceed forward 
with the actual testing of any of those hypotheses, so why worry about 
what we do infer? Finally, I stand by my claim that comparing cladograms 
inferred from different sets of premises is a futile, unscientific 
exercise. The nature of inductive logic is on my side on that one. ;-)

Long live nihilism.

Kirk

On 10/3/2012 3:32 PM, Pierre Deleporte wrote:
> Many thanks, Kirk
>
> for me, the crucial part of your post is :
>
> "We need to think about the why-questions we ask,
> HOW BEST TO INFER CAUSAL EXPLANATIONS FOR OUR OBSERVATIONS, *
> *and the extent to which relevance issues need to be considered."
>
> I think that "relevance" is dependent on "how best to infer"
> so I focus my reflection on: "how best to infer',
> which, in my view, appeals to: "MODELS OF CHARACTER EVOLUTION"
>
> the more we will know about general evolutionary processes (if any)
> for all kinds of characters in all lineages,
> or for different kinds of characters in different lineages,
> the best we will be able to implement this knowledge
> in our formal procedures for
> "SKETCHY HISTORICAL EXPLANATION OF CHARACTER OCCURRENCE AND DISTRIBUTION"
> (my personal formulation of the Fitzhughistic view),
> also known as "PHYLOGENY INFERENCE" (more repanded popular formulation,
> if well conceived)
>
> some approaches make use of explicit models of character evolution,
> some approaches make use of formal rules with implicit models of
> character evolution
> (I exclude explicitly assumed purely formal - 'strictly pattern' -
> approaches as non-biological) -
>
> for me the crucial question is:
> what do biologists know of evolutionary processes (= "models"),
> that they can confidentially use for historical explanatory inference?
>
> because I think that nobody can logically infer any history of sorts,
> without knowing at least some vague notion about the rules of historical
> processes (if any)
>
> I think that "VERY VAGUE" is also a crucial point in your post -
> as applied to causal accounts, in my view it implicitly qualifies
> the underlying explicit or implicit evolutionary models (if any)
>
> as for data partition:
> in my view we can either treat data sets separately
> and then combine them afterwards
> according to their respective evolutionary rules (="weights" or "costs"),
> or analyse them in a single analysis of a single data matrix
> while applying simultaneously different models
> relevant to different subsets of this inclusive data matrix
>
> I'm just trying to propose a positive perspective,
> beyond: "don't compare separate analyses"
>
> everybody "compares" them anyway,
> and you can't do anything against this -
> but if "compare" means" combine", however roughly and intuitively,
> I think it's not at odds with the logical requirement of using total
> relevant evidence
> for inductive historical inference / explanation
>
> I an agree that we better face this logical necessity,
> and make it more explicit, formalized, and biologically supported as
> best as possible
>
> my two cents - and please forgive my poor English (if any!)
> Pierre
>
> post scriptum:
> my compliments to people able to purposedly find this discussion
> under the label "dinosaurs and wolves" -
> TAXACOM is a refreshing intellectual wonderland   :-)
>
>
> Le 03/10/2012 22:00, Kirk Fitzhugh wrote:
>> Hi Pierre,
>>
>> Cladograms are nothing more than abstract devices that imply very vague
>> causal accounts of character origin/fixation in ancestral populations
>> and subsequent population splitting events. We (abductively) infer these
>> hypotheses as answers to questions regarding the occurrences of features
>> among groups of individuals to which different species hypotheses refer
>> (those hypotheses also abductively inferred, though separately from the
>> phylogenetic hypotheses represented by cladograms). But prior to making
>> such inferences, we decide whether or not the explanations of one set of
>> observations have an effect on the explanations of another set of
>> observations. Certainly in the context of cladograms the same sorts of
>> causal events are applied and by default explaining one set of
>> observations would be relevant to explaining another set of observations.
>>
>> I think the greatest problem is that we don't specifically acknowledge
>> the why-questions we're asking, for which we contend cladograms entail
>> answers. Personally, I think answering a question of the form, "Why do
>> these individuals have an A at position 546 in contrast to a G?", in the
>> same causal context as "Why do these individuals have 15 skull bones in
>> contrast to 18?" is very naive. But this means we need to stop thinking
>> only in terms of tossing data into a computer and getting 'trees.' We
>> need to think about the why-questions we ask, how best to infer causal
>> explanations for our observations, and the extent to which relevance
>> issues need to be considered. Simply continuing with the standard
>> approach of 'partitioned analyses' and the empirically meaningless
>> discussions of 'congruence' just don't work.
>>
>> Kirk
>>
>> On 10/3/2012 12:27 PM, Pierre Deleporte wrote:
>>> Hi Kirk,
>>>
>>> just a question :
>>> how do you propose to combine both classes of observations
>>> for an epistemically meaningful phylogenetic explanation?
>>> I mean the details of the procedure, and why so ...
>>>
>>> not a trap, a real interrogation -
>>> you know that I agree with your central point,
>>> I'm just questioning what's the way out
>>>
>>> in other words:
>>> beyond denegation, what positive perspective?
>>>
>>> Best,
>>> Pierre
>>>
>>>
>>>
>>> Le 03/10/2012 20:44, Kirk Fitzhugh a écrit :
>>>> It might have been some consolation to the authors had they acknowledged
>>>> that comparing 'molecular' and 'morphological' trees is epistemically
>>>> meaningless. Then it's a matter of deciding whether or not explaining
>>>> one class of observations is relevant to explaining another class of
>>>> observations.
>>>>
>>>> Kirk
>>>>
>>>> On 10/3/2012 11:06 AM, Wolfgang Wuster wrote:
>>>>> The paper itself ("Assembling the Squamate Tree of Life: Perspectives
>>>>> from the Phenotype and the Fossil Record") is well worth reading,
>>>>> containing a host of new morphological characters, and also discussing
>>>>> the extreme lack of congruence between molecular and morphological data,
>>>>> particularly in relation to the position of the Iguania in squamate
>>>>> phylogenies.
>>>>>
>>>>> Wolfgang
>>>>>
>>>>>
>>>>>> Hi Wolfgang,
>>>>>>
>>>>>>            I haven't seen the newest phylogeny (by Gauthier et al., 2012),
>>>>>> but I assume that it further solidifies the general consensus that
>>>>>> Pachyrhachis and relatives are macrostomatan snakes (not sister group
>>>>>> to all snakes).  However, even that would not necessarily mean that
>>>>>> they re-evolved legs.  Isn't there also a general consensus that it
>>>>>> only indicates that snakes lost their legs numerous times (in
>>>>>> different lineages)?
>>>>>>
>>>>>>                   -------------Ken
>>>>>>
>>>>>> ---------------------------------------------------------------------------------------------------------
>>>>>>> Date: Wed, 3 Oct 2012 07:47:21 +0100
>>>>>>> From: w.wuster at bangor.ac.uk
>>>>>>> To: taxacom at mailman.nhm.ku.edu
>>>>>>> Subject: Re: [Taxacom] dinosaurs and wolves
>>>>>>>
>>>>>>> On 03/10/2012 04:58, Ken Kinman wrote:
>>>>>>>> Stephen,
>>>>>>>> I didn't say anything about reversals requiring reactivation of
>>>>>> genes. I certainly know of no snakes or marine mammals reactivating
>>>>>> leg genes and the reinvention of legs.
>>>>>>> Actually, there is reasonable evidence that simoliophiid snakes
>>>>>>> (Pachyrhachis, Haasiophis) may have re-evolved hind limbs, based on
>>>>>>> their possession of these appendages and their nesting deep in the
>>>>>>> ophidian phylogeny. See Gauthier et al., Bulletin of the Peabody Museum
>>>>>>> of Natural History 53(1), April 2012.
>>>>>>>
>>>>>>> Wolfgang
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> Dr. Wolfgang Wüster - Lecturer
>>>>>>> School of Biological Sciences Bangor University
>>>>>>> Environment Centre Wales
>>>>>>> Bangor LL57 2UW Wales, UK
>>>>>>> Tel: +44 1248 382301 Fax: +44 1248 382569
>>>>>>> E-mail: w.wuster at bangor.ac.uk
>>>>>>> http://pages.bangor.ac.uk/~bss166/
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
J. Kirk Fitzhugh, Ph.D.
Curator of Polychaetes
Invertebrate Zoology Section
Research & Collections Branch
Natural History Museum of Los Angeles County
900 Exposition Blvd
Los Angeles CA 90007
Phone: 213-763-3233
FAX: 213-746-2999
e-mail: kfitzhug at nhm.org
http://www.nhm.org/site/research-collections/polychaetous-annelids
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~




More information about the Taxacom mailing list