On Thursday 02 April 2009, Pedro C. Marijuan wrote:
> It conduces to realizing that the
> information processing of living cells is not of the same class than the
> processes of formal, predicative nature (computation). 

Why computational processes should be necessarely predicative?
Girard's Type Theory (System F) is impredicative yet "fully computational" (it 
computes exactly the recursive functions provably total in second order 
A detailed critique of Rosen's analysis is in:
Matteo Mossio, Giuseppe Longo, John Stewart.  Computability of closure to 
efficient causation.   To appear,  J. of Theoretical Biology, 2009.   

> Rather, 
> biomolecular processing is a “tactilizing” phenomenon based on a myriad
> of specific “molecular recognition” events (Conrad, 1998; Marijuán,
> 2003);

yes, this is the point: the structure of interactions, also beetwen different 
and entangled levels of organization, poses major modelling problems, well 
beyond the computational or the (minor) impredicativity issue (e. g. 
incompatibility between type-free and differentiable manifolds). Rosen was 
wrong on the circularity and impredicativity issue, yet he was right as for 
the global perspective (see the above paper).
Giuseppe Longo 
Laboratoire et Departement d'Informatique 
CNRS et Ecole Normale Superieure 
et CREA, Ecole Polytechnique 

fis mailing list

Reply via email to