Um cálculo da finalidade (propósito) dos fenômenos biológicos

quinta-feira, julho 17, 2014

A Calculus of Purpose

Arthur D Lander

Published: June 15, 2004DOI: 10.1371/journal.pbio.0020164

Why is the sky blue? Any scientist will answer this question with a statement of mechanism: Atmospheric gas scatters some wavelengths of light more than others. To answer with a statement of purpose—e.g., to say the sky is blue in order to make people happy—would not cross the scientific mind. Yet in biology we often pose “why” questions in which it is purpose, not mechanism, that interests us. The question “Why does the eye have a lens?” most often calls for the answer that the lens is there to focus light rays, and only rarely for the answer that the lens is there because lens cells are induced by the retina from overlying ectoderm.

It is a legacy of evolution that teleology—the tendency to explain natural phenomena in terms of purposes—is deeply ingrained in biology, and not in other fields (Ayala 1999). Natural selection has so molded biological entities that nearly everything one looks at, from molecules to cells, from organ systems to ecosystems, has (at one time at least) been retained because it carries out a function that enhances fitness. It is natural to equate such functions with purposes. Even if we can't actually know why something evolved, we care about the useful things it does that could account for its evolution.

As a group, molecular biologists shy away from teleological matters, perhaps because early attitudes in molecular biology were shaped by physicists and chemists. Even geneticists rigorously define function not in terms of the useful things a gene does, but by what happens when the gene is altered. Molecular biology and molecular genetics might continue to dodge teleological issues were it not for their fields' remarkable recent successes. Mechanistic information about how a multitude of genes and gene products act and interact is now being gathered so rapidly that our inability to synthesize such information into a coherent whole is becoming more and more frustrating. Gene regulation, intracellular signaling pathways, metabolic networks, developmental programs—the current information deluge is revealing these systems to be so complex that molecular biologists are forced to wrestle with an overtly teleological question: What purpose does all this complexity serve?

In response to this situation, two strains have emerged in molecular biology, both of which are sometimes lumped under the heading “systems biology.” One strain, bioinformatics, champions the gathering of even larger amounts of new data, both descriptive and mechanistic, followed by computerbased data “mining” to identify correlations from which insightful hypotheses are likely to emerge. The other strain, computational biology, begins with the complex interactions we already know about, and uses computer-aided mathematics to explore the consequences of those interactions. Of course, bioinformatics and computational biology are not entirely separable entities; they represent ends of a spectrum, differing in the degree of emphasis placed on large versus small data sets, and statistical versus deterministic analyses.

Computational biology, in the sense used above, arouses some skepticism among scientists. To some, it recalls the “mathematical biology” that, starting from its heyday in the 1960s, provided some interesting insights, but also succeeded in elevating the term “modeling” to near-pejorative status among many biologists. For the most part, mathematical biologists sought to fit biological data to relatively simple mathematical models, with the hope that fundamental laws might be recognized (Fox Keller 2002). This strategy works well in physics and chemistry, but in biology it is stymied by two problems. First, biological data are usually incomplete and extremely imprecise. As new measurements are made, today's models rapidly join tomorrow's trash heaps. Second, because biological phenomena are generated by large, complex networks of elements, there is little reason to expect to discern fundamental laws in them. To do so would be like expecting to discern the fundamental laws of electromagnetism in the output of a personal computer.