The ID Report

By Robert Deyes

In April of 2008 I was fortunate enough to attend a seminar entitled 'Evolving Intelligent Behavior In Digital Organisms' presented by Michigan State University philosopher Robert Pennock (1). This seminar was part of the larger Bioethics Forum that this year carried the noteworthy title 'Evolution In The 21st Century' and included the likes of NCSE executive director Eugenie Scott and theologian John Haught in its star-studded cast. Pennock kicked off his seminar by talking about Paley's famous 'watchmaker' quote, subsequently claiming that Darwin had shown how complex biological phenomena could occur naturally through the process of evolution (1). He provided examples of several computer-based models that apparently demonstrated evolution in action including John Holland's 'genetic algorithm', Tony Ray's 'Tierra' and Chris Adami's and Titus Brown's AVIDA (1). Indeed AVIDA was the model that Pennock focused on for most of the rest of his talk describing it as a 2-D world containing organisms with virtual genomes and a so-called 'logic-based metabolism' (1). Organisms were selected for according to their ability to perform basic logic functions such as inverting and copying binary sequences.

In their 2003 Nature paper entitled 'The evolutionary origin of complex features' Richard Lenski and colleagues likewise presented AVIDA as a computer program that simulated biological evolution using 'digital organisms' that over time developed and were selected for their ability to perform increasingly complex logic commands (ie: NOT, AND, OR, EQU) (2). These logic commands were performed on 32-bit strings of information which supposedly paralleled the survival experiences that creatures might encounter in our own world. Just as with real creatures, digital organisms competed for food in the form of energy 'tablets' called SIPs (Single Instruction Processing Units). The more SIPs a digital organism was able to consume the more energy it had to both copy its genome and develop logic-based phenotypes that would allow it to compete. And yet what was most striking about this paper was the repertoire of already-existing capabilities that these organisms were provided with before beginning life in AVIDA. To be precise, organisms started off already endowed with virtual genomes, the ability to replicate and one of the necessary Logic commands that would enable them to function and be selected for. As Lenski and colleagues wrote:

"Experiments began with an ancestor that could replicate.... All organisms were identical and obtained equal energy to execute their genomic programs, including the copy commands by which a genome replicates itself one instruction at a time."(2)

Equally striking were the computational merits that were awarded to digital organisms whenever they performed particular logic functions. NOT and AND functions for example were given a computational merit of 2 while 'EQUALS' was given a merit of 32. It seems that even evolutionary biologists would have a problem here since their claim is that in real life, what constitutes evolutionary fitness is never pre-specified. No single trait gives an organism an absolute merit value for every type of situation it encounters. So rather than demonstrating Darwinian natural selection, AVIDA appeared to show what happens in a teleological world where goals and purpose are front-loaded into the fabric of life.

Of course, front loading a system so that it tends towards a desired end through a selective process has its practical uses. Indeed, in recent years there has been a push to develop systems- particularly electronic circuits- that are able to resist extreme changes in conditions (3). No more is this true than for the circuits of spacecraft such as Cassini where extreme environmental conditions are part of reality in space. As science writer Anil Ananthaswamy has pointed out, spacecraft like Cassini and the Galileo orbiter can suffer tremendously from radiation that sends their electrical circuitry haywire (3). Engineers are now looking at ways in which such systems can heal themselves by including the use of evolutionary software based on genetic algorithms that repairs damaged circuits (3). Together with his colleagues Didier Keymeulen and Ricardo Salem Zebulum, Adrian Stoica from the Jet Propulsion laboratory has shown how computers can be made to search for alternative circuits whenever a circuit is broken or disrupted (3). In short, computers have been designed to find new, fitter circuits in the event of catastrophe. But to what extent does this example represent a paradigm for Darwinian evolution? For sure, the evolutionary software loaded onto the computers searches through genetic algorithms to find new, 'fitter' circuits that can replace ones that have been damaged. The work of Stoica and his colleagues is thus truly a marvel of science for it does show how a system can repair itself without recourse to human intervention. Yet to say that it parallels the process of natural selection is to overstate the mark given the specified information that is front loaded into the software that allows it to direct the evolving circuit to a specified configuration. If anything what we see here is an example of goal directed evolution- precisely what Darwinian evolution purports not to be.

At the end of Pennock's talk I asked him what his thoughts were on William Dembski's demonstration that across an entire fitness landscape, evolutionary solutions are actually quite rare and therefore unlikely to be found through a simple random search (4). His answer was that while organisms in AVIDA could never be expected to search an entire fitness landscape, they do eventually find evolutionary solutions. I remained unconvinced after having watched a world imbued with purpose unfold before my very eyes. Pennock had after all done a quick demonstration of AVIDA during his seminar. AVIDA seemed to rely on an established 'fitness function' (5) in the form of merit scores, being preset into its programming. With its teleological foundations and its Logic-based methodology one could say that AVIDA represented nothing more than a Teleo-LOGIC model of life.

References And Notes
1. 7th Annual International Bioethics Forum: Evolution in the 21st Century, BioPharmaceutical Technology Center, Madison, WI, April 17-18, 2008

2. Richard E. Lenski, Charles Ofria, Robert T. Pennock, Christoph Adami, The Evolutionary Origin Of Complex Features, Nature, Volume 423, pp. 139-144

3. Anil Ananthaswamy, (2001), Space Babies, New Scientist 3rd February, 2001 pp. 26-30

4. William Dembski (2002), No Free Lunch: Why Specified Complexity Cannot Be Purchased without Intelligence, Rowman & Littlefield Publishers, Inc, Lanham, Maryland, pp. 212-213

5. In 'No Free Lunch: Why Specified Complexity Cannot Be Purchased without Intelligence', William Dembski describes a fitness function as "the univalent measure defined with reference to the phase space and needing to be optimized" with the phase space being a reference class of possible solutions across an evolutionary landscape. As Dembski subsequently notes, "the task of an evolutionary algorithm is to locate a possible solution where the fitness function attains at least a certain level of fitness", (See Ref 4, p. 187).