Is sex really necessary?
Provacative title for a fairly dry subject. The issue is this: in the Evolutionary Computing literature, there is a slow-burning religious war over the effectiveness of the mutation operator versus the crossover (recombination) operator. Each of these operators has been shown to be effective in certain domains, and new papers are being produced every month that provide results that overturn those of a previous paper.
Several people have taken a stab at characterizing the operators over a variety of problem spaces. The problem is that this is gathering tiny points of data in a vast sea of problem spaces. We need theories about when to apply the operators to which problems that are founded in objective (preferably mathematical) bases so that we can support or refute them with experimental data.
I got to thinking about how mutation (and only mutation) is successfully applied in nature by single-celled organisms. Asexual reproduction is the only form of reproduction for single-celled organisms (that I know of, please correct me if I am wrong!). And bacteria are among the most successful species on the planet, having adapted to every environment we have. Score 1 for mutation.
Yet every advanced organism (I can think of no multicellular organisms that do not) use sexual reproduction (i.e. crossover). Score 1 for crossover.
Theory 1: single-celled organisms don't move far, so they have one fairly fixed environment. Macro organisms can move across vastly different environments. Therefore, there may be some survival potential in taking genes from a creature that has successfully traversed terrain A and also from a creature that has successfully traversed terrain B, yielding (if you're lucky) a creature that can navigate either terrain.
Theory 2: crossover represents a disruptive force on the genomes of both partners, which is detrimental unless there is some balancing benefit (like the terrain diversity in theory 1). This balancing benefit is not felt by bacteria and other single-celled organisms, so all single-cellular animals that tried sexual reproduction died out trying to compete with their simpler brethren.
Theory 3: once upon a time, there were two types of single-celled organisms; those that reproduced asexually, and those that reproduced sexually. The poor bastards who had to find mates didn't do very well, until they mutated adaptations to allow them to swim faster, sense better, etc. Those adaptations could only go so far on a unicellular chassis, so pretty soon, unicellular became multicellular.
Theory 4: God just designed it that way.
Several people have taken a stab at characterizing the operators over a variety of problem spaces. The problem is that this is gathering tiny points of data in a vast sea of problem spaces. We need theories about when to apply the operators to which problems that are founded in objective (preferably mathematical) bases so that we can support or refute them with experimental data.
I got to thinking about how mutation (and only mutation) is successfully applied in nature by single-celled organisms. Asexual reproduction is the only form of reproduction for single-celled organisms (that I know of, please correct me if I am wrong!). And bacteria are among the most successful species on the planet, having adapted to every environment we have. Score 1 for mutation.
Yet every advanced organism (I can think of no multicellular organisms that do not) use sexual reproduction (i.e. crossover). Score 1 for crossover.
Theory 1: single-celled organisms don't move far, so they have one fairly fixed environment. Macro organisms can move across vastly different environments. Therefore, there may be some survival potential in taking genes from a creature that has successfully traversed terrain A and also from a creature that has successfully traversed terrain B, yielding (if you're lucky) a creature that can navigate either terrain.
Theory 2: crossover represents a disruptive force on the genomes of both partners, which is detrimental unless there is some balancing benefit (like the terrain diversity in theory 1). This balancing benefit is not felt by bacteria and other single-celled organisms, so all single-cellular animals that tried sexual reproduction died out trying to compete with their simpler brethren.
Theory 3: once upon a time, there were two types of single-celled organisms; those that reproduced asexually, and those that reproduced sexually. The poor bastards who had to find mates didn't do very well, until they mutated adaptations to allow them to swim faster, sense better, etc. Those adaptations could only go so far on a unicellular chassis, so pretty soon, unicellular became multicellular.
Theory 4: God just designed it that way.
posted by Simuloid | 10:00 PM
0 Comments:
Post a Comment
<< Home