-------------------------------------------------------------------------------- -------------------------------------------------------------------------------- A Set of Questions with Answers with Courtesy of Steve Grand -------------------------------------------------------------------------------- -------------------------------------------------------------------------------- Why does the hybridisation process either produce a Norn or a Grendel? I thought, it would behave always the same. The difference between norn genes and our own is that we are diploid and they are haploid. We have two copies of every gene, as you know, and therefore the copies can interact to produce dominant and recessive behaviour. That would be very difficult to model in artificial creatures, and so norns have only one copy of each gene. When two norns mate, the two genomes line up side by side and cross-over at one or more points. One of the resultant two genes gets destroyed and the other goes on to make a child. Therefore, the child inherits some genes from the mother and some from the father. -------------------------------------------------------------------------------- What are the information provided by the genes' screen [it's in the medecine kit and just before the brain icon] ? 1. This is the total number of genes in that creature's genome. Mutations during breeding may cause genes to be duplicated or lost, so the number can change over generations. Each gene codes for a useful structure, so any losses are likely to be harmful, but not necessarily so. Duplications could lead to new structures (especially chemical or brain structures) forming, which may evolve new functions. 2. This is the creature's internal name, which allows it to track its graphics / gene files. Every creature on your machine has a unique "name". 3. Finger print - um, I've no idea! The person that wrote this applet is away so I can't ask him! 4, 5, 6, These items show the number of genes of given important types. To start with, it shows the number of chemo-receptors (objects that respond to the levels of certain biochemicals). However, by pressing the grey bar (just above the numbers) you can scroll through various gene types and show the numbers of genes of that type. The "Number" entry is how many genes of that type there are, and the "Nucleotides" entry shows how many "codons" are involved - i.e. the longer the gene and the more genes there are, the more nucleotides. 7. This tells you how many neurones there are in the creature's brain (this is genetically defined, so may change after a mutation) 8. This is the number of brain lobes. A lobe is a region of brain cells with consistent properties, that performs a given function or part-function. There are initially nine lobes, but a gene duplication might cause new lobes to exist, which might eventually perform useful brain functions. 9. This is the number of dendrites in the brain (the number of connections between neurones) The main use for all this information is to see what broad changes there have been between parent and child. This could explain any odd behaviour, for example. A much more detailed analysis is possible, if you had the right tools, but at the moment we are not releasing those tools to the public. -------------------------------------------------------------------------------- What's the probability of one mutation and the percentage of mutations during breeding? I forget. I think there's a high probability of one to two mutations per generation. Smaller probabilities for omission / duplication. I had to have a high enough rate to drive reasonable change, but low enough to prevent too many still-born children. -------------------------------------------------------------------------------- Is your naming algorithm undeterminist? It uses random numbers and a cross-check. I wanted to minimise the number of collisions when people export their norns to each other, for efficiency. Picking a random ID causes far less collisions than choosing them consecutively. -------------------------------------------------------------------------------- Why two numbers for each types (for all my pets : Nucleo = 8 x genes for each types) The second number is supposed to be the number of codons (bytes) in that type of gene x the number of genes. It shouldn't be 8 in each case, as some genes are longer than others. Perhaps it's a bug! Still, it's not important and doesn't really tell you anything worthwhile. -------------------------------------------------------------------------------- In 7 and 8, you suggest that numbers could change after mutation. Have you seen such a thing ? do you know the probability of such an event ? No, and no. We haven't had the time to breed many generations during testing, as each new version of the program made the previous creatures obsolete, so we really don't know what will/might happen. You are our researchers! Potentially, changes in the brain could create structures that aren't there at the moment but I can already see the use for. For example, there is no concept of "arousal" in the current network, but Icould have genetically programmed such a thing in, so it's possible that evolution will do it for me! Likewise, their attention-directing lobes were intended to learn the relative significance of objects for themselves, but I didn't get round to doing that, so it might happen naturally. -------------------------------------------------------------------------------- The number of dendrites is a range and thus seems to be just "for fun" No, it's not just for fun. The brain genes specify the numbers of dendrites as a range and a distribution curve. These are important. For example, a very large region of the brain must have a population of neurones with either 1, 2 or 3 dendrites (and in my model it should have more 2s than it has 1s or 3s). A single gene defines that lobe of the brain, yet can specify the distribution and the minimum/maximum number of dendrites needed using only two codons. Evolution might select a better range and/or distribution than the one I picked. -------------------------------------------------------------------------------- Does your system handle variation of neurones and dendrites during the life No, it would have caused big problems, because of the way brain genes work - expressing a new brain gene late in life would wipe out all the memories and learning previously absorbed. For that reason, brain genes are always expressed before birth, regardless of what their "switch-on time" has mutated to. -------------------------------------------------------------------------------- Is there any reason for Grendels to be "Slug Brained" ? "Slug brained" is exactly right! They are not so clever as norns (not that norns are very clever either!) because their brain and brain chemistry genes have been edited and made less than optimal. It is a bit sad, but then grendels are quite happy in their own way! -------------------------------------------------------------------------------- What are the norn.gno and grendel.gno files for ? Those files just got left in the product by mistake - they aren't needed. When I created the creatures' genes, I used an editor that allowed me to name each gene and write notes about it, so that I could remember what it was for later. This information was written into the .GNO files (one file per species) in such a way that I could take any future genome (even if it had mutated, lost and duplicated genes), load it into the editor and still see what each gene was (originally) for. So, there's quite a lot of interesting information in those files, really! --------------------------------------------------------------------------------