Generation+of+novel+invariant+features+(CHIMPs)

Prev: Self-organisation in moving groups Next: DODOM

TODO list:
 * REF Harcourt? male bands
 * REF Goodall, travel distances

Generation of novel invariant features (CHIMPs)
Here we take a look at social structure. In particular we will look at primates (chimpanzees), in which the following things have been observed (Harcourt (REF)): We will make a model to study this social structure. However, lets not pre-specify things, but lets have a look whether the patterns we observe could be //side-effects of things that individuals have to do anyway//. Thus our modeling strategy is one where in the model definition we do not implement the behaviour we are interested in, but rather a minimal structure of the system that we are interested in, and then //observe// how such behaviour could emerge.
 * There are all-male groups, but (almost) no all-female groups.
 * Females live solitary much more often than males.
 * Males travel further than females.

To do this te [|Boekhorst and Hogeweg (1994)] made a model of CHIMPs where individuals: Furthermore:
 * go to the nearest fruit tree and eat until they are satisfied or until the food is exhausted
 * may take other CHIMPs into account when looking for food
 * want to mate: males check for receptive females
 * Females need more protein (for reproduction), so they eat protein-food that is not eaten by males.
 * All this takes place in an environment with FRUITs and PROTs: estimates were made on how much fruit and protein is available and in what kind of chunks the food is available (in terms of chimpanzee hours; to get into the right spatial and temporal scale)

The results of this model can be summarized as follows: Moreover: These observations are very similar to the observations in real chimpanzees!
 * There were random paths of movement and individuals eat the nearest food
 * Various groups formed with interesting compositions
 * There were clearly more lone females than lone males; and
 * There were more all-male groups than all-female groups.

Originally, the occurence of all-male groups had been explained by the idea that males band together to defend their territory and engage in warfare with neighbouring groups. This was thought to be both functional and adaptive, and and advanced feature (i.e. closer to humans than to other animals). However, we see the same in the simpe CHIMP model, with even more exaggerated male grouping. Thus we can see: > it just happens to be the case. But then, why does it happen (how does this epiphenomenon arise)? First of all, if the food occurs in clumps and individuals move towards the nearest group, this can generate grouping. Moreover, males check for receptive females and may move to the same female, which again enhances male grouping. On the other hand, females also need PROTs that are more easily depleted, causing females to split up in search of PROTs. (Note that in orangutans food distribution dependent travel bands arise according to similar reasoning ([|te Boekhorst and Hogeweg 1994]))
 * Chimpanzees do not need to want to be in groups or be solitary
 * Rather, the social structure is an **epiphenomenon**:
 * in terms of speculation about its purpose (function, evolutionary pressures), it is nonsensical to think that this grouping structure is advantageous and should give a higher fitness.

So far, we have shown //one// matching pattern (observable) between simulations and real life, but that is of course quite minimal. However, several other observables also show a good correspondence between the model and real chimpanzees, making a much stronger case: > This is because males are more often in groups, which deplete food faster and therefore also move on faster.
 * In terms of walking distances, males walk much further than females in our model.
 * In real chimps males walk about 4km/day while females on average 2.7 km/day (Goodall REF)
 * The model is actually able to quantitatively reproduce this result.

The moral of this story is therefore as follows:
 * Chimpanzees are not entities that are stupid.
 * However, some observables are so basic, that from those observables one cannot infer that chimpanzees are anything other than stupid.
 * Given that you do certain basic things (eat, follow mates), other behaviours will emerge as side-effects.
 * In that light, the practice of taking an arbitrary feature and then trying to find out how it is optimally evolved, cannot be done because there are many side-effects!
 * We can use models to study how different behaviours are related to each other and can how some may emerge as side-effects from others.

Conclusion

 * We have seen that in TODO behaviour is determined by local information (opportunities) and that behavioural patterns can arise as side-effects.
 * We have furthermore clearly seen a contrast between opportunity-based and optimality-based approaches (the latter largely ignores opportunities), where many behaviours could be explained as side-efffects,i.e. they are a consequence, and //automatic adaptation// (one does not need an adaptive process to not eat what is not there).

Next, we will take a look at TODO in combination with internal change within individuals (i.e. memory).

Next: DODOM

[|**te Boekhorst IJA & Hogeweg P** (1994) Self-structuring in artificial "chimps" offers new hypotheses for male grouping in chimpanzees. Behaviour 130 (1-3).] [|**te Boekhorst IJA & Hogeweg P** (1994)] [|Effects of tree size on travelband formation in orang-utans: data analysis suggested by a model study, in] [| Artificial life IV: proceedings of the Fourth International Workshop on the synthesis and simulation of living systems] [|(RA Brooks, P Maes), MIT Press.]
 * References**

(CHANGELOG 2014-2015)

- Removed from conclusion, because I do not see how it adds to the story (2nd point, opportunity vs optimality):
 * thus if the environment is changed when can get shared memory (cf blackboard in artificial intelligence, [|stigmergy] in ethology)
 * in complex behaviour we get mapping onto new behaviours
 * from optimality point of view gives a multi-peaked landscape
 * these side-effects are an alternative explanation to optimality and functionality,