2010-09-13

Methodological Notes

Scientific research has long been defined by the balance between observation of phenomena of
interest, theoretical predictions (abstractions derived from observation), and experimental 
proof of these predictions; with the advent of computing, model-making ("simulation") has
become a strong third base for scientific exploration. In several interviews, V. Flusser defines
"reality" as a vanishing point which human experience research can asymptotically
approximate; in this sense, simulation is creating alternative realities which can be equally
valid for their respective context.

In biology in particular, this view leads to interesting perspectives: Large groups of people find
experiments with living beings ethically problematic, as they are sentient beings (the role of
perceived evolutionary proximity is interesting here). The gradualist view of modern Darwinism
finds that cutting a clear distinction between whatever cognitive or other skills define "human
specialness" and the general animal kingdom is constantly shifting: every time a new criterion
is erected, a species turns up that demonstrates the same skill.

After genomes for many species have been completely read, the idea that all relevant
information for a living being is encoded in genomes has turned out to be simplistic; this is not
a problem of genetic text interpretation. A lot of the data seem irrelevant (maybe this is
genetic code shrouding?), many phenomena happen in symbiosis with other life forms such as
pervasively present bacteria, and where higher order behavior “comes from” is even more
unclear (e.g., the idea that one could find a "homosexuality gene" is very naive).
Since Popper, scientific theories are not absolute truths, but temporarily considered the best of
our knowledge, and only correct until falsified. For possible epistemic strategies, Considering
how many formerly absolute truths have been toppled already, this calls for more epistemic
humility. One new heuristics based on these insights is to assume less and hypothesize more,
ideally in falsifiable ways - Speaking with Robert Musil, one can develop a sense of the
possible, a “Möglichkeitssinn”.

This is the background for the approach of the present work, which could be called exploratory
or simulative zoosystematics. Of course the pioneer work by Vilem Flusser and Louis Bec
haven been major sources of inspiration here; however, rather than modeling individual specimens in rich detail, we have concentrated on modeling their shared behavior, mainly
commonalities in communicative behavior across species.

The Communication Model

We posit that the basic pattern of communicative behavior shared across species in the
environment we model works as follows:

• An individual waits for incoming elementary signals, be they acoustic, light, or
electromagnetic radiation. As each individual generally often also sends such signals,
there is a relatively constant stream of signals coming in.
• Of these signals, it collects those it finds interesting, based on its own set of
preferences (which can be personal or species-based, or both)
• The signals it keeps in short term memory are arranged into longer strings of signals,
based on local context (i.e. an already assembled partial message in working memory).
• When it deems a word complete, that word is expressed in some form: as a sound
pattern, as light hues flashing over the animal's skin, as a rhythmic motion pattern, as
electromagnetic radiation patterns, or as combinations of several modalities.

This produces a characteristic distribution of activity: When one observes an individual being, it
will seem largely inactive most of the time; however, an observer sees and hears light and
sound patterns by other animals nearby. When the observed animal becomes active, its sound
and light patterns will be both individualized (it chooses its own preferred style of self-
expression) and shared: some properties, like distance patterns in what could be called
expressive parameters, and especially relationships in time patterns (“rhythms”) will reveal 
themselves to the attentive observer.

Sending signals and full messages costs energy; receiving them generates energy within each
animal. When its energy sinks very low, it goes into a hibernation-like mode where it does little
until it reaches high energy again; at high energy, it becomes hyperactive until its energy is
near its average again. Within the group interaction, this leads to overlapping cycles of activity
that create remarkable variety in the overall sound/light-scape of activity.

Note that for this modeled behavior, the precise nature of the individual atomic signals and
their assemblages, as well as their meanings, are irrelevant: to observe communication
behavior in a general sense, it is not necessary be able to understand the details. Thus, for
practical reasons of simplifying understanding, we chose alphabetical letters as the individual  signals, and strings of letters as full messages (“words”).

We further posit that there is second level of “objects of interest”: Each animal is assumed to
have a background of several “magic words” it finds particularly attractive; incoming “word”-
messages are compared to these magic words, and when they are similar enough, lead to
heightened activity - this mode has been affectionately nicknamed “song and dance”, or “party
mode”.

Preliminary conclusions

The behavioral richness of a modeled submarine communication ecosystem is quite rewarding
for the careful observer: One can switch perspectives between activity at different distances.
Here, the similarity of the location the piece has been developed for (Grosser Wasserspeicher
Berlin) to the deep sea has been inspirational: it is very dark, very quiet, and has very long
reverb decay. This preserves sound energy so well that one can still hear sounds from 100m
away in the silent phases of the nearer creatures.

Maybe the most interesting insight is that even with large species-based and individual
differences in expression of the discourse of messages, a careful observer can develop a fine
ear for the shared vocabulary of elements.

Future work

Many aspects of the models have only been touched, and warrant further consideration:
Some individuals have only been sketched roughly, and fleshing out more details will make the
overall ensemble behavior richer.

The animal's reactions to environmental influences are quite simple at the current state of
development – in essence, most of them find intrusive observers a source of irritation. Here, a
mix of curiosity and reluctance could produce intriguing interactions, not just with their known
habitat neighbours, but outside influences too.

Introducing long-term memory is quite likely to produce interesting longer cycles of behavioral 
evolution; logging long runs of ensemble activity will create very rewarding material for
analysis, generating further ideas for next directions to experiment with.

No comments:

Post a Comment