mccool@dgp.toronto.edu (Michael McCool) (11/03/90)
kpc00@JUTS.ccc.amdahl.com (kpc) writes: >I am very interested in the reasoning here. Any comments? Followups >are redirected to c.a.p. >In article <1990Oct25.100748.2501@jarvis.csri.toronto.edu> >mccool@dgp.toronto.edu (Michael McCool) writes: > >>It is really an interesting question of whether birds (in flight) > >>show intelligent behaviour or are just purely constrained by the > >>physical laws of flight (wind streams etc). > I seem to recall windstream has nothing to do with it; the > ====== >From the simulations? (Underscoring is mine.) I am recalling the results of the paper, which I don't have handy, so I might be wrong. I believe these were the constraints used. The point was that constraints between members gave rise to the flocking collective behaviour; there wasn't an *external* force of any kind, or any mechanism of global organisation. Using tropisms etc. is an addition to *control* the simulation for purposes of animation, and is not necessary to exhibit the collective behaviour; the purpose of "maintaining a minimum speed and altitude" would do. Presumably birds would have some "reason" to remain airborne. > collective behaviour is a result of the birds desire to "remain > ================= > together" balanced against a desire to avoid collision with each > ======== ========================= > other and objects. >I don't mean this as a rhetorical device; just as a question: what >kinds of reasoning did you use here, and what are the names for the >elements of reasoning involved? (For example, had you made an I suppose something like this *was* how I reasoned: 1. a hypothesis is stated 2. the hypothesis is verified against the real system The hypothesis that birds want to avoid collision is not too far fetched :-) and the reason, ecologically, that schools and flocks exist is for common protection, breeding, etc. therefore you would expect them to want to stay together. The simulation is a verification of the theory, and although a point-to-point comparision to actual flock behaviour is difficult the motion, qualitatively, is similar. There is a problem with using a simulation as evidence for a theory, although it is used in other fields where observing the real system is difficult (i.e. astrophysics). > 1 A is the reason for B > B is like C (it was made for the sake of being like C) > There is probably a reason, of the same sort as A, for C >implies > 2 A is, or is probably, the reason for C This is, essentially, how I reasoned. There is the addition, however, that the rules for B were chosen to be analogs to rules that are expected to exist for C. We want the "fundamental" rules for the behaviour of C; so we try to construct a system with as few rules as possible, that are also expected to apply to C, such that a simulated system acts "like" C. The rules for A just didn't appear out of thin air, and then a resemblance of B to C noted; this was the result AIMED for. >This doesn't necessarily map onto the above example, but did > 1 seeing that an algorithm based on simulation of what, if it were > human agents instead of graphical objects, would be desires and > goals, simulates a flock of birds's salient group behavior fairly > well >lead you to believe that > 2 birds have these desires or analogues or homologues of these > desires I have used the word _want_ for lack of a better term; substitute tropism. I suppose the answer is yes. Although I expect people, in a flocking situation, might :-) use other forms of organization (committees to decide by representation the next turn to take? Descision by an autocratic ruler?). Mobs, I expect, might be an exception, if the people were stressed enough to not act like intelligent agents. By using the word "want" I don't mean to imply that the units of the flock are intelligent, or have any mental model we could recognize as a "desire", or are "aware" of their desires. They just respond to the environment in a certain way. A detailed description of my reasoning: 1. seeing that a set of rules generates behaviour that is qualitively like flocking behaviour and 2. having an expectation that the rules used are based on constraints that could reasonably exist in a real flock and 3. knowing that fish and birds don't have the most highly organized nervous systems and therefore probably follow simple rules and 4. observing that my rules are simple (they don't even require memory) I conclude 5. The proposed rules have a high probability of driving flocking behaviour in a real flock. >or perhaps instead that > 2 avoidance and cohesion strategies is the best, or a very good, way > to describe the flock's behavior as an entity I would use "explain" rather than "describe". I would conclude that avoidance and cohesion are an excellent candidate as a *mechanism* for the flock's collective behaviour. This behaviour is still not explained; if it is emergent, we would see behaviour that cannot be predicted except by simulation. By the way, as far as I know, I am an organic reasoning system and not a simulation :-) Of course, any computer trying to pass the Turing Test would say that, wouldn't it? Michael McCool@dgp.toronto.edu
steve@grian.cps.altadena.ca.us (Steve Mitchell) (11/06/90)
mccool@dgp.toronto.edu (Michael McCool) writes: >kpc00@JUTS.ccc.amdahl.com (kpc) writes: >> 2 birds have these desires or analogues or homologues of these >> desires >I have used the word _want_ for lack of a better term; substitute tropism. Just to inject a bit of natural history into this discussion, for several years I lived with an English (house) sparrow. In my extensive observations of her I was forced to the conclusion that she did indeed "want" various things. As an example, I kept a variety of treat foods on hand. These included lettuce, jerusalem artichoke, cashew nuts, steamed rice, etc. When she wanted one of these things she would fly to the room I was in, give a particular call, and fly back to her usual feeding station. I would go there, and start naming the various treat foods currently on hand. When I got to the one she wanted, she'd start wiping her beak on her perch, staring at me, and wiping her beak. I would then get what she wanted and feed it to her. On occasion when I brought something she had not selected, she would reject it and go back to the staring-and-beak-wiping routine. She knew what she wanted! -- - Steve Mitchell steve@cps.altadena.ca.us grian!steve@elroy.jpl.nasa.gov ames!elroy!grian!steve "God is licht, an in him there is nae mirkness ava." -- 1 John 1:5
mccool@dgp.toronto.edu (Michael McCool) (11/08/90)
steve@grian.cps.altadena.ca.us (Steve Mitchell) writes: >mccool@dgp.toronto.edu (Michael McCool) writes: >>kpc00@JUTS.ccc.amdahl.com (kpc) writes: >>> 2 birds have these desires or analogues or homologues of these >>> desires >>I have used the word _want_ for lack of a better term; substitute tropism. >... for several years I lived with an English (house) sparrow. In my >extensive observations of her I was forced to the conclusion that she >did indeed "want" various things. I didn't mean to imply that a mental model didn't exist, just that it wasn't necessary to the behaviour. Anchovies can show "flocking" behaviour, yet are not exactly mental giants. I guess behaviouralism was based on "if a behaviour can be explained without a mental model, then mental models don't exist". This is not what I meant to say. Seeing as how I'm getting *way* out of my field, I'll shut up now. Michael McCool@dgp.toronto.edu --- I saw the pale student of unhallowed arts kneel down beside the thing he had put together... --- Mary Shelly
greenba@gambia.crd.ge.com (ben a green) (11/08/90)
Michael McCool say "I guess behaviouralism was based on 'if a behaviour can be explained without a mental model, then mental models don't exist.' This is not what I meant to say." I think it was (and still is): If a behaviour can be explained without a mental model, then why bother with a mental model? -- Ben A. Green, Jr. Q: What shall we do about ignorance and apathy? A: I don't know and I don't care.
sarima@tdatirv.UUCP (Stanley Friesen) (11/10/90)
In article <GREENBA.90Nov7142313@gambia.crd.ge.com> greenba@gambia.crd.ge.com (ben a green) writes: >I think it was (and still is): If a behaviour can be explained >without a mental model, then why bother with a mental model? Because there are other data to be explained beyond mere behavior. For instance, we have considerable data on the activity of neurons within the brain. When this sort of info is taken into account, some sort of internal modelling is required. -- --------------- uunet!tdatirv!sarima (Stanley Friesen)
brossard@sicsun2.epfl.ch (Alain Brossard EPFL-SIC/SII) (12/19/90)
In article <GREENBA.90Nov7142313@gambia.crd.ge.com>, greenba@gambia.crd.ge.com (ben a green) writes: > > I think it was (and still is): If a behaviour can be explained > without a mental model, then why bother with a mental model? > > --