Paul A. Fishwick wrote:
> It is not always clear when one needs to create a fully artificial human with simulated
> cognition, and when it suffices to create only the usual queuing models and models
> based on spatial, temporal and procedural (trained) constraints.
Very rough answer:
You do not need a 'fully artificial human with simulated cognition' at
either end of the constraint spectrum. If people are very constrained
in their action you can (usually) rely on the cognition and hence the
actions to reflect these constraints, hence you do not need the
cognition itself in the model. If the people are so unconstrained that
there is no pattern at all to their behaviour the modeller may be able
to package this as randomness or noise and still maintain a simple
model.
In between these extremes you will likely not be so lucky. People have
been wrong by assuming optimistically that one is at one of these
'simple' extremes: traffic modellers have in the past wrongly assumed
that behaviour is sufficiently constrained to avoid the complexities of
driver cognition and economists continue to assume that much human
'irrational' behaviour can be reified out as random noise.
Corrollary:
If you are at the constrained end of the spectrum, adding constraints
may well make the situation more predictable, if you are at the other
end adding constraints may well make it more unpredictable.
Regards.
--------------------------------------------------
Bruce Edmonds,
Centre for Policy Modelling,
Manchester Metropolitan University, Aytoun Bldg.,
Aytoun St., Manchester, M1 3GH. UK.
Tel: +44 161 247 6479 Fax: +44 161 247 6802
http://www.cpm.mmu.ac.uk/~bruce
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|