p.3 The starting point in most models of physical pattern formation is a set of equations of motion for
which a uniform solution is assumed to exist. The equations can be realistic descriptions of the system in question, or a
mathematical model chosen because its solutions mimic the behavior of the system one wishes to study.
p.9-10 we would like to know which simplifying assumptions are safe to make without losing the essence
of the process, and which simplifying assumptions lead to a model of little biological relevance... The model should,
of course, incorporate as much biological information as possible.
p.10 Perhaps the most important thing that should be required of a model is not that it
reproduces a pattern faithfully, but that with small quantitative changes in parameters values, it can produce the
evolutionary diversity present in that pattern, and the effects of perturbation experiments and mutations
on the pattern... Finally, a model should not only produce the final pattern but mimic in its time
evolution the succession of patterns seen during ontogeny of the system.
p.14 Learning requires some level of a priori knowledge about the task to be learned
p.190 the evolution of life as well as natural selection can also be modeled by complex dynamical systems
with fixed rules, although this modeling will be extremely difficult
p.190 we need two sets of dynamical systems: at the higher level, there is a dynamics of the rules, and
at the lower level, there is a dynamics of the entities. Many evolutionary models are of such nature. The complexity of the
system results from the interplay between higher-level and lower-level dynamics.
p.192 Besides the dynamical rule, the wiring diagram of a network also plays an important role in determining
the dynamics.
p.290 In simulation the theoretical model grasps and accurately summarizes the principles behind the process
being simulated, while in mimicry the model is wrong even though it produces the right kind of pattern... Unfortunately, much
modeling in theoretical developmental biology appears at present to be mimicry.
p.291 In order to have a reasonable assurance that a model has captured the essence of a process, it must
produce a pattern whose details resemble those of the morphology being modeled, it must also reproduce in its dynamics
reasonable portions of the ontogenetic transformation that the real pattern undergoes... it must be able to produce...
a range of diversity of the pattern identical to that found to occur in nature. Few models meet these expectations.
p.350 To make a forecast if the [underlying deterministic] equations are not known, one must find both the
rules governing system evolution and the actual state of the system... the goal of modeling is to find a
description that accurately captures features of the long-term behavior of the system.
