66 Informational environments
regularities, in turn, spatio-temporally local maxima and minima can be detected
in many cases. If there are such local maxima and minima, these may serve as
indicators for informational relations that concern individuals, populations, biological species or other kinds of entities whose identity is determined historically
or genealogically. The laws of nature will provide some boundary conditions that,
however, taken by themselves, will be insufficient to determine the identity of
individuals and species or their possibility of existence – which may turn out to be
the one thing relevant to R. Or those laws will not be sufficient to determine the
specific local character of a bundle of conditions – why some individual or species
not only could but had to exist here and now.
Basically, both ‘hard’ and ‘soft’ natural information are relevant to perception
and action, but in different respects, on different levels: information L concerns
basic invariants in the environment, and the conditions for the perception of
objects in the ambient optic array (what basic substance characteristics, what size,
how distant, how transformed in visual in appearance when the perceiver moves).
These conditions are, except under the most extraordinary circumstances – which
we are most likely to find in some philosopher’s possible worlds – not specific to
any domain. Information L , for its basic, foundational character, does not change
across domains but first enables the tracking of individuals or kinds of things
throughout their specific domains. Without some core regularities, environments
would be capricious and unpredictable. The validity of information C , in turn,
depends on both the concrete informational domains that the perceiving organism is accustomed to, and the relative reliability of the signals that make up these
domains.
Hence, information L and information C will also differ with respect to their reference classes (see NI-2 in the previous section): information L comprises all and
only those detectable relations in the environment which are governed by strict,
and universally applying, natural regularities. Information C , however, comprises
relations that depend on less strict and less universal regularities by their very
nature, and hence have their reference classes constrained by where and how
those regularities apply.
It might seem that domains, as defined previously (especially in NI-5a), are thus
a special case of reference classes, as defined previously (especially in NI-2b), so
that D ⊂ C. There is a difference though: a domain is a domain of a type of signals, bound together by what is being tracked, or supposed to be tracked, by R (an
individual, a population of individuals, a set of entities) and the I e -regularities that
respectively apply. A reference class, in turn, comprises whatever set of proximal
r and distal s entities and a regularity I between them, which together shall serve
to determine a probability with which some r within that set signals some condition at s. The choice of the reference class is, like the identification of a domain,
dependent on the purposes, constitution and abilities of R. Hence, the reference
class of a certain type of informational relation might be partly coextensive with
the domain of a type of signals (namely in comprising these signals), but it makes
no reference to why it should be circumscribed in a certain way, whereas a domain
of some type of signal is explicitly defined with respect to the circumscribing
regularities, in turn, spatio-temporally local maxima and minima can be detected
in many cases. If there are such local maxima and minima, these may serve as
indicators for informational relations that concern individuals, populations, biological species or other kinds of entities whose identity is determined historically
or genealogically. The laws of nature will provide some boundary conditions that,
however, taken by themselves, will be insufficient to determine the identity of
individuals and species or their possibility of existence – which may turn out to be
the one thing relevant to R. Or those laws will not be sufficient to determine the
specific local character of a bundle of conditions – why some individual or species
not only could but had to exist here and now.
Basically, both ‘hard’ and ‘soft’ natural information are relevant to perception
and action, but in different respects, on different levels: information L concerns
basic invariants in the environment, and the conditions for the perception of
objects in the ambient optic array (what basic substance characteristics, what size,
how distant, how transformed in visual in appearance when the perceiver moves).
These conditions are, except under the most extraordinary circumstances – which
we are most likely to find in some philosopher’s possible worlds – not specific to
any domain. Information L , for its basic, foundational character, does not change
across domains but first enables the tracking of individuals or kinds of things
throughout their specific domains. Without some core regularities, environments
would be capricious and unpredictable. The validity of information C , in turn,
depends on both the concrete informational domains that the perceiving organism is accustomed to, and the relative reliability of the signals that make up these
domains.
Hence, information L and information C will also differ with respect to their reference classes (see NI-2 in the previous section): information L comprises all and
only those detectable relations in the environment which are governed by strict,
and universally applying, natural regularities. Information C , however, comprises
relations that depend on less strict and less universal regularities by their very
nature, and hence have their reference classes constrained by where and how
those regularities apply.
It might seem that domains, as defined previously (especially in NI-5a), are thus
a special case of reference classes, as defined previously (especially in NI-2b), so
that D ⊂ C. There is a difference though: a domain is a domain of a type of signals, bound together by what is being tracked, or supposed to be tracked, by R (an
individual, a population of individuals, a set of entities) and the I e -regularities that
respectively apply. A reference class, in turn, comprises whatever set of proximal
r and distal s entities and a regularity I between them, which together shall serve
to determine a probability with which some r within that set signals some condition at s. The choice of the reference class is, like the identification of a domain,
dependent on the purposes, constitution and abilities of R. Hence, the reference
class of a certain type of informational relation might be partly coextensive with
the domain of a type of signals (namely in comprising these signals), but it makes
no reference to why it should be circumscribed in a certain way, whereas a domain
of some type of signal is explicitly defined with respect to the circumscribing