The domains of natural information 57
views, which to undertake is the purpose of this chapter. I will outline the concepts
of natural information and of informational domains that, I suggest, are the best
way to go, and then very briefly recapitulate the lessons to be learned from Dretske’s and Gibson’s notions of information.
Natural information and reference classes
Let me try to provide some theoretical definitions that, although certainly not
aspiring to constitute a fully fledged theory of natural information in its own
right, or to provide a genuine conceptual analysis of “natural information”, still
integrate and synthesise the insights from the preceding discussion in a way that
describes natural information as both objective and context-sensitive enough to be
useful to its consumers under the manifold conditions under which it is encountered. Above all, it shall help telling apart the variant from the invariant aspects of
what a perceiving organism encounters in his environment. The basic idea to be
promoted here is that natural information is both a nomologically governed relation and domain specific in relevant respects.
(NI-1) The invariance condition: informational relations obtain if and only if
distal conditions F and their transformations at s are matched by detectable proximal conditions and transformations at r in accordance with a
strict natural regularity I that requires a uniform explanation.
(NI-2) The reference class condition: the probability of a certain informational
relation between r and s being F to obtain in accordance with regularity
I is determined with respect to a certain reference class C, where
(a) C may universally comprise all real and possible situations of I-regular
covariance between type-r signals and s-conditions – the upper
boundary case;
(b) C may comprise a spatio-temporally bounded region in which typer signals covary with F-conditions at s under I, with other conditions concerning r, s or F obtaining outside that region – the case of
locally bounded information;
(c) C may comprise a regionally bounded or global set of type-r signals of which only a subset {r 1
i
,. . ., r n
i
} covaries with s-conditions
under I, with no means of further discrimination available – the
case of probabilistically bounded information.
(NI-3) The receiver condition: the definition of the reference class C is dependent on the concrete situation, purposes, constitution and abilities of a
receiver R.
Part NI-1 should be general and basic enough to capture Dretske’s central claims
on the nature of information discussed in Chapter 2 while omitting his reference
to knowledge and conditional probabilities. Only in NI-2, do receivers and probabilities begin to make an explicit appearance. In the case of NI-2a, the receiver
views, which to undertake is the purpose of this chapter. I will outline the concepts
of natural information and of informational domains that, I suggest, are the best
way to go, and then very briefly recapitulate the lessons to be learned from Dretske’s and Gibson’s notions of information.
Natural information and reference classes
Let me try to provide some theoretical definitions that, although certainly not
aspiring to constitute a fully fledged theory of natural information in its own
right, or to provide a genuine conceptual analysis of “natural information”, still
integrate and synthesise the insights from the preceding discussion in a way that
describes natural information as both objective and context-sensitive enough to be
useful to its consumers under the manifold conditions under which it is encountered. Above all, it shall help telling apart the variant from the invariant aspects of
what a perceiving organism encounters in his environment. The basic idea to be
promoted here is that natural information is both a nomologically governed relation and domain specific in relevant respects.
(NI-1) The invariance condition: informational relations obtain if and only if
distal conditions F and their transformations at s are matched by detectable proximal conditions and transformations at r in accordance with a
strict natural regularity I that requires a uniform explanation.
(NI-2) The reference class condition: the probability of a certain informational
relation between r and s being F to obtain in accordance with regularity
I is determined with respect to a certain reference class C, where
(a) C may universally comprise all real and possible situations of I-regular
covariance between type-r signals and s-conditions – the upper
boundary case;
(b) C may comprise a spatio-temporally bounded region in which typer signals covary with F-conditions at s under I, with other conditions concerning r, s or F obtaining outside that region – the case of
locally bounded information;
(c) C may comprise a regionally bounded or global set of type-r signals of which only a subset {r 1
i
,. . ., r n
i
} covaries with s-conditions
under I, with no means of further discrimination available – the
case of probabilistically bounded information.
(NI-3) The receiver condition: the definition of the reference class C is dependent on the concrete situation, purposes, constitution and abilities of a
receiver R.
Part NI-1 should be general and basic enough to capture Dretske’s central claims
on the nature of information discussed in Chapter 2 while omitting his reference
to knowledge and conditional probabilities. Only in NI-2, do receivers and probabilities begin to make an explicit appearance. In the case of NI-2a, the receiver