28 Informational environments
the informational relation as such. Millikan (2004, 71f ) argues that, if there are
enough successful instances of signalling a certain world affair to outweigh the
effects of wasted effort or damaging mistakes, so that the ratio between the effects
of the successful instances and those of the failures contributes to an explanation
of the individual’s and/or its species’ persistence and reproduction, the successful
instances in which r indeed signals s being F will suffice to determine the informational content of the signal.
In turn, the separation of the probabilities for the source and the channel conditions highlighted earlier is designed to specifically account for the possibility
of failing instantiations that are due not to unreliable conditions at the source or
dysfunctions in the receiver but to uncooperative environmental conditions that,
varying over space and time in kind and intensity, may negatively affect the transmission. Under real-world conditions, less-than-perfect consumers of information will have to make do with abundant but incompletely trackable information
in less-than-noise-free environments. Formal idealisations of the kind Dretske
appears to endorse may contribute to missing this point.
Despite these differences, a common concern for Dretske, Skyrms and Millikan lies in providing an account of the possibility of meaningful yet false, misrepresenting signs. They diverge on the question whether there can be such a
thing as false natural information to begin with. Where Dretske holds that, if it
is false, it is not information, Millikan maintains that, if it can be false, it is not
natural but already intentional information, while Skyrms settles for a kind of
laissez-faire probabilism that allows for all degrees of reliability of information,
the only threshold being the possibility of a breakdown of the hypothetical or real
sender-receiver game. In fact, Millikan claims that the mark of distinction of the
entire teleosemantic programme does not as much lie in the theories of meaning
it proposes but in its specific set of theories of misrepresentation (Millikan 2004,
Chapter 5). Depending on whether one counts Dretske and Skyrms as teleosemanticists, the variety of such theories within that camp seems to be fairly broad.
As we will see in the next section, these differences concerning false information
may ultimately betray a mostly implicit disagreement about a less obvious but
more foundational set of philosophical presuppositions.
Natural information and the roots of intentionality
The fact that Dretske, when discussing informational content, treats probabilities
at the source and probabilities concerning the channel conditions separately and
sets the standards for the former improbably high is, with some likeliness, systematically grounded: it is the very detachment of knowledge-conveying information
from meaning and its characteristic side effects of intensionality and error on which
Dretske’s approach builds. In his original definition of informational content, the
consumer of information appears only as the unnamed bearer of the knowledge
k in relation to which, under certain conditions, a signal r becomes the information that s is F. Under some, but not all, circumstances, k will become relevant
to the informational relation, and only under some, but not all, circumstances,
the informational relation as such. Millikan (2004, 71f ) argues that, if there are
enough successful instances of signalling a certain world affair to outweigh the
effects of wasted effort or damaging mistakes, so that the ratio between the effects
of the successful instances and those of the failures contributes to an explanation
of the individual’s and/or its species’ persistence and reproduction, the successful
instances in which r indeed signals s being F will suffice to determine the informational content of the signal.
In turn, the separation of the probabilities for the source and the channel conditions highlighted earlier is designed to specifically account for the possibility
of failing instantiations that are due not to unreliable conditions at the source or
dysfunctions in the receiver but to uncooperative environmental conditions that,
varying over space and time in kind and intensity, may negatively affect the transmission. Under real-world conditions, less-than-perfect consumers of information will have to make do with abundant but incompletely trackable information
in less-than-noise-free environments. Formal idealisations of the kind Dretske
appears to endorse may contribute to missing this point.
Despite these differences, a common concern for Dretske, Skyrms and Millikan lies in providing an account of the possibility of meaningful yet false, misrepresenting signs. They diverge on the question whether there can be such a
thing as false natural information to begin with. Where Dretske holds that, if it
is false, it is not information, Millikan maintains that, if it can be false, it is not
natural but already intentional information, while Skyrms settles for a kind of
laissez-faire probabilism that allows for all degrees of reliability of information,
the only threshold being the possibility of a breakdown of the hypothetical or real
sender-receiver game. In fact, Millikan claims that the mark of distinction of the
entire teleosemantic programme does not as much lie in the theories of meaning
it proposes but in its specific set of theories of misrepresentation (Millikan 2004,
Chapter 5). Depending on whether one counts Dretske and Skyrms as teleosemanticists, the variety of such theories within that camp seems to be fairly broad.
As we will see in the next section, these differences concerning false information
may ultimately betray a mostly implicit disagreement about a less obvious but
more foundational set of philosophical presuppositions.
Natural information and the roots of intentionality
The fact that Dretske, when discussing informational content, treats probabilities
at the source and probabilities concerning the channel conditions separately and
sets the standards for the former improbably high is, with some likeliness, systematically grounded: it is the very detachment of knowledge-conveying information
from meaning and its characteristic side effects of intensionality and error on which
Dretske’s approach builds. In his original definition of informational content, the
consumer of information appears only as the unnamed bearer of the knowledge
k in relation to which, under certain conditions, a signal r becomes the information that s is F. Under some, but not all, circumstances, k will become relevant
to the informational relation, and only under some, but not all, circumstances,