24 Informational environments
being or becoming F at t 1 would be p = 1, so that no binary decisions are required
and no information would be generated. However, to the extent that this signal is
tied to that condition at the source unequivocally and lawfully, it is this kind of
relation that is relevant to Dretske, in that only such a relation is able to ground
knowledge. In turn, the transmission conditions investigated by the mathematical
theory determine how strong a signal has to be, how much redundancy is required
and hence what degree of reduction of possibilities at the source has to obtain
in order for it to successfully reach its receiver once it relates to conditions at
the source in an appropriate way. Both conditions have to be fulfilled in order to
afford knowledge to the receiver of information. In fact, all the remarkably steep
conditions Dretske imposes upon his notion of informational content are best read
in the light of his ultimate purpose, which is a genuinely and rather traditionally
epistemological one: what are the conditions of knowledge?
On Dretske’s account of informational content, the probability in question is
explicitly designed as an all-or-nothing affair: if p is not 1, it is bound to be 0
(Dretske 1981, 60). It is either true or false that s is F when some r-token that
signals s being F is produced and transmitted, and if it is false, no information is
transmitted. Only under these clear-cut conditions, Dretske continues to argue,
can information afford knowledge. If information were allowed to be false or
uncertain, and still be information proper, there would be no knowledge – notwithstanding the possibility of poor channel conditions or faulty mechanisms of
information uptake that add uncertainty to its use. We could not even be properly
mistaken about something in the first place if the underlying informational relations were already equivocal. Hence, the probabilities concerning informational
content and those concerning information transmission uptake should be clearly
distinguished – a distinction that is not always clear in Dretske.
To return to the example of frog vision that opened this chapter: a certain pattern
on the frog’s retina will elicit a certain behavioural response, namely the tongue
darting out to catch and eat the object so projected. Under normal conditions in
frog habitats, and under normal conditions of functioning for the frog’s perceptual apparatus, the respective patterns are virtually always caused by insects, and
they are so caused in accordance with fairly robust natural regularities. Given
that insects provide nourishment to the frog, one would be entitled to say that the
patterns on the frog’s retina convey information on the presence of insects-asnourishing-objects in his vicinity. However, if there is something dysfunctional
about the frog’s perceptual apparatus or if unusual external conditions intervene,
the informational relation in question cannot be tracked. The perceptual mechanisms involved may well maintain their functions even if they fail to perform it. If
an experimenter tosses lead pellets across the frog’s visual field, any object causing a sufficiently similar pattern on the frog’s retina would be met with the same
response. The experimenter would be tampering with the channel conditions, adding seeming signals that do not have their proper source. Hence, the visual pattern
would not convey information on the presence of insects anymore (Dretske 1981,
33–35).
5
If, however, the original informational relation were not firmly in place,
the frog could not react appropriately, even under optimal channel conditions.
being or becoming F at t 1 would be p = 1, so that no binary decisions are required
and no information would be generated. However, to the extent that this signal is
tied to that condition at the source unequivocally and lawfully, it is this kind of
relation that is relevant to Dretske, in that only such a relation is able to ground
knowledge. In turn, the transmission conditions investigated by the mathematical
theory determine how strong a signal has to be, how much redundancy is required
and hence what degree of reduction of possibilities at the source has to obtain
in order for it to successfully reach its receiver once it relates to conditions at
the source in an appropriate way. Both conditions have to be fulfilled in order to
afford knowledge to the receiver of information. In fact, all the remarkably steep
conditions Dretske imposes upon his notion of informational content are best read
in the light of his ultimate purpose, which is a genuinely and rather traditionally
epistemological one: what are the conditions of knowledge?
On Dretske’s account of informational content, the probability in question is
explicitly designed as an all-or-nothing affair: if p is not 1, it is bound to be 0
(Dretske 1981, 60). It is either true or false that s is F when some r-token that
signals s being F is produced and transmitted, and if it is false, no information is
transmitted. Only under these clear-cut conditions, Dretske continues to argue,
can information afford knowledge. If information were allowed to be false or
uncertain, and still be information proper, there would be no knowledge – notwithstanding the possibility of poor channel conditions or faulty mechanisms of
information uptake that add uncertainty to its use. We could not even be properly
mistaken about something in the first place if the underlying informational relations were already equivocal. Hence, the probabilities concerning informational
content and those concerning information transmission uptake should be clearly
distinguished – a distinction that is not always clear in Dretske.
To return to the example of frog vision that opened this chapter: a certain pattern
on the frog’s retina will elicit a certain behavioural response, namely the tongue
darting out to catch and eat the object so projected. Under normal conditions in
frog habitats, and under normal conditions of functioning for the frog’s perceptual apparatus, the respective patterns are virtually always caused by insects, and
they are so caused in accordance with fairly robust natural regularities. Given
that insects provide nourishment to the frog, one would be entitled to say that the
patterns on the frog’s retina convey information on the presence of insects-asnourishing-objects in his vicinity. However, if there is something dysfunctional
about the frog’s perceptual apparatus or if unusual external conditions intervene,
the informational relation in question cannot be tracked. The perceptual mechanisms involved may well maintain their functions even if they fail to perform it. If
an experimenter tosses lead pellets across the frog’s visual field, any object causing a sufficiently similar pattern on the frog’s retina would be met with the same
response. The experimenter would be tampering with the channel conditions, adding seeming signals that do not have their proper source. Hence, the visual pattern
would not convey information on the presence of insects anymore (Dretske 1981,
33–35).
5
If, however, the original informational relation were not firmly in place,
the frog could not react appropriately, even under optimal channel conditions.