96 Informational environments
the fact that environments are normally shared between individuals and require
coordinated activities.
A third and possibly stronger reason for assuming that signalling within an organism and the purposeful placement of signals and informational artefacts in the environment are disanalogous affairs is provided by Sterelny (2003, Chapter 2): the
production of signals by an organism, especially when directed towards organisms
outside one’s own population, may not always and not even typically serve cooperative purposes. Although the uptake of natural information on physical conditions in
the environment is a straightforward affair, Sterelny argues, everyone who, like the
proponents of behaviour-based AI, focuses on this kind of information misses the
importance of signals that are made with the purpose of deception. A potential prey
may use mimicry in order to trick a predator into mistaking him for a poisonous or
otherwise unpleasant animal, or a predator may use camouflage in order to making
him being mistaken for a piece of a plant or otherwise benign object in the environment. In such cases, the function of the signals or structures in question is to actually
reduce reliability and unequivocality of the information available to the receiver,
and hence deprive the signals of their informational, that is specifying character.
Only if and when means of finer discrimination or additional cues are or become
available to the receiver, can the signals in question be disambiguated.
Hence, Sterelny concludes (in explicit reference to Godfrey-Smith 1996a), one
key function of cognition is to match less-than-reliable cues encountered in a
multifarious, “informationally translucent” environment against each other. Precognitive “detection systems” that are confined to one or a few non-integrated
information channels will be a viable option for organisms who are relatively
more robust and specialised in their constitution and, as the flip side of the same
coin, live in less variable, “informationally transparent” environments than those
generalists for whom a complex array of conditions will matter and who have
a rich repertoire of responses to match variation in their environments. For the
latter, “robust tracking” will be the best solution. An “informationally opaque”
or, in the terms referred to earlier, chaotic environment will, in terms of adaptive
economics, not warrant that effort but favour fixed responses and resilience. The
environment whose informational characteristics are most suitable to the evolution of complex cognitive functions, as Sterelny argues, is neither transparent
(or dull) enough to always and in all respects provide reliable cues nor opaque (or
chaotic) enough to automatically frustrate any cognitive effort.
What informational environments are
On the background of the preceding discussion, let me now introduce an account
of informational environments that caters for the various shades of informational
characteristics of environments as outlined by Sterelny (2003), from opaque to
transparent, while reserving the character of being genuinely informational to
those signals which actually serve an adaptive function for its receivers. In departure from the distinction between natural cues and sender-receiver games proposed by Godfrey-Smith, my account shall equally cover natural and purposefully
the fact that environments are normally shared between individuals and require
coordinated activities.
A third and possibly stronger reason for assuming that signalling within an organism and the purposeful placement of signals and informational artefacts in the environment are disanalogous affairs is provided by Sterelny (2003, Chapter 2): the
production of signals by an organism, especially when directed towards organisms
outside one’s own population, may not always and not even typically serve cooperative purposes. Although the uptake of natural information on physical conditions in
the environment is a straightforward affair, Sterelny argues, everyone who, like the
proponents of behaviour-based AI, focuses on this kind of information misses the
importance of signals that are made with the purpose of deception. A potential prey
may use mimicry in order to trick a predator into mistaking him for a poisonous or
otherwise unpleasant animal, or a predator may use camouflage in order to making
him being mistaken for a piece of a plant or otherwise benign object in the environment. In such cases, the function of the signals or structures in question is to actually
reduce reliability and unequivocality of the information available to the receiver,
and hence deprive the signals of their informational, that is specifying character.
Only if and when means of finer discrimination or additional cues are or become
available to the receiver, can the signals in question be disambiguated.
Hence, Sterelny concludes (in explicit reference to Godfrey-Smith 1996a), one
key function of cognition is to match less-than-reliable cues encountered in a
multifarious, “informationally translucent” environment against each other. Precognitive “detection systems” that are confined to one or a few non-integrated
information channels will be a viable option for organisms who are relatively
more robust and specialised in their constitution and, as the flip side of the same
coin, live in less variable, “informationally transparent” environments than those
generalists for whom a complex array of conditions will matter and who have
a rich repertoire of responses to match variation in their environments. For the
latter, “robust tracking” will be the best solution. An “informationally opaque”
or, in the terms referred to earlier, chaotic environment will, in terms of adaptive
economics, not warrant that effort but favour fixed responses and resilience. The
environment whose informational characteristics are most suitable to the evolution of complex cognitive functions, as Sterelny argues, is neither transparent
(or dull) enough to always and in all respects provide reliable cues nor opaque (or
chaotic) enough to automatically frustrate any cognitive effort.
What informational environments are
On the background of the preceding discussion, let me now introduce an account
of informational environments that caters for the various shades of informational
characteristics of environments as outlined by Sterelny (2003), from opaque to
transparent, while reserving the character of being genuinely informational to
those signals which actually serve an adaptive function for its receivers. In departure from the distinction between natural cues and sender-receiver games proposed by Godfrey-Smith, my account shall equally cover natural and purposefully