3 Varieties of perception
One implication of Fred Dretske’s nomological, objectivist view of information
is that his aforementioned “suitably placed observer” is not necessarily identical
with the actual receiver or consumer of the information in question. He might be
the observer of an informational situation in which a different receiver is involved.
The receiver in that situation may or may not learn something about X, and he, she
or it may or may not have access to information about the reliability of the signals
received. Presumably, the frog belongs to the less privileged group. The human
observer is taken to be in a better position, for being able to assess informational
situations in a way that other organisms are not. This epistemic privilege seems
to include, again in principle at least, the informational situations he, she or it is
involved in – even if no ultimate verification of the relations of our perceptions
to their distal objects is in reach. This epistemic privilege seems to be assumed
without further justification, and it places human agents outside the environmental
constraints that affect other organisms.
A second implication of Dretske’s informational semantics is that, despite all
his reference to the mathematical theory of information, his concept of information is at variance with notions of information in mathematics or computer science
in at least one important respect: it may work as a remedy against the symbol
grounding problem that keeps haunting AI, but provides no account of how natural information is processed. Conversely, AI, along with the mathematical theory
of communication, often confines itself to the tasks of information processing,
leaving the grounding of the symbols so processed to take care of itself, or assuming it to be someone else’s business anyway. This observation does not rule out the
possibility of a division of labour between computer scientists and AI researchers
on the one side and approaches to cognition based on theories of natural information on the other. However, the very paradigm of a psychology of perception
that is based on an account of environmentally rooted natural information, a paradigm endorsed by Dretske (1981, Chapter 6), is expressly sceptical of AI, in
assuming that no processes of a computational kind are involved in perception to
begin with: James Jerome Gibson’s ecological theory of visual perception (1979).
Although other proponents of theories of natural information are not as resolutely
disinclined towards the computational realm, Gibson’s stance may count as symptomatic of a systematic incompatibility.
One implication of Fred Dretske’s nomological, objectivist view of information
is that his aforementioned “suitably placed observer” is not necessarily identical
with the actual receiver or consumer of the information in question. He might be
the observer of an informational situation in which a different receiver is involved.
The receiver in that situation may or may not learn something about X, and he, she
or it may or may not have access to information about the reliability of the signals
received. Presumably, the frog belongs to the less privileged group. The human
observer is taken to be in a better position, for being able to assess informational
situations in a way that other organisms are not. This epistemic privilege seems
to include, again in principle at least, the informational situations he, she or it is
involved in – even if no ultimate verification of the relations of our perceptions
to their distal objects is in reach. This epistemic privilege seems to be assumed
without further justification, and it places human agents outside the environmental
constraints that affect other organisms.
A second implication of Dretske’s informational semantics is that, despite all
his reference to the mathematical theory of information, his concept of information is at variance with notions of information in mathematics or computer science
in at least one important respect: it may work as a remedy against the symbol
grounding problem that keeps haunting AI, but provides no account of how natural information is processed. Conversely, AI, along with the mathematical theory
of communication, often confines itself to the tasks of information processing,
leaving the grounding of the symbols so processed to take care of itself, or assuming it to be someone else’s business anyway. This observation does not rule out the
possibility of a division of labour between computer scientists and AI researchers
on the one side and approaches to cognition based on theories of natural information on the other. However, the very paradigm of a psychology of perception
that is based on an account of environmentally rooted natural information, a paradigm endorsed by Dretske (1981, Chapter 6), is expressly sceptical of AI, in
assuming that no processes of a computational kind are involved in perception to
begin with: James Jerome Gibson’s ecological theory of visual perception (1979).
Although other proponents of theories of natural information are not as resolutely
disinclined towards the computational realm, Gibson’s stance may count as symptomatic of a systematic incompatibility.