Organizers: Anne Baril and Orlin Vakarelov
Abstract:
In this project I investigate what minimal conditions can allow a sys-
tem to be regarded as a cognitive agent, i.e. what the subject of cognitive
science is. An agent is a highly organized complex system which has in-
trinsic goals (in its minimal sense, persistence within viability boundary)
and whose dynamical organization supports an informational description.
Every agent, it follows, is informationally deprived. This deprivation raises
a design problem: How can the organization begin to overcome this in-
formational limitation? I claim that cognition is the general strategy for
resolving this design problem. More precisely, I define cognition as follows:
"Cognition is the set of the mechanisms/organizational constraints of
an autonomous agent that allow lowering of the conditional information
entropy of selected important informational sources in the environment on
the control structure of the agent, so that the agent can improve the se-
lection of actions to produce successful behavior in light of its information
gathering and carrying limitations."
I suggest that standard cognitive capacities — learning, memory, fea-
ture detection, representation, reasoning, etc. — can be viewed as special
cases of this general strategy.
1:40 p.m.
Anne Baril
A new approach to the problem of significant truths
Abstract:
Appealing to apparently non-epistemic considerations to solve problems in
epistemology may seem, well, unappealing. Yet if we allow such appeals we
make possible some very natural and plausible solutions to some stubborn
problems in epistemology. In this talk I explain one such solution to one such
problem. I propose a eudaimonist solution to the problem of significant
truths: the problem of explaining why some truths are more significant than
others. I argue that an appeal to a conception of human flourishing can be
part of a solution to the problem of significant truths without an
objectionable reduction of the epistemic to the non-epistemic.
2:40 p.m.
Orlin Vakarelov
Informational Networks: A Meta-architecture for Situated Cognition
Abstract:
The talk attempts to provide a conceptual basis for allusions to in-
formation processing in discussions about cognition. It deines the no-
tions of informational medium as a subsystem within a larger dynamical
system, the notions of information processing and information manage-
ment operations, and the notion of informational network — a network
of informational media. It suggest that situated cognitive systems can be
modeled with the help of informational networks at a high level of abstrac-
tion. Finally, it examines various properties that informational media may
need to posses to participate in an informational network. Understand-
ing those properties helps understand some of the design constraints on
various cognitive/AI mechanisms.
3:20 p.m.
END