Current Research My current research ranges from agent communication to computational models of human cognition. My past work has included empirical and theoretical analyses of human belief revision, non-monotonic reasoning, expert systems, and machine learning. I bring to these studies a strong interdisciplinary perspective that unites elements of artificial intelligence, philosophy, and cognitive psychology. Common Sense, Reasoning, and Rationality explores themes of rationality for both humans and machines, from these interdisciplinary views. |
Agent
Communication for Cooperative Action
This research project is concerned with
the design and implementation of models for normative agent communication,
from which error conditions for agent communication can be
defined—cases in which an agent generates a not-understood or error message. Such a model specifies task and agent interdependencies,
agent roles, and predicate properties at a domain-independent level of
abstraction. The theoretical perspective taken here is that communication
acts are actions upon a state, and that agent state is usefully modeled as
consisting of beliefs, desires, and intentions. In the case of agent
communication, the state is the internal state of the receiving agent. The normative communication model
defines what are legal actions upon specific elements of this state, and
hence define which agent ‘beliefs’ are allowed to be updated, revised, or
accessed via a communication act. The interaction model can also set policies
for belief revision as a response to a not-understood message, which may be necessary when task
allocation or coordination relationships change during run time. Recent
research work has extended and applied this framework to the web services domain. |
Cognitive architectures,
deliberation, and action The term belief-desire-intention (BDI) has been used to denote a
position on theoretically useful mental state distinctions, particular models
of how these mental states affect reasoning, and a genre of architectures or
frameworks for developing software agents. Beyond the software agent
community, elements of this BDI perspective receive a distinctly different
realization in computational theories of human cognition. My interest in this
area considers whether and how a general cognitive architecture achieves the
functionality attributed to a BDI model of agency, particularly in matters
related to deliberation, goal formation, goal values, and the meta-level
choice between deliberation and action. Finding the same themes of rational
agency emerging at architectures specified at different levels of abstraction
reinforces the intuitions driving these theories and provides new insights
into how these intuitions translate into process models of cognition.
Students with a strong background and interest in cognitive science should
contact me about possible work in this area and are encouraged to explore the
ACT-R architecture. |
Belief Revision When a reasoner
recognizes that some newly-acquired information causes a conflict with a set
of currently-held beliefs, there may be different ways of resolving that
conflict, and these different ways correspond to abandoning (or at least
calling into question) different ones of the currently-held beliefs. One
issue that arises under this characterization concerns the principles by
which such a decision is made, i.e., the decision to abandon belief i rather than belief j, so that the new information can be incorporated into a
consistent situational model. Belief revision is a crucial part of everyday
reasoning, scientific reasoning, and even agent communication. My objective
in this line of work is to understand how relevant knowledge is brought to
bear in constructing alternative accounts or belief sets; There are various
metrics to choosing among alternative belief sets once they are constructed.
It is the controlled construction of these states that is most interesting to
me, for without specifying the process of belief state construction and how
it stops, a reasoner will be forever mired in deliberation. Hence, belief
revision is strongly related to control issues for intention setting and
re-deliberation. |
Previous Research Areas My previous work has included experimental and
systems work related to belief revision, case based learning, inductive
reasoning, and other issues in cognitive science. I was also a principal
investigator on the Orlando
Project |