Cite: New York Sun,
Tuesday, December 14, 2004, p. 14
CTI's mission: To create and apply technologies for knowledge,
reasoning, dialogue, and decision - based on cognitive science,
decision theory, information technology, and practical experience.
Founded in 1990 by Dr. Marvin S. Cohen,
CTI is an independent small business located in Arlington,
Virginia. The firm pursues research in human-centered technologies
in order to develop innovative and effective applications
for training and computer support for cognitive skills. CTI's
staff combines expertise and practical experience in:
A Trimodal Theory of Decision Making. Researchers in many areas of decision making have concentrated on narrow slices of the decision making domain, e.g., choice in classical decision theory or recognition in naturalistic research. We attempt to distill a more accurate and comprehensive conception of decision making from recent empirical research and theory, to elucidate some of its normative and descriptive features, and to draw conclusions for the cognitive engineering of decision aids.
Time and Uncertainty in Naturalistic Decision Making. This chapter focuses on a critical issue in NDM research: expertsí management of the time they take to gather information and verify assessments and decisions. We critique NDM theories that equate expertise entirely with substantive knowledge, trivilizing the issue of time management. Satisfactory management of time for deliberation requires not only substantive domain knowledge but also the acquisition of strategic thinking skill. We cite both conceptual arguments and experimental evidence with commercial airline pilots for this hypothesis, and we present a model of a relatively automatic process for allocating time baseed on animal foraging. The foraging for surprise model predicts that thinking strategies will improve as decision makers spend varying amounts of time deliberating about tasks that vary in known uncertainty, novelty, and cost of delay. We contrast this model with a "stopping rule" proposal that imples fixed, pre-existing knowledge structures (Klein, et al.'s Data-Frame theory). We sketch some implications for training and decision aiding from an NDM perspective.
and Critical Thinking. Critical
thinking is a purposive dialogue with oneself or others,
about alternative possibilities. Critical thinking
usually takes place in a practical context, where its purpose
is to improve situation understanding or decisionmaking -- by
proposing, critiquing, evaluating, and improving alternative hypotheses,
interpretations of evidence, predicted outcomes, goals, and courses
of action. Dialogue theory enables us to identify dialogue roles
and phases, and moves and constraints associated with each, that
characterize successful critical thinking dialogue, and which
enable participants to achieve their purposes within the time
and Leadership. Leadership skills are displayed in the way individuals
communicate with one another in the context of a collaborative
practical activity. An integration of cognitive and communicative
concepts leads to a conceptualization of leadership as the
orchestration and improvisation of appropriate dialogues to achieve
group objectives. Leaders must learn to recognize the
types of dialogue that are appropriate to different situations
and understand how to recognize and implement the associated expectations
and roles. They must also acquire skills for overcoming obstacles
to successful dialogue and for managing the flow of conversation
without violating trust.
Making and Critical Thinking. We have learned from
studies of proficient real-world decision makers that they are
distinguished by (1) their recognitional skills -- the ability
to recognize familiar patterns and to rapidly retrieve appropriate
expectations, goals, and responses -- and (2) by their ability
to apply their experience in unfamiliar, novel, or anomalous situations,
which do not match previous patterns. Under the latter conditions,
recognition needs to be supplemented by metacognitive processes
that monitor the products of recognition and, when problems are
found, adopt strategies to resolve them. CTI's Recognition
/ Metacognitionmodel provides a systematic
and precise account of processes that decision makers use to understand
and plan in such novel and uncertain situations.
Here are some basic readings about this model and its applications:
Models and Critical Thinking.
According to the Recognition / Metacognition model, effective
decision makers monitor their own evolving mental models. When
they find different types of uncertainty (e.g., gaps, conflicts,
or unreliable assumptions), they draw from a repertoire of strategies
for addressing that type of uncertainty. Refinement, modification,
or replacement of mental models is therefore problem-driven, subject
to regulation by the stakes of the problem and the time available.
(CTI received recognition for its mental models work by being
named of the five best Small Business Innovative Research projects
Critical Thinking. A computer-based implementation
of the Recognition / Metacognition model has been developed .
This work represents a synthesis of our model with work on rapid
parallel inference by Dr.
Lokendra Shastri of the International Computer Science Institute,
affiliated with the University of California at Berkeley. The
tool, called Shruti,
has two levels: (1) It performs rapid recognitional inferences
and planning within a large expert belief network, utilizing temporal
synchrony to maintain dynamic object binding and support complex
relational reasoning.(2) It executes a metacognitive control process
that monitors recognitional results for different types of uncertainty
and shifts attention within active memory to activate new long-term
knowledge and adjust assumptions. A simulation of a tactical Army
battlefield scenario, as well as a Navy tactical scenario, has
been implemented in this system, and are suitable for use in training
systems and decision aids.
Mental Models, and Critical Thinking. CTI has developed
several decision aid prototypes based on the Recognition / Metacognition
model. All these aids enable users both to (1) view and manipulate
graphical representations of mental models, and (2) annotate or
revise the models to highlight problems and solutions discovered
in critical thinking. In addition, some of the systems also support
collaborative critiquing and improving of mental
models, as well as real-time retrieval of information that is
relevant to improving or correcting the model under construction..
CTI has studied the way users interact with advanced decision
aiding systems and has developed methods for adapting decision
aid displays and reasoning strategies to human cognitive strengths
CTI developed a decision analytic theory of trust in
decision aids that conceptualizes it as a situation-specificexpectation of system performance, over different
time periods and conditional on different phases and circumstances
of use. The model makes predictions about the dynamic evolution
of trust and the appropriateness of different user interaction
strategies (e.g., when monitoring is or is not appropriate). We
have developed training for critical thinking
by decision aid users based on the model.