Eric Horvitz and Misha Pavel
Information Science and Technology Study Group (ISAT)
Defense Advanced Research Projects Agency
The ISAT group is charged with providing the Information Technology Office (ITO) of DARPA with continuing, independent assessments of information sciences and technologies. The group identifies opportunities for innovations in computation, communications, and related applications, and performs strategic appraisal of technologies and overall new research directions for DARPA.
At the ISAT Woods Hole meeting in the summer of 2000, Eric Horvitz used the phrase augmented cognition to define a promising direction for DARPA research, while serving on an ISAT working group exploring new directions for DARPA. He used the phrase to describe a potentially fruitful multipronged research endeavor that would explore opportunities for developing principles and computational systems that support and extend human cognition by taking into explicit consideration well-characterized limitations in human cognition, spanning attention, memory, problem solving, and decision making. The proposed research included opportunities for developing fast-paced and fluid computational methods that could react to cognitive state sensed or inferred in a variety of ways, as well as behavioral and functional designs developed to complement limitations in human cognition. Research and prototypes from Microsoft Research on mixed-initiative interaction and the use of Bayesian attentional models to manage interruptions, and a recent ISAT study on the recall of memories, highlighted broad opportunities in this area. Low-hanging fruit for continuing research were deemed to include methods for managing interruption and recovery, assisting with multitasking, enhancing informational bandwidth via the best mix of channels, providing reminders or assistance at the right time, and addressing biases in judgment. He presented Augmented Cognition as a key direction to DARPA Director Frank Fernandez and program managers in July 2000.
In distinction to related efforts in HCI that have drawn broadly on principles of human cognition and models of cognition as information processing to support design, augmented cognition focuses on the precise targeting of one or more specific cognitive bottlenecks within larger applications—and on opportunities to build out carefully from these specific components, potentially bringing together multiple approaches to cognitive support within single systems.
We can draw a variety of examples of opportunities for augmented cognition applications from different subspecialties in psychology. For example, there is opportunity to leverage models and experimental results from cognitive psychology on divided attention and disruption in peripheral awareness and monitoring applications. Such results can be used to control if, how, and when information about a monitored system is presented to users focusing on another, more central task. As another example, results from visual search and attention might be leveraged in designs for information display that have the ability to relay with increased efficiency the most critical information to users in time-critical settings.
Beyond building and testing real-world applications, augmented cognition includes explicit efforts to enrich knowledge about cognitive mechanisms by taking advantage of new kinds of user studies during attempts to build, deploy, and test systems designed to address cognitive limitations. For example, new psychological studies and results might be developed by studying how users react to different alerting systems or policies in specific contexts. As another example, it may be feasible to investigate how different visual representations of actions alternatives and uncertain outcomes might help in debiasing well-characterized errors of judgment in fusing multiple findings and decision making under uncertainty.
The goal of the ISAT study on Foundations of Augmented Cognition has been to take a pass at mapping out the key challenges and opportunities, as well as initial directions that promise to be successful or to provide valuable learning experiences. The effort is aimed at further clarifying challenges and opportunities, and to support DARPA program management with an assessment of directions and thoughts on developing programs related to this topic.
As follow up on the work on DARPA directions last summer, we held an
workshop on directions for augmented cognition in Seattle in
November, 2000, organized by Eric Horvitz, Lee Kollmorgen, and Dylan
Schmorrow. This meeting was followed up
by an ISAT breakout group on augmented cognition during
the adjacent Fall 2000 ISAT meeting.
Slides from the briefing created by this group are available via a link
below. The ISAT breakout group led to
two ISAT efforts. The first, is a
formal 2001 study on “Mixed-Initiative
Control and Monitoring within Human-'Bot Systems.” The second exploration is
an incubation effort centering on an exploration of the terrain, set of
opportunities with augmented cognition, referred to as the “Foundations of
Augmented Cognition Technologies” incubation effort. We are holding the conference at the National Academy of Science as part of this effort. We view the output of the ISAT study as
providing support for the bidding and definition of one or more future DARPA programs in Augmented
Cognition.We are coordinating with Dylan Schmorrow, a program manager at DARPA. Dylan has
been pursuing possibilities for putting together a formal DARPA program focusing on some aspect of augmented cognition. During the coarse of the ISAT study, Dylan has pursued the funding of a set of
seedling projects in this area.
Breakout groups Participants will
receive a message from Rich Entlich on details about travel and lodging. ISAT contact for
arrangements: Institute for
Defense Analyses 1801 North
Beauregard Street Alexandria, VA
(tel.), 703-845-6848 (fax)
Several early presentations
Travel and lodging
Rich Entlich (firstname.lastname@example.org)
Participants will receive a message from Rich Entlich on details about travel and lodging.
ISAT contact for arrangements:
Institute for Defense Analyses
1801 North Beauregard Street
Alexandria, VA 22311-1772
703-845-6648 (tel.), 703-845-6848 (fax)