Human Failure Modes - Center for Software Engineering
Download
Report
Transcript Human Failure Modes - Center for Software Engineering
Human Failure Modes
Dr. Azad M. Madni
Professor , Epstein Department of Industrial and Systems Engineering
Director, SAE Program
Co-Director, CSSE
March 6, 2012
Copyright © 2012 Azad Madni
Outline
■
■
■
■
■
■
■
■
Human Failure Modes
Demanding Systems Requirements
Implications for Humans
Evolving Human Roles
Systems Engineering Mindset
The Remarkable Human Brain
Human Error Sources
Potential Remedies and Opportunities
Copyright © 2012 Azad Madni
Human Failure
■ Comprises
human errors, which are unintentional behaviors
violations, which are willful disregard of rules and regulations
■ Human errors fall into specific categories
slips, lapses of memory
mistakes in following rules and procedures
mistakes in understanding
Copyright © 2012 Azad Madni
Demanding System Requirements
■
■
■
■
Adaptability
Reconfigurability
Composability
Resilience
These requirements pose formidable challenges for
humans that work with and within complex systems
Copyright © 2012 Azad Madni
Implications for Humans
■ System adaptability implies changing contexts and potential
changes to human-system interactions
■ System reconfigurability implies potential changes to human
roles and human-system function allocation
■ System resilience implies potential dynamic changes to
human role and attendant changes to cognitive load
■ System composability (as in SoSs) implies potential changes
in collaborators (lack of shared conceptual model)
These changes can increase likelihood of human error.
Copyright © 2012 Azad Madni
Evolving Human Roles
■ From that of operator outside
system to that of agent within
an adaptable system
decision maker
supervisor
monitor with override authority
re-assignable participant
(peer, assistant)
These roles require new behaviors.
Copyright © 2012 Azad Madni
Systems Engineering Mindset
• Humans are suboptimal job performers that need to be
shored up and compensated for during task performance
• This perception leads to systems that are inherently
incompatible with human conceptualization of work
• The resulting mismatch inevitability creates human reliability
issues that show up as human error
This mindset fails to capitalize on human ingenuity
Copyright © 2012 Azad Madni
The Remarkable Human Brain
■ Yuor Barin Can Raed This
■ For emaxlpe, it deson’t mttaer in waht oredr the
ltteers in a wrod aepapr, the olny iprmoatnt tihng
is taht the frist and lsat ltteer are in the rghit pcale.
The rset can be a toatl mses and you can sitll raed
it wouthit pobelrm.
■ S1M1L4RLY, Y0UR M1ND 15 R34D1NG 7H15 4U70M471C4LLY
W17H0U7 3V3N 7H1NK1NG 4B0U7 17.
■ How ?
Source: LiveScience.com
Copyright © 2012 Azad Madni
Human Error Sources (Examples)
■
Erroneous/Incomplete Mental Model
often traceable to poor design- results in mistakes
lack of complete info causes user to make
unwarranted assumptions about system state
also results from misrecognition of cues/state info
■
Drop in Vigilance/Arousal during Monitoring
occurs with infrequent stimulus leading to
missed cue detection
■
Loss of Focus during Task Performance
results in slips (execution errors) arising from inattention
■
Cognitive Overload
causes: multi-tasking, context switching, decision making under stress
can lead to suboptimal behaviors and human errors (mistakes and slips)
Copyright © 2012 Azad Madni
Key Findings
■
■
■
■
Humans change cognitive strategies under overload
Inverted-U relationship: performance & stress
Humans unable to distribute attention under stress
Adaptability of human-in-the-loop system is
upper-bounded by acceptable human error rate
■ System inspectability facilitates human intervention
and avoids having to make erroneous assumptions
■ For robust performance
need to minimize multitasking and context switching
employ alerting/automation to monitor and flag rare events
need to understand cognitive strategies under overload for effective aiding
Copyright © 2012 Azad Madni
Potential Remedies
■ Design human work to avoid multi-tasking and frequent context
switching to the extent possible
■ Assign rare event monitoring to automation or alerting mechanisms
■ Provide decision aiding and performance support for decision
making under stress
■ Design appropriate incentives to counter risk compensation
tendency
■ Employ automation and dynamic function allocation to assure
manageable cognitive load
Most complex problems will require a combination of
many of these remedies.
Copyright © 2012 Azad Madni
Potential Opportunities
■ Exploit human ingenuity and creativity in:
adapting to shifting contexts
generalizing from specifics
recognizing novelty and improvising
aggregating information in the absence of an algorithm
detecting and filling gaps (e.g., in narratives)
Most complex problems will require a combination of
human creativity and ingenuity.
Copyright © 2012 Azad Madni
So,….Is Human Error a Cause or
Consequence?
Thank You
Copyright © 2012 Azad Madni
My References
■ Madni, A.M. “Integrating Humans with Software and Systems: Technical Challenges and a
Research Agenda,” Keynote Presentation, 22nd Annual Systems & Software Technology
Council, Salt Lake City, Utah, April 26–29, 2010.
■ Madni, A.M. “Integrating Humans with Software and Systems: Technical Challenges and a
Research Agenda,” INCOSE Journal of Systems Engineering and, Vol. 13, No. 3, 2010.
■ Madni, A.M. “Integrating Humans with Software and Systems: Technical Challenges and a
Research Agenda,” Keynote Presentation, INCOSE 2010 LA Mini-Conference, Loyola
Marymount University, October 16, 2010.
■ Madni, A.M. “Integrating Humans With and Within Software and Systems: Challenges and
Opportunities,” (Invited Paper) CrossTalk, The Journal of Defense Software Engineering,
May/June 2011, “People Solutions.”
Copyright © 2012 Azad Madni