Influencing Trust in Cybersecurity by Hacking the Human Factor
Abstract
The “prevalence effect” describes the fact that signal rarity diminishes human capability to detectand respond. In email cybersecurity this means that as defense systems reduce attack emails, theuser has more difficulty detecting and reporting what attack emails remain. When humansperform more poorly as automation is more successful, major implications for trust are assured.Despite no previous efforts directly investigating this paradox, the supposition is supported bypast research into automation, autonomy, and trust. In human-technology interaction, trust anddistrust arbitrate both initial decisions to engage in use, and subsequent adaptive or maladaptivepatterns of use. The present work seeks to explore the potentially paradoxical relationship ofprevalence effects to operator trust in 1) defensive autonomy 2) participants’ own ability, and 3)the digital environment. The direction and magnitude of resultant changes to these dimensions oftrust will be explored and modeled algorithmically in the context of both cyberdefense andcyberattack. The proposed effort will employ the Email Testbed (ET), an email cyberattacksimulator previously built in cooperation with the Air Force 711th HPW. Outcomes will besubjected to signal detection analysis, and used to generate algorithmic approaches andcountermeasures toward proper trust calibration and maximum performance. The ultimate goal islogic for trust-aware cyberdefensive autonomy better protecting the warfighter and public.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 11, 2018
- Source ID
- FA95501810337
Entities
People
- Benjamin Sawyer
Organizations
- Air Force Office of Scientific Research
- Massachusetts Institute of Technology
- United States Air Force