CAREER: Understanding the Principles of Working Memory in Phishing Decision Making

网络钓鱼是欺骗人们披露敏感信息或不适当地授予对安全系统的访问权限的做法。虽然网络钓鱼攻击在互联网上非常猖獗，但由于自动化，个人每天或每周遭遇攻击的可能性很小。然而，人们希望发现自动化遗漏的罕见攻击。这些攻击往往是新颖的和有针对性的，因此很难被发现。过去的研究将不注意归咎于人类对网络钓鱼易感性的重要线索，但阻止人们注意网络钓鱼消息中的指标的潜在认知功能尚未得到很好的理解。该项目正在开发认知过程的基础理论，以解释和预测人类信任或怀疑网络钓鱼攻击的决定。该项目将为新的电子邮件安全和培训解决方案的开发提供信息，以提高公众检测网络钓鱼攻击的能力，这样，该项目具有广泛的潜在社会影响。这项研究是使用实验室实验和计算模型的发展来揭示网络钓鱼检测的关键认知功能。该研究应用自然语言处理方法来确定人们在决策过程中编码到记忆中的网络钓鱼消息中的特征，以及这些特征对成功检测的影响。这项研究还研究和表征了对记忆过程至关重要的功能，这些功能会影响个人对网络钓鱼攻击的敏感性。还研究了应用认知模型和强化学习模型提供个性化网络钓鱼培训体验的挑战。从这个项目的活动中产生的知识和方法的影响超出了网络钓鱼。它们为错误信息检测和威胁检测中的相关挑战提供了信息。该项目的长期目标是建立一个多学科的安全研究计划，并通过包容性的研究，教育和推广活动，教育多样化的工程队伍。该奖项反映了NSF的法定使命，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

Phishing is the practice of deceiving people into disclosing sensitive information or inappropriately granting access to a secure system. Although phishing attacks are rampant on the Internet, due to automation, the likelihood of an individual encountering an attack daily or weekly is small. Yet people are expected to detect the rare attacks that automation misses. These attacks tend to be novel and targeted and, therefore, difficult to detect. Past research has blamed inattention to important cues for human susceptibility to phishing, but the underlying cognitive functions that prevent people from paying attention to indicators in a phishing message are not well understood. This project is developing foundational theories of cognitive processes to explain and predict human decisions to trust or suspect phishing attacks. The project will inform the development of new email security and training solutions to improve the general public’s ability to detect phishing attacks, and in this way, the project has broad potential societal impact. This research is conducted using laboratory experiments and the development of computational models to reveal cognitive functions critical to phishing detection. The research applies natural language processing methods to determine the features in phishing messages that people encode to their memory during decision-making, and the impact of these features on successful detection. This research also studies and characterizes functions critical to memory processes that impact individual susceptibility to phishing attacks. The challenges of applying cognitive models and reinforcement learning models to provide personalized phishing training experiences are also investigated. Knowledge and methods produced from the activities of this project have implications beyond phishing. They inform related challenges in misinformation detection and threat detection. The long-term goal of this project is to establish a multi-disciplinary research program in security and educate a diverse engineering workforce through inclusive research, educational, and outreach activities.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.