INTL 712 ethical implications of behavioral psychology, and continuous evaluation programs within

Transcript

Excellent. Thank you for that and I'll I'll definitely let you know when to change. Uh thank you guys again uh not only for your time today but uh we're going to talk um a little bit today uh with the EP with the ethical implications of behavioral psychology and continuous evaluation. So interdisciplinary approach uh to where we examine uh the psychological, legal and operational dimensions of the program and uh some policy recommendations, legal discussion, literature discussion um to make sure that our national security efforts uh remain, you know, aligned with our democratic values and human rights. And I really uh that really resonated with what Vince was saying about you know a little bit more the democratic values and human rights and the things that we were talking about earlier.

So [clears throat] to understand uh let's go to the next slide. Okay. So to understand uh you have to understand this in context you know specifically uh with uh continuous evaluation and insider threats. Um as we as we see you know in literature that our insider threats are becoming uh more subtle and more dangerous within inside of the IC. our our traditional uh methods that we use for security clearance investigations and reinvestigations were were no longer sufficient to detect you know those patterns and nuance behaviors of of the potential for insider threats which is where uh CE came into play and of course you know CE being real time and data driven uh is a platform uh that monitors personnel for those types of risks.

So uh with the with the implementation of CE it really represented a paradigm shift in certain security practices uh moving from a reactive to proactive monitoring proactive monitoring but also the shift um from the traditional way to CE caused uh you know it introduced a good amount of significant ethical concerns when predictive behavioral analytics um which are being used to assess uh psychological traits and behaviors. Uh let's move to the next slide. Sorry, [clears throat] it's a little post-nasal. I apologize. So, with predictive behavioral analytics in CE, we're going to dive down into uh you know, the predictive analytics within within the CE systems.

And this is where behavioral psychology enables these systems to be able to proactively uh detect the threats uh through analyzing things like stress indicators, personality traits and even uh things like uh emotional cues. And these models that are used uh that are described actually that are used they they ingest digital footprints, social interactions, behavioral patterns to generate a risk assessment. [gasps] But even though these predictive capabilities allow for early intervention, uh it raises a significant concern uh especially when I was taking a look at this when we consider both the psychological and ethical dimensions. But we'll also have to start with how these models really work uh particularly CE systems that use behavioral data uh to build profiles that estimate risk. So that one will kind of hang that on the back of our mind a little bit.

But um you know in the literature we have uh where uh there's there's a convergence of the utility of these models specifically uh two scholars Caputo uh and Hussein as well as Russell Russo and Torres. Um there's rich argument um in convergence where that predictive uh analytics allow security professionals to intervene uh before a threat materializes. So with that discussion uh it's really seen as a proactive approach to see it's seen as a major advancement over traditional reinvestigation cycles which miss those subtle behavioral shifts. Um but on the opposite side of that there's also divergence in the literature where there is uh ethical and psychological implications uh where they diverge sharply particularly in two cases um in Ajimni and um actually in the American psychological association a recent one that was put out in 2025 where there's this there's this not only caution to the wind but almost like a warning or a beacon uh that the models are inherently probabilistic and they are prone to mclassification. So without human context um the algorithms may misinterpret benign behaviors like emotional withdrawal due to grief, emotional withdrawal due to stress uh and perceiving those as as a threat or a flag.

And really in those two divergences, you see that it leads to false positives, uh, reputational harm and additional psychological distress. And really from a a theoretical standpoint, when you back up a little bit to attribution theory, it becomes really relevant when you paint in the details because attribution theory informs and explains how individuals interpret the causes of behaviors, whether they're internal traits or external circumstances. and predictive models um in CE lack that fine nuance um to distinguish between the causes leading to flawed assessments. So on another note as well, cognitive appraisal theory which also states that you know emotional responses are shaped by individual interpretations of events. Um it's leads to the understanding with the with the with the divergence of literature that the behavioral data alone cannot fully capture uh psychological uh state of mind or the psychological state of that current individual.

The second point where it diverges into literature is that there are there's concerns of psychological linguistics u uh linguistic analysis as sentiment direction but there the tools can identify the emotional states and shifts of tone but they can also there's a high risk I don't want to say high just going to say that there's a risk of um the models oversimplifying human expression. So there's instances that we're all familiar with with sarcasm, you know, even cultural idioms or emotional suppression that are uh misread by algorithms. Um it leads, you know, [snorts] that misreading leads to inaccurate conclusions and bird in 2023 also emphasizes the needs for explanability and validation in those models. you know warning in in the literature there was a there was a real warning that popped out that without transparency you know predictive analytics uh can fail to meet uh the ethical standard or ethical baseline. So the psych uh let's go to the next slide please.

Yes. Um I'm going to prelude this one and then we'll go into that. So with that um you know to address those concerns you know the literature continues to really call for an increase in uh ethical oversight as well as transparency in the algorithms and the continued integration of human judgment into the CE systems and there's also the growing consensus that predictive analysis uh predictive analytics must be guided by rigor scientific rigor and ethical governance. But really that divergence on how much autonomy these systems should have and how much discretion should really be left to be left to the human human analysts. And the same thing as we go you know through into the ethical concerns and you know even from from my paper discussion you know with the with the literature in this when we take a look at the ethical landscape surrounding predictive behavioral analytics uh specifically in the case of continuous evaluation the literature is really rich u but [snorts] also with that richness comes division and it's a complex it's really a complex reflection of the tension between the technical innovations that we have with CE and the continued need to protect individual rights and even to the extent of protecting um psychological well-being for the individual and even in the heart at the heart of that debate that there's a rich ethical debate that you know the use of those probabilistic models in CE to assess human behavior that's really where those models while they're powerful they're definitely prone to bias and they and they lack transparency.

Um you know the data sets that inform those models range everything from financial records to social media activity um to infer psychological traits and uh to predict future behavior. So in the in that in other dis in the discussions as well there's also convergence that if those systems if they're left unchecked can lead to again mclassification uh stigmatization and as well as the erosion of trust. So without really rigorous validation and contextual integr u contextual interpretation rather uh these systems are these systems may disproportionately flag individuals based on things like uh as simple as race, gender and even socioeconomic status. So where those convergences happen, there's also divergence as well in the ethical permissibility place of where the use of psychological data inside a predictive analytics occur. Um this debate of whether or not that um any attempt to quantify or predict human behavior in a high stakes environment even like the IC is riddled inherently and it's in its problem inherently problematic.

um you know the the the citations you know the the sources and literature they argue about autonomy and uh consent uh as well as the potential for psychological harm and those are really grounded in uh dental ethics as well as uh the prioritization and the protection of individual rights over ones that would be uh utilitarian outcomes. So, um, can we move to the next slide? Let me make sure I'm moving on task here. Um, let's stay back on the previous one. Sorry about that. My apologies.

[clears throat and cough] But the also with the um the the advocacy of hold on, let me back up. I just had a brain freeze. um with the where the literature also is informing us is that there is also a consequentialist view. So that is where we see this rich argument suggesting that things like uh predictive behavioral analytics are ethically justified if they prevent harm uh and protect national security. And these scholars really come together to advocate for the development of uh robust uh we'll say oversight mechanisms and transparency protocols as well as ethical review boards to mitigate those types of risks uh with the right safeguards in literature.

It's the CE systems can be more effective and ethically sound and at the same time you know psychological theory also informs informs that debate. helps kind of paint in the lines or color in the picture for us. Um, in social identity theory, we see for example the how uh we highlight how surveillance can affect group dynamics and individual self-concept. um when individuals specifically feel they're being unfairly monitored or watched we'll say or singled out it leads to the in-group outgroup tension as well as I we'll call it I don't want to call it degradation of u we'll call it degradation of uh cohesion but main we'll call it a reduced organizational cohesion if that makes sense. So with that um you know the CE system still they operate without clear communication or consent that undermines certain psychological lead psychological needs as well as uh leading to disengagement and even resistance.

Okay. Uh let's move to the next slide. So for the legal frameworks and oversight um as well we'll take a look um there's legal frameworks that govern CE um as well as the the backbone of how CE was you know is authorized implemented uh regulated across the IC and really at the core um are the statutory and executive directives in EO 13467 and seed 6 which process which they outline the operational standards for CE and the types of data that can be collected. Uh those types specifically are criminal records, financial records, uh financial anomalies, uh foreign travel and even social media activities. And really these legal instruments provide the legitimacy of CE but it also raises significant concerns about privacy um significant concerns about due process and the scope of surveillance.

really in the privacy act in 1974 when you really take a dive down into it with this with a with a CE lens you take a look at the relevance since it governs you know the collection maintenance and dissemination of PII um by each federal agency and its mandate is pretty transparent as far as uh consent and the right to access the data but CES often we'll say often operate in ways that challenge those principles of the privacy act. Um especially when we take a look at uh behavioral data that is being inferred rather than being explicitly provided in the literature. There's also convergence um with a congressional research group in Burke and White that's uh that agree that national security operations have to be conducted uh within the bounds of constitutional protection. But there's also divergence of where the current legal frameworks are necessary or are sufficient already uh to address the complexities and things like the privacy act and there's five USC 522 um that offer you know that state that there's adequate safeguards already and enforced. So um you know if with these safeguards that if they're properly enforced in the literature you know it wouldn't contend with the idea that the statutes are uh you know that they are you know quote unquote designed now to handle the nuances of of algorithmic decisionm and psychological profiling as an example.

But really from a legal theory perspective, it's really taking a look at uh you know at the statutes and also that convergence and divergence of literature. You know where the ongoing debate is legally uh you know it is specifically where there's emphasis in in strict adherence to codified law. So with that there may they may be legally authorized but are they ethically justified? And that's that's still the the ongoing debate that where we see [snorts] you know gaps in oversight you know where CE is is subject to you know things like internal review but ODNI and even the PCL um the enforcement mechanisms are really limited. So there's still that growing call for independent review boards and reporting requirements and judicial oversight to ensure accountability. So there's a lot of richness to churn in that soil um in that in that area especially in the in the legal front but also you know it's one of those questions is how much is enough as well.

Um let's move on to the next slide. So the impacts on trust and mental health. Uh this is a this is a big one um for individuals. We're all human beings participating in the intelligence community as well as the the CE program as part of our you know ongoing background profession and one of the profound areas that we see on individuals um you know we've even seen this you know over time but is how that feeling of constant surveillance false you know really fosters a culture of suspicion which leads to added stress and anxiety. as well as in the worst cases Jos jobs had dissatisfaction where those inferred traits are flagged by CE but they are not they're not concrete actions.

So with that with those unintentional flaggings or intentional flaggings it really erodess morale and undermines trust in the workforce and the ethical management of the CE systems. uh they you know it's essential to mitigate these harms to foster a supportive IC um by promoting both transparency and fairness but CE should become a tool where that it not only protects security but also upholds the psychological well-being the health profile of personnel that are that are enrolled in it. Uh the next slide please. So some policy recommendations you know based on on those findings in in review of the literature. Um there are some policy recommendations um on the need again for ethical review boards that composed of you know psychologists even ethicists and legal scholars and civil liberty advocates to mandate transparency in you know that that type of decision making where algorithms are involved and also to promote the uh individual the opportunity with the right to appeal and those can come you know those protocols can come from you know psychological data governance that can be developed with individuals like we had mentioned on one of the early slides with the American Psychological Association's ethical standards.

Um, and that people that operate as analysts for CE receive training, adequate training and ethical decision making regarding mental health sensitivity and making sure that the program is both ethically sound and that it's ethically effective. All right. Uh, next slide, please. And we'll take a look here too um at the at the psychological implications for those that are underneath of the CE umbrella. And really one of the most human- centered dimensions of the of the of the CE framework in totality.

Um taking a look at this in extensive detail. Um there is a psychological toll of being under constant surveillance. uh this this is not a a theoretical it's it's a deeply personal thing and it's operationally significant you from a psychological standpoint uh that CE environment or umbrella it introduces that persistent state of hypervigilance but it also uh we also come to understand that it also aligns with uh things and principles in uh general ad adaptation syndrome which describes how prolonged exposure to these type of ex to these types of stressors rather um lead to exhaustion and really breakdown and in in the CE environment it's not just a single you know a one-time thing or a single event and it and it's ongoing it's an invisible presence of that surveillance just ongoing over and over and over again. So with the uh with that information in mind, we had to really balance the harms and benefits um of the CE system uh with psychological resilience in mind of you know of our workforce you know over the number of years and and that means really integrating not only mental health support but also ensuring the fairness of of the algorithmic uh portion of this and that clear channels uh exist for redress address and also for the opportunity to appeal. But also it means for those again we had talked us talked about a little bit about the training of the CE analysts to recognize the difference behavior anomalies and psychological distress and then be able to respond uh both with empathy and discretion.

Um next slide please. Uh yeah, there we go. Thank you. So, we talked about the individual a little bit and we're going to dive back into the harms and benefits the government as well uh for the uh the for the CE system. uh that the consequences the harms and benefits to the government as far as uh you know CE have it clearly offers an operational advantage and they enable real-time detection of insider threats you know reduce resilience on outdated investigation cycles and align with uh certain modern risk management [clears throat and cough] excuse me risk management strategies where we see in in the era today um that these or we have an increase you know we're seeing sophisticated adversaries on the increase as well as asymmetric threats and CE does definitely provide a proactive mechanism to safeguard that information sensitive information as well as national security and literature converges you know in that as well um about the behavioral shifts the financial irrele irregularities and um the social patterns that that definitely indicate risk And the integration of behavioral analytics into the CE platform is a strategic enhancement because it allows agencies to respond both with agility and precision.

But again you know on the divergence side we see that uh you know a lot of that items that come out of there is uh you know the benefits to operational security comes at the at the risk of the culture of suspicion and the feeling of being constantly scrutinized or under even undervalued. um reduces the you know reduces employee performance as well as increased turnover and undermines uh the uh it undermines the uh the very security that CE uh aims to protect. Okay, next slide. Um on this conclusion for future directions uh the of course the CE program today you know it the CE programs offer significant potential um regarding insider threats and detection uh for national security purposes but it does it does not need it should not need it should not come at the expense of civil liberties and psychological well-being specifically the cost of civil liberties and the cost of psychological well-being. uh with that uh you know future our future research should really take a look at the long-term psychological impacts of CE and continuing to examine not only internally here in the US but also how other nations are balancing the security and ethics uh the security and ethics in similar programs that they would have maybe across the five eyes or even 14 nations and uh to see how they're balancing both the ethics in a you in similar programs.

So they may not call it CE, they may call it something else or it may be a fragment or a compartment of of multiple things. All right. And I think that comes to the end. The rest is all the references for all the interdisciplinary discussion. That's more than welcome for your reading pleasure.

You're more than welcome to review any of those. And I appreciate the time.