How to address privacy, ethical and regulatory issues: Examples in cognitive enhancement, depression and ADHD

Editor's Notes

• #10: Where neuroscientists find themselves pondering questions that have perennially plagued philosophers Conceptual framework for self: neuro-essentialism Mapping the brain’s networks = a feat no less equated to mapping human identity “the wiring that makes us who we are” (Seung, 2012).”
• #11: Where neuroscientists find themselves pondering questions that have perennially plagued philosophers Conceptual framework for self: neuro-essentialism Mapping the brain’s networks = a feat no less equated to mapping human identity “the wiring that makes us who we are” (Seung, 2012).”
• #12: Where neuroscientists find themselves pondering questions that have perennially plagued philosophers Conceptual framework for self: neuro-essentialism Mapping the brain’s networks = a feat no less equated to mapping human identity “the wiring that makes us who we are” (Seung, 2012).”
• #15: “pervasive neurotechnology applications include brain-computer interfaces (BCIs) for device control or real-time neuromonitoring, neurosensor-based vehicle operator systems, cognitive training tools, electrical and magnetic brain stimulation, wearables for mental wellbeing, and virtual reality systems.” Ienca and Andorno, 2019
• #17: Catastrophic mistakes could include impairing agency.
• #18: “Research shows that the combination of neuroimaging technology and artificial intelligence allows to “read” correlates of mental states including hidden intentions, visual experiences or even dreams with an increasing degree of accuracy and resolution.”—Ienca 2019; Forbes, 2009 Wolpe
• #19: “A recent experiment recorded functional MRI (fMRI) data from volunteers while they were watching video clips. From their brain activity, computers were able to partially reconstruct some of the images that the volunteers saw. That is, from the information gathered by the brain scans it was possible, to a certain extent, to watch what they were watching. This is not the same as watching what was "in their minds," so to speak, but watching what their minds were watching. Still, quite an amazing achievement.” npr, 2011
• #20: Privacy. Research is dependent on participant trust which is a precious commodity. Opportunity for continuous (vs. episodic) recording in ‘patient’s ecosystem’. As I and others have said, data can be collected and easily passively and actively input throughout the patients day. And even with the best intentions, this raises issues of privacy.   Privacy risks. As with genomics, wide scale use of neuroscience data could lead to re-identifiability of participants.  “Neuroimaging data coupled with layers of descriptive meta-data may mean that “sulcal and gyral fingerprints” or even BOLD activity patterns could compromise participant confidentiality, even when the data has been “anonymized” in ordinary ways. “  -Choudhury, 2014 pg. 5. But imagine in this instance where we can cross link GPS, mood, purchases, social media feeds, AND brain activity.   Even having a GPS stamp and two expensive purchases from a typical store, people can be re-identified, so what with be the potential for maintaining de-identification with brain recording data?   Importantly, Neural data *may* feel qualitatively different because the types of information that is inferred. We know from some interviews with participants about BCI and with some researchers that privacy is an issue of concern.   From GNS paper: The public has expressed some privacy concerns as collected via outreach seminars in Europe which focus on brain computer interfaces or BCI (Jebari and Hansson, 2013) about the acquisition of private information and similar concerns have been discussed amongst surveyed scientists in Japan, for example (Higashijima et al., 2011); and in the scholarly literature around concerns of appropriate informed consent with BCI research (Klein and Ojemann, 2016). Not unlike genomics, in the case of increasingly large “big data” set analysis in neuroscience, it may be the case that maintaining de-identification (i.e. stripping information that will link data back to the individual) is a lost cause (Choudhury et al., 2014). The more likely possibility with large data sets or small numbers of initial participants is re-identification. In such cases, consent for participation in studies may need to acknowledge these realities. Finally, we show that even data sets that provide coarse information at any or all of the dimensions provide little anonymity and that women are more reidentifiable than men in credit card metadata. These metadata are generated by our use of technology and, hence, may reveal a lot about an individual (16, 17). Making these data sets broadly available, therefore, requires solid quantitative guarantees on the risk of reidentification. A data set’s lack of names, home addresses, phone numbers, or other obvious identifiers [such as required, for instance, under the U.S. personally identifiable information (PII) “specific-types” approach (18)], does not make it anonymous nor safe to release to the public and to third parties.
• #22: New unanticipated/legacy use make predicting risk challenging. Beyond the clinic might predictive information go to employers, insurance agencies, etc.? What new kinds of information may be deduced/disclosed in analysis? With big data, might we revisit meaning of privacy anyway? Privacy intrusions and re-identification may become an inevitable norm. Harms extend beyond the participant with big data. One can experience networked harms from afar. Perrhaps a better discussion is how we do not want information to be used, rather than guarding the flow of information such as for insurance or to employers?   In general, online data collection is happening in many ways unbeknownst to the consumer. For example, apps that ask for access to stored data on devices to work, not only access private information like your contacts, calendar information, and your device’s unique id, but also can even save content on your device. Data is often shared actively and passively. Actively though mechanisms like social media, but also passively through connections of app families such as logging into apps via Facebook or Gmail. App partners may share data with their partners and so on. Finally, data can be combined for new uses beyond what the user expected. The app’s networks may tap into large entities like Facebook, Twitter, or Apple and can amplify consumer-generated data. Even phone numbers can link to wide sources like users social networking sites and users credit card. In the end, re-identification is a strong possibility if not inevitable. We also see harms move beyond an individual. we see “network harms” where harms can affect non-participants in studies by virtue of being in a network.
• #23: New unanticipated/legacy use make predicting risk challenging. Beyond the clinic might predictive information go to employers, insurance agencies, etc.? What new kinds of information may be deduced/disclosed in analysis? With big data, might we revisit meaning of privacy anyway? Privacy intrusions and re-identification may become an inevitable norm. Harms extend beyond the participant with big data. One can experience networked harms from afar. Perrhaps a better discussion is how we do not want information to be used, rather than guarding the flow of information such as for insurance or to employers?   In general, online data collection is happening in many ways unbeknownst to the consumer. For example, apps that ask for access to stored data on devices to work, not only access private information like your contacts, calendar information, and your device’s unique id, but also can even save content on your device. Data is often shared actively and passively. Actively though mechanisms like social media, but also passively through connections of app families such as logging into apps via Facebook or Gmail. App partners may share data with their partners and so on. Finally, data can be combined for new uses beyond what the user expected. The app’s networks may tap into large entities like Facebook, Twitter, or Apple and can amplify consumer-generated data. Even phone numbers can link to wide sources like users social networking sites and users credit card. In the end, re-identification is a strong possibility if not inevitable. We also see harms move beyond an individual. we see “network harms” where harms can affect non-participants in studies by virtue of being in a network.
• #25: “Research shows that the combination of neuroimaging technology and artificial intelligence allows to “read” correlates of mental states including hidden intentions, visual experiences or even dreams with an increasing degree of accuracy and resolution.”—Ienca 2019; Forbes, 2009 Wolpe
• #28: Confucianism is a biological, social, and metaphysical reality. ”Family reflects the deep structure of the universe namely a union of the universe, yin and yang..[and] carries a profound sense of necessity and normativity for human life Finally, I’d like to mention that out of Confucianism and many of the religions I’ve described like Buddhism, relational identities is a fundamental belief about construction of self and one’s movement throughout the world. Preliminary work in China to explore public values about cognitive enhancement have shown that while Westerner typically cite concerns of violations of authenticity and integrity for the self Taiwan and Chinese populations cite concerns about fairness, whether everyone or who will have access reflecting concerns about the relational self. Both equally cite concerns of safety.   Of note is that the relational self is a feature of many collectivist cultures found in East Asia and much of the world. As an aside, 70% of the world’s population are collectivist. We might ask, How might regulation and development differ across differences in cultural values, particularly as we think about broader uses of many of the neurotechnologies being developed now that will often go beyond the lab and beyond the clinic into commercial and legal domains. Core of American bioethics is protecting the autonomy of the individual and many of concerns expressed with these technoloiges 70% worlds populations are collectivist Differing views on moral decision making
• #30: ALSO INS?
• #35: Why is it important to think about oversight here?