Knowing Neurons
Science PolicyNeuropolicy Paper Competition 2023Neuroscience Technologies

The Truth is in Your Eyes: Regulating the Covert Use of Pupil Biofeedback

SECOND PLACE WINNER!

Policy Memo by Erin Morrow

INTRODUCTION

Within the last decade, consumer technologies have become increasingly capable at recording and analyzing physiological data. For instance, smartwatches can now monitor real-time heart rate, detect gait changes, and even conduct electrocardiograms, making recommendations based on these measures. These developments follow the meteoric rise of precision medicine, which aims to improve quality of life and medical care by collecting a massive amount of health data from each individual (see Jeong, Bychkov, & Searson, 2019). However, a notable movement in extended reality (XR) technology has seen companies begin collecting, storing, and using biometrics for purposes unrelated to health (Plopski et al., 2022). 

Apple can predict how engaged a user is with a displayed item and optimize content accordingly.

Announced in June 2023, Apple’s Vision Pro headset will utilize eye tracking to shape how users interact with an XR interface. Indeed, recent Apple patents outline a mechanism by which their device can track pupil size in order to infer a user’s attentional state, and – based on this data – change how content is displayed. For example, if a user’s pupil typically dilates more in response to a certain color, the device will choose to use that color more often (Crispin, Yildiz, & Mulliken, 2020, 2021). In this way, Apple can predict how engaged a user is with a displayed item and optimize content accordingly. Although this feature was likely designed to improve user experience, it also poses an unprecedented risk to consumer autonomy. That is, users may be unaware that sensitive biometric data are being used to actively manipulate content and potentially drive third-party profit. For instance, the device may ‘learn’ that a certain typeface is more effective at eliciting pupil dilation to a certain advertisement and continue to use this typeface to encourage purchases. Yet, the nearly 10-minute extended trailer for Vision Pro includes no mention of biofeedback capabilities, nor how this information might be shared. Meanwhile, some emphasize that the Vision Pro could act as a rudimentary predecessor to consumer brain-computer interfaces (BCIs), or machines that can be operated entirely using neural signals (Crispin, 2023; VK, 2023). Indeed, possibilities for combining XR and BCI technology have already begun to be explored (Arpaia et al., 2022). Although the Vision Pro does not reach this level of function – relying upon a variety of non-neural inputs – it is critical to anticipate and regulate the deceptive use of biofeedback technology as innovation accelerates.

ETHICAL IMPLICATIONS: SIMILARITIES TO AND DIFFERENCES FROM EXISTING APPROACHES

Using real-time eye tracking to gauge attention raises unique ethical concerns compared to mainstream techniques that measure digital engagement. Namely, pupil data offers a more direct window into human mental states. Pupil size indexes several aspects of neural processing, including cognitive load/effort (Wierda et al., 2012), attention (Hong, Walz, & Sajda, 2014; Kang, Huffer, & Wheatley, 2014), and arousal (Bradley et al., 2008; Clewett, Gasser, & Davachi, 2020). More specifically, brain regions that support these processes – such as the locus coeruleus – activate autonomic pathways that control smooth muscles in the iris, directing the pupil to expand or contract (Bradley et al., 2008; Liu et al., 2017; Steinhauer et al., 2004). Given the accessibility of the eye within this network, pupillometry has been exploited as a relatively inexpensive, noninvasive method of inferring psychological states since the 1960s (Laeng, Sirois, & Gredebäck, 2012). In contrast, more indirect metrics – such as the time one spends on a social media post – can fail to capture true user engagement. For example, current algorithms cannot discriminate between a user that attentively watches video content and another that leaves a video open in the background while performing a separate task. By eliminating this ambiguity, pupil biofeedback provides tech companies with more accurate estimates of users’ moment-to-moment reactions to content. Given evidence that pupil dilation may relate to consumer preference (Ramsøy et al., 2017), devices could then alter content based on this information to influence the user’s decision-making, potentially without their knowledge or consent.

pupil data offers a more direct window into human mental states

These risks extend an existing debate on the ethical implications of neuromarketing. Neuromarketing refers to the use of neuroscience techniques to better understand consumer behavior – that is, how consumer preferences can be represented, decoded, and influenced (Murphy, Illes, & Reiner, 2008; Stanton, Sinnott-Armstrong, & Huettel, 2017). Previous work has used a wide range of methodologies – including functional magnetic resonance imaging (fMRI), electroencephalography (EEG), and eye-tracking – to investigate how the brain responds to different products and advertisements (see Stanton, Sinnott-Armstrong, & Huettel, 2017 for review). These techniques provide the appeal of hard science to marketing research by swapping traditional focus groups and self-report measures for ‘objective’ neural data. However, if this neural signal is manipulated to achieve some goal, one’s cognitive liberty – or the right to “think one’s own thoughts” (pp. 2301; Rainey et al., 2021) – could be considered at risk (Rainey et al., 2021; see Farahany, 2018; Sententia, 2006). Although the potential loss of cognitive liberty is sometimes overstated given the current state of the science (Stanton, Sinott-Armstrong, & Huettel, 2017), stealth neuromarketing constitutes a unique challenge by influencing decision-making without an individual’s awareness. Indeed, the public seems particularly sensitive to manipulations of consumer behavior that are unwitting (Murphy, Illes, & Reiner, 2008; see Karremans et al., 2006; Moore, 1982). The covert use of pupil biofeedback seems to fit squarely in this category; in this case, users provide sensitive data about their own mental states that is unknowingly used to tune content and maximize engagement. In order to safeguard consumer autonomy, it is imperative to ensure users are aware of the ways in which their biometrics influence their experience.

POLICY RECOMMENDATIONS

Policy Option 1: No change. Although maintaining the legislative status quo is a possibility, this option leaves little recourse for potential threats to autonomy. With a company as large and ubiquitous as Apple embracing the eye-tracking space – following industry giants such as Microsoft (Stellmach et al., 2023) and Google (Calderone, 2015) – pupil biofeedback could soon be in the hands of millions. Although the $3,500 price tag is likely too high for the average iPhone user, future iterations are expected to increase affordability and access. Although failing to go as far as the Neuralink chip –which reads electrical signals directly from the brain and must be surgically implanted – the Vision Pro represents a cheaper, more accessible, and less invasive approach that is likely to see broader implementation. Allowing this innovation to proceed unregulated risks deception and exploitation by third-party advertisers that can access sensitive information about users’ mental states and use it to drive business. Given our positionality at a turning point in consumer neurotechnology development, we must consider policy alternatives.

This novel access permits companies to manipulate content based on information that is both sensitive and largely involuntary

Policy Option 2: Amend the definition of deception in the FTC Act. An alternative course of action is to expand the definition of deception in the Federal Trade Commission (FTC) Act. The FTC Act allows the forenamed agency to bring action against companies that deceive consumers (Federal Trade Commission Act, 2018) – that is, “[represent], [omit], or [perform a] practice that is likely to mislead” (pp. 1; FTC Statement on Deception, 1984). Notably, the FTC has previously investigated the use of deception in online behavioral advertising, in which companies personalize advertisements for an individual user based on their digital activity (Federal Trade Commission, 2009). Although the covert use of pupil biofeedback may fall under this category and thus be subject to existing FTC guidelines, it likely represents a new and unique risk by incorporating real-time biometrics. As explained earlier, pupil data assess mental states more directly than other measures of digital engagement. This novel access permits companies to manipulate content based on information that is both sensitive and largely involuntary (Loewenfeld, 1993). Therefore, broadening the definition of deception to include the covert use of biometric data would be beneficial. Section 57a of the FTC Act permits the creation of these new guidelines (Federal Trade Commission Act, 2018); indeed, recent legislation has prohibited deceptive practices related to COVID-19 (COVID-19 Consumer Protection Act of the 2021 Consolidated Appropriations Act, 2020) and online ticket sales (Better Online Ticket Sales Act, 2016), for example. Thus, precedent exists to secure additional protections for consumer autonomy.

Policy Option 3: New legislation to address covert manipulation based on neural data. Given the rapid expansion of consumer neurotechnology in recent years, new legislation altogether may be warranted. Indeed, emerging products – such as neurostimulation devices, wearables, brain-computer interfaces, and other extended reality technology – likely fall outside of the purview of existing guidance (Kreitmar, 2019; Wexler & Reiner, 2019) and could benefit from their own regulatory framework. Thus, a second option is to draft a new bill regulating any covert manipulation of content based on neural data. This manipulation could extend beyond advertisement and other marketing tactics. For example, imagine a future wearable device that could assess an individual’s mood and subsequently alter the news stories or social media posts they are shown without their knowledge. This scenario is reminiscent of the controversial ‘emotional contagion’ experiment, in which experimenters covertly manipulated the news feeds of over 500,000 Facebook users to affect their mood (Kramer, Guillory, & Hancock, 2014). Although complaints were brought to the FTC after this study was published (Gibbs, 2014; Rotenberg & Horwitz, 2014), the use of neurotechnology to produce similar effects has a much smaller precedent. Therefore, new legislation could be tailored to require companies to disclose and explain how neural data is used to affect user experience. Ideally, these disclosures should be provided 1) outside of lengthy terms and conditions, 2) in easy-to-understand language (i.e., that avoids unnecessary jargon), and 3) in a location that is easily and continuously accessible to the user. By prioritizing consumer education, requirements such as these surpass more general guidelines for non-deceptive practices.     

Although the covert use of pupil biofeedback may fall under this category and thus be subject to existing FTC guidelines, it likely represents a new and unique risk by incorporating real-time biometrics

CONCLUSION     

As neural data becomes an increasing target for consumer devices, it is essential to preserve autonomy. Using Apple’s Vision Pro as a case example, I describe how eye-tracking offers an inexpensive and accessible way for companies to gather information about users’ mental states. However, it may be unclear to the user how their pupil biometrics are being used to shape future content. Here, I briefly outline three potential policy alternatives, including: 1) no change, 2) amending the definition of deception in the FTC act, and 3) drafting new legislation to address any covert manipulation based on neural data and require clear disclosure practices. These options offer a starting point for policymakers to consider how to protect consumer autonomy in the age of new forms of sensitive data.

~~~

Written by Erin Morrow
Edited by Zoë Dobler and Kayla Lim

~~~

Become a Patron!

   

References

Arpaia, P., Callegaro, L., Cultrera, A., Esposito, A., & Ortolano, M. (2021). Metrological characterization of consumer-grade equipment for wearable brain–computer interfaces and extended reality. IEEE Transactions on Instrumentation and Measurement71, 1-9. https://doi.org/10.1109/TIM.2021.3127650

Better Online Ticket Sales Act, 15 U.S.C. § 45c (2016). https://www.ftc.gov/legal-library/browse/statutes/better-online-ticket-sales-act

Bradley, M. M., Miccoli, L., Escrig, M. A., & Lang, P. J. (2008). The pupil as a measure of emotional arousal and autonomic activation. Psychophysiology45(4), 602-607. https://doi.org/10.1111/j.1469-8986.2008.00654.x

Calderone, J. (2015, January 1). Eye tracking in Google Glass: A window into the soul?. 

Scientific American. https://www.scientificamerican.com/article/eye-tracking-in-google-glass-a-window-into-the-soul/

Clewett, D., Gasser, C., & Davachi, L. (2020). Pupil-linked arousal signals track the temporal organization of events in memory. Nature Communications11(1), 4007. https://doi.org/10.1038/s41467-020-17851-9

COVID-19 Consumer Protection Act of the 2021 Consolidated Appropriations Act, Pub. L. No. 116-260, 134 Stat. 1182, Division FF, Title XIV, § 1401 (2020). https://www.ftc.gov/legal-library/browse/statutes/covid-19-consumer-protection-act-2021-consolidated-appropriations-act

Crispin, S. R. [@sterlingcrispin]. (2023, June 5). I spent 10% of my life contributing to the development of the #VisionPro while I worked at Apple as a… [Image attached] [Tweet]. Twitter. https://twitter.com/sterlingcrispin/status/1665792422914453506

Crispin, S. R., Yildiz, I. B., & Mulliken, G. H. (2020). Biofeedback method of modulating digital content to invoke greater pupil radius response (European Patent Application No. EP3891585A1). European Patent Office. https://patents.google.com/patent/EP3891585A1/en

Crispin, S. R., Yildiz, I. B., & Mulliken, G. H. (2021). Biofeedback method of modulating digital content to invoke greater pupil radius response (U.S. Patent Application No. US20210349536A1). U.S. Patent and Trademark Office. https://patents.google.com/patent/US20210349536A1/en

Farahany, N. A. (2012). Incriminating thoughts. Stanford Law Review, 64(2), 351-408.

Federal Trade Commission (2009). Self-Regulatory Principles for Online Behavioral Advertising. https://www.ftc.gov/sites/default/files/documents/reports/federal-trade-commission-staff-report-self-regulatory-principles-online-behavioral-advertising/p085400behavadreport.pdf

Federal Trade Commission Act, 15 U.S.C. §§ 41-58 (2018). https://www.ftc.gov/legal-library/browse/statutes/federal-trade-commission-act

FTC Statement on Deception, 103 F.T.C. 174, 175 (1984) (appended to Cliffdale Assocs., Inc., 103 F.T.C. 110 (1984)) (“Deception Policy Statement”). https://www.ftc.gov/system/files/documents/public_statements/410531/831014deceptionstmt.pdf

Gibbs, S. (2014, July 4). Privacy watchdog files complaint over Facebook emotion experiment. 

The Guardian. https://www.theguardian.com/technology/2014/jul/04/privacy-watchdog-files-complaint-over-facebook-emotion-experiment

Hong, L., Walz, J. M., & Sajda, P. (2014). Your eyes give you away: prestimulus changes in pupil diameter correlate with poststimulus task-related EEG dynamics. PloS One9(3), e91321. https://doi.org/10.1371/journal.pone.0091321

Jeong, I., Bychkov, D., & Searson, P. C. (2018). Wearable devices for precision medicine and health state monitoring. IEEE Transactions on Biomedical Engineering, 66(5), 1242-1258. https://doi.org/10.1109/TBME.2018.2871638

Kang, O. E., Huffer, K. E., & Wheatley, T. P. (2014). Pupil dilation dynamics track attention to high-level information. PloS One9(8), e102463. https://doi.org/10.1371/journal.pone.0102463

Karremans, J. C., Stroebe, W., & Claus, J. (2006). Beyond Vicary’s fantasies: The impact of subliminal priming and brand choice. Journal of Experimental Social Psychology42(6), 792-798. https://doi.org/10.1016/j.jesp.2005.12.002

Kramer, A. D., Guillory, J. E., & Hancock, J. T. (2014). Experimental evidence of massive-scale emotional contagion through social networks. Proceedings of the National Academy of Sciences of the United States of America111(24), 8788. https://doi.org/ 10.1073/pnas.1320040111

Kreitmair, K. V. (2019). Dimensions of ethical direct-to-consumer neurotechnologies. AJOB Neuroscience10(4), 152-166. https://doi.org/10.1080/21507740.2019.1665120

Laeng, B., Sirois, S., & Gredebäck, G. (2012). Pupillometry: A window to the preconscious?. Perspectives on Psychological Science7(1), 18-27. https://doi.org/10.1177/1745691611427305

Liu, Y., Rodenkirch, C., Moskowitz, N., Schriver, B., & Wang, Q. (2017). Dynamic lateralization of pupil dilation evoked by locus coeruleus activation results from sympathetic, not parasympathetic, contributions. Cell Reports20(13), 3099-3112. https://doi.org/10.1016/j.celrep.2017.08.094

Loewenfeld, I.E. (1993). The Pupil: Anatomy, Physiology, and Clinical Applications. Wiley-Blackwell. 

Moore, T. E. (1982). Subliminal advertising: What you see is what you get. Journal of Marketing46(2), 38-47. https://doi.org/10.1177/002224298204600205

Murphy, E. R., Illes, J., & Reiner, P. B. (2008). Neuroethics of neuromarketing. Journal of Consumer Behaviour: An International Research Review7(4‐5), 293-302. https://doi.org/10.1002/cb.252

Plopski, A., Hirzle, T., Norouzi, N., Qian, L., Bruder, G., & Langlotz, T. (2022). The eye in extended reality: A survey on gaze interaction and eye tracking in head-worn extended reality. ACM Computing Surveys (CSUR)55(3), 1-39. https://doi.org/10.1145/3491207

Rainey, S., Martin, S., Christen, A., Mégevand, P., & Fourneret, E. (2020). Brain recording, mind-reading, and neurotechnology: ethical issues from consumer devices to brain-based speech decoding. Science and Engineering Ethics26, 2295-2311. https://doi.org/10.1007/s11948-020-00218-0

Ramsøy, T. Z., Jacobsen, C., Friis-Olivarius, M., Bagdziunaite, D., & Skov, M. (2017). Predictive value of body posture and pupil dilation in assessing consumer preference and choice. Journal of Neuroscience, Psychology, and Economics10(2-3), 95. https://doi.org/10.1037/npe0000073

Rotenberg, M., & Horwitz, J. (2014). Complaint, Request for Investigation, Injunction, and Other Relief. Electronic Privacy Information Center. Complaint before the Federal Trade Commission. https://epic.org/wp-content/uploads/privacy/ftc/facebook/Facebook-Study-Complaint.pdf 

Sententia, W. (2004). Neuroethical considerations: cognitive liberty and converging technologies for improving human cognition. Annals of the New York Academy of Sciences1013(1), 221-228. https://doi.org/10.1196/annals.1305.014

Stanton, S. J., Sinnott-Armstrong, W., & Huettel, S. A. (2017). Neuromarketing: Ethical implications of its use and potential misuse. Journal of Business Ethics144, 799-811.  https://doi.org/10.1007/s10551-016-3059-0

Steinhauer, S. R., Siegle, G. J., Condray, R., & Pless, M. (2004). Sympathetic and parasympathetic innervation of pupillary dilation during sustained processing. International Journal of Psychophysiology52(1), 77-86. https://doi.org/10.1016/j.ijpsycho.2003.12.005

Stellmach, S., McClister, C., Zhang, B., Buck, A., Tieto, V., Jahiu, D., v-chmccl, Arya, H., Sharkey, K., Ferrone, H., Eveleigh, K., magdavuko, Coulter, D. (2023, March 2). Eye tracking on HoloLens 2. Microsoft Learn. https://learn.microsoft.com/en-us/windows/mixed-reality/design/eye-tracking

VK, A. (2023, June 14). Apple’s Vision Pro might be the world’s first brain chip interface (BCI). Analytics India Magazine. https://analyticsindiamag.com/apples-vision-pro-might-be-the-worlds-first-brain-chip-interface-bci/

Wexler, A., & Reiner, P. B. (2019). Oversight of direct-to-consumer neurotechnologies. Science363(6424), 234-235. https://doi.org/10.1126/science.aav0223

Wierda, S. M., van Rijn, H., Taatgen, N. A., & Martens, S. (2012). Pupil dilation deconvolution reveals the dynamics of attention at high temporal resolution. Proceedings of the National Academy of Sciences, 109(22), 8456-8460. https://doi.org/10.1073/pnas.1201858109

Author

  • Erin Morrow

    Erin is a PhD student in cognitive psychology at the University of California, Los Angeles. She uses behavior, pupillometry, and neuroimaging techniques to investigate how emotional arousal affects the structure and content of human memory. Erin also has strong interests in brain and mental health policy, neuroethics, and science communication. She currently serves on the Student/Postdoc Committee of the International Neuroethics Society. Previously, she earned a B.S. in Neuroscience and Behavioral Biology from Emory University. Outside of the lab, Erin enjoys music, graphic design, and calligraphy.

Erin Morrow

Erin is a PhD student in cognitive psychology at the University of California, Los Angeles. She uses behavior, pupillometry, and neuroimaging techniques to investigate how emotional arousal affects the structure and content of human memory. Erin also has strong interests in brain and mental health policy, neuroethics, and science communication. She currently serves on the Student/Postdoc Committee of the International Neuroethics Society. Previously, she earned a B.S. in Neuroscience and Behavioral Biology from Emory University. Outside of the lab, Erin enjoys music, graphic design, and calligraphy.