You are on page 1of 4

Neuron

NeuroView

Internet-Based Brain Training Games,


Citizen Scientists, and Big Data: Ethical Issues
in Unprecedented Virtual Territories
Ryan H. Purcell1,* and Karen S. Rommelfanger2,3,4,*
1Department of Pharmacology
2Department of Neurology
3Department of Psychiatry and Behavioral Sciences
4Center for Ethics Neuroethics Program

Emory University, 1531 Dickey Drive, Atlanta, GA 30322, USA


*Correspondence: ryan.purcell@emory.edu (R.H.P.), krommel@emory.edu (K.S.R.)
http://dx.doi.org/10.1016/j.neuron.2015.03.044

Internet brain training programs, where consumers serve as both subjects and funders of the research, repre-
sent the closest engagement many individuals have with neuroscience. Safeguards are needed to protect
participants’ privacy and the evolving scientific enterprise of big data.

Brain Training: A New Frontier of ment (Alloway and Alloway, 2010). WM training task (Duckworth et al., 2011).
Cognitive Enhancement? training has also been associated with Clearly, there are many factors involved
For a consumer base of tens of millions neurophysiological changes in brain that may complicate efforts to definitively
worldwide, brain training programs activation patterns, structural changes, demonstrate the effectiveness of BTPs.
(BTPs) offer the enjoyment of a video and alterations in dopamine function Nonetheless, the market for BTPs remains
game or puzzle with the promise of (for review see Jonides et al., 2008). healthy and growing with companies
increased intelligence and an overall Much of these data have been used to highlighting (often their own) published
better brain. Unlike pharmaceutical support the basis of commercial BTPs: studies while promising a product that
‘‘enhancement,’’ these BTPs and their global improvement through influencing effectively alters the brain’s neuroplastic-
purveyors currently enjoy a privileged neuroplasticity. ity (Chancellor and Chatterjee, 2011).
status, protected from public scrutiny by The efficacy of BTPs remains contro-
virtue of their non-invasive nature and versial and has been reviewed and Novel Ethical Issues and
perceived requirements for the users analyzed extensively (Rabipour and Raz, Vulnerabilities Presented by
to engage in effortful, yet fun, mental 2012). Recently, 74 scholars issued a Big Data, Citizen Scientists,
‘‘work’’ to enhance their cognitive abili- statement of skepticism regarding com- and New Virtual Territories
ties. Through repeated use of these mercial BTPs (Allaire, 2014) expressing The era of big data has resulted in the
computerized tests, commercial program particular concerns for those who might recruitment of a new cadre of researchers
developers purport that training specific choose to use BTPs therapeutically. One who act as both subjects and ‘‘citizen sci-
cognitive domains with their products of the signatories, Susanne Jaeggi, was entists’’ (Marx, 2013). MIT neuroscientist
will offer broad benefits to other untrained lead author of a breakthrough 2008 Sebastian Seung has recruited tens of
tasks, called the ‘‘transfer effect,’’ the holy study, which reported that challenging thousands of such citizen scientists to
grail of cognitive training research. adaptive WM training increased mea- reconstruct visual neurons, which has
Cognitive function can be broken down sures of fluid intelligence in a dose-depen- already resulted in an innovative, high-
into specific dimensions such as atten- dent manner, a theoretical basis for many profile paper (Kim et al., 2014). The possi-
tion, memory, language, executive func- BTPs (Jaeggi et al., 2008). Redick et al., bilities for crowdsourcing technical and
tion, and visuospatial abilities. Working who have argued that Jaeggi’s 2008 study experimental help in an environment
memory (WM) is a construct used to suffered from a small sample size, of dwindling funds, especially with the
describe the role of short-term memory attempted and failed to replicate these outcome of high-quality peer-reviewed
in complex, higher-order cognition findings with a larger cohort and active research, are extremely enticing. This
(Jaeggi et al., 2008). Those training inter- control groups (Redick et al., 2013). model’s potential impact for research
ventions that seem to have the greatest Others suggest transfer effects may be is still far from being fully realized, but
benefits toward globalized improvements constrained by age (van Muijden et al., already has high visibility and enormous
in cognitive function share the common 2012), and that any patterns of generaliz- public enthusiasm.
element of improving WM (Jaeggi et al., able effects on cognitive improvements In a similar way, BTP companies
2008), and WM measures have been are more associated with emotional provide a larger service beyond poten-
shown to rival typical measures of intelli- states such as motivation, arousal, and tially enhancing consumer abilities—
gence in determining scholastic achieve- reward than are attributable to the brain advancing science while cultivating a

356 Neuron 86, April 22, 2015 ª2015 Elsevier Inc.


Neuron

NeuroView

publically inclusive research process. In tional data such as hours of sleep and in addition to cognitive ability as has been
the case of commercial BTPs, progress alcohol consumption). As they state in demonstrated in Lumosity’s recent publi-
is made possible by citizen scientists the article, ‘‘we have only scratched the cation (see Sternberg et al., 2013).
who provide the data as well as help surface of what the further study of this Website privacy policies in general
fund the research through consumer/ dataset might uncover.’’ They conclude have been critiqued for giving users insuf-
membership fees. The availability of by extending invitations to researchers ficient information to understand a site’s
BTPs online allows opportunities for the to partner with them for future studies, policies (Kahn et al., 2014). Were it even
public to engage in cognitive training extolling the virtues of their large dataset, possible to review all policies at websites
and for companies to conduct indepen- unconstrained by the typical demo- visited, it remains difficult to predict
dent research without the constraints graphic afforded by the college partici- how data that are collected now may be
and oversight of a laboratory setting. pant pool, unfettered by time-consuming secured, interpreted, and used in the
Given the rapid rate of collection of and prohibitively expensive facilities and near and long term given evolving security
highly sensitive cognitive performance staffing needed in a conventional psy- threats and rapidly advancing methods
data through BTPs, we feel that concerns chology laboratory. While such benefits of data analysis. Further protections
with BTPs move beyond validating effi- are not unfounded, collaborations would for the legacy use of these data and the
cacy and into the realm of privacy. Con- likely increase the flow and distribution types of research or application are also
cerns about the lack of protection for of vulnerable information beyond what unclear. For example, should a BTP com-
neurodata have been focused on data many users today may understand. pany fail, there is no guarantee that per-
gathered from fMRI and EEG studies— Furthermore, researchers and partici- sonal information will not be sold off along
which collect indirect measures of brain pants may not agree upon evaluations with other assets. Posit Science explicitly
activity (Tovino, 2005). While we may not of which data are considered vulnerable states in its privacy policy for BrainHQ
fully understand fMRI and EEG data in and what research might be considered that users’ information may be treated
the present (which are largely correlative harmful to participants. as a business asset and therefore sold
with behavior), it is conceivable that new The popular media and general public in the event of bankruptcy (BrainHQ,
insights into the subject’s personality, expressed outrage at Facebook and its 2014). The US Federal Trade Commission
cognitive capacity, and future behaviors academic collaborators over a recent (FTC) recently demonstrated concern in
might someday be gleaned from the study (on ‘‘massive-scale emotional a similar matter and moved to stop
same data with evolving methods of contagion’’) (Kramer et al., 2014). Approx- the sale of personal information and stu-
interpretation. Unlike brain imaging, phys- imately 700,000 users unknowingly dent records from the now defunct
iological biomarkers, or even genetic participated in the study, had no opportu- website ConnectEdu (http://www.ftc.gov/
data, BTP data are being interpreted nity to opt out, and consented by way of news-events/press-releases/2014/05/
as current demonstrations of existing agreeing (with a click) to Facebook’s ftc-seeks-protection-students-personal-
behaviors and predispositions, and not Data Use Policy upon creating an account information-education), but it remains to
just correlations or future predictions of with the company (which could have be seen if this precedent will continue to
human cognitive capacity and perfor- occurred more than a decade ago). As a hold or be applied to research contexts
mance. Yet, the vulnerability of cognitive commercial entity, even in collaboration for BTP data.
performance data collected from BTPs with an academic university, Facebook
has been overlooked, and we believe was not subject to federal regulations for Recommendations and
the rapid consumption of such games human subject research. Instances such Conclusions
warrants a sense of immediacy to safe- as these highlight the uncertainties and For a public eager to be self-empowered
guarding these data. lack of clear definitions for privacy in this and to maximize its cognitive potential,
Lumosity has recently published, on new frontier of virtual big data research the possibilities of brain training are tanta-
their website (Sternberg, 2013), data on (Kahn et al., 2014). lizing. Similarly, for scientists in a climate
the ‘‘smartest universities in the US’’ Internet users are accustomed to of dwindling funds, crowdsourcing data
based on correlates of Lumosity aggre- trading personal information for access is enticing. We see such data as having
gate scores from university consumers to information and services as is evinced great innovation and value, not just as a
that were widely circulated by the by the Facebook study and emerging research model, but also for exploring
popular press. Another collaboration BTP companies. According to the privacy cognitive function more generally and as-
with Lumosity’s Lumos Labs resulted in policy of Lumosity.com (Lumos Labs, sessing whether or not BTPs can be effec-
a report entitled, The largest human 2013), the parent company Lumos Labs tive. In order to ensure this work is fruitful
cognitive performance dataset reveals collects information about users’ date and the benefits of such a research model
insights into the effects of lifestyle factors and time accessing the site, the websites are to be fully realized, there is an immedi-
and aging (Sternberg et al., 2013). Stern- visited before and after Lumosity, as ate need to evaluate and establish the
berg et al. collected data on 36 million well as performance on games. These standards of conduct for this research.
users from 231 country codes and were data, along with the personal infor- Use of such cognitive performance
able to study the health, lifestyle, and mation provided at registration, can data falls into a new territory somewhere
cognitive performance of their users produce an intimate depiction of a user’s between commercial, research, and edu-
(some users volunteered to submit addi- lifestyle, habits, and geographical location cational domains, and to date protections

Neuron 86, April 22, 2015 ª2015 Elsevier Inc. 357


Neuron

NeuroView

exist only within those realms. Existing companies’ more lucrative asset than the US began the Human Genome
protections for consumer telecommuni- the BTPs themselves are the enormous Project. As the US and EU embark on
cations data such as FTC oversight datasets they are amassing from personal the next large-scale scientific endeavors
and the US Personal Data Privacy and information about consumers around the now focused on the brain (US BRAIN
Security Act of 2014 recognize that world. These companies are currently Initiative, EU Human Brain Project), it is
data misuse can ‘‘cause serious or irrepa- able to share these data with researchers important that neuroprivacy concerns
rable harm to an individual’s livelihood, needing access to extensive banks of are addressed in a more timely manner,
privacy, and liberty and undermine effi- cognitive performance data from users perhaps through the creation of a
cient and effective business and govern- worldwide, beyond the common aca- neuroscience information nondiscrimi-
ment operations,’’ and its misuse is demic research populations of college nation act that regulates the flow of
punishable (http://www.congress.gov/bill/ students. Online marketers will be eager such information and determines which
113th-congress/senate-bill/1897). Educa- to access and consume such data as parties have access and even who might
tional records are heavily protected by well. What are the rights of the consumers own such data (i.e., the consumer, the
the US Family Educational Rights and to own their cognitive performance data researcher, or the commercial entity).
Privacy Act, which allows individuals and (perhaps for their own future legacy use) In April of 2013, with the announcement
family members of those under 18 access and what obligations, if any, do commer- of the BRAIN Initiative, President Obama
to educational records and the right to cial BTP companies have to safeguard directed his Commission for the Study of
request correction of those records, and it? In this (r)evolution, all stakeholders— Bioethical Issues to ‘‘consider the potential
limits schools’ abilities to disclose those including researchers, participants, and implications of the discoveries that we
records to third parties without written academic publishers—will need to take expect will flow from studies of the brain,
consent from parents and/or students. care to evaluate the process. and some of the questions that may be
Data from online BTPs, which arise from A regulatory model for big data raised by those findings and their applica-
tests of cognitive ability in its many research between commercial and aca- tions—questions, for example, relating
manifestations, fit somewhere between demic partners must be in place to to privacy.questions about stigmatiza-
these consumer and educational do- keep pace with the evolving research tion and discrimination based on neuro-
mains, warranting a comparable degree environment. The US Department of logical measures of intelligence or other
of protection. Commerce’s National Telecommunica- traits’’ (Obama, 2013). Now is the time for
Perhaps most challenging to deter- tions and Information Administration government commissions, academics,
mine, given evolving definitions and un- (NTIA) questioned how big data collection as well as BTP researchers and policy
certainties related to privacy, is how one will impact President Obama’s Consumer makers to take a closer look at how indi-
can actually ‘‘consent’’ to participate in Privacy Bill of Rights (http://www.ntia. vidual privacy can best be protected in
such research. As others have argued, doc.gov/press-release/2014/ntia-seeks- the context of online brain training and
the standard approach to website privacy comment-big-data-and-consumer-privacy- also Internet-based research and crowd-
policies is grossly insufficient for informed bill-rights). The US federal regulations for sourcing in neuroscience research. At
consent (Fiske and Hauser, 2014; Kahn human subjects research is in a state of stake here is, on a large scale, the cre-
et al., 2014). Not only is it impractical to revision (Fiske and Hauser, 2014). These dibility of neuroscience research, but
expect users to read and fully understand include numerous provisions including also—perhaps even more importantly—a
the standard online terms and policies of clarifying expectations of privacy (or lack pivotal precedent in terms of an individ-
every website with which they interact, thereof) and redefining minimal risks and ual’s cognitive liberty and the right to own
many users may assume it is safe or feel harms, particularly in light of new models and control the flow of one’s neurodata.
coerced facing a lack of options to of crowd-sourced online research. Yet, it
choose otherwise. A relatively simple shift is difficult to fully understand risks, and ABOUT THE AUTHORS
toward a requirement to ‘‘opt in’’ to a in turn inform participants of the risks, at
study, rather than having the onus on the this juncture because the evolving sci- Ryan H. Purcell is a PhD candidate in the Neurosci-
user to fully read a policy and ‘‘opt out’’ ence, the legacy of these data, and their ence Graduate Program at Emory University and
(if that option is even given), could be a ownership remain unclear. The question studies the signaling and regulation of adhesion
major improvement. New federal guide- driving this research guideline’s revision G protein-coupled receptors linked to synapto-
genesis with an interest in the molecular basis of
lines will need to take into account new is evaluating what constitutes human
memory. He serves as an editorial intern for the
informed consent logistics and hurdles, research. Because online BTPs clearly American Journal of Bioethics Neuroscience.
what actually constitutes human subjects collect identifiable, sensitive information Karen S. Rommelfanger is an assistant professor
research online, privacy and concerns for on living individuals through direct inter- in the Departments of Neurology and Psychiatry
attendant harms, as well as blurred lines actions and are being used systematically and Behavioral Sciences, the Neuroethics Pro-
for responsibility for regulatory oversight to create generalizable knowledge about gram director at Emory University’s Center for
whether that be across commercial and human cognition, BTP data collection is Ethics, and neuroscience editor-in-residence at
the American Journal of Bioethics Neuroscience.
academic domains or transnational human research.
Trained as a neuroscientist and ethicist, her
boundaries (Kahn et al., 2014). In 2008, US President George W. Bush research explores how evolving neuroscience
Another unaddressed issue is data signed the Genetic Information Nondis- and neurotechnologies challenge societal defini-
ownership. Arguably, commercial BTP crimination Act into law 18 years after tions of disease and medicine.

358 Neuron 86, April 22, 2015 ª2015 Elsevier Inc.


Neuron

NeuroView
REFERENCES Jaeggi, S.M., Buschkuehl, M., Jonides, J., and news/Charge%20from%20President%20Obama.
Perrig, W.J. (2008). Proc. Natl. Acad. Sci. USA pdf.
105, 6829–6833.
Allaire, J.C. (2014). A Consensus on the Brain
Training Industry from the Scientific Community. Rabipour, S., and Raz, A. (2012). Brain Cogn. 79,
Jonides, J., Lewis, R.L., Nee, D.E., Lustig, C.A., 159–179.
http://longevity3.stanford.edu/blog/2014/10/15/
Berman, M.G., and Moore, K.S. (2008). Annu.
the-consensus-on-the-brain-training-industry-from-
Rev. Psychol. 59, 193–224. Redick, T.S., Shipstead, Z., Harrison, T.L., Hicks,
the-scientific-community-2/.
K.L., Fried, D.E., Hambrick, D.Z., Kane, M.J., and
Kahn, J.P., Vayena, E., and Mastroianni, A.C. (2014). Engle, R.W. (2013). J. Exp. Psychol. Gen. 142,
Alloway, T.P., and Alloway, R.G. (2010). J. Exp. Proc. Natl. Acad. Sci. USA 111, 13677–13679. 359–379.
Child Psychol. 106, 20–29.
Kim, J.S., Greene, M.J., Zlateski, A., Lee, K., Ri-
Sternberg, D.A. (2013). Lumosity’s Smartest Col-
chardson, M., Turaga, S.C., Purcaro, M., Balkam,
BrainHQ (2014). BrainHQ Privacy Policy. http:// leges 2013. http://blog.lumosity.com/wp-content/
M., Robinson, A., Behabadi, B.F., et al.; EyeWirers
www.brainhq.com/privacy. uploads/2013/11/SmartestColleges2013_white-
(2014). Nature 509, 331–336.
paper.pdf.
Chancellor, B., and Chatterjee, A. (2011). AJOB Kramer, A.D., Guillory, J.E., and Hancock, J.T.
Neurosci. 2, 18–27. (2014). Proc. Natl. Acad. Sci. USA 111, 8788–8790. Sternberg, D.A., Ballard, K., Hardy, J.L., Katz, B.,
Doraiswamy, P.M., and Scanlon, M. (2013). Front.
Lumos Labs (2013). Privacy Policy. http://www. Hum. Neurosci. 7, 292.
Duckworth, A.L., Quinn, P.D., Lynam, D.R.,
lumosity.com/legal/privacy_policy.
Loeber, R., and Stouthamer-Loeber, M. (2011). Tovino, S.A. (2005). J. Law. Med. Ethics 33,
Proc. Natl. Acad. Sci. USA 108, 7716–7720. 844–850.
Marx, V. (2013). Nat. Methods 10, 1069–1074.

Fiske, S.T., and Hauser, R.M. (2014). Proc. Natl. Obama, B. (2013). Charge from President van Muijden, J., Band, G.P., and Hommel, B.
Acad. Sci. USA 111, 13675–13676. Obama. http://bioethics.gov/sites/default/files/ (2012). Front. Hum. Neurosci. 6, 221.

Neuron 86, April 22, 2015 ª2015 Elsevier Inc. 359

You might also like