Transcript, Meeting 17, Session 3

Date

June 9, 2014

Location

Atlanta, GA

Presenters

Giorgio A. Ascoli, Ph.D. 
University Professor, Molecular Neuroscience Department
Founding Director
Center for Neural Informatics, Structures, and Plasticity
Krasnow Institute for Advanced Study
George Mason University

 
Helen Nissenbaum, Ph.D.
Professor, Media, Culture, and Communication, and Computer Science
Director, Information Law Institute
New York University

Download the Transcript

Transcript

DR. GUTMANN: Our next session will focus on data sharing and access in neuroscience. This is an issue we've been keenly interested in. We will hear from Dr. Giorgio Ascoli. Dr. Ascoli is University Professor in the Molecular Neuroscience Department and Founding Director of the Center for Neural Informatics at Krasnow Institute for Advanced Study of George Mason University. He created and curates NeuroMorpho.org, a publicly accessible collection of three-dimensional digital reconstructions of neurons. Dr. Ascoli serves on review panels for the National Institutes of Health, the National Science Foundation, and the Intel Science Talent Search, national and international scientific advisory boards, and editorial boards of numerous biomedical journals. Welcome.

DR. ASCOLI: Thank you very much, Dr. Gutmann, and the whole Commission for giving me this opportunity to brief you on a topic that is near and dear to my heart, that of data sharing in neuroscience. And I understand that talking about bioethics of data sharing can be controversial. It isn't for me. I view it as a pretty clear stand. It's ethical to share data. It's unethical not to share data. Let me explain what I mean.

The data that I will be illustrating is mostly not about patients. It's mostly not about live human beings with their consciousness. It's by and large about nonhuman animals. A lot of the research that is done in neuroscience is on animal models where there are no issues of privacy but there are tremendous implication for our understanding of health and disease as well as what the implication of understanding how the brain computes can have for our society in general.

The brain has been hailed as the most complex object of the universe, and much of the complexity lies in the connectivity among its hundred billion or so neurons. Quite a bit of the data that I will show is precisely about the structure of those neurons, and that's something that has been on the forefront of neuroscience research since the early days, Ramon y Cajal more than a hundred years ago, and we are still struggling with that. I'll give you a brief illustration of what those data entail.

This data is hard won. It's very expensive to collect both in terms of labor and money, and it can be shared. There are no technical barriers. There are no privacy issues. There are still tremendous social barriers. Neuroscientists do not seem to be good at sharing. And I think that right now there is still no consensus as to the fact the data should be shared, even if it might look apparent to us. So the tool or the database that I'm going to use to illustrate my point is NeuroMorpho.org. It's something we started under the NIH program, or in fact pan-agency program that was called the Human Brain Project, the U.S. Human Brain Project, not to be confused with the European Human Brain Project that has been launched last year, and the idea was that now that the data is all acquired in digital form and there are digital files, it is no longer images and words but actual numbers very much like gene sequences or protein structures that can be stored in computers and therefore in databases.

The reconstructions of neurons are essentially very much like three-dimensional structure of proteins. I'm going to show you an example of this. This is one of the neurons reconstructed in my own labs, so I will not risk upsetting anyone by looking at their data. So this is a reconstruction of a neuron. The data is available online and it can be opened through any normal browser. This is a single neuron. There are about 10,000 of these that are deposited in NeuroMorpho.org. It can be moved, manipulated, zoomed. What you see here in green is a dendrite, which is the input structure of the nerve cell, and in gray is the axon, which is the output structure of the cell. The input structure I opened up in a colorful view here for clarity. The green and the purple are different aspects of the tree, but it essentially goes down to pretty painstaking detail of the tracing. And a human operator started from histological preparation in this case from a rat and observed this under the microscope and reconstructed it. Each of these neurons is annotated with a variety of metadata: the species, the brain region, the experimental preparation that was used. There is a link to the literature. This is in fact a reciprocal link. You can go to PubMed and from PubMed come back up directly to the data, as well as morphometric measurements, which are the geometric properties of these neurons by which the neuron can be searched.

And just to give you a sense, this is the underlying data. Each of these lines is essentially an X, Y, and Z coordinate, as well as the connectivity within that structure, and the idea here is it's a lot of data points, a lot of data points for a single neuron.

So in this particular neuron it is 8,255 mouse clicks that a neuroscientist, in this case one of my post docs, did under a microscope to reconstruct this neuron. It took about six months of work. So if you go to the main page of the database you can see how many times these files have been downloaded. You can see how many neurons there are and how many hits the site has had, and you can look at the detailed statistics of this where it says where the data is going. It's all around the world. It says how many times it was accessed and when, the hits by quarter, by IP addresses, and most importantly, if you scroll down enough, by neuron types, by brain regions, and by contributing labs.

And the point here is not that this data is beautiful. It is, but the main point here is that this data is useful. Obviously every time that individual labs reconstruct neurons, they usually publish a paper, two, three, they renew their grants, they make a discovery, but a lot of discoveries have occurred with the data that have been downloaded from the database, very much like a lot of discoveries today in biology are made by analysis of gene bank and not just by sequencing genes in one's lab.

So what is the issue? The issue is here. There are no privacy barriers on these particular data. We do have human data but it's mostly postmortem. There are no technological barriers either. We take care of this resource with the generous NIH funding. We take care of populating the database and making it accessible, widely and freely. People do not deposit their data. We are asking for their data as of yet, but even the data that we ask, if you go on the literature coverage, it's actually eye opening. We mined through over 20,000 research articles. The technology to reconstruct neurons have been invented about 30 years ago, so we don't have infinite papers to go through. We have a finite status. Of course, every month there are new publications.

When you look at the papers, we actually color code them. The ones that are in green are the ones for which the authors agreed to share the data and the data is in the repository. The ones in red or pink are the ones for which the data is not available. And the ones in yellow are the ones that we are still struggling and requesting and negotiating the access to the data. And there are many thousand papers here. The summary of this you can actually see in this tab, and the summary here is pretty simple. Of the papers that we mined, about a third to a quarter of people have agreed to share their data. These are the number of neurons that are available, and about two-thirds to three-quarters they have not. And you can actually -- the data is public. They are on PubMed. You can look at the reasons that they give for declining the data. And it goes from not responding at all to our requests to, well, short of, essentially the dog ate the hard drive, and lost the data. People have moved, and simply the contact information is lost, but a very large number are flat out declined. "We don't have time. We have better things to do, and we are publishing our next paper. We are constructing more neurons for our lab."

So this in my mind is a tremendous loss for neuroscience. This is something that is not happening in other fields. Data from physics, both particle physics and astronomy, is widely shared. Data from molecular biology, both gene sequencing and protein structures, is widely shared. I understand that there are issues of privacy when sharing fMRI data and human data. There are truly no such barriers for sharing a huge amount of research data that is valuable on animal models.

DR. GUTMANN: Thank you very much. We will now hear from Dr. Helen Nissenbaum. Dr. Nissenbaum is Professor of Media, Culture, and Communications, and Computer Science at NYU where she is also Director of the Information Law Institute. Dr. Nissenbaum has written and edited many books including Values at Play in Digital Games, and Privacy in Context: Technology, Policy, and the Integrity of Social Life, and her research publications have appeared in journals of philosophy, politics, law, media studies, information studies, and computer science.

I have to say as a matter of personal privilege and pleasure that Dr. Nissenbaum served as Associate Director of the University Center of Human Values at Princeton University where I was Director, and she was fabulous in that role and has had abiding interest and expertise in issues of privacy. So welcome, Helen, and thank you for being with us.

DR. NISSENBAUM: Thank you very much It is an honor to be here. It's really a privilege to be able to present some of these ideas to you. As you may guess, if you know anything about my work, I'm going to be talking about data sharing and access from a privacy perspective, and so I'm going to start by posing the question, what kind of problem is a privacy problem? Because I suspect that the answer that I would give is going to be different from the answer that many people in the field would give about privacy. So what I'm going to do is spend some time talking about the problem that I see as the privacy problem and just a quick few minutes on some of the standard solutions, and then I'm going to finally conclude with some thoughts about how this thinking may guide or help or shape some of the work, really important work that you are doing here in the Commission on neuroscience.

So as you see, what I've considered to be privacy is appropriate flow of information. Many people will talk about privacy as secrecy like nondisclosure. They'll say there are losses of privacy whenever you disclose information, or they may insist that information subjects should be able to control flow of information. I think that the ethical value of privacy is that information should flow appropriately in society, and then the theory of contextual integrity that I'm going to talk about briefly is that we model this idea of appropriateness through this construct of context-specific informational norms.

Now, these information norms, I want to just give you a kind of intuitive grasp of what these norms can look like just by some examples that hopefully you'll find somewhat plausible, just a kind of intuitive sense of what these norms look like. But in fact the norms, what the theory suggests, is that the norms have a very particular structure. And what I mean by that is that when we are thinking about appropriate flow, when we want to understand whether a flow is appropriate or not, we need to hear three things. One is who the actors are, who is the center, and now we are talking about in certain capacities, acting in certain capacities; who is receiving the information; and who is the subject, again, acting in a certain capacity. Is it a physician, a patient, is it a nurse, and so forth. What type of information are we talking about, and briefly, attributes; and then finally something that I've called the transmission principle which we can think about as constraints in flow.

Now, some of the critique I have of other work is that often we'll ask about, oh, is it possible to disclose such and such kind of information without saying disclose to whom, and so we get into a fair bit of trouble when we don't specify all the elements of the norm.

Now, the history of science and technology, or you could say the history of privacy as we've experienced it, I mean, even if you go back to Warren and Brandeis, but I would rather say the last half century, is that science and technology has disrupted flows, and that's why we are experiencing these threats to privacy. But what I've argued is it's not enough simply to recognize that the flows have been disrupted by information, science and technology, and I mean, it could be in the biosciences, but a lot of this has happened in the digital sciences, information, technology, digital media, mobile networks, and so forth.

But we want to understand, we really want to frame how the disruption is taking place. Sometimes new technologies allow information to flow to different actors. Sometimes the technology enables new types of information, and I think that in the realm of neuroscience that is a lot of what is happening. There's new types of information. We don't really know what to do about it. It's disturbing. And sometimes the concerns or the disruptions are in these transmission principles. And those of you who followed, for example, Kyllo and thermal imaging, Kyllo v. United States, or Jones v. United states, these are Fourth Amendment cases, the issues at stake are whether thermal imaging, constituting a new kind of transmission principle, abides by the Fourth Amendment constraints or whether attaching a GPS device has contravened the Fourth Amendment prohibitions. And so that's where a lot of the action is taking place.

So I think there's an important matter here, which is when we look at these disruptions, we need to identify where the disruptions are taking place. Now, often disruptions -- and sometimes it's important just to reveal the disruption, because often, for example in some of the work that I've done on posting court records online, people will say, "Oh, court records, they're public, so what's the big deal? Post them online for open access." And some of the work is just to systematically and carefully go through and show where the disruptions are actually taking place, what's different about posting this information online.    

Now, sometimes disruptions are good and we want to replace the old practices we've been engaged in with the new disruptive flows of information. This is another and really important part of what we need to do, which is evaluate the disruption. And in the theory of contextual integrity I've offered three factors when we consider these disruptions. One is that we consider interests, preferences, and desires that are served, you know, vested interests, stakeholders, and so on, economic style of analysis. Second we need to consider the ethical and political values, and there's been a lot of excellent work in the privacy community that has centered on these two tasks. But the third is something that I think I want to really dwell on, which is these context-specific ends and values.

And so if you are thinking about, for example, education and the introduction of third-party education providers or these direct-to-consumer tests, we need to ask not only who is harmed, who is benefited, but what are the goals of education, how are these served by these new flows of information. Because philosophically what I really tried to say about privacy is that, if we are not doing it justice as a value that should be protected, if we are only thinking about it in terms of the harms, for example, to the information subject, what we really need to understand is how appropriate information flows serve to promote and preserve social and institutional integrity. That's where privacy comes in as a societal value, and that's one of the most important reasons why we need to protect privacy as an important value.

Now, when we look at the questions of neuroscience and privacy -- and I'm mindful of my time rushing away. I know I'm not going to get to the end of this. I realize that some of the privacy issues have to do with intrusiveness of neuroscientific activity, but what I'm focusing today on is the idea of access to data. And a lot of the solution has immediately gone to anonymity and informed consent, and I noticed how much we talked about informed consent even in this morning's discussion. So a lot of attention is paid to how do we anonymize data so that we can make it available and make it productive, or how do we effectively operationalize notice and consent.

I'm pessimistic, particularly in the area of information, that anonymity and consent are not going to be the solutions for us in the realm of data access. First of all, namelessness itself does not preserve the underlying value for which we've pursued anonymity; and second of all, I think informed consent is irrelevant.

The conversation we had about diminished capacity I think should be seen as an opportunity for us, because when we think about how to treat people with diminished consent, then something kicks in. We have to take responsibility for that individual. Proxy consent and so on I think is not the solution. We have to say what should we be doing, what are our responsibilities in relation to these individuals. And with privacy I don't think we can shove the responsibility onto the shoulders of the data subject.

So, only a couple of slides. We need to do all of this work when we think about privacy in relation to access to data. We need to identify the patterns of disruptive flow and evaluate the informational norms with a view to context-specific ends and values. These are just some brief proposals that I will just leave in front of you while I end. Thanks a lot.

DR. GUTMANN:  Thank you. We asked for a brief presentation so we can have a full conversation, and I open it up for questions from anybody. Dan, why don't you begin.

DR. SULMASY: Just a brief sort of scientific question for Dr. Ascoli, and I wanted to know how much variation there is within species in these maps. I mean, is it sort of equivalent to variation genetically such that most is conserved in each brain in terms of the mapping of various neurons to others, or is there tremendous variation that comes during development? And part of what this does is help give us an idea of the size of the data need, because the more variation obviously the more data you need.

DR. ASCOLI: It's a great question. The short answer is it's not known. There are some elements that can help. There is difference in the differences. Invertebrates, so for example the worm C. elegans, appear to be more stereotypical, so you can look at maps from one individual to the next and they tend to be much more similar; whereas, in mammals there is essentially absolute variability such that the only descriptors that appear to be sensible are statistical in nature. So nobody has ever figured out a way how to compare a given neuron from a mammal to another given neuron from a member of the same species but a different individual. But this is done at the population level. The same is true from variability within individual. So every one piece of the cortex, for example, in humans, has tremendous internal variability as well.

So the size of the data is not known yet that is necessary. This, if anything, stresses even more the need for sharing. What is clear is that the kind of research that can be done by pooling the data, for example, from these databases, is of a different type than the kind of research that can be done by individual labs. They are both worthwhile, but it's not just that you're accelerating the pace of research. You are enabling research that cannot otherwise be done.

DR. GUTMANN: Anita.

DR. ALLEN: Thank you. I had just a question for Dr. Ascoli. I took the central message of your talk to be that data sharing in your context is good and that there are no privacy issues. And I was struck by how definitively you stated, "There are no privacy issues. Neuronal morphology is essentially devoid of ethical and legal privacy issues." I was curious about why you are so definitive about that. What is your analysis of why there are no privacy and ethical issues? Because I think that if you have Helen Nissenbaum's kind of approach to privacy where privacy is all about appropriate flows, maybe you wouldn't say there are none, but you might explain it in terms of there not being any inappropriate flows. I just wanted to know what is your analysis of why there are no privacy issues.

And here is what I'm thinking about. So you might have three different kinds of data that might flow. You might have, say, a neuron, a single neuron image, right? Or you might have a placenta, and there was a famous case of a nursing student who posted someone's placenta on Facebook. It was not identified but there was a question about privacy. Can you just -- is someone's placenta something which should not be shared? And then you might have an fMRI. So take the spectrum, neuron, placenta, fMRI, and help us to understand more about why you think it's so clear that there's no privacy issue when it comes to the neuronal morphology.

DR. ASCOLI: The first issue I'll try to clarify. I'm saying that there is no privacy issue for the vast majority of the data that this research is about, which is on animal models.

DR. ALLEN: Putting aside the rats. But you talked about some of the samples come from deceased people. I just wanted to know what is your analysis of why the neurons of a dead person don't matter, why the neurons of an alive person don't matter, in the context of the kind of data sharing that you are doing and recommending.

DR. ASCOLI: I am not saying they don't matter. They certainly matter. Right now there is no technology to look at individual neurons in a live person, so that's not in the present or foreseeable near future of scientific technology. We can look at neurons in genetically-modified animals with fairly invasive techniques or in invertebrates, live, but in humans we cannot quite do that. So the only way that we can see neurons in humans is indeed from excised tissue. And it's not only postmortem; it's sometimes taken from surgery. And when that is shared, it is when the patients have agreed to either donate their brains to scientific research or to have the tissue removed analyzed for scientific research. I mean, that's within the same realm by which people agree to have their sequences posted and their brain images posted. So I'm not reducing the importance of that analysis in no means. And for human data those neurons would ultimately fall under the same purview as fMRI and genetics. But right now much of what we know about individual neurons as opposed to whole brains, and much especially of what we do not know and we are actively researching at the level of individual neurons, is emphatically not on human tissue; it's on animal models.

DR. GUTMANN: Just for the record of the Bioethics Commission, in our previous work -- and this is relevant to what Dr. Nissenbaum has argued -- in our previous work we have made it very clear that consent is important but it's not the only thing that's important.

Even if you could get consent, there are other ethical standards, which you characterized, Helen, as context flow, which remain important. And so everything you say is -- you've now made clear, Dr. Ascoli, I think that you're not disagreeing with that. It's just that what you are working with doesn't run up against those issues.

DR. ASCOLI: Exactly. Yes.

DR. NISSENBAUM: Could I just respond a little bit? I don't know that it has much -- it doesn't necessarily apply to what you have in your database, but one of the reasons that big data is such a problem for those of us who are thinking about privacy issues is that sometimes the information, let's say about a dead person, the information we're getting from a dead person -- and I know you've done your reports on genetic, you know, the human genome. The information I can get from one study can reflect on other people, so I can learn something about you from information about [someone else]. And this emerges as a privacy problem, and that's the end run around consent, because I get your consent, but I don’t get [someone else's consent] and nevertheless I'm able to learn something about you. The way out of this, the only way out of this that I can see is to go to those other values that you mentioned, which is to do all the hard work of understanding, evaluating whether this information collection and provision of access is worthy and ethically justifiable within society, not whether we got somebody's consent for it.

DR. GUTMANN: But you are not -- I just want you to clarify. You are not denying that getting the consent is important in the first place.

DR. NISSENBAUM: No, I'm not denying  that. I'm only --

DR. GUTMANN: It's not sufficient, necessary but not sufficient.

DR. NISSENBAUM: And the reason I'm pushing so hard is that not in the bioethical realm where I think there are these other protections, beneficence, justice, and I think there should be more with data, but in the realm of privacy, consent has -- and if you look at the federal trend, that has become the measure.

DR. GUTMANN:  Understood. Understood. Nita.        

DR. FARAHANY: It's on this same question of consent where I was struck by what you said. And I was imagining a model that I've heard discussed in the genomics context, which is -- and maybe this fits nicely within what you're suggesting -- which is to provide the ability first to recognize anonymity as kind of a losing game, but then to provide more control to the individual over the data. And so a model that's been suggested is to create kind of a repository where individuals have the ability to decide either the level of consent that they might provide overall, like this is a model in which I'm willing to share all, I'm willing to share some, I'm willing to share only a little, or you have to come ask me each time there's a new study and I can just select a few things on my computer and decide whether or not I share. That to me seems like a true consent model. It's about flow of information but it empowers the individual to choose when they share and with whom they share and for which purposes they share and what model of sharing that they might adopt.

And so with your emphasis on should we collect the information to begin with, once we have collected the information, do you think that kind of consent model is a sensible one or is your statement that consent is not a meaningful value? Do you mean it to apply even to that kind of a model of sharing?

DR. NISSENBAUM: I do mean to apply it to that. Maybe going back to what Amy had said, which is that I don't think -- and I'm looking at privacy issues, I'm not talking about other biomedical interventions for which there may be other sorts of reasoning that elevate consent and make it more important -- but in the realm of privacy, consent is not the be all and end all. Sometimes information flows because it promotes certain societal values for that information to flow. And simply because that information is about you does not give you the right to consent or not to consent. It is a societal question to ask what information should be under your control and why, and what information should not be under your control and why. I don't think the presumption is control. I mean, it's a longer story. I won't take more of your time, but that's where I see the direction.

DR. GUTMANN: I have Jim next.

DR. WAGNER: Dr. Ascoli, I confess to be a little confused by the thesis of your presentation, and if you could help straighten that out, I would appreciate it. I agree with you that sharing of data and access to data in common formats is going to be enormously important as we need to distribute the analysis of these huge data set problems that Dan refers to. I think you and I are on the same page. But I'm confused, it would seem that the lack of participation, though, is not a consent issue in your case. After all, these PIs are saying [they] would rather spend time publishing, which is of course also disclosing data, as opposed to contributing to NeuroMorpho.org. What am I missing? Is it just something in the culture that says they don't value having data at a common location of similar format? What's the concern?

DR. ASCOLI: I think ultimately it's an issue of competition. If they share the data, the data is usable by peers. Typically what people fear is, "They are going to find issues with my data. They are going to publish and compete at the next study section where my grant is going to be reviewed and now my competitors are going to have two more papers. And maybe my next graduate students will see things that my previous graduate students did not see in this same data set and I should be the one publishing in three years as opposed to someone else." I think the issue of "I don't have time" is personally more as an excuse. That's, of course, my opinion. And as societies we would not accept people to say you published the data but you didn't have time to write up the methods. We would say no, you have to disclose the methods, for many reasons. I would say at this point in these days and age sharing the data is really part of the method disclosure.  

DR. WAGNER: It's disappointing but understandable, I guess, in competition. But one of the things we ought to consider is the ethical dimension of ensuring the quality of data that are available. In other discovery sciences, and you mentioned cosmology, for example, in other discovery sciences there's a very high threshold to the acknowledgment that a discovery is credible. And so I do see that as being perhaps a threat to those who would like to generate large volumes of questionable data, but I think there is an ethical issue also. If we are going to be trying to draw conclusions from, as I say, from the many parts coming from many directions of these large data sets, there may be an ethical question about ensuring something about the quality of those data.

DR. ASCOLI: Absolutely. So the standard, my understanding of the standard in science right now is based on peer review, quality assurance, so we only request data that has been published or we request data that relies on studies that have been published. And the assumption is if the peer and the community have agreed that this data is useful for some discovery, it may be useful for additional discovery. We do occasionally receive data from someone who said, "Well, we collected this data. It's the kind that you store. Please put it up." If it has not undergone peer review, we will not consider it.

DR. GUTMANN:  You are putting in high relief the value, the public value of sharing peer-reviewed findings and sharing the data behind them. And the authors who said that their data sets were lost are really violating a norm, an ethical norm of publication, which is to have the actual data that underlie your conclusions. So to get to neuroscience which we are dealing with, we would -- we are dealing with this issue only because there is a reason to believe there is a high public value in having data sharing moving forward, which is to anticipate what Dan said before, but the value is, in your response to Anita's question, the value is to have an ethically valid way of data sharing. And I have no doubt that there is a way of doing that, and that's what we are discussing.

And John has the last question before we break for lunch and give both of you an opportunity to have the last word.

DR. ARRAS: Thank you, and thank you both. So hats off to Professor Nissenbaum for sort of breaking open a pretty stale debate on privacy and consent. I really found your paper in the briefing book to be very stimulating, okay? But as somebody weaned on notions of consent and anonymization, I'm still having trouble wrapping my head around this different way of looking at things, okay? So you are saying that instead of focusing on consent and anonymization we should be looking at a host of other issues. Information flow is one of them. But that seems to me to be the answer to the question, like how should the information be flowing, right? That's the conclusion that we would want to come to.

Now, with regard to the way in which big data can render obsolete the concern for consent and anonymization, I'd like you to get a little bit more concrete with us about exactly what sort of alternative analysis you would propose. Because, I mean, one way to put it might be that the emphasis on consent is really a kind of rights-based language that the proper social policy should be the one that acknowledges and respects individual rights. So if I understand you correctly, and I'm not at all sure that I do, when you talk about these contextual issues, you are really in a way sort of transmuting the notion of consent into a kind of interest of individuals that should be taken into account but not in the same way that a rights-based account would. Is that a fair statement? Because you clearly are concerned about impacts on the individual, right? In your paper you talked about the tyranny of the minority and the disturbing ways in which these companies and social media networks can generate information about us even if they're just gleaning it from other people's data, right? So you see that is a threat to the individual's interests, right? So could you just -- I guess I'm -- for a philosopher this is an atypical question but I'm asking you to be a little bit more concrete. Sorry.

DR. NISSENBAUM: No, no. Thank you very much for asking those things. The reason I'm pessimistic about consent is because I don't think it can achieve anymore what it was supposed to achieve. Some of that was resonating in the comments made over there in that corner of that room. And so because -- and the same thing with anonymity. We couldn't achieve it in some shallow sense of it, but we want to say why was anonymity something we went for in the first place? Why are we agonizing about how to express notice, how to implement or operationalize consent? Not because we think they're ends in themselves, but they are kind of procedural proxies for what we want to achieve at the end of the day.

So I'm not objecting to rights based, and I think the individual is crucially part of this story, but because of big data, those things we used to be able to achieve with anonymity and consent we no longer can guarantee. And so either we are going to shut down data completely, and I don't think that's a good idea at all, or we need to actually look directly at the question and say are we justified in getting the data that we are getting about individuals, and confront that question. Is it important enough to be gathering that kind of information, to be providing access to that information?

 By the way, data access, you can provide access to data in a whole variety -- that's what information technology can do for you too. It doesn't necessarily have to be placed on the open web with no constraints whatsoever. We can fine print in those things. So I keep pushing back to -- I think that the Commission's work is really -- I would like to, you know, see that as, when dealing with questions of data, to ask, what are the goals that are being served and how important are they. We need to answer those questions as we would for someone who is impaired, for someone who is cognitively impaired. Those are the kinds of questions we need to ask now and not just to shove it all off onto the individual.

DR. GUTMANN: Giorgio, would you like to have the last word?

DR. ASCOLI: Yeah. I would like to just note the issue of who owns the data to start with. And in case of human data, we can consider the subject in the picture. In the case of elements of our society that do not have legal ownership status, such as rats and worms and zebrafish, then the question is whether the data is owned by the researcher or by society. And I think that the question might be both in equal divisions and equal aspects of it.

But I think that part of the issue is how we reward the sharing of the data. So right now, for example, in genomics, if I discover a new gene, it would be crazy for me not to post it on GenBank because of my chance to win the Nobel Prize 20 years later as a link to the fact that this gene will eventually cure cancer, it doesn't really matter whether I'm the one that discovers the drug for cancer based on that gene or someone else does, it's still the gene that I deposited and I'll probably share the Nobel Prize with the person who discovers the drug. Whereas that shift in culture of rewarding the sharing of data by providing credit for the data that was shared at all levels from tenure committees in universities -- I know we have two presidents of universities at least at the table -- all of the way down to IP issues and financial remunerations, and, of course, academic credit, has not quite broken through in neuroscience. So I think the issue is really an issue of how to reward the sharing of that ownership for both society  and the principal investigators.

DR. GUTMANN:  Thank you very much, and on behalf of all of us thank Drs. Nissenbaum and Ascoli.

This is a work of the U.S. Government and is not subject to copyright protection in the United States. Foreign copyrights may apply.