Intellectual Property Policy at The Neuro, an Open Science Institute

In 2016 The Neuro, otherwise known as the Montreal Neurological Institute and Hospital, became the world's first open science institute. The Neuro's definition of and approach to open science is captured in its Open Science Principles. While most of the goals embedded in the principles have received increasing approval from researchers, funders, and other institutes, The Neuro's stance on intellectual property continues to see resistance. In this paper I describe how The Neuro conceives of the relationship between open science, neuroscience, and intellectual property; how it came to its current policy by examining both broad and Neuro specific innovation trends; and why it believes minimizing academic intellectual property is key to the future of neuroscience research.

1 . Op e n S ci e n ce a n d I n t e l l e ct u a l P r op e r t y a t T h e 1 . Op e n S ci e n ce a n d I n t e l l e ct u a l P r op e r t y a t T h e N e u r o N e u r o When folks from T he Neuro talk about open science, no topic is more likely to receive pushback than its stance on intellectual property ("IP"). Data sharing, great; open access publishing, absolutely; sharing code, protocols, and reagents, all for it; not filing for patents or enforcing restrictive IP rights over research outputs, well, wait a minute now.
T o bring some clarity to T he Neuro's position, and emphasize its importance, I want to explain how T he Neuro understands the relationship between neuroscience research, open science, and intellectual property. T o do so I begin by discussing some fundamental concepts and historical examples, move on to an explanation of T he Neuro's policy towards IP, and finish by discussing IP-based innovation trends both generally and specific to T he Neuro.

The Neuro and The Need for Open Neuroscience
T wo things define T he Neuro's institutional identity. One is a commitment to uncovering the mysterious workings of the central nervous system through research; the other is providing world-class clinical care to patients with neurological disorders. T hese two aspects are, of course, deeply intertwined. Following the vision of T he Neuro's founder, the inimitable Wilder Penfield, clinical experience influences research, the insights from which are in turn integrated into the clinic.
As grand as this vision of a feedback loop between neuroscience research and clinical practice was -indeed unique in 1934 when T he Neuro was founded -it has become overwhelmingly obvious that, when it comes to understanding the nervous system, it is not enough. With ~86 billion neurons in the adult human brain [1] , each having potentially thousands of dynamic, synaptic connections to its peers (not to mention interacting in largely unknown ways with vast numbers of glial cells), no single institute can possibly generate more than a fraction of the discoveries necessary to understand and treat neurological disorders. In light of this, T he Neuro has come to fully embrace the idea that "it takes the world to understand the brain" [2] and that the only way to effectively collaborate on a global scale is to adopt radical sharing through open science.
T he need for open science becomes yet more acute when one brings into the equation the current reproducibility (and replicability) crisis plaguing much of the scientific research world [3] [4] . Communicating findings through research manuscripts may have been an astounding leap forward when it first arose in the 17th century, but now clearly fails to Qeios, CC-BY 4.0 · Article, July 28, 2020 Qeios ID: OMUWEL · https://doi.org/10.32388/OMUWEL 2/20 meet science's own exacting standards. A paper alone is, simply put, too often inadequate to enable others to reproduce, replicate, remix, and re-use reported results in the ways needed to tackle the complexity of the human nervous system. Doing so requires, at least, open access to and use of the data, research methods, materials, and software underlying the reported results.
What T he Neuro came to realize, in sum, is that the way neuroscientists work together needs to once again evolve if progress in understanding the human nervous system and treating neurological diseases is to be more than piecemeal and unreliable. Just as the founding of T he Neuro created a new species of institute engaged in neuroscience research, and sharing results through papers and journals was a leap in scientific communication, the time has come for a new evolutionary epoch in neuroscience.
History will come to know that epoch as the age of open neuroscience.

Intellectual Property and Open Neuroscience
T he connection between open science and intellectual property often seems daunting in its complexity, but taking a step back can be reduced to a relatively simple set of ideas.
Open science requires that the sharing and subsequent use of all information, tools, and knowledge relevant to an experimental result be as frictionless as possible; that, in other words, others are able to access and use everything that went into and came out of an experiment with a minimum of restrictions. Intellectual property exists precisely to restrict access to or use of information and knowledge. Patents restrict how knowledge is applied to produce a result; copyright restricts how knowledge and information are copied and disseminated; technological protection measures coupled with anti-circumvention laws (aka DRM) and trade secret laws restrict access to information; and trademarks and certification marks restrict how information about an entity's identity can be used.
T here are limits and restrictions on the breadth of various IP instruments -for example the inability to patent abstract knowledge, equations, and naturally occurring phenomena; or the fact that copyright does not cover facts but how they are creatively expressed -but the above description will serve as a useful and accurate heuristic for understanding how IP impacts knowledge generated through research. As a final note here, I am primarily addressing IP relevant to Canada and the United States, and so will not be addressing things like database rights, though the concerns herein apply to many non-North-American species of IP as well. T o state it baldly, the presence of most forms of IP coupled with enforcement activities aimed at preventing others from accessing and using knowledge hampers the most effective dissemination and use that knowledge. While it is possible to have IP and simply not engage in enforcement activities, doing so injects unnecessary uncertainty -it is always possible that the IP will be sold or exclusively licensed to an entity that will enforce it or that the rights holder may simply change their mind -and in the case of patents leads to high costs for essentially no gain. When the results so restricted are the basic building blocks of further work the cost is large and, as the history of university technology transfer and the research publishing industry has demonstrated, the benefits few and unevenly distributed. In view of these issues T he Neuro adopted, through its Open Science Principles, a policy under which it will not claim restrictive IP rights over any of its research outputs nor will it support its researchers in doing so.

Standing on the Open Shoulders of Giants
T hankfully, T he Neuro did not have to start from a blank slate when designing its policy approach to IP. When it comes to patents and trade secrets, T he Neuro could adapt the approach taken by the Human Genome Project ("HGP"). Under the Bermuda Rules [5] , Qeios, CC-BY 4.0 · Article, July 28, 2020 Qeios ID: OMUWEL · https://doi.org/10.32388/OMUWEL 4/20 those engaged in decoding the human genome agreed not to patent any DNA sequences and to publicly release those sequences within 24 hours. T his approach was necessary if the required level of collaboration was to be achieved; bickering over ownership and multiple labs racing to sequence the same potentially high value target would simply not get the job done.
T he connection between the HGP and the current effort to solve the neural code (or "neural connectome") is entirely apposite. Both are collective journeys of basic scientific discovery tackling a problem much too large for any single lab, institute, or even country to take on itself. T he Structural Genomics Consortium, engaged in large part in decoding the humanproteome, has long followed its own open science philosophy. Each of these efforts is tied together by their ambition to solve a different facet of complex human biology, and each has recognized the importance of coordination and eliminating IPbased barriers to collaboration, dissemination, and reuse.
Advances in the diagnosis and treatment of human maladies is based fundamentally in our basic understanding of biology, which, given the complexity of biological systems, in turn depends on the collaboration and open sharing. Previous research practicesensiled and uncoordinated research, sharing research results solely through (mostly paywalled) manuscripts that lack details crucial to replication and reproduction, and patenting basic discoveries -are clearly not working.
In the neuroscience context, this conclusion is amply demonstrated by the continued reliance on treatments discovered in the 1950s for Parkinson's disease [6] and the high failure rate of new potential treatments for Alzheimer's Disease [7] [8] [9] , Huntington's Disease [10] , and ALS [11] . We need better models, theories of disease mechanisms, and biomarkers if we are to have a real impact on the suffering caused by neurological disorders. T he only real hope of gaining these fundamental insights when faced with a problem as complex as the human nervous system is through an open, maximally IP-free approach.

The Neuro's Position: No Restrictive Intellectual Property
T he Neuro's approach to intellectual property, implicit throughout its Principles but found explicitly in Principle 4, is one they have called "no restrictive intellectual property." It means, in short, that T he Neuro's researchers should not apply for or enforce any form of IP over research outputs unless doing so is to protect the interests of research participants or contribute to the vitality of the scientific research ecosystem.

Non-Restrictive IP Rights
It will be useful to explain in more detail what is meant by IP being "non-restrictive". T he idea breaks down into two general categories.

Protecting the Interests of Research Participants
T he first and most important category relates to circumstances where a shared resource involves concerns around research ethics and the interests of research participants. T his is most clearly the case with individually identifiable data. If, for example, MRI data is bundled with clinical information that could be used to identify a participant, it may be appropriate to include some technological protection measure or contractual restriction such that that identifying information can only be accessed with a password, within a secure system, by registered users, or after evaluation of a data access request.
T he tiered access model developed for T he Neuro's C-BIG Repository is a good example of how these ideas are being put into practice. Data which pose little risk to participants' interests (like metadata) are openly accessible, those posing some risk require registration and agreement to a terms of use and a set of community norms, while access to individually identifiable data and samples is controlled by C-BIG's T issue and accepted because they align with the core values of science and its unique incentive structure. As a large bolus of literature -starting with Merton's norms of science [14] and perhaps explained best in economic terms by Paul David [15] -has recognized, the research ecosystem has its own unique set of norms and incentive structures that motivate discovery. T hese norms and incentives, however, are not those of wealth maximization, property, and economic transactions. T hey are instead grounded in collective ownership of the fruits of intellectual labour, universal validity, organized skepticism, and disinterested experimentation (for norms) and reputation within your field, career advancement, the joy and challenge of discovery, and contributing to the store of public knowledge (for incentives). while applying a minimum of legal restrictions. T hey also speak directly to some of the concerns expressed by those in T he Neuro's research community during its buy-in process [16] .
6. Doing our Research on IP and Innovation T he Neuro's decision to adopt an approach that avoids restrictive intellectual property was informed by more than the historical and theoretical considerations discussed so far.
It also considered the current relationship between IP and university research generally as well as the innovation history of T he Neuro. When these basic knowledge outputs are subject to the restrictions imposed by patents because of potential (and in the vast majority of cases unrealized [17] [18] ) commercial potential, they pose at worst a serious risk of hindering downstream research and future commercial development [19] and at best contribute little to downstream innovation [20] and the public good [21] . In a world with ubiquitous interconnected information technology, and the ease such technology brings to sharing information, the best way of making sure important basic knowledge gets to where it needs to go is to make it FAIR and get out of the way. T he all too likely result of increased university patenting practices [22] is simply a contribution to a growing patent thicket, like that routinely commented on in nanotechnology [23] . Non-commercial research is, unfortunately, not legally exempt from the resulting tangle of patent rights [24] , meaning that both private and academic researchers must overcome a prolix nest of IP rights and monopolistic prices, or waste time designing around the thicket.

Contributing to T hickets
If, as is so often the case, researchers lack sufficient time or financial resources to obtain a license or buy equipment with a high monopoly-dictated price, they may be forced to abandon research directions all together. Alternatively, they may need to resort to extralegal measures like pirating research software or obtaining reagents directly from colleagues without institutional permission to do so, risking infringement claims and institutional disciplinary actions.

Perverse Incentives
T he prevalence of patenting at universities has negative impacts throughout the institutional hierarchy. At the level of the researcher, the possibility of a patent over a research output injects a direct financial incentive into what is supposed to be unbiased discovery. Doing so can lead to counterproductive practices such as not sharing the tools others need to replicate and build on published results [25] [26] [27] . It can also lead to significant publication delays as researchers attempt to avoid novelty killing public disclosures which would make their research outputs unpatentable.
Almost certainly the biggest cost, however, is not due to researcher secrecy or egos, but to institutionally caused delays in sharing resources due the ongoing realignment of institutional values towards commercialization through IP. T he (very small) possibility of a windfall through successful licensing or creation of spin-off companies has slowly introduced complex and unnecessarily restrictive IP clauses in material transfer agreements and research collaboration agreements. T hese IP clauses routinely pose the greatest negotiation sticking-points, and require the most negotiation time, both for transfers [28] and for industry collaboration contracts. Eliminating such clauses gives up little and would have a multiplicatively beneficial effect to collaboration.
T he resulting delay interferes with researchers' ability to rapidly share their tools and effectively collaborate [29] . Despite the fact that such interference in each individual case may be small -though still often on the order of weeks or months -the cumulative effect represents years of waste. What is worse, the majority of academic researchers have

Patent Litigation and T rolls
With the increasing incidence of patent litigation generally [30] , including by universities [31] , concern around real and potential negative impacts of patents on academic research mounts. A good example of the situation T he Neuro wants to avoid is the current patent battle over CRISPR [32] . Forcing the transfer of perhaps the greatest tool genetic science has ever known through the bottlenecks of IP-based considerations will not help science advance. T he historical cases of the Oncomouse [33] , involvement of universities in cases defending patents over genes [34] , and the ongoing dispute involving a CAR-T therapy [35] also present significant cause for concern.
It is also extremely troubling that university patent practices can either feed patent trolls [36] (also see this 2016 dataset of university patents owned by Intellectual Ventures) or, worse, cause universities to act like them [37] . Assertion of patent rights is big business with the potential for substantial returns, but only at the cost of blocking the fluidity of ideas that is the lifeblood of science. I then looked more deeply into the currently active patents: T wo were for an ROI concerning an antisense oligonucleotide with potential utility in treating cancer, three related to two ROIs regarding software for analyzing brain morphology, and the remaining eight were for a single ROI relating to a gene associated with epilepsy.

Commercialization of Patents Over Neuro Research
T he antisense oligonucleotide patents were subject to an exclusive license agreement that was terminated before leading to a product. T he three software patents constitute Qeios, CC-BY 4.0 · Article, July 28, 2020 Qeios ID: OMUWEL · https://doi.org/10.32388/OMUWEL 12/20 a portion of the assets owned by a Neuro spin-off that has been running for approximately 15 years but which, according to discussions with the founder, has had to divert much of its revenue into IP related legal and maintenance fees and, because of an IP inventorship dispute, is now unable to sell the company.
T he eight gene related patents, applied for many years ago by T he Neuro's Director Dr.
Guy Rouleau and some of his colleagues, consist of 1 Australian patent licensed to Bionomics Ltd, 1 Canadian patent, 1 Japanese patent, and 5 US patents based on continuation or divisional applications to the USPT O -the multiple applications likely due to a scramble to recover patentable subject matter in the wake of the Myriad decision [40] . None of these patents have led to a successful diagnostic or treatment product. As noted in Dr. Rouleau's recent article in Cell [41] , this failure to produce any useful product is a significant part of his motivation in transitioning T he Neuro to an open science model.

Innovation Activities through Neuro Researcher-Entrepreneurial
Activities and Spin-Offs 8. T h e I m p ossi b i l i t y of Op e n i n g M cG i l l -I P f or t h e 8. T h e I m p ossi b i l i t y of Op e n i n g M cG i l l -I P f or t h e COV I D-1 9 P a n d e m i c COV I D-1 9 P a n d e m i c disseminate materials (e.g. Addgene) at a fraction of the cost of proprietary models.
T hese efforts are largely driven by members of the scientific community itself and represent a slow but seismic shift away from companies engaged in linear, restrictive, and secretive development and towards a model based on distributed, facilitated collaboration between members of the relevant scientific community.
Other possibilities, including tool and technique vetting services (e.g. Y-Char-OS), synergy between open science and artificial intelligence [43] , open science drug development [44] , T he great strength of all of the above efforts is the ability to capitalize on or facilitate global collaboration and rapid, iterative development of products and services. By delivering value to user-collaborators and allowing rapidly iterative modification of products and services without relying on IP they can avoid the often weighty costs and heavy restrictions associated with obtaining and enforcing IP all while contributing to the well of public knowledge (not to mention remaining financially sustainable).

Conclusion: Trying, Testing, and Sharing
T he Neuro is confident that the approach is has developed is -given its institutional context, values, and history -the best way to encourage basic science, the commercial development of diagnostics and treatments, and neuro-based innovation generally. It is always possible, however, that the approach will need to be changed or refined.