Developing a 21st Century Neuroscience Workforce: Workshop Summary (2015)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Training in Transdisciplinary Research Key Highlights Discussed by Individual Participants • Effective collaboration among disparate disciplines is critical (e.g., molecular biology and bioengineering or electrophysiology and physics); in this space innovation will arise and there are opportuni- ties to train scientists to work toward a sum greater than the collected parts and spur innovation (Federoff, Sejnowski, and Steward). • There is a distinction between training of neuroscientists and training in neuroscience, and several participants raised the question of whether graduate programs should focus on one or the other (Steward). • As neuroscience expands in scope, adding various tracks might be beneficial (Landis, Litt, Raman, Sejnowski, and Steward). • Forming successful transdisciplinary collaborations requires time and involves a considerable amount of risk, and can be encouraged with increased incentives with regard to grant review and hiring/promotion decisions (Sejnowski). • Opportunities for transdisciplinary training come from a variety of sources, including National Science Foundation grants and BRAIN Initiative short courses (Ferrini-Mundy and Litt). NOTE: The items in this list were addressed by individual participants and were identified and summarized for this report by the rapporteurs. This is not intended to reflect a consensus among workshop participants. As neuroscience has evolved as a discipline and incorporated many types of science—microbiology, genetics, statistics, animal behavior, optics, engineering, computational biology, etc.—it has become increas- 43 

44 DEVELOPING A 21st CENTURY NEUROSCIENCE WORKFORCE ingly less likely that any one individual or laboratory will have all the expertise needed to tackle higher-level problems. Many workshop partic- ipants noted that teams of scientists from disparate disciplines are neces- sary for improving fundamental neuroscience knowledge, developing new treatments, building the next powerful tool, and revolutionizing im- aging technology. But it is not a matter of getting out a checklist and making sure each team has a biologist, a physicist, a mathematician, an engineer, and a chemist. Rather, each of these disciplines needs to learn how to work effectively with the others. According to several workshop participants, in this space innovation will arise and there are opportuni- ties to train scientists to work toward a sum greater than the collected parts. Oswald Steward, director of the Reeve-Irvine Research Center at the University of California Irvine School of Medicine, enumerated several decision points for graduate programs to consider when training students to engage in transdisciplinary research. Terry Sejnowski emphasized the need for better incentives to encourage scientists to collaborate on transdisciplinary projects. Howard Federoff, executive dean of the Georgetown University School of Medicine, provided an overview of how best to enable transdisciplinary teams to do translational science. Finally, Dennis Choi, director of the Neurosciences Institute at Stony Brook University, called for improving collaboration between basic neu- roscientists and clinicians. DEFINING TRANSDISCIPLINARY NEUROSCIENCE Steward challenged participants to think about the requirements for successful transdisciplinary collaborations and what impact those re- quirements can have on training students. In his opinion, a transdiscipli- nary team should be composed of specialists, not people with a general knowledge of their discipline. For example, if a team requires a neuro- scientist, he noted that this person needs to be a card-carrying neurosci- entist, that is, someone with a Ph.D. in neuroscience who regularly does neuroscience research. Similarly, if a neuroscientist were putting together a team requiring a mathematician, the neuroscientist would want a card- carrying mathematician, not a neuroscientist with some knowledge of math. Although several participants had differing opinions on the type of neurosci- entist needed for such transdisciplinary collaboration—particularly given that most individuals engaged in the neuroscience workforce would not be 

TRAINING IN TRANSDISCIPLINARY RESEARCH 45 considered a card-carrying neuroscientist—Steward noted that there is still a need for experts in the discipline of neuroscience. Related to the question of specialization versus generalization, Steward discussed the differences between the training of neuroscientists—training in the core knowledge of a discipline that qualifies one to be called a neu- roscientist—and training in neuroscience—training for individuals in other disciplines that would allow them to be partners in the greater enterprise of neuroscience research. Steward’s opinion was that a single graduate pro- gram probably could not effectively accommodate both of these needs. Instead, programs should carefully consider these differences when devel- oping their training goals and either define themselves as institutes, which are cross-disciplinary, or departments, which have a specialized focus. As for the training of neuroscientists, a number of participants asked, what does every card-carrying neuroscientist need to know? While there were no clear answers, many participants agreed there should be a limit to how many additional courses should be required of students given the already long average time to earn a degree, even if that limit means sacri- ficing breadth of knowledge. According to a few workshop participants, one way around this class time limit is to offer micro- or nano-courses, rather than semester-long courses, to give students a chance to sample relevant topics. In addition, several workshop participants suggested that many neuroscience courses could be more effective as a series of coordi- nated hands-on exercises or demonstrations rather than traditional di- dactic lectures. More importantly, graduate students need to be trained in how to do rigorous science (as noted in Chapter 3) and establish effective transdisciplinary collaborations. Another potential training-related choice that Steward pointed out is training students to be Renaissance scientists or goal-directed scientists. The former operate in a mode of pure exploration and discovery, while the latter can be plugged into teams to solve specific problems and de- velop treatments for disorders. Again, Steward suggested that different tracks are needed to train each type of scientist. Reconsidering the One-Size-Fits-All Approach to Training Several workshop participants also expressed the need for separate tracks in neuroscience training, although the dimensions along which to separate varied. A comment by Landis captured a common sentiment expressed throughout the workshop: “There’s too much neuroscience, it’s not one thing anymore.” She asked how much cellular and molecular 

46 DEVELOPING A 21st CENTURY NEUROSCIENCE WORKFORCE neuroscience training is necessary for someone who is doing brain map- ping at the macro level? How much magnetic resonance imaging infor- mation does someone who is working at the cellular and molecular level need to have? Does a neuroscientist 10 years from now need to be fully articulate in all of the areas of neuroscience? Chesselet noted training programs should be designed in a way in which there is a balance be- tween trainees being aware of a topic compared to having the knowledge (often best acquired in a laboratory setting) to apply it to a research pro- ject. Landis suggested that one solution to narrowing students’ focus was to look for lessons from the field of neurology, where there is a core res- idency program, followed by subspecialty training. Several workshop participants stated that separate tracks might allow for more focused courses—and possibly less overall class time for trainees—and might encourage the development of mini-courses that address particular prob- lems in a certain subspecialty. Potential tracks that could be created in graduate neuroscience programs include electrophysiology, optical imag- ing, fMRI, cellular and molecular neuroscience, translational neurosci- ence, neuroengineering, theory and modeling, and systems neuroscience. Trainees could also be split into theoretical or experimental tracks, sug- gested Litt. Such tracks could potentially be organized around current faculty expertise and available core resources and infrastructure at each institution. Several workshop participants noted that these subspecialty tracks might serve as the basis of tight-knit communities among students and alumni that might be advantageous when seeking internships and employment opportunities. Indira Raman suggested stratifying trainees according to their interested career pathway as well (i.e., academia and non-academic careers). Finally, Sejnowski and Steward suggested that questions about spe- cialization and the formation of graduate school tracks may ultimately be dictated by outside forces; what kind of workforce do employers need, and, to a lesser degree, what kind of training will funding agencies sup- port? Will there be more jobs in goal-directed science? Is the workforce trending toward large teams? Will these teams need a certain ratio of specialists to generalists (who might be in a better position to support or manage all the moving pieces in a lab)? Will workforce needs vary across subfields of neuroscience? Understanding these needs will be crit- ical for optimizing graduate neuroscience training, said Steward. 

TRAINING IN TRANSDISCIPLINARY RESEARCH 47 Challenges to Cooperative Science Sejnowski mentioned two primary challenges to overcome to en- courage more collaborative science: trust and acknowledgment. Forming collaborations with scientists in other fields is similar to getting married, he noted. You have to have trust, get to know the person, and work to- gether long enough to develop a common language. According to Sejnowski, seeking out scientists from other fields to collaborate and be on the same page on large, complex problems can take such a considera- ble amount of time and risk that collaboration may not be viewed as worth the effort. These factors could especially dissuade new faculty, who are eager to publish high-impact articles and obtain grant funding in order to secure tenure. The field needs a better way, according to Sejnowski, to reward scientists for building cooperative, transdiscipli- nary teams. One way to do this is to convince academic departments and funding agencies to assign collaborative projects more weight when making decisions about promotions and grants, akin to the suggestion for the sharing index and the data citation mentioned in Chapter 2. NIH, for its part, has begun to recognize the importance of taking risks to encour- age interdisciplinary research with the advent of its Common Fund High Risk, High Reward program,1 which supports the Early Independence Award,2 the New Innovator Award,3 the Pioneer Award,4 and the Trans- formative Research Award.5 Initiatives for Cooperative Science Transdisciplinary collaboration was a vital part of the discussion among the NIH BRAIN Initiative Working Group, according to Sejnowski, a member of the working group. One of the seven core prin- ciples of the initiative that is listed in the BRAIN 2025 report6 is cross boundaries in interdisciplinary collaborations (NIH, 2014). Within the report, potential collaboration scenarios to facilitate the BRAIN Initia- tive’s goals were discussed: 1 See http://commonfund.nih.gov/highrisk/index (accessed October 29, 2014). 2 See http://commonfund.nih.gov/earlyindependence/index (accessed October 29, 2014). 3 See http://commonfund.nih.gov/newinnovator/index (accessed October 29, 2014). 4 See http://commonfund.nih.gov/pioneer/index (accessed October 29, 2014). 5 See http://commonfund.nih.gov/TRA (accessed October 29, 2014). 6 BRAIN 2025: A Scientific Vision. See http://www.braininitiative.nih.gov/2025/BRAIN 2025.pdf (accessed October 28, 2014). 

48 DEVELOPING A 21st CENTURY NEUROSCIENCE WORKFORCE • “The physicists and engineers who develop optical hardware should partner with the biologists and chemists who develop new molecular sensors. • The tool builders who design new molecules for sensing or regu- lating neurons should partner with neuroscientists who will rig- orously examine their validity in neurons and brains. • The theorists who develop models for understanding neuronal dynamics should partner with experimentalists, from initial ex- perimental design to execution to interpretation. • The clinicians and neuroscientists who develop sophisticated im- aging methods in humans should partner with scientists working in animal models who can relate imaging signals to the underly- ing cellular mechanisms with great precision” (p. 51). Finally, Sejnowski mentioned the imperative for transdisciplinary re- search held by another science initiative of which he is also an organiz- ing member, the National Academies Keck Futures Initiative (NAKFI). NAKFI brings together scientists from disparate fields to work intensive- ly over 2 days in small teams to address issues related to important sci- ence problems. A few years ago, NAKFI conducted a survey of 600 stakeholders to examine how research could be more innovative. The survey asked individuals to rate the importance and ubiquity of several factors that are integral to interdisciplinary collaborations, including data accessibility, institutional support, responsive funding, ingenuity/risk taking, incentives, and education/training (see Figure 4-1). The results revealed that the more critical gaps—those factors rated high in im- portance and low in ubiquity—were responsive funding, incentives, and ingenuity/risk taking (see Figure 4-2). Katja Brose, editor of Neuron, cautioned workshop participants about moving into the direction where it is all about team science. In her opinion, what makes neuroscience spe- cial is the diversity of topics and approaches. There are instances in which collaboration at the investigator level or between laboratories is more desired than team science, in which several experts come together for a common goal. 

TRAINIING IN TRANSD DISCIPLINARY RESEARCH R 49 FIGUR RE 4-1 A conceptual model of the variou s factors that ccontribute to een- hancingg innovative reesearch. SOURC CE: Terry Sejnowski presen ntation, Salk Innstitute for Biological Studiies, Octobeer 28, 2014. OP PPORTUNIT TIES FOR TR RANSDISCIIPLINARY T TRAINING Thhroughout thee workshop, several particiipants discusssed a number of opporttunities to sup pport transdissciplinary traiining, to incluude incorporaat- ing coourses about collaboration c in neurosciennce graduate programs. D De- scriptions of a few w of these funding oppportunities annd courses aare providded below. Ak kil and Chessselet stressedd that education about tranns- discipllinary researcch should not stop in ggraduate schoool, but rathher shouldd be modeled after continu uing medical eeducation couurses for physi- cians, in which education e is incorporatedd throughoutt the traineee’s lifespaan. 

50 DE EVELOPING A 21st 2 CENTURY Y NEUROSCIEN NCE WORKFORCE FIGUR RE 4-2 Resultts from a surv vey of scientissts across manny disciplines rre- spondinng to question ns about the immportance of a variety of fa factors (and thheir prevaleence) for increaasing innovatio on in science. SOURC CE: Terry Sejnowski presen ntation, Salk Innstitute for Biological Studiies, Octobeer 28, 2014. NSF Research Traineeeships NRRT7 supports the developm ment of innovvative traininng programs ffor teams of graduate students within a single uniiversity or insstitution arounnd a cross-disciplinaryy theme relatted to nationnal research ppriorities. NRRT replacees NSF’s IGERT grants, the last of w which were iissued in 20113. Joan Ferrini-Mundy F y, assistant director d of thhe Directoratee for Education and Human Resourrces at NSF, noted that beetween 1998 and 2013, 278 IGERT 1 lead univversities in suupport of 6,500 T awards werre issued to 100 graduaate students. Neuroscience N e projects accoounted for 155 percent of tthe IGERT T awards. Preevious neuroscience-relatedd IGERT awaards focused on themess such as neu uroimaging ofo non-humann primates, neuroprosthesees, 7 See http://www.nsf.gov/funding/pgm m_summ.jsp?pim ms_id=505015 (accessed Octobber 28, 2014 4). 

TRAINING IN TRANSDISCIPLINARY RESEARCH 51 and computational neurobiology. The initial priority research theme of the new NRT program is Data-Enabled Science and Engineering. How- ever, proposals are encouraged on any other crosscutting, interdiscipli- nary theme. Ferrini-Mundy added that time to degree was not slowed for students doing interdisciplinary research on IGERT grants. She also mentioned that students would report anecdotally that their ability to communicate and to get other people excited about their science improved as a result of their IGERT experience. Integrative Strategies for Understanding Neural and Cognitive Systems (NSF-NCS) The NSF-NSC grants8 support transformative science and engineer- ing efforts to accelerate knowledge of neural and cognitive systems. Be- cause the complexities of the brain and behavior touch on many aspects of science and engineering, these grants will cut across NSF’s various directorates. For 2015, the NSF-NSC grants are organized around two research themes: (1) Neuroengineering and Brain-Inspired Concepts and Designs and (2) Individuality and Variation. Within each theme, projects will address general advances in theory and methods, technological in- novations, educational approaches, enabling research infrastructure, and workforce development. BRAIN Initiative Short Courses In recognition of the critical role cross-disciplinary research will play in developing the next generation of tools and computational approaches for studying the brain, the NIH BRAIN Initiative plans to sponsor short courses9 for training graduate students, medical students, postdoctoral scholars, medical residents, and/or early-career faculty. Courses will be offered to neuroscientists as well as to scientists from other disciplines. One course will focus on tools to classify cell types, reconstruct neural pathways, and record from and manipulate neural circuits using electrical 8 See http://www.nsf.gov/funding/pgm_summ.jsp?pims_id=505132 (accessed October 24, 2014). 9 BRAIN Initiative short course about data analysis/handling. See http://grants.nih.gov/ grants/guide/notice-files/NOT-MH-15-005.html (accessed October 28, 2014). BRAIN Initia- tive short course about innovative tools. See http://grants.nih.gov/grants/guide/notice- files/NOT-MH-15-006. html (accessed October 28, 2014). 

52 DEVELOPING A 21st CENTURY NEUROSCIENCE WORKFORCE and optical techniques. The other course will focus on quantitative meth- ods for analyzing high-dimensional imaging, electrophysiological, anatom- ical, and behavioral datasets. Aside from using these relatively low-cost short courses as a training tool, the staff organizing the NIH BRAIN Initia- tive hopes the courses will expose physical and information scientists to the projects in the initiative and help foster cross-discipline collaborations. National Academies Keck Futures Initiative Seed Grant Program10 NAKFI has hosted an annual meeting over the past 10 years on how to push forward innovations through interdisciplinary interactions among many different fields of science and engineering. Previous meetings with a strong neuroscience focus include signaling, complex systems, imaging science, the digital brain, and collective behavior. Participants are invited to self-organize into cross-disciplinary teams to apply for seed grants to further pursue ideas stimulated by conversations and breakout sessions that occur over the course of the 2-day meetings. University of Pennsylvania’s Brain-Computer Interface Course One challenge in bringing multiple disciplines together to work on a neuroscience problem is communication, said Litt. Every field has its own special language, and to some extent, its own worldview. Litt de- scribed his “Brain-Computer Interface” course, which brings together students in neuroscience, physical science, and engineering, as a model for training disparate groups to communicate and collaborate with one another. In addition to classroom lectures on topics such as modeling and simulation, the students work in interdisciplinary teams on 10 hands-on programming projects. Project examples include modeling visual cortex orientation tuning columns, controlling robot arms driven by motor units, classifying speech, and designing cochlear implants. CROSS-TRAINING IN CLINICAL NEUROSCIENCE Despite the historic intersection between neuroscience and clinical science, there is little cross-training between the two disciplines, said Choi. The clinical interface is central to neuroscience for at least two rea- 10 See http://www.keckfutures.org/grants (accessed October 29, 2014). 

TRAINING IN TRANSDISCIPLINARY RESEARCH 53 sons, he added: (1) the potential for medical benefit is a key source of inspiration, purpose, funding, and public support; and (2) the interface represents the necessary experimental platform for investigating, and ultimately understanding, the human mind. Although the clinical dimen- sion in neuroscience training has been around longer than more novel and emerging dimensions, such as genetics, engineering, and informatics, Choi noted that basic neuroscience training typically provides limited exposure to principles of clinical medicine, clinical research, and overall disease biology. The converse is true as well; clinical training typically provides scant exposure to the scientific method. Choi cited four primary consequences of the lack of cross-training in these areas. First, neuroscientists without clinical training are vulnerable to what Landis referred to as “pseudo-translation”—studies that combine disease models of uncertain value with interventions unlikely to ever be applicable to patients. These ideas can gain inappropriate traction and lead to dilution of resources and disappointment for myriad stakeholders, said Choi. Second, he noted that clinicians without basic neuroscience knowledge are vulnerable to unsubstantiated claims about the brain and are prone to adopting dogmatic approaches based on anecdotes rather than available evidence. Third, without cross-training and collaboration, opportunities to exploit the clinical setting for studies in basic neurosci- ence can be missed. Choi used work by Edward Chang at the University of California, San Francisco, as an example of such research. Chang took advantage of intraoperative electrocorticography for patients about to undergo epilepsy surgery to test theories about how speech is perceived and generated by the brain. Choi added that as new tools and technolo- gies provide answers to research questions the neuroscience enterprise cannot afford the divide between basic and clinical science. Fourth, the lack of exposure to, and appreciation of, the clinical method creates situ- ations where researchers might not truly understand clinical data. There are fundamental risks, Choi said, for bias in clinicians taking clinical his- tory and in performing examinations of the nervous system. Once those biases are incorporated into databases where phenotype is linked with genetic and imaging information, Choi stated that they become intracta- ble and can greatly affect study outcomes. Another obstacle to maximizing the potential of the clinical interface is the historical balkanization in the way patients and diseases are man- aged in medical centers today, noted Choi. The primary example of these long-standing divisions is between neurology and psychiatry. Despite the fact that the distinctions between their missions are blurring, they remain 

54 DEVELOPING A 21st CENTURY NEUROSCIENCE WORKFORCE separate departments and their cultures and associated training programs remain fully segregated. Balkanization occurs throughout clinical sci- ence, Choi continued. Diseases of the nervous system are typically man- aged by disparate clinical departments and not just psychiatry, neurology, and neurosurgery: • Medicine (Alzheimer’s disease, fibromyalgia, sleep disorders) • Pediatrics (cerebral palsy, genetic/metabolic disorders) • Radiology (stroke) • Anesthesia (pain) • Ophthalmology (macular degeneration) • Otolaryngology (tinnitus, hearing loss) • Orthopedics (stroke, traumatic brain injury) • Obstetrics (hot flashes, seizures, hyperemesis) • Rehabilitation (stroke, traumatic brain injury) • Emergency medicine (stroke, traumatic brain injury) • Oncology (central nervous systems cancers, radiation encepha- lopathy) • Surgery (neurological intensive care) Because of this balkanization, Choi said opportunities for collabora- tion are challenging, but offered several suggestions for improving cross- training between neuroscience and clinical science (see Box 4-1). In addition, a few workshop participants discussed unique challenges for physician-scientists. For example, several speakers noted that M.D./Ph.D. students often accelerate through their Ph.D. coursework to go into their clinical training. Landis asked whether there was a special role for them, given that in principle they could speak to both communities (neuroscience and clinical science). BOX 4-1 Challenges and Opportunities for Improving Cross-Training Between Neuroscience and Clinical Science Challenges • The interface between basic research and clinical science pro- vides both a key source of inspiration, purpose, funding, and pub- lic support as well as the necessary experimental platform for investigating and ultimately understanding the human mind. 

TRAINING IN TRANSDISCIPLINARY RESEARCH 55 • Neuroscience training typically provides limited exposure to prin- ciples of clinical medicine, clinical research, and overall disease biology, while clinical training typically provides scant exposure to the scientific method. • Opportunities for improving cross-training between neuroscience and clinical science include increased availability of neurobiology of dis- ease courses and clinical rotations for neuroscience trainees, and completion of research projects and journal clubs for clinical trainees. Opportunities • For neuroscience graduate students: o Increase the availability and strength of neurobiology of dis- ease (NBD) courses. This type of course could have a greater impact if delivered online due to the labor-intensive steps of bringing in patients and performing all of the clinical interven- tions (e.g., Society for Neuroscience NBD workshop, National Institute of Neurological Disorders and Stroke/Child Neurology NBD in Children website). o Augmented online clinical courses (e.g., Stanford Online, HarvardX) with readings and discussions with faculty—even tests and recognition of achievement. o Increase the availability of clinical courses (pathophysiology, clinical research) for neuroscience graduate students. o Increase the availability of clinical rotations to neuroscience Med- ical Engineering and Medical Physics (HST MEMP) program. • For clinical students: o Increase opportunities for exposure to basic neuroscience, as well as physics, informatics, and engineering, for selected medical students (e.g., HST MEMP program) o Increase graduate students along the lines of those imple- mented by the Harvard-Massachusetts Institute of Technology Health Sciences and Technology requirements for the training of clinical residents in core scientific methods, focusing on skills needed to read the clinical literature with a critical eye. o Require completion of a research project. o Journal clubs focusing on the critical analysis of key papers. • For both neuroscience and clinical students: o Build interdisciplinary and interdepartmental teams around shared clinical research or clinical care goals, involving both M.D.s and Ph.D.s (e.g., Parkinson’s disease centers). o Merge neurology and psychiatry training—and eventually, departments. SOURCE: Dennis Choi presentation, Stony Brook University, October 29, 2014.