An IBRO Neuroscience School.
Part of the EU Neuro-IT Network of Excellence
August 21 - September 15, 2000
International Centre for Theoretical Physics, Trieste Portugal
Evaluation Report Made By The Course Students
This report summarises the impressions of the students participating in the "EU Advanced Course in Computational Neuroscience" in Trieste, Italy, August 21st to September 15th, 2000. It is meant to provide the funding bodies of the course with an independent evaluation. This assessment has been done by the students only and has been communicated to the funding agencies directly. The students of this course were Ph.D. students and young post-doctoral researchers from various disciplines (physics, computer science, mathematics, neurobiology, medicine, and psychology) spanning the wide field of neuroscience.
Overall concept of the course
The course schedule was divided into three different sections: lectures, tutorials, and individual project work of the students.
The lectures that were held by the tutors and guest lecturers introduced techniques, problems and models of computational neuroscience on the various levels of neural organisation ranging from subcellular processes to system-level operations. At the beginning of the course the lectures were divided into a dual track for experimentalists and theoreticians, where the experimentalists were introduced to mathematics, UNIX and artificial neural networks, while the theoreticians had introductions to neural morphology, anatomy and membrane properties. This was done in order to fill out the gaps in knowledge caused by the different backgrounds, making it possible to benefit equally from the rest of the course.
The tutorials addressed more methodical aspects, e.g. concerning the use of software for analysis and simulation (e.g. GENESIS, NEURON, XPP), theoretical neuron models, bifurcation analysis or statistics.
In their projects, the students investigated problems of their own interest, with each student being assigned to one tutor for supervision and help. These projects dealt with problems on all levels of neuroscience and comprised both modelling approaches and mathematical analyses.
The students were very satisfied with this structure, as it both encouraged individual work, discussion with the faculty and joint learning as a group. The lectures and tutorials held high standards and also introduced the students to many subjects that they would otherwise not have been exposed to. The result was a broadening of horizons beyond particular subjects, making the task of bridging the many sub-fields of computational neuroscience a bit easier. This is especially relevant as computational neuroscience is by its nature a very multidisciplinary field, encompassing both biological, mathematical and to some extent engineering reasoning, and in great need of integrative approaches that go beyond specialities. By showing the full range of methods, levels of complexity and different modes of thinking that can be applied the course made it more possible for the students to move in a multi-disciplinary direction, or learn to network with people from different specialities so that their diverse skills can be used in a complementary way.
The dual track lectures introduced this year were a welcome addition, as they helped students who were predominantly experimentalists get up to speed on the mathematical background and the theoretical/modelling students get acquainted with the neural systems. However, it was felt among the students that the dual track lectures held a somewhat too high level for at least some of the students. At the same time, background knowledge varied widely among the students, and it may be impossible to create a lecture track that will fit all students. It might be useful to introduce this dual track as a series of elective tutorials rather than lectures, making it possible for each student to tune their introduction to compensate for their varying background. It might also be beneficial to add an introduction to basic neuroanatomy earlier in the theoretician track schedule than was done this year.
Projects
As in previous years, there was a broad consensus among students that individual project work constitutes an essential part of the course, perhaps even the backbone linking together the different lectures and activities. Applying the new techniques learned to concrete problems, or trying one's own traditional techniques on new problems results in a better understanding of both the power and difficulties of the computational approaches. Modelling differs both from experimental and purely theoretical approaches in crucial ways that are best learned by experimenting with the creation and exploration of models, i.e. learning-by-doing. In this exploration the presence of a knowledgeable tutor is important to guide the experimenting and learning in a useful direction.
The tutors were generally accessible and open for scheduled or informal discussions on the projects. Their support was found to be inspiring and important. The students sharing the same tutor also got to know each other better and in some sense acted as informal tutors to each other. A minor problem was that some tutors this year did not stay the whole time but were replaced by other tutors, which created a slight interruption in the project work as the students had to explain their projects to them and might be getting new advice. The resulting project presentations demonstrated a wide range of inspiring projects, many of which will no doubt be extended into full-length papers or conference submissions.
Quality of teaching
The students found the quality of teaching to be very high, both among the lecturers and tutors. As they were mainly speaking of their own areas of expertise and research they were motivated, knowledgeable and in general able to present the material in a clear and enthusiastic manner. The presence of many active researchers at the course that were at central positions in their fields made it possible for the students to get acquainted with the current issues in a way that would not have been possible in a more conventional course. Questions and discussions flowed freely, often involving both faculty and students and continuing over lunch or dinner. Many guest lecturers took interest in student projects where they helped with advice or guidance.
Infrastructure and technical equipment
An intensive course like this demands a careful balance between an environment supporting concentrated study and work, and the possibility of restoring one's energy by doing something different. The ICTP site is ideal from this perspective, as it both has all necessary facilities - accommodations, computers, lecture halls, food etc. - within a short distance from each other and is sufficiently far from Trieste to make city distractions less disruptive. At the same time the surroundings were relaxing and excursions during the weekends possible.
The lecture halls, library and computer environment was fully adequate for the projects and research.
General atmosphere
A sense of community quickly emerged among the students, who got to know each other well over the course regardless of nationality or field of study. Plans for reunions in connection to major neuroscience conferences were drawn up; the course appears to have been very successful in building a community among young computational neuroscientists that will persist long after the course. This sense of community extended to the faculty, creating a fertile ground for further information exchange and cooperation. This community building is especially important for students from nations that have fewer opportunities for international contacts. The course can hopefully act to break their isolation and increase the chances of international collaboration.
Final recommendations
Computational neuroscience holds great promise to apply mathematical and computer science methods in neuroscience, and thus enable a better understanding of brain structure and function. This approach requires that users of both experimental and theoretical methods understand the methodologies of each other and have at least some familiarity with both areas. Without this understanding experimentalists will not be able to draw on the powerful generalisations of theory and modelling, and more theoretical researchers will not be aware of relevant experimental work or its importance or limitations. To achieve the broad understanding needed close interaction between students of different fields is necessary, as well as a willingness to learn about the issues in different methodologies.
The Trieste course achieves this goal admirably. After spending a month in close contact with enthusiastic fellow young scientists from different fields, experiencing the excellent lectures and tutorials and building a shared community the students are well prepared for the trans-disciplinary challenges of computational neuroscience. The course is a unique opportunity to educate and network the next generation of computational neuroscientists. We therefore strongly recommend future funding for the Trieste course and hope that other courses may follow its example.
On behalf of the participants of the "EU Advanced Course in Computational Neuroscience", held 21.08.00 - 15.09.00 in Trieste,
Anders Sandberg
Department of Numerical Analysis and Computer Science
Royal Institute of Technology
100 44 Stockholm
Sweden