Education Objectives

EUCog will put a particular emphasis on education, as a means of furthering future scientific progress. This will involve two aims. First, rather than focussing on individual research projects, this initiative therefore promotes activities that will enable young researchers to build the expertise that will allow them to move fluently between the contributing domains and become productive and creative scientists. Second, rather than taking on the education ourselves, we will support extant university education structures by providing material.

Towards the first aim, it is necessary to offer specialized training events on particular advanced aspects of cognitive systems. Building non-trivial artificial cognitive systems requires a high degree of technical sophistication. We therefore propose to hold one summer-school per year on the transfer of technical know how.

Towards the second aim, we propose several activities that support conventional academic education. One aim is to define tutorials on selected aspects of cognitive systems research on a European level, both on the undergraduate and on the postgraduate level (based on the euCognition initiative NA 044-4, lead by Joanna Bryson). This aim will involve many educational institutions in the EU and will result in about two tutorials per year.

Responsible project partners: Andreas Engel, Rolf Pfeifer, Gregor Schöner.

Tutorials for cognitive systems – 3.1

Writing good tutorials on cognitive systems is both a great challenge and one of the most efficient ways to motivate and prepare future PhD-students for a career in cognitive systems research. With resources available in this project, we propose to make significant steps towards this goal, on the basis of work in WP 1 (state of the art), WP 2 (challenges) and this WP 3 (education). These steps will consist in tutorials on selected aspects of cognitive systems (e.g. speech processing). These tutorials are complementary to the model curriculum and the courseware activities. We envisage that they contribute to a future textbook project, based on P. Hendriks, N.A. Taatgen & T.C. Andringa (eds): Breinmakers & Breinbrekers: Inleiding Cognitiewetenschap (Addison Wesley Longman, 1997) and on Rolf Pfeifer and Christian Scheier: Understanding Intelligence (MIT Press 1999).

It will be investigated whether these tutorials can plausibly contribute to a standardized description of a university programme in cognitive systems. In this case, we will formulate Dublin cycle level descriptors for University Programmes that will define the objectives and final qualifications for a range of academic degrees in Cognitive Systems. It may turn out, however, that a unified description is not achievable or not even desirable.

Educational robotic toolkit – 3.2

Recent developments in artificial intelligence and cognitive science suggest that behaviour is not only the result of centralized control, but also emerges from the interaction of central control, body (morphology and materials), and the environment. Existing robotic toolkits (such as Lego Mindstorms) are still based on classical control theory (with central controller, rigid materials, etc.).

We propose to develop the blueprint of a robotic toolkit that makes use of more recent concepts developed in the field of embodied artificial intelligence, such as the importance of material properties, passive dynamics, morphological computation, self-organization, emergence and distributed control. In order to further development of this toolkit we will pursue networking activities to unite interested research groups in a joint effort. A robotic toolkit could enable researchers to employ the synthetic methodology while freeing them from having to deal with issues of designing low-level electronics and mechanics; the robotic toolkit provides for a “rapid prototyping tool”.

In robotics, the synthetic methodology – “learning by building” – has proven enormously powerful and appealing to young people, so this toolkit will be a significant step ahead for the education in cognitive systems.

Summer-schools – 3.3

Building non-trivial artificial cognitive systems requires a high degree of technical sophistication and young scientists need training in the hot-topics of the field. A well tried and tested method is to hold intensive summer schools where students can learn from established experts in hands-on experience. Several partners have long experience in running this kind of events.

Summer schools will show a high degree of interdisciplinarity, bringing together scientists from disciplines as diverse as robotics, computer science, cognitive science, psychology, neuroscience, biology or philosophy. The goal of each school is to offer around 30 participants the opportunity to learn about relevant studies through lectures given by senior scientists, discuss topics of particular interest in interactive workshops, to present their own research, and to exchange ideas with and to establish new contacts to colleagues in the field. Each school will deliver several types of documented materials that are made available for the publication on the EUCog website, which will constitute valuable resources for a broader audience: (1) Speakers will be asked to supply a publishable version of their lecture slides and participants will be asked to write a brief summary of each lecture. (2) The key discussion points of the workshops will be summarized. (3) All participants will present a poster on their current research. These posters will be published on the website.

On-site training courses – 3.4

Through network actions we will also provide more focussed training classes, lasting about a week, for smaller groups of PhD candidates and post-docs. These classes will normally be held on specific topics at institutes hosting laboratories or training facilities (e.g. robotics labs, BCI labs, neuro-imaging facilities) that are equipped for on-site tutorials and courses in specific techniques and approaches.