- Private group -

Technologists are often naïve when appropriating themselves of topics that have been studied for ages in other disciplines. Knowledge is an important case in point. The rational of this initiative is to renew ties between the different disciplines that are studying knowledge, the phenomena of knowing (especially beyond the 'declarative' kind of knowledge) and related issues (e.g., learning, context, knowledge transfer, knowledge as a social construct,…) from various perspectives (e.g., neural, behavioural, social, epistemological). This is an initiative for exploring the interdisciplinary fundamentals of knowledge, knowing, doing and being, tightly linked to the conception of future knowledge technologies.

What are we looking for?
•    What should be the orientation of research on this topic? As stated, do you feel it is too broad or, on the contrary, too narrow?
•    Have any recent scientific results been obtained relevant to this topic? Is there already a well-established community on this?
•    Do you know of related initiatives, for instance at national level, or in other continents?
•    What is needed at this point to advance this? More exploration of different ideas? More coordination among groups or related initiatives? A strong push for a precise technological target and, if so, which one? Anything else?

Background: Following the last FET consultation during 2012-13, 9 topics were identified as candidates for a FET Proactive. This topic has been selected for inclusion in the FET Work Programme for 2014-15. Comments are invited on whether this topic is still relevant, or if any changes would be necessary to take account of recent research results. We are also trying to understand better how to advance these areas.

To participate to the consultation:
- register to the group (create an ECAS login if you don't have one yet);
- then "log in" and enter your contribution in the "Add new comment" box, at the vey bottom of the page.
You can also participate by commenting on submitted ideas and/or voting for them.

If you wish to share with us additional documents or have any questions about the process, please send them to our FET mailbox.


0 users have voted.


nkasteka's picture

very interesting topic!

Dear Beatrice Marquez-Garrido, the topic scetched above sounds very interesting and timely to me! I have been working on epistemic cultures for the past 15 years, recently shifting to a new conceptual frame, namely that of technoepistemic cultures. This approach focuses on presumable changes when sciences morph into technosciences (as is currently happening in the contexts of synthetic biology or neuroscience) and bears relevance not only for the philosophy and sociology of science, but also for the performance and organisation of (techno)science itself as well as its societal governance. In this wake, I would suggest to add "designing" to "knowing", "doing" and "being".
Besides a few preliminary publications regarding the concept of technoepistemic cultures authored by myself, I first and foremost recommend work by Bernadette Bensaude-Vincent and Alfred Nordmann on the concept and conceptualisation of technoscience.
With kind regards and best wishes for this topic!
Karen Kastenhofer, Austrian Academy of Sciences and University of Vienna

0 users have voted.
nsalmeri's picture

important multidisciplinary topic, great future potential

Noninvasive imaging techniques allow a direct window to the workings of the human brain. The pivotal developments over the past two decades were made possible by a strong interdisciplinary effort between physics, mathematics, medicine and psychology. For many aspects of sensory, motor and cognitive processing, imaging techniques can now offer a fairly consistent view of neural activation. The field is now getting to the point where we can ask, for real, what type of information the brain signals contain and convey. Recent developments in computational science are making it possible to directly interrogate the brain about the way it processes and organizes information and knowledge. For example, based on the neural activation to a certain subset of words (e.g., butterfly, house, pen etc) one can learn a model that not only links the learned words and their corresponding brain activation patterns but can predict brain patterns of previously unencountered words or, based on the measured brain pattern, suggest a word the person saw or heard. Such models, and relationships between neural representations of concepts, could provide brain-level descriptions of knowledge representation. This type of novel approaches open up a wealth of possibilities and new types of questions, such as: While brain representations of concrete concepts are probably fairly similar across individuals, languages and cultures (e.g., dog and cat are more similar to each other than either one to fork), would there be more variability for abstract concepts (e.g., freedom, democracy) and how would that manifest in communication? Could one imagine some standardized way of mapping individual brain bases of knowledge and finding a transfer function to better achieve the intended meaning in communication (kind of à la Star Trek)? Important steps in this latter direction are currently being taken (in the USA…), by finding transfer functions between individual brain representations when the subjects watch a movie (rather than matching the brain anatomy). While big data is currently a big buzz (for a good reason), the individual brain is probably what we are all most intrigued about. Europe has played a major role in development of brain imaging techniques. It has the relevant depth and width of knowledge as well as the culturally and linguistically versatile environment to make a strong impact in this field. If we have the resources to efficiently combine neuroimaging, behavioural sciences, computational sciences etc to describe the organization of knowledge in individual brains and minds, we could ideally think of ways for more efficient and individualized learning programmes (which will be needed all the more in our rapidly changing world!), improved interaction/communication between individuals (and nations?) as well as a wealth of novel human technologies.

Good luck with this important topic!
Best wishes,
Riitta Salmelin, Aalto University, Finland

11 users have voted.
nmuldthe's picture

embodied cognition

An important development in cognitive neuroscience has been the notion of embodied cognition, the fact that much of our knowledge is not represented in the brain in abstract forms, but in a bodily form. For instance knowledge about tools is not abstract, but in part represented in brain regions involved in encoding the motor acts involved in using a tool. Knowledge about the inner states of others (their motor intentions or their pain for instance), is also encoded in part in brain regions involved in our own motor actions and our own pain. These results, stemming from neuroimaging and neuroscience more generally, have given new impulses in cognitive sciences more generally to take the 'body' seriously, even when it comes to knowledge. What we need now, is a multidisciplinary effort to bring this embodied cognition realization to a state in which embodied cognition is a well understood basis for knowledge and a model on how to create intelligent artifacts that can cooperate with humans in intelligent and interactive ways. For that, we need an effort to understand the developmental processes that provide biological systems with embodied cognition, including human and animal studies. We also need an effort to implement these principles in robotic platforms, rather than in traditional artificial intelligence systems that lack a body. How neural nets, exploiting recent trends in deep learning, can then generate embodied cognition in artificial systems can then work side-by-side with human and animal studies of knowledge development over the life span. This promises (a) a more rigorous understanding of knowledge in embodied systems and (b) new approaches to artificial intelligence that would lend themselves to robots that can learn and interact seamlessly with humans to resolve some of the important societal problems of healthcare in the elderly.

Christian Keysers,
Netherlands Institute for Neuroscience
Amsterdam, The Netherlands.

0 users have voted.
nmarcojs's picture

Very long term Knowledge Preservation

Knowledge in science is usually taken to be preserved through publications.
In our days more and more of this knowledge is linked to data (data collected and analyzed), simulation,
and the software tools used for it. There are many projects on data preservation, and even on software preservation.
But there is no easy way to capture all assumptions, intermediate steps, conditions, that are critical to get final results in many cases. And also there is no way to assure that this knowledge will be alive in 100 years from now, when the authors are no longer around (well, who knows!).
The proposal is to articulate with a very long term (100 years) scope the issue of knowledge preservation,
assuming that up to know this was the role mainly of museums and libraries, but that new and complex knowledge results will require a more complex framework.

28 users have voted.
nlaschca's picture

Soft Robotics

It has been widely recognized in the last years that intelligence requires a body. This brought the disciplines of artificial intelligence and robotics to merge, in so-called embodiment. Further than this, modern views of artificial intelligence attributes a stronger role to the body and its interaction with the environment, in so-called embodied intelligence, or morphological computation.
Knowing, doing, being cannot be considered as functions of the brain, only; they are not just given by reasoning and computing, but they are given, in part, by the body itself, by the way it interacts with the environment and the way it reacts to such interactions.
In this perspective, while robotics has demonstrated to be an effective tool for studying neuroscience and intelligence so far, it becomes difficult to use most current robots in investigating this new view of intelligence. A physical body is needed, indeed, but it also needs adequate interactions and reactions with the environment. In one word, it needs compliance, to put in place and investigate the way how control, sensory-motor behaviour, and ultimately intelligence are given by the body itself. From robots based on rigid links and controlled actuators, there is a need to go towards robots with compliant behaviour, soft materials, and embodied intelligence, according to the recent paradigms of soft robotics.
A key point in understanding knowledge, and the functions of knowing, doing, and being, is then having a ‘soft’ robotic body.

11 users have voted.
nhofstge's picture

Identifying foundational Hilbert problems for Social Simulation

Technology, knowledge and our social systems are fast becoming more intertwined. Social simulation aims to create working computational models of social systems according to the idea “Grow it to show it”, allowing to link micro-behaviours (e.g., behaviours of individuals) to macro-patterns (at the level of e.g., society) through mechanisms that capture the dynamics. Because models link these levels of aggregation they could play a unifying role in the social sciences, and because they are actually “grown” they could be empirically validated. However, at present the social simulation field just reflects the vast dispersion of the social sciences.
In 1900, David Hilbert posed 23 problems in Mathematics that were all unsolved at the time and have been influential in guiding the field. We propose to assemble a multi-disciplinary team to do the same for social simulation, assembling fundamental issues that 1) are contested across the social sciences and that have proven to be un-decidable solely empirically, 2) for which social simulation can provide a solution if the problem is solved at the modelling level, 3) that help to advance the social simulation methodology in general, 4) whose resolution is of great social importance. In honour of Nigel Gilbert, pioneer of social simulation, we propose to call them ‘Gilbert problems’.
Gert Jan Hofstede, Wageningen University
Tina Balke, University of Surrey
Marina De Vos, University of Bath
Harko Verhagen, University of Stockholm

37 users have voted.
nzambrda's picture

Neuromorphological Computation in Robotics

The principle of knowing, doing, and being is strongly related with the interaction of a specific agent with the real world, aiming at understanding how some high level functions, like attention, memory or learning, could arise from the low level processing of, for example, neural computation. Most of the computational neuroscientific works, indeed, are based on large-scale simulations of thousand of neurons showing how their activities correlate with a certain stimuli. Even simulations in complex virtual environments do not provide sufficient validity at the behavior that such “smart agent” could have in a real environment, at real time, with multiple sources of information.
Bringing neuroscientific evidences into a real agent implies the ability for the robot to orchestrate a multitude of asynchronous sensory inputs and motor outputs. And, moreover, to learn at multiple timescales and to deal with control line delays, measurement uncertainty and the synchronization of several control loops. There are a least two key advances that are needed: firstly, all the processing made by the robot has to match the asynchronous and parallel computation of biological systems – for example by using Spiking Artificial Neural Networks; secondly, the design of sensors and actuators has to take advantage of the computation nested in the system morphology – a principle known as Morphological Computation.
The general principle of the integration of higher cognitive functions and low level computation in real environments is of utmost importance for both neuroscience and for a new generation of advanced robots.

10 users have voted.
nsuchond's picture

European focus

This is a great topic, with many opportunities for cutting edge research, but it lacks European focus.

For instance, among the most advanced world economies (USA, Europe, Japan and China), Europe is unique in its varied multilingual character. Consequently, Europe has a rich and long-standing tradition in language research. It seems reserving a subtopic in a future call that would be specifically language related could improve long term competitiveness of European economy and strengthen cohesion of European society.

12 users have voted.
nramonja's picture

The challenge of more global knowledge

Often, domains of knowledge are studied in isolation. More and more borders between domains of knowledge are disappearing. One can see this in the increasing multidisciplinarity of sciences, and the fact that a lot of recent databases and knowledge bases are network structured, hence connecting every piece of knowledge with allmost everything else.

We currently have techniques for data storage, reasoning with knowledge, discussing, optimizing, experimenting, computing, understanding, feeling, analyzing, exploiting knowledge and securing knowledge, many of which allow to treat knowledge at a shallow level. A major challenge is to revisit these fundamental techniques, and develop new ones able to treat 'global' knowledge. In tackling this challenge, radical changes in principles and standards may be considered and debated. For humans it may be hard to understand conclusions based on billions of facts. It may be hard to discuss or feel about what we don't understand. It may be infeasible to compute in one of the classic ways what we cant to know even if all basic facts are known. .

There are many lines of research focusing on new technologies, computing paradigmas (quantum, dna, neural, ...), materials, acting (robotic, human-computer-interface, ...), , ... There aer many application domains with each their own specificity. They all have a place in programs such as H2020. What I like in this topic title is the word "fundamental", referring to the possibly more abstract questions I raised above. Investigating those without the burden of requirements to perform one of the 'new technologies' could clearly improve the way we treast knowledge (and researchers handling knowledge).

0 users have voted.