Dynamical Structures in Complex Systems: an information theoretic perspective (DySCS)

25 September 2014

IMT Institute for Advances Studies Lucca Campus

Information theory can provide powerful tools for studying, understanding and designing complex systems. Examples span from physics and biology to social and artificial systems. Some attempts to formalize self-organization and emergence have also been undertaken from an information-theoretic perspective. Indeed, complex systems can be studied with the aim of enlightening the processes and the structures involving information elaboration within the system and between the system and its environment. In particular, the dynamics of a complex system is often driven by the interplay among its subsystems, which may also emerge during the evolution of the system and change the overall dynamics. Indeed, an intriguing aspect is the "sandwiched" nature of these phenomena: while past researches were almost exclusively focused on bottom-up emergence in two-level systems, researchers are becoming increasingly aware that in the most interesting cases the new entities do emerge between pre-existing levels. So, once appeared, in some cases these entities could have fostered the formation of new levels within the system. The paradigmatic example may be that of organs and tissues in multicellular organisms: both the lower (cellular) level and the upper one (organism) predate theappearance of the intermediate structures. Other examples come from the physical, chemical and socio economical fields. The identification of this kind of structures in dynamical systems is a not an obvious task and a major challenge, both for natural and artificial systems.

Despite the presence of these structures, complex systems show an high level of coherence: probably the most fascinating example of this sort of unity is the human brain, which show an immense variety of behaviors by maintaining at the same time the internal awareness of a single experience. Interestingly, recent outcomes in neurosciences suggest that an information theory approach is a good candidate to investigate these particular emergent dynamical organizations inside complex systems. We think that the generalization of the approach in different disciplines is of paramount importance to catch the foundation of the synergy among the different complex systems parts beyond the peculiarities of the distinct contexts.

This workshop is aimed at fostering the discussion on information-theoretic perspectives to the study of the dynamics of complex systems, with emphasis on the detection of relevant integrated structures or subsystems and the understanding of their relation with the dynamics of the whole system.

Topics of interest

Contributions solicited cover a variety of topics including but not limited to the use of information theoretical measures and information theory in general for:

  • detecting integrated dynamical structures
  • understanding the dynamics of integrated subsystems
  • discovering, designing and driving hidden dynamical structures
  • providing tools for complex systems design
  • understanding emergence and self-organization
  • studying the dynamics of meso-level structures
  • studying the emergence of meso-level structures
  • studying the emergence of boundaries, sensors and actuators
  • providing models of the dynamics of systems internally showing both integrated and segregated subsystems
  • investigating the emergence of level between levels (“sandwiched” emergence)

Preliminary Program

14:00-14:10 Welcome

14:10-14:50 Invited talk: Daniel Polani, “Informational Principles in the Perception-Action Loop”

Session 1

14:50-15:10 Analysis of emergent competence networks in a regional innovation system: what can we learn from alternative methods? (Caloffi et al.)

15:10-15:30 Information adaptation: toward a cognitive-information theoretic approach to cities as complex adaptive systems (Portugali and Haken)

15:30-15:50 Information theory for complex systems: what is the entropy? (Thurner et al.)

15:50-16:10 Searching for dynamically relevant subsets in complex systems (Filisetti et al.)

16:10-16:30 Uncovering individual node's contribution to the segregation and integration of information in neural networks (Zamora-López et al.)

16:30-16:50 Does training lead the formation of modules in threshold networks? (Nicolay et al.)

16:50-17:30 Break

Session 2

17:30-17:50 A dynamical approach to assess synthetic modularity (M. Zattoni)

17:50-18:10 Topological Shape and Formal Languages for Studying RNA Folding (Mamuye and Merelli )

18:10-18:30 A self-adaptive method for detecting brain pathologies: the epilepsy case study (Merelli et al.)

18:30-18:40 Wrap-up and conclusion of the workshop

 

Invited speaker: Daniel Polani, University of Hertfordshire, UK

Abstract: Informational Principles in the Perception-Action Loop

Ashby's Law of Requisite Variety (1956) and, in last years, especially its later rediscovery and extension by Touchette and Lloyd (2000, 2004) have indicated that Shannon information acts as fundamental "currency" constraining the potential organisation and "administration" of cognitive tasks. In particular, there is increasing evidence that decision processes in biological organisms in fact exploit the limits implied by aforementioned work, and their cognitive operation can therefore be subject to analysis with respect to information-theoretical optimality principles.

Under this hypothesis, many aspects of biologically plausible cognitive processing can be treated informationally, requiring only high-level constraints without having to specify detailed mechanisms.

This gives rise to novel tools not only for high-level analysis of biological cognitive systems, but also for purposes of prediction and construction of biologically plausible artificial cognitive models.

The talk will give an introduction into the question and methodology and demonstrate its operation with a number of examples.