Introduction

LOCEN is a research group working within the Istituto di Scienze e Tecnologie della Cognizione (ISTC), situated in Rome, in turn part of the Consiglio Nazionale delle Ricerche (CNR; the Italian National Research Council). LOCEN was founded in 2005 and is one of the most young, dynamic and innovative research groups of ISTC-CNR.

Since its birth, LOCEN has been capable of self-funding both its members' salary and research on the basis of European Projects (Gianluca is the only one to have gotten a permanent position, in 2009). In particular, the group got funded through the following European Projects:
1. MindRACES. Period:  11/2004-10/2007,120.000 euros.
2. ICEA. Period: 01/2006-12/2009, 630.000 euros.
3. IM-CLeVeR (LOCEN is the Coordinator). Period: 01/2009-04/2013, 1.681.479 euros.

LOCEN aims at investigating brain, behaviour and society through computational models which have a close dialogue with data (and researchers) of neuroscience, psychology, and social sciences (see below LOCEN mission, research topics, and research method).

People

Coordinators
Gianluca Baldasarre (Researcher III Level)
Marco Mirolli (Researcher III Level)

Researchers
Francesco Mannella (PhD Student)
Vincenzo Fiore (PhD Student)
Stefano Zappacosta (Postdoc)
Fabian Chersi (Posdoc)
Daniele Caligiore (PhD Student)
Valerio Sperati (Research Assistant)
Simona Bosco (management)

Past Members
Dimitri Ognibene (PhD Student)
Alberto Venditti (Master Student)
Massimiliano Schembri (Master Student)
Angelo Rega (Master Student)

Mission

LOCEN aims at understanding how brain generates behaviour of organisms by interacting with the body and with the environment via sensors and effectors, by using robotic and computational models.

Research topics

Complex systems:

We are interested in acquiring the conceptual and formal tools related to the study of complex systems as brain-body-environment form a complex system, so we are interested in:
Complex systems, non-linear dynamical systems, hierarchical levels, positive and negative feedback, emergence, information theory, entropy.

Computational tools:
We are interested in developing 'computational tools' to develop computational models, so we are interested in:
Genetic algorithms, neural networks, reinforcement learning, unsupervised learning, supervised learning, Hebb learning.

Aspects of brain of interest:
We are interested in the following brain aspects, mechanisms and areas:
Neuron functioning, LTP, LTD, STDP, amygdala, basal ganglia, hippocampus, orbitofrontal cortex, dorsal prefrontal cortex, medial pre-frontal cortex, anterior-cingulate cortex, supplementary-motor cortex, premotor cortex, motor cortex, parietal cortex, visual cortex, ventral and dorsal visual pathway, muscles, muscle sensors, muscle control, limb dynamics and kinematics.

Behaviours of interests:
We are interested in the following aspects of behaviour:
Classical conditioning, instrumental conditioning, motivation, affective regulation of learning, action selection; navigation; eye-arm-hand coordination, grasping, reaching, action hierarchy and compositionality, motor babbling, learning of affordances; development of sensorimotor skills, curiosity-driven learning, novelty detection, focussing on zone of proximal development; active vision, where/what/how, attention; prediction, planning.

The computational modeling research approach we use (see details below), has its roots in the approaches of:
Computational neuroscience, artificial life, cognitive neuroscience, developmental robotics, machine learning, artificial intelligence, cognitive science.

Research method

LOCEN founds its research on an innovative research method (drawing its roots in many computational approaches), based on two fundamental meta-principles, and five specific principles. Note that we are aware that this methodology is not the only way of using computational models and produce valuable knowledge. However, we think that among the existing computational methods we have been developing one which is rather powerful for understanding how brain and behaviour work as it possesses a strong cumulativity (see below). The principles of the method are as follows:

...2 Meta-principles...

a. Seek explanations within the theoretical framework of evolution.
The biological theory of evolution is the best theoretical framework to uderstand behaviour and brain. Organism's behaviour is as it is as it evolved to increase their survival and reproductive chances. Brain is as it is as it evolved to produce such behaviour (and satisfy internal constraints of metabolic efficiency, space, etc.). So when you try to understand behaviour and brain always ask: what is the function of this? (of course, bear in mind that biology teaches us that evolution is dirty, recicles, has path dependencies, exaptations, etc.: all these are important to find sound explanations).

b. Seek theoretical cumulativity.
The halmark of science is cumulativity of knowledge. This means that a good scientific method is one which assures the cumulativity of knowledge in time, by allowing you to rank theories and hypothesis on reality and discard those that explain less of the phenomena of interest. All the five principles which follow are directed to strengthened the cumulativity of the reasearch method used by LOCEN.

...5 Principles...

1. Use computational models to understand brain, body and behaviour.

In fact: (a) they form a complex system and computational models and robots are the the only way you can use to understand its emergent properties; (b) neuroscience and psychology are producing a huge amount of data but they fail to produce unified pictures of brain and behaviour: in this respect, computational models and robots have an exceptional theoretical catalyst power as they force researchers to translate theories into explicit coherent models which produce quantitative detailed testable predictions.

2. Constrain models by requiring that they reproduce specific quantitative data on behaviour, furnished by psychology, ethology, and other disciplines.
In fact anecdotal and qualitative comparisons of models with data are not enough compelling to select models.
 
3. Constrain the assumptions on the architecture, functioning, and learning mechanisms of models on the basis of data on brain furnished by neuroscience.
In fact you can always build different models to reproduce an observed behaviour: using neuroscientic constraints greatly aids the selection of models.

4. Test your models within embodied systems, endowed with realistic bodies and interacting with realistic environments via noisy and quantitative sensors and actuators.
This constrain allow developing and selecting models which are capable of scaling to reproduce the complexity of behaviour observed in real organisms. It also forces models to be inerently quantitative (i.e. not symbolic) and robust in the face of variety of input and noise.

5. Reproduce learning processes which lead to the target behavious.
This is important to understand not only how behaviour is structured, but also the mechanisms which lead to its acquisition. Indeed, a substantial part of behaviour (say 20%) and brain structure (say 50%) is as it is because it has to let the system to acquire with experience and learning processes the behaviours it needs to increase fitness.

Research collaborations

International collaborations

Collaborations with common publications:
1.    Prof. Redgrave Peter & Prof. Gurney Kevin (Dept. of Psychology, University of Sheffield, England): integrated model of amygdala, nucleus accumbens, hippocampus, superior colliculus for the regulation of learning and acquisiton of actions.
2.    Prof. Butz Martin (Dept. of psychology, Univ. of Wurzburg, Germany): theoretical analysis of anticipatory behaviour.
3.    Balkenius Christian (Dept. Cognitive Science, Lund University, Lund, Sweden): neural models of eye-hand coordination and of attention.

Other collaborations:
1.    Prof. Barto Andrew (Univ. of Massachusetts, Amherst, Boston, USA): intrinsic and extrinsic motivations, hierarchical reinforcement learning.
2.    Prof. Ziemke Tom (Dept. of Cognitive Science, Univ. of Skovde, Skovde, Sweden): integrated model of amygdala and orbitofrontal cortex for the regulation of learning.
3.    Prof. Schmidhuber Jergen (Istituto Dalle Molle per l\rquote Intelligenza Artificiale, Lugano, Svizzera): intrinsic motivations and reinforcement learning.
4.    Prof. Triesch Jochen (Goethe University, Frankfurt Institute for Advanced Studies, Germany): neural models of vision.
5.    Prof. Lee Mark (Department of Computer Science, Aberystwyth University, England): neural models of leaning and development.

National collaborations

Collaborations with common publications, european projects, shared PhD students:
1.    Sandini Giulio (Istituto Italiano di Tecnologia; Prof. Bioengineering, Universit\'e0 di Genova) & Prof. Metta Giorgio (Istituto Italiano di Tecnologia; Prof. Bioengineering, Universit\'e0 di Genova): exploitation of the humanoid robot iCub  (a common PhD, collaboration within IM-CLeVeR project).
2.    Guglielmelli Eugenio (Pro. Bioengineering, Campus Biomedico di Roma) e Flavio Keller (Prof. Developmental neuroscience, Campus Biomedico di Roma): Mechatronics, developmental robotics, rehabilitation, (a shared PhD, collaboration within IM-CLeVeR).
3.    Puglisi-Allegra Stefano (Prof. Biopsycology, Dean Psicologia II, Universit\'e0 di Roma La Sapienza) & Cabib Simona (Prof. Biopsycology, Universit\'e0 di Roma La Sapienza): models of stress coping and neuromodulators action (two shared PhDs).
4.    Borghi Anna (Senior Lecturer of Psychology, Universit\'e0 di Bologna): models of affordances and tool use (common publications).

Within-institute collaborations

Collaborations with common publications, european projects, shared PhD students:
1.    Research group LARAL (\ldblquote Laboratory of Autonomous Robotics and Artificial Life\rdblquote , Coordinator Nolfi Stefano): evolutionary robotics, autonomous robotics.
2.    Research group PSY (\ldblquote Theoretical psychology\rdblquote , Coordinator Castelfranchi Cristiano): (a) Colabroation within the project MINDRaces; (b) Common publication in collective robotics and the evolution of language.
3.    Research group UCP (\ldblquote Unit of Cognitive Primatology\rdblquote , Coordinator Visalberghi Elisabetta): (a) Collaboration within the project IM-CLeVeR; (b) Common publications on models of categorisation in monkeys.
4.    Research group related to EU project ROSSI (Coordinators Anna Borghi and Domenico Parisi): affordances, eye-hand coordination, and language.

Other results

Invitations for publications

Type, Number
Book chapters    3
Conference presentations    5
Summer school courses    1

Committes

Organisation commette
OMLL 2007 (Chair)
EvoLang 2006
SAB 2006
WIVA 2005 (2 Co-chairs)
ABiALS 2006 (Co-chair)
ABiALS 2008 (Co-chair)

Scientific commettee
Comitato scientifico    ICDL 2008      ICDL 2009
Epirob 2008       Epirob 2009
    SAB 2006            SAB 2008
IEEE CEC 2009

Promotion commettee
WIVA 2005
WIVA 2006
WIVACE 2007
WIVACE 2008
WIVACE 2009

Peer-Reviewing

International Journals
Frontiers in Neurorobotics
Adaptive Behavior
Connection Science
IEEE Trans. Syst. Man Cyb.
IEEE Trans. in neural netw.
Cognitive Processing
Mind & Society
Sistemi Intelligenti
Interface

International Conferences
IJCNN
SAB
CogSci
EuroCogSci    EpiRob
ICDL
CEC
EvoLang

PhDs

Name: Dimitri Ognibene

Type:

Period: 2004-2008
Date degree:
Supervisor: Giulio Sandini
Co-supervisor: Gianluca Baldassarre
Institute of degree: Dipartimento di Informatica Sistemistica e Telematica, Universita\rquote di Genova,  Genova, Italia

Name: Daniele Caligiore
Type:
Period: 2007-2010
Date degree:
Supervisor: Eugenio Guglielmelli
Co-supervisor: Gianluca Baldassarre
Institute of degree: Universita\rquote Campus Biomedico, Roma, Italia

Name: Vincenzo Fiore
Type:
Period: 2008-2011
Date degree:
Supervisor: Stefano Pugliesi-Allegra
Co-supervisor: Gianluca Baldassarre
Institute of degree: Facolta\rquote di psicologia, Universita\rquote degli Studi di Roma \ldblquote La Sapienza\rdblquote , Roma, Italia
Dottorato in psicobiologia

Name: Francesco Mannella
Period: 2006-2009
Type: Dottorato in psicobiologia
Date degree:
Supervisor: Stefano Pugliesi-Allegra
Co-supervisor: Gianluca Baldassarre
Institute of degeree: Facolta\rquote di psicologia, Universita\rquote degli Studi di Roma \ldblquote La Sapienza\rdblquote , Roma, Italia

University training (Tirocinio)

Name: Alberto Venditti
Period:  01/01/2007 31/12/2007
Istitute of degree: Facolta\rquote di Psicologia, Universit\'e0 di Roma \ldblquote La Sapienza\rdblquote , Roma, Italia

Visitors

Name, period, Institute of Origin

Rob Lowe:   06/04/2009   09/05/2009    Cognitive Science, University of Skovde, Skovde, Sweden
Ansgar Koene:    26/05/2009  02/06/2009    Department of Psychology, Univesity of Sheffield, Sheffield, UK