Description: State-dependent sequence processing in Spiking Neural Networks

This title appears in the Scientific Report : 2019

State-dependent sequence processing in Spiking Neural Networks

Several organizational principles of the neocortex appear to imply a strong predisposition to acquire temporal structure in a completely incidental/unsupervised manner, a process that iscentral to many core aspects of cognition. In the work I will present, we explore the processesinvolved in implici...

Personal Name(s):	Duarte, Renato (Corresponding author)
Contributing Institute:	Computational and Systems Neuroscience; INM-6 Jara-Institut Brain structure-function relationships; INM-10 Computational and Systems Neuroscience; IAS-6
Imprint:	2019
Conference:	NII Shonan Meeting: "Language as goal-directed sequential behavior", Kanagawa (Japan), 2019-05-19 - 2019-05-23
Document Type:	Conference Presentation
Research Program:	Supercomputing and Modelling for the Human Brain W2/W3 Professorinnen Programm der Helmholtzgemeinschaft Theory, modelling and simulation
	Publikationsportal JuSER

Several organizational principles of the neocortex appear to imply a strong predisposition to acquire temporal structure in a completely incidental/unsupervised manner, a process that iscentral to many core aspects of cognition. In the work I will present, we explore the processesinvolved in implicit, structured sequence learning in biologically-inspired architectures,systems where the current state continuously interacts with and modifies the processingcharacteristics. We demonstrate a prominent role of synaptic plasticity (particularly of inhibitorysynapses) in representational and rule-guided learning, an effect achieved by maintainingcompact dynamic representations and sparse, distributed activity patterns. We highlight a formof sequential metastability as a potential mechanism for sequence processing in neocorticalcircuits. In addition, I will discuss how innate constraints in the patterning of the synapticmachinery throughout the neocortex may bias a circuit’s intrinsic timescales and memorycapacity, while the high degree of complexity and heterogeneity may serve importantcomputational purposes by expanding the circuit’s functional space.