NEUROINFORMATICS:
Eduardo Serrano :
[email protected]Francisco de Borja Rodríguez:
[email protected] INTRODUCTION TO
COMPUTATIONAL NEUROSCIENCE:
• Fundamentals of Neuroscience
• Introduction to neural information processing
• Scales and strategies in neural modeling
Computational Neuroscience
• Computational Neuroscience studies the nervous system
from the perspective of information processing.
• Computational Neuroscience is a multidisciplinary field: it
uses knowledge from biology, biophysics, biochemistry,
computer science, engineering, medicine, psychology….
• What is it useful for?
– To understand the function of the brain
– To study several types of diseases and to design prosthetic
deices and brain machine interfaces.
– To design novel bio-inspired paradigms in artificial intelligence,
robotics and bio-inspired devices.
The brain and the computers
Information Processing Element Processing Type of Fault - Able to Intelligent
Power
Processor elements size speed computation tolerant learn Concious
1014 10-6 m 30 W 100 Hz Parallel, Yes Yes Often
synapses distributed
108 10-6 m 30W 109 Hz Serial, NO A little No
transistors (CPU) centralized (yet)
Nature and human engineering
• Nature uses a different “engineering” strategy than
humans.
• Classic human engineering is based on control
(controlling dynamics, controlling noise, controlling
quality, etc.) and is primarily goal-oriented.
Thus, its capabilities are limited by the imagination of those setting
the goals. Things like noise, oscillation, transients and
combinatorial richness interfere with this imagination and are thus
seen as hazards.
Engineering by Nature
• Nature´s engineering, is based on exploitation (of
oscillations, noise, diversity and combination, etc). As
this natural engineering configures more complex
phenomena, the space of possibilities expands, thus
making even more complex phenomena possible.
• Nature's engineering is open-ended, and things such as
noise, variation, transients and combinatorial richness
are seen as enablers rather than problems.
Levels of information processing
in the brain
• Subcellular
Informational complexity
• Cellular
• Network
• System
• Brain
Mechanisms to generate information at the
cellular level
channel
lipid bilayer
The neuron excitability arises from the electrical
properties of its membrane.
Membranes keep different ionic concentrations between
the inside and the outside of the cell
This creates a voltage difference between the inside and
the outside, and a concentration of changes at the
surface.
Membranes have passive and active electrical properties.
In addition to electrotonic components (voltage-
independent) we have voltage- or concentration-
dependent
There is a combination of fast and slow dynamics
neural diversity
Neurons are the cells that have the largest morphological
diversity
Firing diversity
Action potentials also display a large variety of
waveforms.
Computational neuroscience uses
models of neurons and networks
To study the mechanisms of:
• Information coding
• Information transduction
• Information processing and memory
• Information coordination and creation
• Information execution
Computing with attractors
Stable node: simple attractor in the phase space of
a strongly dissipative dynamical system
Computing with separatrices:
saddle with two stable and two
unstable separatrices
A separatrix is a manifold (surface or curve) that refers to
the boundary separating two modes of behavior in the
phase space of a dynamical system
