Week 1.1 - Introduction

Question 1

Name 5 disciplines within Neuroscience and describe their focus.

Answer 1

• Neurobiology - focuses on the biological components of the neuron, and analyzes its bits and pieces in all its glorious biological variability, such as ion channels, neurotransmitters, neuroanatomy.
• Computational neuroscience — in broad outlines — denotes the use of computers for simulating neurons and neuronal networks, i.e., it’s dynamics. Computational neuroscience is concerned with reproducing essential dynamical phenomena of neurons.
• Theoretical neuroscience - uses statistics and math to quantify and describe neuronal dynamics.
• Cognitive Neuroscience - tackles the problems of cognitive function: ways in which neurons enable organisms to deal with the environment.
• Neurodynamics. A branch of mathematical physics that attempts to explain the spiking behavior of neurons from a physical as well as dynamical perspective.

Question 2

What are some of the main differences between Computational Neuroscience and Neural Computation?

Answer 2

Neural Computation

• Focuses on reproducing neural function via simplified single neurons and abstracted networks.
• Often disregards biological detail; high level of abstraction.

Computational Neuroscience

• Focus on biophysics, the spikes or activity of the neuron and its biological components.
• Uses computers to simulate neurons and neural networks (its dynamics).
• Attempts to reproduce/explain dynamical emergent properties of biological neural networks.

Question 3

What is a model?

Answer 3

A model is a representation of a part of reality.

Some examples:

• a computational model of the neuron
• an animal is a model for another animal
• a model of a cognitive function (e.g., memory)
• a statistical model of the weather
• a physical model of the atom

Question 4

What is a good model?

Answer 4

It depends on the question that you want to answer. Because based on that question, you will look for the appropriate level of abstraction.

But in general, these are the desired properties:

a model should be…

• simple
• accurate
• representative
• explanatory
• predictive
• understandable
• detailed

*Note that these properties can be contradictory. There is a trade-off between complexity (i.e. biologically implausibility and efficiency). Can you spot other contradictions here?

Question 5

What is a pattern?

Answer 5

Some kind of reoccurring structure; statistical regularities in the world. Think about ‘visual’ patterns, ‘auditory’ patterns, patterns of muscle activities, and many others.

Question 6

What are Artificial Neural Networks?

Answer 6

ANN’s are computing systems that are inspired by how neurons in the brain work. Most often they abstract biological complexity to emulate either electrophysiological or cognitive function.

Question 7

What is the difference between ‘emulation’ and ‘simulation’?

Answer 7

Emulate: Reproducing the function of a system. A model ‘emulates’ behavioral function, as in, ‘this model emulates the decision making behavior of mice’

Simulate: Use a computational model to reproduce the physical behavior of a substrate, as in ‘this model simulates the membrane potential’.