Welcome to CSE/NEUBEH 528:

Computational Neuroscience

Instructors:
Rajesh Rao (rao@cs)
Adrienne Fairhall (fairhall@u)
TA: Yanping Huang (huangyp@u)

R. Rao, 528 Lecture 1

Todays Agenda
Course Info and Logistics
Motivation

What is Computational Neuroscience?

Illustrative Examples
Neurobiology 101: Neurons and Networks

R. Rao, 528 Lecture 1

Course Information
Browse class web page for syllabus and course information:

http://www.cs.washington.edu/education/courses/528/
Lecture slides will be made available on the website
Textbooks
Required:
Theoretical Neuroscience:
Computational and Mathematical Modeling
of Neural Systems by P. Dayan & L. Abbott
Recommended:
Tutorial on Neural Systems Modelling
by T. Anastasio

R. Rao, 528 Lecture 1

Course Topics
Descriptive Models of the Brain

How is information about the external world encoded in

neurons and networks? (Chapters 1 and 2)
How can we decode neural information? (Chapters 3 and 4)

Mechanistic Models of Brain Cells and Circuits

How can we reproduce the behavior of a single neuron in a

computer simulation? (Chapters 5 and 6)
How do we model a network of neurons? (Chapter 7)

Interpretive Models of the Brain

Why do brain circuits operate the way they do?

What are the computational principles underlying their
operation? (Chapters 7-10)

R. Rao, 528 Lecture 1

Course Goals
General Goals: Be able to

1. Quantitatively describe what a given component of a

neural system is doing based on experimental data
2. Simulate on a computer the behavior of neurons and
networks in a neural system
3. Formulate computational principles underlying the
operation of neural systems
We would like to enhance interdisciplinary cross-talk

Neuroscience

Computing and Engineering

(Experiments, methods,
protocols, data, )

(Computational principles, algorithms,

simulation software/hardware, )

R. Rao, 528 Lecture 1

Workload and Grading

Course grade (out of 4.0) will be based on homeworks and a

final group project according to:

Homeworks: 70%
Final Project: 30%
No midterm or final
Homework exercises: Either written or Matlab-based

Go over Matlab tutorials and homework on class website

Group Project: As part of a group of 1-3 persons, investigate

a "mini-research" question using methods from this course

Each group will submit a report and give a presentation
R. Rao, 528 Lecture 1

Lets begin

What is Computational Neuroscience?

R. Rao, 528 Lecture 1

Computational Neuroscience
The goal of computational neuroscience is to explain in

computational terms how brains generate behaviors

(Sejnowski)
Computational neuroscience provides tools and methods for

characterizing what nervous systems do, determining how

they function, and understanding why they operate in
particular ways (Dayan and Abbott)
Descriptive Models (What)
Mechanistic Models (How)
Interpretive Models (Why)
R. Rao, 528 Lecture 1

An Example: Receptive Fields

What is the receptive field of a brain cell (neuron)?

Any ideas?

R. Rao, 528 Lecture 1

An Example: Receptive Fields

What is the receptive field of a brain cell (neuron)?
Classical Definition: The region of sensory space that

activates a neuron (Hartline, 1938)

Example: Region of the retina where a spot of light activates a
retinal cell
Current Definition: Specific properties of a sensory stimulus

that generate a strong response from the cell

Example: A circular spot of light that turns on at a particular
location on the retina

R. Rao, 528 Lecture 1

10

An Example: Cortical Receptive Fields

Lets look at:
I. A Descriptive Model of Receptive Fields
II. A Mechanistic Model of Receptive Fields
III. An Interpretive Model of Receptive Fields

R. Rao, 528 Lecture 1

11

I. Descriptive Model of Receptive Fields

Spot of light turned on

Retina

Output
responses
(spike trains)
from a
Retinal
Ganglion
Cell

Retinal Ganglion Cells

R. Rao, 528 Lecture 1

(From Nicholls et al., 1992)

12

I. Descriptive Model of Receptive Fields

Mapping a retinal receptive field with spots of light

On-Center
Off-Surround
Receptive Field

Off-Center
On-Surround
Receptive Field
(From Nicholls et al., 1992)

R. Rao, 528 Lecture 1

13

Descriptive Models: Cortical Receptive Fields

Examples of
receptive
fields in
primary
visual cortex
(V1)
Retina

R. Rao, 528 Lecture 1

Lateral
Geniculate
Nucleus (LGN)

V1

(From Nicholls et al., 1992)

14

Extracting a Quantitative Descriptive Model

The Reverse Correlation Method (Brief intro for now)

Random Bars
Sequence
(white noise
stimulus)
I. Ohzawa

For each output spike, look back and record

stimulus sequence occurring before the spike;
Compute the average sequence

R. Rao, 528 Lecture 1

15

A Quantitative Model of a V1 Receptive Field

Space-Time
Receptive Field

Time (T)

Space (Y)

Spatial Receptive
Field for T = 0-300 ms

Space (X)

Space (X)
(Copyright 1995, Izumi Ohzawa)

R. Rao, 528 Lecture 1

16

II. Mechanistic Model of Receptive Fields

The Question: How are receptive

fields constructed using the neural

circuitry of the visual cortex?

How are
these
oriented
receptive
fields
obtained?

R. Rao, 528 Lecture 1

17

II. Mechanistic Model of Receptive Fields: V1

LGN RF

Lateral
Geniculate
Nucleus (LGN)

R. Rao, 528 Lecture 1

V1

LGN
Cells

V1 RF

V1
Cell

18

II. Mechanistic Model of Receptive Fields: V1

Model suggested by
Hubel & Wiesel in the
1960s: V1 RFs are
created from converging
LGN inputs

(From Nicholls et al., 1992)

Center-surround LGN
RFs are displaced along
preferred orientation of
V1 cell
This simple model is still
controversial!

R. Rao, 528 Lecture 1

19

III. Interpretive Model of Receptive Fields

The Question: Why are receptive

fields in V1 shaped in this way?

What are the
computational
advantages of
such receptive
fields?

R. Rao, 528 Lecture 1

20

III. Interpretive Model of Receptive Fields

Computational Hypothesis: Suppose the

goal is to represent images as faithfully and

efficiently as possible using neurons with
receptive fields RF1, RF2, etc.
Given image I, want to reconstruct I using

RF1

RF2

neural responses r1, r2 etc.:

^

I RFi ri

RF3

Idea: Find the RFi that minimize the

^

squared pixelwise errors: || I I || 2 and are

as independent from each other as possible
R. Rao, 528 Lecture 1

RF4
21

III. Interpretive Model of Receptive Fields

Start out with random RFi and run your algorithm on natural

images

White
=+
Dark
= - R. Rao, 528 Lecture 1

22

III. Interpretive Model of Receptive Fields

Conclusion: The brain may be trying to find faithful and

efficient representations of an animals natural environment

Receptive Fields in V1

R. Rao, 528 Lecture 1

23

We will explore a variety of Descriptive,

Mechanistic, and Interpretive models
throughout this course

R. Rao, 528 Lecture 1

24

Neurobiology 101:
Brain regions, neurons, and synapses

R. Rao, 528 Lecture 1

25

Our 3-pound universe

R. Rao, 528 Lecture 1

26

Major Brain Regions: Brain Stem & Cerebellum

Medulla
Breathing, muscle tone
and blood pressure
Pons
Connects brainstem with
cerebellum & involved
in sleep and arousal
Thalamus

Cerebellum
Coordination of voluntary
Corpus collosum
movements and
Cerebellum
sense of equilibrium

Hypothalamus
Pons

Medulla

R. Rao, 528 Lecture 1

27

Spinal cord

Major Brain Regions: Midbrain & Retic. Formation

Midbrain
Eye movements, visual and
auditory reflexes
Reticular Formation
Modulates muscle
reflexes, breathing &
pain perception. Also
regulates sleep,
wakefulness &
arousal
R. Rao, 528 Lecture 1

Thalamus

Midbrain
Hypothalamus
s collosum
Cerebellum

Pons
Medulla
Spinal cord

28

Major Brain Regions: Thalamus & Hypothalamus

Thalamus
Relay station for all
sensory info (except
smell) to the cortex

Hypothalamus
Regulates basic needsCorpus cCorpus
o los
callosum
fighting, fleeing,
Cerebellum
feeding, and
mating

Thalamus
Hypothalamus
Pons
Medulla
Spinal cord

R. Rao, 528 Lecture 1

29

Major Brain Regions: Cerebral Hemispheres

Consists of: Cerebral

cortex, basal ganglia,

hippocampus, and
amygdala

Cerebrum/Cerebral Cortex

Involved in perception

and motor control,

cognitive functions,
emotion, memory, and
learning
R. Rao, 528 Lecture 1

Corpus collosum
Cerebellum

Pons
Medulla
Spinal cord

30

Enterthe neuron (brain cell)

Cerebrum/Cerebral Cortex

Thalamus

~40 m

Pons
Cerebellum
Medulla

A Pyramidal Neuron

Spinal cord

R. Rao, 528 Lecture 1

Cerebral
Cortex Neuron

Neuron from the

Thalamus

31

Neuron from the

Cerebellum

Neuron Doctrine/ Dogma:

The neuron is the appropriate basis
for understanding the computational
and functional properties of the
brain
First suggested in 1891 by Waldeyer

R. Rao, 528 Lecture 1

From Kandel, Schwartz, Jessel, Principles of

Neural Science, 3rd edn., 1991, pg. 21

32

The Idealized Neuron

Input
Spikes

Output
Spike

(Excitatory Post-Synaptic
Potentials)
R. Rao, 528 Lecture 1

33

What is a Neuron?
A leaky bag of charged liquid
Contents of the neuron enclosed

within a cell membrane

Cell membrane is a lipid bilayer

Bilayer is impermeable to
charged ions such as Na+, Cl-,
K+, and Ca2+
From Kandel, Schwartz, Jessel, Principles of
Neural Science, 3rd edn., 1991, pg. 67

R. Rao, 528 Lecture 1

34

The Electrical Personality of a Neuron

Each neuron maintains a potential

difference across its membrane

Inside is 70 to 80 mV
relative to outside
Ion concentration [Na+], [Cl-]
and [Ca2+] higher outside; [K+]
and organic anions [A-] higher
inside
Ionic pump maintains -70 mV
difference by expelling Na+ out
and allowing K+ ions in

[Na+], [Cl-], [Ca2+]

[K+], [A-]
0 mV

Outside

-70 mV

Inside
[K+], [A-]
[Na+], [Cl-], [Ca2+]

R. Rao, 528 Lecture 1

35

Influencing a Neurons Electrical Personality

How can the electrical potential difference

be changed in local regions of a neuron?

R. Rao, 528 Lecture 1

36

Membrane Proteins: The Gatekeepers

Proteins in membranes act as

pores or channels that allow

specific ions to pass through.
E.g. Pass K+ but not Cl- or Na+

These ionic channels are gated

Voltage-gated: Probability of
opening depends on membrane
voltage
Chemically-gated: Binding to a
chemical causes channel to open
Mechanically-gated: Sensitive to
pressure or stretch

From Kandel, Schwartz, Jessel, Principles of Neural

Science, 3rd edn., 1991, pgs. 68 & 137

R. Rao, 528 Lecture 1

37

Gated Channels allow Neuronal Signaling

Inputs from other neurons

chemically-gated channels (at

synapses) Changes in local
membrane potential

Inputs

This causes opening/closing of

voltage-gated channels in dendrites,

body, and axon, resulting in
depolarization (positive change in
voltage) or hyperpolarization
(negative change)

R. Rao, 528 Lecture 1

Synapse

38

The Output of a Neuron: Action Potentials

Voltage-gated channels cause

action potentials (spikes)

1. Rapid Na+ influx causes
rising edge
2. Na+ channels deactivate
3. K+ outflux restores
membrane potential

Action Potential
(spike)

Positive feedback causes spike

Na+ influx increases

membrane potential,
causing more Na+ influx

From Kandel, Schwartz, Jessel, Principles of Neural

Science, 3rd edn., 1991, pg. 110

R. Rao, 528 Lecture 1

39

Propagation of a Spike along an Axon

R. Rao, 528 Lecture 1

From: http://psych.hanover.edu/Krantz/neural/actpotanim.html

40

Communication between Neurons: Synapses

Synapses are the connections

Spike

between neurons
Electrical synapses (gap
junctions)
Chemical synapses (use
neurotransmitters)
Synapses can be excitatory or

inhibitory
Synapse Doctrine: Synapses

are the basis for memory and

learning

Increase or decrease in
membrane potential

R. Rao, 528 Lecture 1

41

Distribution of synapses on a real neuron

R. Rao, 528 Lecture 1

42
http://www.its.caltech.edu/~mbklab/gallery_images/Neu_Syn/PSD-95%20and%20Synapsin.jpg

An Excitatory Synapse
Input spike
Neurotransmitter
release
Binds to Na
channels (which
open)
Na+ influx
Depolarization due
to EPSP (excitatory
postsynaptic
potential)
R. Rao, 528 Lecture 1

43

An Inhibitory Synapse
Input spike
Neurotransmitter
release
Binds to K
channels
K+ leaves cell
Hyperpolarization due
to IPSP (inhibitory
postsynaptic potential)
R. Rao, 528 Lecture 1

44

Down in the
Synaptic Engine
Room
A reductionists
dream! (or
nightmare?)
Note: Even this is
a simplification!
From Kandel, Schwartz,
Jessel, Principles of
Neural Science, 3rd
edn., 1991

R. Rao, 528 Lecture 1

45

Synaptic plasticity: Adapting the connections

Long Term Potentiation (LTP): Increase in synaptic strength

that lasts for several hours or more

Measured as an increase in the excitatory postsynaptic
potential (EPSP) caused by presynaptic spikes

LTP observed as an increase in size or slope

of EPSP for the same presynaptic input

R. Rao, 528 Lecture 1

EPSP

46

Types of Synaptic Plasticity

LTP: synaptic strength increases after prolonged pairing of

presynaptic and postsynaptic spiking (correlated firing of

two connected neurons).
Long Term Depression (LTD): Reduction in synaptic

strength that lasts for several hours or more

R. Rao, 528 Lecture 1

47

Example of measured synaptic plasticity

(From: http://www.nature.com/npp/journal/v33/n1/fig_tab/1301559f1.html)

R. Rao, 528 Lecture 1

48

Types of Synaptic Plasticity

LTP: synaptic strength increases after prolonged pairing of

presynaptic and postsynaptic spiking (correlated firing of

two connected neurons).
Long Term Depression (LTD): Reduction in synaptic

strength that lasts for several hours or more

Spike-Timing Dependent Plasticity: LTP/LTD depends on

relative timing of pre/postsynaptic spiking

R. Rao, 528 Lecture 1

49

Spike-Timing Dependent Plasticity

Amount of increase or decrease in synaptic strength (LTP/LTD)

depends on relative timing of pre & postsynaptic spikes

pre after post

pre before post

LTP

LTD

R. Rao, 528 Lecture 1

(Bi & Poo, 1998)

50

Comparing Neural versus Digital Computing

Device count:

Human Brain: 1011 neurons (each neuron ~ 104 connections)

Silicon Chip: 1010 transistors with sparse connectivity

Device speed:

Biology has up to 100s temporal resolution

Digital circuits will soon have a 100ps clock (10 GHz)

Computing paradigm:

Brain: Massively parallel computation & adaptive connectivity

Digital Computers: sequential information processing via
CPUs with fixed connectivity

Capabilities:

Digital computers excel in math & symbol processing

Brains: Better at solving ill-posed problems (vision, speech)

R. Rao, 528 Lecture 1

51

Conclusions and Summary

Structure and organization of the brain suggests

computational analogies
Information storage: Physical/chemical structure of
neurons and synapses
Information transmission: Electrical and chemical
signaling
Primary computing elements: Neurons
Computational basis: Currently unknown (but inching closer)
We can understand neuronal computation by understanding

the underlying primitives through:

Descriptive models
Mechanistic models
Interpretive models
R. Rao, 528 Lecture 1

52

Next Class
Descriptive Models
Neural Encoding
Things to do:
Visit course website
http://www.cs.washington.edu/education/courses/528/
Matlab practice: Homework 0 and tutorials online
Read Chapter 1 in Dayan & Abbott textbook

R. Rao, 528 Lecture 1

53