Transcript Arithmetic

Cognitive Neuroscience
and Embodied Intelligence
Imaging the Living Brain
Based on book Cognition, Brain and Consciousness ed. Bernard J. Baars
Janusz A. Starzyk
EE141
1
Introduction
The brain imaging has been a
breakthrough technology for cognitive
neuroscience and cognitive psychology.
 Before these techniques were developed
brain study was based on experiments on
animals, and injured human beings.
 But brain injuries are imprecise, damaged areas are hard to locate, and
often observed post-mortem (as in case of Broca’s and Wernicke’s
patients).
 Brain also compensates for the damage, lesions change over time,
adaptation occurs, so that post mortem examination is very imprecise.
 Animal studies depend on presumed homologies – not very convincing.
 No other animals can speak to communicate clearly what they
experience.

2
EE141
Introduction
The brain study was enhanced by imaging techniques like
electroencephalography (EEG) based on X-rays computer tomography,
positron emission tomography (PET), magnetic resonance imaging
(MRI) etc.
 We can observe functional activity of the brain
 Magnetic imaging technique known as diffusion tractography allows to
3
view
white
(myelinated)
fiber
tracts
from
cortex
to
the
spinal
cord.
EE141

Brain recording



Individual neuron’s activities
can be recorded.
Picture shows spike counts for
a single neuron in response to
various images.
This particular neuron
responds selectively to
images of Jennifer Aniston.
4
EE141
Brain recording





Individual neuron’s recording is
seldom possible in humans due
to ethical reasons.
Instead deep electrode
recordings can be performed on
some primates like macaque
monkey.
This is used to test working
memory shown here in Delayed
Match To Sample task.
The second set is displayed 16
sec later – during this time
monkey must remember the
sample to have a correct match.
Firing of the single neuron during
this 16 sec is responsible for
working memory.
EE141
5
Brain imaging techniques
Electroencephalography, (EEG)
 Magnetoencephalography, (MEG)
 Arteriography or Angiography
 Computerized tomography, (CAT)
 Single Photon Emission Computer Tomography,
(SPECT)
 Positron Emission Tomography, (PET)
 Magnetic Resonance Imaging, (MRI)
 Functional MRI, (fMRI)
 Magnetic Resonance Spectroscopy, (MRS)

6
EE141
Time-space tradeoff
fMRI has good spatial
resolution and poor temporal
resolution.
Magnetoencephalography
(MEG) has a good temporal
resolution but cannot locate
precisely the source of firing.
Some studies combine EEG
and fMRI


Most popular imaging methods are compared for their time vs space
resolution.
They do not have yet resolution to track a single neuron or a cluster
of neurons.
7
EE141
Time-space tradeoff
A coronal slice
 Brain Navigator software shows various slices through the brain.
 It shows precise x,y,z, coordinates into brain locations
8
EE141
Time-space tradeoff
A mid-sagittal slice
 A horizontal slice through the eyeballs
as indicated on the top of the image
 Notice they white matter inside and the
gray outer layer
EE141
9
Single-neuron recording
 Hubel and Wiesel (1962) received Nobel price for single-neuron
activities recording in the cortex of a cat.
 More recent work recoded activities in medial temporal lobes.
 Depth electrodes used in humans only in very special cases – eg.
before surgery in epileptic patients.
 The implants can help to determine location of the seizure onset.
EE141
10
Single-neuron recording
Conscious and unconscious observations
 Single neuron recording gives us only a partial information about the
brain function.
 Other observations like subcellular processes, non-classical cells
and synapses, glial cells participate in neural processing.
 Many scientists believe that brain processes can only be observed
11
on the population of neurons.
EE141
Animal and human studies
 Until recently, studies of macaque
monkeys were dominant source of
information about vision, memory,
attention and executive function of
brain
 Their brains have similar functional
regions with minor anatomical
differences
EE141
12
Electroencephalography (EEG)
 Brain’s large scale electrical activities
can be recorded through the scalp on
the surface of the cortex.
 EEG picks up electrical field
generated by group neurons
activities.
 EEG was discovered in 1929 by Hans
Berger.
 EEG has extremely good temporal
resolution.
 A disadvantage is that the surface
neurons are observed better than
activities of deeper neurons.
 Averaged EEG over a number of
experiments yields regular waveforms
that are easier to interpret.
13
EE141
EEG frequencies and their functions
 Delta is the lowest frequency < 4 Hz and occur in a deep sleep or
vegetative state of brain characterizing an unconscious person.
 Theta has frequency 3.5-7.5 Hz – arise from synchronous firing of
many neurons, observed during some sleep states and during quiet
focus (meditation). They are observed during memory retrieval.
 Alpha waves are between 7.5 and 13 Hz – arise from synchronous
firing of large groups of neurons. They originate from occipital lobe
during relaxation with eyes closed but still awake.
 Beta activity is fast irregular at low voltage. Associated with waking
consciousness, busy or anxious thinking, and active concentration.
 Gamma generally ranges between 26 and 70 Hz. Characterizes
active exchange of information between cortical and subcortical
regions. Seen during the conscious waking state and REM dreams
14
(Rapid
Eye
Movement
sleep).
Overlaps
with
beta
activity.
EE141
EEG frequencies and their functions
 Regular waveforms from different brain regions: alpha indicates alert
state, theta reflects interactions between hippocampus and cortex,
15
and gamma indicates conscious perception.
EE141
EEG observations
 High density array of
EEG electrodes placed
on scalp at precise
locations pick up signals
from dendrites of the
outside layers of cortex.
 Fourier analysis of EEG
signal helps to classify
observed responses.
 EEG reveals patters during sleep, waking
abnormalities, even response to music.
16
EE141
EEG response to evoked potential
 Red color – higher
voltage.
 Explicit recognition
shows stronger
signals that a
‘feeling of knowing’.
 Activities are not
very well localized.
 EEG responds to changes of the evoked potential (ERP) averaging
it over many trials.
 ERP can illustrate, for instance, brain patterns in visual processing.17
EE141
EEG response to evoked potential
 Evoked potential (ERP) in response to music.
 Regular musical sequences are perceived as more pleasant and
show a larger evoked potential.
18
EE141
Magnetoencephalography (MEG)
 Measures magnetic field produced by
brain activities.
 Is has spatial resolution of few
millimeters and temporal resolution of
few milliseconds.
 MEG uses Magnetic Source Imaging
(MSI) to superimpose magnetic activities
onto brain anatomical pictures provided
by MRI.
 MSI is used before brain surgery to
locate vital parts of the brain that must
be protected during surgery.
19
EE141
Magnetoencephalography (MEG)
 MEG is entirely silent
(unlike MRI) and
noninvasive so it is
attractive for use with
children and
vulnerable people
 Pictures at the bottom
show the vector fields
of MEG over the head
of the subject
20
EE141
Magnetoencephalography (MEG)
Magnetic field produced by a neuron
Not sensitive to top or bottom
neuron activities
 Due to magnetic field properties, MEG is sensitive to dendritic
flow at the right angles to the walls of cortical folds (sulci).
EE141
21
Stimulating the brain
Penfield’s
map of
language
perception
and
production
 Noninvasive imaging techniques (fMRI, MEG, EEG) help to
observe brain activities.
 Equally useful information may come from evoking neural activities
directly by applying electrical stimuli.
 Wilder Penfield stimulated brains of his epileptic patients during
surgery.
22
EE141
Transcranial Magnetic Stimulation
 Today it is possible to simulate brain lesions in healthy patients without
surgery using transcranial magnetic stimulation (TMS).
 Brief magnetic pulses either excite or inhibit a small region of cortex.
 For instance, if a hand area in motor cortex is stimulated, the subject’s
hand will suddenly move.
 Applying inhibitory pulse over the same region makes movement of the
23
hand
difficult.
EE141
Transcranial Magnetic Stimulation
 TMS works at the
milliseconds scale so it is a
useful technique to study
contribution of specific brain
regions to cognitive process.
 In this example TMS is
applied to Brocka’s and
Wernicke’s regions in the left
hemisphere.
 TMS is safe at mild levels of
intensity and frequency.
24
EE141
fMRI and PET
 EEG and MEG measure
brain activity directly.
 Currently the most popular
techniques fMRI (functional
magnetic resonance
imaging).
 fMRI measures the oxygen
level in local blood circulation
technique called BOLD
(blood-oxygen level
dependent activity).
25
EE141
fMRI and PET
 After burst of neurons
activity there is a drop in
oxygen and glucose.
 BOLD curve shows the
level of oxygen in the
local region.
 Then and upswing in the
BOLD curve reflects a
wave of a new bloodcurried nutrients to the
active region.
 This wave is used up and the oxygen level returns to normal within few
seconds.
 Because of this time delay the fMRI temporal resolution is not so good.
26
EE141
fMRI principle of operation
 Magnetic field aligns spins of oxygen atoms.
 When the field is turned off spins relax returning to their normal random
orientations.
 This relaxation of nuclear spin is picked up as a signal by sensitive 27
coils
EE141
and localized in 3D.
PET vs fMRI
PET scans showing
speaking, seeing, hearing
and producing words






Positron emission tomography is another indirect brain imaging technique.
It was developed much earlier that MRI.
Provides a measure of metabolic brain activity.
It is very expensive and requires a cyclotron.
Subject must be injected with a radioactive tracer.
PET is still important in medical research since different tracers can be
used to trace different molecules.
28
EE141
For other investigations MRI and fMRI are preferred.
Co-registration
 MRI shows structure but not function.
 fMRI shows function but its spatial resolution is not well defined.
 Typically the two images are superimposed in the process called coregistration to relate brain activity to its anatomical features.
29
EE141
Co-registration
 Brain is active all the times.
 To identify task specific
activities a common method is
to subtract two fMRI images
comparing stimulated activity
and control activity.
 Subtraction can have
unwanted consequences when
the important processing goes
on in both conditions.
 Another approach is known as
parametric variation in which
variance of each variable is
estimated.
30
EE141
Visual experiment with fMRI
 fMRI images were
obtained comparing face
objects to nonface
objects.
 Subjects were supposed
to match faces and their
location.
 Figure shows fMRI of
brain activity in two
different tasks.
 Notice that location
matching activates
different brain area than
face matching.
EE141
31
Cognitive functions and brain
 There is an evidence that using specific
cognitive functions changes not only
the brain efficiency of doing them but
also size of the bran part responsible
for such processing.
 For instance hippocampus is
responsible for navigation and memory
of places and routs.
 Birds and animals that burry or hide
their food have bigger hippocampus
than non-storing animals.
 A study compared size of hippocampus
between London taxi drivers and
reference population and showed that
their hippocampus is bigger.
32
EE141
Summary
Brain imaging techniques can illustrate activities of a
single neuron, large cortical structures, dynamic brain
activity, and neurons connectivity.
We learned about a number of most important methods
for brain imaging and discussed their properties.
Brain imaging transformed study of human cognition.
Combination of methods is used to enhance observation
accuracy in time and space.
New methods are constantly being produced.
33
EE141