Folie 1

1
EEG Oscillations One of the basic findings in
brain research is that timing of neural activity
is extremely precise and occurs at a millisecond
range. A critical question in brain research,
thus, is what kind of processess allow for this
precise timing within short AND long
distances. Oscillations may provide an answer.
In the human EEG there are three kinds of (more
or less) well described dominant ongoing,
state/task dependent oscillations Alpha (about
7 12 or 13 Hz) Sleep Spindles (about 12.5 15
Hz) (frontal midline) Theta (around about 6 Hz)
2
Basic Terms
-
Amplitude
Time

1 Period Number of periods per second
Frequency in Hz e.g., an oscillation has a period
of 200 ms 5 periods per second 5 Hz
3
Strogatz,S.H., Stewart, I.(1993).Coupled
Oscillators and Biological Synchronization. Scient
ific American, December, 68-75.
4
Strogatz,S.H., Stewart, I.(1993).Coupled
Oscillators and Biological Synchronization.
Scientific American, December, 68-75.
5
Strogatz,S.H., Stewart, I.(1993).Coupled
Oscillators and Biological Synchronization. Scient
ific American, December, 68-75.
6
Strogatz,S.H., Stewart, I.(1993).Coupled
Oscillators and Biological Synchronization.
Scientific American, December, 68-75.
7
Encoding processes and action potentials (APs)
Basic findings The sequence of APs encodes
information. The example below shows APs
recorded from five neurons. The APs apparently
occur in irregular intervals and, thus, each
neuron encodes different information.
8
This oscillation enables information processing.
It is the driving force of triggering APs.
9
Synchronization of a network by an oscillation
Synchronized, rhythmic activation
Not synchronized activation of a network
10
Oscillations play an important role in the timing
of action potentials (APs)
APs Threshold Subthreshold oscillation
Hyperpolarizing Depolarizing
APs Threshold Suprathreshold oscillation
Hyperpolarizing Depolarizing
11
The development of neural coding the functional
meaning of oscillations and background
noise. After Braun et al. (1994). Oscillation
and noise determine signal transduction in shark
multimodal sensory cells. Nature, Vol. 367, 20
january, 270-273.
12
33 x 4 132 ms
33 x 1 33 ms
33 x 2 66 ms
33 x 3 99 ms
Histogram of interspike distances (ID) Sensory
response at 18 C
0 20 40 60 80 100 120
140 160 180
13
(No Transcript)
14
At 33 C ID 6.99 ms 143 Hz
At 8 C ID 71,4 ms 14 Hz
Interspike Interval Number of
spikes per second
15
(No Transcript)
16
Analysing EEG Oscillations Different types of
event-related changes in oscillatory activity
Stimulus Phase reset
Random phase prior to stimulus onset
Amplitude change
ERS
Intertrial variabilty of phase angle
ERD
Low phase reset Large Phase Locking index (PLI)
Single trials
High no phase reset Small Phase Locking
Index (PLI)
Whole Power (average over single trials) Evoked
Power (ERP)
ERP
Time window of phase reset and amplitude change
may be different
17
Findings about AlphaIndividual Alpha
Frequency (IAF)and event-related
desynchronization/synchronization (ERD/ERS)
18
Even in a sample of young, age matched healthy
subjects, IAF varies to a large degree, between
about 7.5 and 13.5 Hz. Thus, frequency bands
should be adjusted individually using IAF
IAF
Reference (baseline) power Test power
(performance of a task)
A
Lower-1 ?
Lower-2 ?
Upper ?
Theta
Hz
4 6 8 10 12
B
A subject with an individual alpha frequency
(IAF) of 7 Hz. The lower alpha band of this
subject already falls in the theta range if fixed
frequency bands are used
Hz
4 6 8 10 12
A subject with an individual alpha frequency
(IAF) of 12 Hz. The upper alpha band of this
subject already falls at least partly outside the
alpha range if fixed frequency bands are used
C
Hz
4 6 8 10 12
19
Absolute power, Recording site O1
Hz
Hz
Power spectra for 7 and 12 year old children
(Klimesch, 1999 data replotted from ref. ?139?,
p. 201 with permission for reprint). As compared
to young children, older children show a
pronounced increase in upper alpha power and a
strong decrease in theta and delta power. Power
suppression during eyes opening is smaller for
young (B) as compared to older children (C).
20
Age (A) and performance related (B) differences
in IAF
A
B
IAF increases and declines with age just as
cognitive performance and brain volume does
Hz
Hz


10
10


Good memory performers

5
Good memory performers
5
Bad memory performers
Bad memory performers
Alzheimer (65 years)
Young Adults (25 Years)
1 2 3 4 5 6 7 8 9 11 13 15
20
30
40
50
60
70
Age (years)
Interindividual differences in alpha frequency
are large and vary with age and memory
performance. (A) From early childhood to puberty,
alpha frequency increases from about 5.5 to more
than 10 Hz but then starts to decrease with age.
(B) As compared to bad memory performers, good
performers have a significantly higher alpha
frequency, even in Alzheimer demented subjects.
21
Brain volume and IAF are negatively correlated
13.5 12.5 11.5 10.5 9.5 8.5 7.5 6.5 5.5
Hypothetical regression lines for head size
and alpha frequency as predicted by
Nunez (1995) and supported by
experimental findings.
Frequency in Hz
Ve 9.5 m/s Ve 8.5 m/s Ve 7.5 m/s Ve 6.5
m/s Ve 5.5 m/s
11 12 13 14
15 16 17 18
Radius in cm
According to Nunez (1995), global (in contrast to
local) alpha frequency is interpreted in terms of
standing waves and is estimated by f(alpha)
Ve/2?R where Ve is the velocity of action
potentials (in cortico-cortical axons) and R is
the brain radius (in cm) of an idealized spehere
(cf. Nunez, 1995, p. 82ff). A radius of R 14 cm
represents a brain with a cortical surface of
2463 cm² which is about the mean value for young
adults. Because alpha frequency is about 10 Hz in
young adults, the corresponding velocity Ve of
action potentials is about 8.5 m/sec. Differences
in the thickness of the myelin sheath play an
important role for Ve and, thus, have a strong
effect on alpha frequency. A difference in Ve of
1 m/sec is related to a difference in alpha
frequency of more than 1 Hz on the average. The
lower alpha frequency in young children may be
due to differences in the myelin sheath.
22
IAF influences (underlies?) time judgement
Data from Dzhebrailova, T.D. (1995). Perception
and reproduction of time intervals in different
a- and b-rhythm characteristics. Human
Physiology, 21(4), 367-370. A) Measurement of
subjective time First press the on button and
then, after you think 20 sec elapsed, the off
button. Subject with fast alpha
Objective time
On Off
Hypothesis Subjects use the number of cycles
to measure time
Underestimation of time
On
Off
Subject with slow alpha
Overestimation of time
Result The duration of measured time correlates
NEGATIVELY with alpha frequency r - .50 r -
.52 and r -. 54 at P3, O1 and O2 respectively.
23
IAF influences (underlies?) time judgement
B) Subjects have to judge the duration of time
between an on and off signal
On Signal Off
Hypothesis Subjects use the number of cycles
to measure time
Subject with fast alpha gives a high
estimate Subject with slow alpha gives a low
estimate
Objective time
Result The duration of judged time correlates
POSITIVELY with alpha frequency r .49 and r
.58 at P3 and O2 respectively.
24
Individual Alpha Frequency (IAF), Conclusions
IAF varies as a function of- age- brain volume
and conduction velocity- memory and cognitive
performance QuestionIs IAF a passive
resonance phenomenon or an active process e.g.,
controlling the timing of cognitive processes?
25
A)
Lower alpha 8 - 10 Hz O1 correctly remembered
pictures
Evoked alpha
-1 µV
ERD ERS 50 0
50
-1000 -500 0 500
1000 ms
B)
Upper alpha 10 - 12 Hz O1 correctly remembered
pictures
-1 µV
Evoked alpha
ERD ERS 50 0
50
-1000 -500 0 500
1000 ms
26
Upper Alpha Semantic judgment task, Left
Hemisphere Klimesch, W., Doppelmayr, M.,
Pachinger, Th., Russegger, H. (1997).
Event-related desynchronization in the alpha band
and the processing of semantic information.
Cognitive Brain Research, 6 (2), 83-94.
Figure 9. Time course of event-related
desynchronization (ERD) in the upper alpha band
during a semantic judgment task ?79?. Subjects
had to judge whether a feature word (presented
during t2) is semantically congruent with a
concept word (presented during t4). Each interval
(t1 - t5) represents a time period of 500 ms. The
results show that upper alpha desynchronization
(as measured by ERD) is largest during t5 which
is that time period in which the semantic
judgment process actually takes place. Note the
strong left hemispheric advantage (larger
ERD-values over the left as compared to the right
side of the scalp) particularly during t5.
27
Upper Alpha Semantic judgment task, Right
Hemisphere
28
Evidence for hypothesis that semantic memory is
related to upper alpha ERDFrom Röhm, D.,
Klimesch, W., Haider, H., Doppelmayr, M. (2001).
The role of theta and alpha oscillations for
language comprehension in the human
electroencephalogramm. Neuroscience Letters, 310,
137-140.
Experimental design
READING TASK Subjects were instructed to
silently read and to pronounce the sentence right
after a question mark would appear.
SEMANTIC TASK Subjects were instructed to read
the sentence in order to search a super-ordinate
concept for the noun of the third chunk and to
pronounce the super-ordinate concept after the
question mark appeared.
Sample 22 right handed volunteers (8 males, mean
age 22.88 SD 3.34 14 females, mean age
23.79 SD 4.41). Each subject had to perform
first the reading and then the semantic task.
29
Evidence for hypothesis that semantic memory is
related to upper alpha ERDSignificant increase
in upper alpha ERD during retrieval from semantic
long-term memory and semantic processing -
although sentences were already presented in the
preceding reading task.
Build up of contextual constraint
Conclusions An event-related increase in theta
band power appears to reflect working memory
demands. An event-related decrease in upper alpha
band power appears to reflect semantic memory
demands.
30
Event-related increase in alpha (ERS). Example
from Jensen, et al. (2000).
31
Frontal midline theta
The power spectral average over multiple trials
for the 10 subjects during retention. We observe
a peak in the alpha band over posterior regions,
and a peak in the theta band over the frontal
midline. Do these peaks correlate with memory
load?
Posterior alpha
5 10 15 Hz
32
The power in the alpha band for memory load 2
(blue), 4 (green) and 6 (red). Grand average for
10 subjects. The power in the alpha band
increases systematically with memory load over
posterior regions. Note that the increase in high
frequency alpha (10, 11 Hz) is larger as compared
to slow frequency alpha (9 Hz)
9 10 11 Hz
Largest memory load effect in upper alpha
33
Increase in alpha power from S2 to S6
Retention 0 - 3 sec
Retrieval
Note ERD during retrieval
0 0.5 1 1.5 2 2.5
3 3.5 sec
34
Theta upper alpha phase coupling in a Sternberg
task (load 2 and load 4) Schack, Klimesch
Sauseng (2005). Internat. Journal of
Psychophysiology, in press.
F7, set size 4
O2, set size 4
5
10
Hz
Whole Power
Upper Alpha Sync. during retention
15
logarithmic scale for power
20
ms
ms
-500
0
500
-1000
-500
0
500
5
10
Hz
Evoked Power
15
20
ms
ms
-500
0
500
-1000
-500
0
500
5
PLI
PLI
10
Hz
15
20
ms
-1000
-500
0
500
ms
-500
0
500
35
Difference of phase-locking index at 6 Hz load 4
load 2
permutation test (1000 perm.) for PLI at F7
(0-400 ms) tsum0.038 no univariate differences
36
Difference of phase-locking index at 12 Hz load
4 load 2
permutation test (1000 perm.) for PLI at O2
(100-500 ms) tsum0.022 No univariate sign.
diff. 200-500 ms
37
With increasing load alpha PLI decreases at O2
but theta PLI decreases
38
Significant theta and upper alpha PLI and
significant phase coupling suggest nested
oscillations as illustrated below.
ERPs Sternberg yes response, load 2 (blue) and
load 4 (red)
F7
-4 -2 0
2 4 µV
250 500 ms
ERPs generated (in part) by nested theta and
upper alpha. Phase reversal between left
frontal and right posterior sites. F7 dominated
by theta O2 dominated by upper alpha and theta
O2