Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging Pascal Fries Attentional selection selective synchronization. through
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Pascal Fries
Attentional selection
selective synchronization.
through
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Selective attention implementsselective interactions among neuronal groups.
V4 (“lower” visual area)
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Selective attention implementsselective interactions among neuronal groups.
V4 (“lower” visual area)
PFC
TEO (“higher” visual area)
?
?
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Selective attention implementsselective interactions among neuronal groups.
V4 (“lower” visual area)
TEO (“higher” visual area)
PFC
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Selective attention implementsselective interactions among neuronal groups.
TEO (“higher” visual area)
V4 (“lower” visual area)
PFC
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
“Higher” (visual) area
“Lower” visual area Time
Fries, TICS, 2005.
The Communication Through Coherence (CTC) hypothesis:
Selective interaction is implemented mechanistically by selective synchronization.
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Fries et al., Science, 2001.
Attention enhances local gamma-band synchronization.
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Fries et al., Science, 2001.
Attention enhances local gamma-band synchronization.
20 40 60 80 1000
100
200
300
Frequency (Hz)
Coh
eren
cez
Attended
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
20 40 60 80 1000
100
200
300
Frequency (Hz)
Coh
eren
cez
Fries et al., Science, 2001.
Attention enhances local gamma-band synchronization.
Unattended
Attended
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Bar touch
Time0 ms-2000 ms 500-5000 ms
Prestimulusinterval
Stimulus interval Responseinterval
Stimulus
The monkey attention paradigm
+ + +
Fixation
Womelsdorf et al., Nature, 2006.
Gamma-band activity in monkey V4 predicts change detection efficiency.
Womelsdorf et al.,Nature, 2006.
+
x10-3
2.6
2.2
1.8
fast
Time (ms)-150-450 150-300 0
RF stim. is attended
Gam
ma
pow
er
slow
Time (ms)
0.18
0.16
0.20
-150-450 150-300 0
Gam
ma
SFC
Time (ms)
0.7
0.5Firin
gra
te
-150-450 150-300 0
0.6
0.4
0.8
-150 150-300 0
+
x10-3 slowfast
2.6
2.2
1.8
-150-450 150-300 0
RF stim. is ignored
0.18
0.16
0.20
-150-450 150-300 0
Time (ms)-150-450 150-300 0
0.7
0.5Firin
gra
teG
amm
apo
wer
Gam
ma
SFC
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
+ *
Time0 ms-1000 ms 50-3000 ms
Prestimulusinterval
Stimulus interval Responseinterval
Stimulus
The visual MEG paradigm
Hoogenboom et al., Neuroimage, 2005.
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
The spatial selective tactile delayed matching to sample task.
Bauer et al., JNS, 2006.
A B C D E
The different tactilepatterns used:
Time
D
D A D
EAB
C
C
E
35m
s
1000m
s
500m
s
200m
s500
ms
1000m
s
35m
s
1000m
s
35m
s
1000m
s
35m
s
1000m
s
35m
s
1000m
s
35m
s
1000m
s
35m
s
1000m
s
35m
s
Button
press(both
thumbs)
unattended non-targets
attended non-targets
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
-0.2 0 0.2 0.4 0.6 0.840
60
80
100
120
140
160
180
-6
-4
-2
0
2
4
6
Cum
ulat
ive
Z-sc
ore
Time rel. to stim. onset [s]
Freq
uenc
y[H
z]
Tactile stimulation leads to an increase of gamma-band activity mainly in contralateral S1.
Bauer et al., JNS, 2006.
L RL R A P
STI
MR
IGH
TS
TIM
LEFT
t-valuet-value
Gamma-band: Stimulation effect
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Attention enhances the somatosensory gamma-increase.
Time rel. to stim. onset [s]
Cum
ulat
ive
Z-sc
ore
0 0.2 0.4 0.6 0.8
0
1
2
Left attended
Right unattended
Left unattended
Right attended
Bauer et al., JNS, 2006.
L RL R A P
STI
MR
IGH
TS
TIM
LEFT
t-valuet-value
Gamma-band: Attention effect
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
+ *
Time0 ms-1000 ms 50-3000 ms
Stimulus interval
Stimulus
Force
Desiredforcelevel
Prestimulusinterval
Responseinterval
The MEG/EMG paradigm
Schoffelen et al., Science, 2005.
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
0
2.5
5
0 1 2
Haz
ard
rate
(%)
8
6
4
2
0
0 1 2
UP-schedule DOWN-schedule
The hazard rate modulates reaction times.
Time rel. to stim. onset (s) Time rel. to stim. onset (s)
Schoffelen et al., Science, 2005.
Rea
ctio
ntim
e(m
s)
410
390
370
350
400
380
360
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
0 1 2
20
40
60
80
Freq
uenc
y(H
z)
-40-200204060Coherence change (%)
Time (s)0 1 2
20
40
60
80
Freq
uenc
y(H
z)
-40-200204060Coherence change (%)
Time (s)
UP-schedule DOWN-schedule
The hazard rate modulatescortico-spinal coherence.
Schoffelen et al., Science, 2005.
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
0 1 2-10
0
10
20
30
Time rel. to stim. onset (s)
Gam
ma
-coh
eren
cech
ange
(%)
0
20
40
0 1 2Time rel. to stim. onset (s)
UP-schedule DOWN-schedule
The hazard rate modulates cortico-spinalgamma-band coherence.
Schoffelen et al., Science, 2005.
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
20 40 60 80
1
-1
0
Frequency (Hz)
Cro
ssco
rrela
tion
coef
ficie
nt(C
oher
ence
XH
azar
dra
te)
20 40 60 80
1
-1
0
Frequency (Hz)
The hazard rate selectively correlateswith cortico-spinal gamma-coherence.
Schoffelen et al., Science, 2005.
UP-schedule DOWN-schedule
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Time
Towards a direct test of “Communication Through Coherence”.
Womelsdorf et al.,Science, 2007.
0π
0π
“Good” phase relation
“Bad” phase relation
0π
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Neuronal interaction depends on neuronal synchronization.
MUA-MUAphaserelations
Womelsdorf et al.,Science, 2007.
Time
0.4
0.2
0
0 π−πMUA-MUA Phase relation
MU
A-M
UA
Am
plitu
deco
rrela
tion
0 π−π
Time
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Neuronal interaction depends on neuronal synchronization.
MUA - MUAamplitude correlation
Phase relation(0 = “good”)
Womelsdorf et al.,Science, 2007.
MUA - LFPamplitude correlation
Phase relation(0 = “good”)
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
CC (phase relation X amplitude corr.):
Good phase relation precedes strong amplitude correlation.
Womelsdorf et al.,Science, 2007.
Peak lag = -5 ms (p < 0.001)
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
General summary
Gamma-band synchronization is dynamically modulated and predicts behavior.
This suggests an important functional role. Synchronization seems to be one mecha-nism through which cognitive operations like attention control neuronal interactions.
Neuronal synchronization has an mechanistic influence on neuronal interactions.
Groups of neurons influence each other stronger when they are synchronized as com-pared to when they are out of synchrony.
Mechanistically, this is likely due to the fact that rhythmic synchronization within a neu-ronal group implements rhythmic changes in input gain to that group.
Leiden, 26 May 2008 F.C. Donders Centre for Cognitive Neuroimaging
Links: The data analysis software is available in the FieldTrip open source Matlab tool-box: www.ru.nl/fcdonders/fieldtrip.
Positions:Three postdoctoral positions and one PhD position are open in my lab.Details at: www.ru.nl/fcdonders.
Jan-MathijsSchoffelen
ThiloWomelsdorf
WolfSinger(MPI)
BobDesimone(MIT)
RobertOostenveld
Gijs vanElswijk
Thanks to:
NienkeHoogenboom
FemkeMaijDick
Stegemann
ConradoBosman
MarkusBauer