phase dynamics of neural populations under impact of ... · phase dynamics of neural populations...
TRANSCRIPT
Phase dynamics of neural populations under Phase dynamics of neural populations under impact of statistical feedbackimpact of statistical feedback
Andreas DaffertshoferAndreas Daffertshofer
Research Institute MOVEResearch Institute MOVEFaculty of Human Movement SciencesFaculty of Human Movement Sciences
VU University Amsterdam, The NetherlandsVU University Amsterdam, The Netherlands
Leiden, May 24, 2009Leiden, May 24, 2009
Experimental approaches Experimental approaches ……
Manipulate M/EEG Manipulate M/EEG dynamics via cognitive factorsdynamics via cognitive factors
Changes in Changes in corticocortico--cortical and cortical and corticocortico--spinal synchronizationspinal synchronization
Conceptual framework: movementConceptual framework: movement (in(in--)stabilities)stabilitiesCognitive factors, e.g., attentionCognitive factors, e.g., attention(Motor) learning(Motor) learning
…… Mathematical modeling of information exchange between Mathematical modeling of information exchange between neural ensemblesneural ensembles
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
MEG CentreAmsterdam
MEG CentreAmsterdam
-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4
-100
-50
0
50
100
150
[0.10,-117.27]
[0.10,140.01]
time [sec]
LC15LP11
motor eventmotor event
Experimental framework Experimental framework ………… methods, procedures, analysesmethods, procedures, analyses
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Learning rhythmic bimanual tappingLearning rhythmic bimanual tapping
Houweling, Daffertshofer et al., Houweling, Daffertshofer et al., NeuroImageNeuroImage 20082008
1:1
2:3
ββ--amplitude amplitude modulation modulation
corticocortico--spinal spinal
entrainmententrainmentin in ββ--bandband
learninglearningtim
etim
e
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Which part of activity is transferred to the spinal cord?Which part of activity is transferred to the spinal cord?
Van Wijk, Daffertshofer, et al., Van Wijk, Daffertshofer, et al., Cerebral Cortex Cerebral Cortex (2009)(2009)
--4s 4s --2s 2s --0s0s
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
““Example descending (red) and ascending (blue) pathways that coulExample descending (red) and ascending (blue) pathways that could mediate d mediate corticomuscularcorticomuscular coherence.coherence.””(after Baker, (after Baker, Current Opinion in BiologyCurrent Opinion in Biology 2007)2007)
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Experimental approaches Experimental approaches …… conclusionsconclusions
Neural Neural dynamics can be altered via cognitive factors dynamics can be altered via cognitive factors
Changes in Changes in corticocortico--cortical and cortical and corticocortico--spinal synchronization spinal synchronization are functionally relevantare functionally relevant
MovementMovement (in(in--)stabilities provide excellent window into dynamical )stabilities provide excellent window into dynamical changes changes
(Motor) learning alters (Motor) learning alters corticocortico--spinal entrainmentspinal entrainment
…… Mathematical modeling of information exchange between Mathematical modeling of information exchange between neural ensemblesneural ensembles
KNAW-MEG CentreAmsterdam
KNAW-MEG CentreAmsterdam
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
( ) ( ) ( ) ( )( )
( ) ( ) ( )( )
( ) ( ) ( ) ( )( )
( ) ( ) ( )( )
, ,
, ,
, ,
, ,
, d , , d , , EXT
, d , , d , ,
{ }
{ }d e d i
d e d i
e ie
p x t p x t
e ii
p x t p x t
E x t S Xf x X E X t t Xf x X I X t t
I x t S Xf x X E X t t Xf x X I X t t
Ξ Ξ
Ξ Ξ
′ ′= − − − +
′ ′= − − −
∫ ∫
∫ ∫
Jirsa & Haken, Jirsa & Haken, Physical Review LettersPhysical Review Letters 19961996
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
•• EE((xx,,tt)) / / II((xx,,tt)) are pulse (action potential) densities of excitatory / inhibitoare pulse (action potential) densities of excitatory / inhibitory ry ((sub)populationssub)populations at time at time tt and position and position xx
•• ff(e/i)(e/i)(x,X(x,X)) are the densities of excitatory and inhibitory connections are the densities of excitatory and inhibitory connections between sites between sites xx and and XX
•• ppdd is the pulse densityis the pulse density
•• tt’’=|=|xx--X|/cX|/c with with cc as transmission velocity along the axon (6as transmission velocity along the axon (6--9 9 m/sm/s) )
( ) ( ) ( ) ( )( )
( ) ( ) ( )( )
( ) ( ) ( ) ( )( )
( ) ( ) ( )( )
, ,
, ,
, ,
, ,
, d , , d , , EXT
, d , , d , ,
{ }
{ }d e d i
d e d i
e ie
p x t p x t
e ii
p x t p x t
E x t S Xf x X E X t t Xf x X I X t t
I x t S Xf x X E X t t Xf x X I X t t
Ξ Ξ
Ξ Ξ
′ ′= − − − +
′ ′= − − −
∫ ∫
∫ ∫
Jirsa & Haken, Jirsa & Haken, Physical Review LettersPhysical Review Letters 19961996
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
( ) ( ) ( ) ( )( )
( ) ( ) ( )( )
( ) ( ) ( ) ( )( )
( ) ( ) ( )( )
, ,
, ,
, ,
, ,
, d , , d , , EXT
, d , , d , ,
{ }
{ }d e d i
d e d i
e ie
p x t p x t
e ii
p x t p x t
E x t S Xf x X E X t t Xf x X I X t t
I x t S Xf x X E X t t Xf x X I X t t
Ξ Ξ
Ξ Ξ
′ ′= − − − +
′ ′= − − −
∫ ∫
∫ ∫
Jirsa & Haken, Jirsa & Haken, Physical Review LettersPhysical Review Letters 19961996
Freeman, Freeman, Societies of BrainsSocieties of Brains 1975, 1975, ErmentroutErmentrout & Cowan, & Cowan, Biological CyberneticsBiological Cybernetics 19791979Robinson et al., Robinson et al., Physical Review EPhysical Review E 19981998
( ) ( ) ( ) ( )
( ) ( ), ,
, . ,
with 0 & d 1
, , d
, ˆ
t
n m n m
t
dn n n m d m
w z w t
x t p x t w t pT x
τ τ
κκ τ τ τ
−∞
−∞
≥ − =
Ψ == −
∫
∫
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Jirsa & Haken, Jirsa & Haken, Physical Review LettersPhysical Review Letters 19961996
Freeman, Freeman, Societies of BrainsSocieties of Brains 1975, 1975, ErmentroutErmentrout & Cowan, & Cowan, Biological CyberneticsBiological Cybernetics 19791979Robinson et al., Robinson et al., Physical Review EPhysical Review E 19981998
Frank, Daffertshofer et al.,Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
( ) ( ) ( ) ( )( ) ( ) ( )
( ) ( ) ( ) ( )( ) ( ) ( )
,
, ,
,
,
, d , ,
d , , EXT
, d , ,
d ,
ˆ
ˆ ˆ
ˆ
ˆ ,
{}
{}
e e
e i e e
i
ee
i
ee
i
ii i
E x t S Xf x X E X t t
Xf x X I X t t
I x t S Xf x X E X t t
Xf x
T
T T
X I X t t
T
T
κ
κ κ
κ
κ
Ξ
Ξ
Ξ
Ξ
′= − −
′− +
′= − −
′−
∫∫
∫∫
( ) ( ) ( ) ( )
( ) ( ), ,
, . ,
with 0 & d 1
, , d
, ˆ
t
n m n m
t
dn n n m d m
w z w t
x t p x t w t pT x
τ τ
κκ τ τ τ
−∞
−∞
≥ − =
Ψ == −
∫
∫
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
( ) ( ) ( )1 1 dˆ, , 1 ,ddp x t T x t x ttτ
− ⎛ ⎞= Ψ = + Ψ⎜ ⎟⎠
⇒⎝
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica D D 20002000
( ) ( ) ( ) ( )( ) ( ) ( )
( ) ( ) ( ) ( )( ) ( ) ( )
,
, ,
,
,
, d , ,
d , , EXT
, d , ,
d ,
ˆ
ˆ ˆ
ˆ
ˆ ,
{}
{}
e e
e i e e
i
ee
i
ee
i
ii i
E x t S Xf x X E X t t
Xf x X I X t t
I x t S Xf x X E X t t
Xf x
T
T T
X I X t t
T
T
κ
κ κ
κ
κ
Ξ
Ξ
Ξ
Ξ
′= − −
′− +
′= − −
′−
∫∫
∫∫
( ) ( ),1 expm n
zw z w zτ τ
⎧ ⎫= = −⎨ ⎬⎩ ⎭
( ) ( ) ( ) ( )
( ) ( ), ,
, . ,
with 0 & d 1
, , d
, ˆ
t
n m n m
t
dn n n m d m
w z w t
x t p x t w t pT x
τ τ
κκ τ τ τ
−∞
−∞
≥ − =
Ψ == −
∫
∫
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
( ) ( ) ( ) ( ) ( )( ) ( ) ( )
( ) ( ) ( ) ( ) ( )( ) ( ) ( )
d 1, , d , ,d
d , , ext
d 1, , d , ,d
d , ,
{}
{}
ee
i
ei
i
x t x t S Xf x X X t tt
Xf x X X t t
x t x t S Xf x X X t tt
Xf x X X t t
κτ
κ κ
κτ
κ
Ξ
Ξ
Ξ
Ξ
′= − + −
′− − +
′= − + −
′− −
∫
∫
∫
∫
E E E
I
I I E
I
ˆ ˆ ˆ, and ext EXTTE TI T= = =E I
( ) ( ){ }( ) ( ){ }, ,
, ,
d 1 extdd 1d
e ik k e k l l k l ll l
e ik k i k l l k l ll l
S f ft
S f ft
κ κ κτ
κ κτ
= − + − +
= − + −
∑ ∑
∑ ∑
E E E I
I I E I
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Wilson & Cowan, Wilson & Cowan, BiophysBiophys. J.. J. 19721972
( ) ( ){ }( ) ( ){ }, ,
, ,
d 1 extdd 1d
e ik k e k l l k l ll l
e ik k i k l l k l ll l
S f ft
S f ft
κ κ κτ
κ κτ
= − + − +
= − + −
∑ ∑
∑ ∑
E E E I
I I E I
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
+ noise
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
( ) ( ) { }( ) ( ) ( )( )rot rot rotro,
1
t 21 sin;N
n n n m n mn nn nm
j nd Kdt N
h t Qω α μ=
Φ = + − Φ −ΦΦ +−Ω − Γ∑
~ ~ KuramotoKuramoto--SakaguchiSakaguchi equationequation + external forces+ external forces
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Dynamics of the rotating phase of each oscillatorDynamics of the rotating phase of each oscillator
SakaguchiSakaguchi, , ProgProg. Theo. Phys.. Theo. Phys. 1988; 1988; SakaguchiSakaguchi, , ShinomotoShinomoto & Kuramoto, & Kuramoto, ProgProg. Theo. Phys.. Theo. Phys. 1988 1988
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
( ) ( ) { }( ) ( ) ( )( )rot rot rotro,
1
t 21 sin;N
n n n m n mn nn nm
j nd Kdt N
h t Qω α μ=
Φ = + − Φ −ΦΦ +−Ω − Γ∑
,
1 for 1, ,,
1 otherwisen m n m l
l MK Kv v v
= …⎧= = ⎨−⎩
( ) ( ) with d
d 2sVh h V V π= = − Φ = Φ +Φ
( )
( )
rot 12
rot 32
for 1, ,
otherwisen
n
n
t n M
t
πφ
π
⎧Φ − −Ω = …⎪= ⎨Φ − −Ω⎪⎩
heterogeneityheterogeneity
nonnon--rotating phaserotating phase(unimodal freq. dist.)(unimodal freq. dist.)
external forceexternal forceattracts phaseattracts phase
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Dynamics of the rotating phase of each oscillatorDynamics of the rotating phase of each oscillator
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
{ }( ) ( )1
; sin 2N
n n j n m nm
d Kh Qdt Nφ φ μ φ φ
=
= − − + Γ∑
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Phase oscillatorsPhase oscillators
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
{ }( ) ( )1
; sin 2N
n n j n m nm
d Kh Qdt Nφ φ μ φ φ
=
= − − + Γ∑
( ) { }( ) ( ) ( ) ( ) ( )2 2
20
, ; sin , d , ,jP t h K P t P t Q P tt
π
φ φ μ φ χ χ χ φ φφ φ⎡ ⎤∂ ∂ ∂
= − − − +⎢ ⎥∂ ∂ ∂⎣ ⎦∫
Mean field approximation yields a FokkerMean field approximation yields a Fokker--Planck equationPlanck equation……
……that is nonthat is non--linear in the probability densitylinear in the probability density
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Phase oscillatorsPhase oscillators
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
{ }( ) ( )1
; sin 2N
n n j n m nm
d Kh Qdt Nφ φ μ φ φ
=
= − − + Γ∑
{ }( ) ( ) ( )14; , cos cos 2V φ α β α φ β φ= − +⎡ ⎤⎣ ⎦
( ) { }( ) ( ) ( ) ( ) ( )2 2
20
, ; sin , d , ,jP t h K P t P t Q P tt
π
φ φ μ φ χ χ χ φ φφ φ⎡ ⎤∂ ∂ ∂
= − − − +⎢ ⎥∂ ∂ ∂⎣ ⎦∫
Mean field approximation yields a FokkerMean field approximation yields a Fokker--Planck equationPlanck equation……
……that is nonthat is non--linear in the probability densitylinear in the probability density
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Phase oscillatorsPhase oscillators
x1 x2
Haken et al., Haken et al., BiologBiolog. . CybernCybern.. 19851985SchSchöönerner et al., et al., BiologBiolog. . CybernCybern.. 19861986Kay et al., Kay et al., J. Exp Psychol.J. Exp Psychol. 1987, 19911987, 1991Fuchs et al., Fuchs et al., BiologBiolog. . CybernCybern.. 19961996Daffertshofer et al., Daffertshofer et al., PhysicaPhysica DD 19991999Beek et al., Beek et al., Brain & Brain & CognCogn.. 20022002……
sin sin 2d A Bdtφ φ φ= − −
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Frank, Daffertshofer et al., Frank, Daffertshofer et al., PhysicaPhysica DD 20002000
( ) { }( ) ( ) ( ) ( ) ( )2 2
20
, ; sin , d , ,jP t h K P t P t Q P tt
π
φ φ μ φ χ χ χ φ φφ φ⎡ ⎤∂ ∂ ∂
= − − − +⎢ ⎥∂ ∂ ∂⎣ ⎦∫
φ
φ
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
( ) { }( ) ( ) ( ) ( ) ( )2 2
20
, ; sin , d , ,jP t h K P t P t Q P tt
π
φ φ μ φ χ χ χ φ φφ φ⎡ ⎤∂ ∂ ∂
= − − − +⎢ ⎥∂ ∂ ∂⎣ ⎦∫
Nonlinear FokkerNonlinear Fokker--Planck EquationsPlanck Equations•• can be related to noncan be related to non--extensive entropies via extensive entropies via MaxEntMaxEnt principles (Frank & principles (Frank &
Daffertshofer, Daffertshofer, PhysicaPhysica A, 1999); A, 1999);
e.g., for e.g., for TsallisTsallis generalized entropy we find generalized entropy we find
( )
( ) ( ) ( ) ( )2
2
1( ) 1 ( ) 1
1 , , ,
q
S p p x dxq
P x t h x P x t P x tt x xε
= −−
∂ ∂ ∂= − + ⎡ ⎤⎣ ⎦∂ ∂ ∂
∫
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Frank, Frank, NonNon--linear Fokkerlinear Fokker--Planck equationsPlanck equations Springer 2004Springer 2004
•• often obey poweroften obey power--laws, i.e. they mimic longlaws, i.e. they mimic long--range correlationsrange correlations
Frank, Daffertshofer & Beek, Frank, Daffertshofer & Beek, J. Biological PhysicsJ. Biological Physics 20002000
coupling via coupling via PP((xx,,tt)) = statistical feedback: example = statistical feedback: example ‘‘simulated annealingsimulated annealing’’
( ) ( )
2d 1d 1 ,
n
x h xt P x t
βα
⎡ ⎤= + Γ⎢ ⎥
+⎢ ⎥⎣ ⎦
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Frank, Daffertshofer & Beek, Frank, Daffertshofer & Beek, J. Biological PhysicsJ. Biological Physics 20002000
coupling via coupling via PP((xx,,tt)) = statistical feedback: example = statistical feedback: example ‘‘simulated annealingsimulated annealing’’
( ) ( )
2d 1d 1 ,
n
x h xt P x t
βα
⎡ ⎤= + Γ⎢ ⎥
+⎢ ⎥⎣ ⎦
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
Frank, Daffertshofer & Beek, Frank, Daffertshofer & Beek, J. Biological PhysicsJ. Biological Physics 20002000
coupling via coupling via PP((xx,,tt)) = statistical feedback: example simulated annealing= statistical feedback: example simulated annealing
1, 1, 10n α β= = = 1, 100, 10n α β= = =
• Experimental approacheso Overviewo Design
• Synchronizationo Corticalo Spinal
• Phase dynamicso Jirsa & Hakeno Sakaguchi & Kuramoto
• Statistical feedbacko Mean fieldo Simulated annealing
( ) ( )
2d 1d 1 ,
n
x h xt P x t
βα
⎡ ⎤= + Γ⎢ ⎥
+⎢ ⎥⎣ ⎦
Thanks for your attentionThanks for your attention
Peter BeekPeter BeekLieke PeperLieke PeperTill FrankTill FrankTjeerd BoonstraTjeerd BoonstraSanne HouwelingSanne HouwelingAlistair VardyAlistair VardyBernadette van WijkBernadette van Wijk
Kees StamKees StamBob van DijkBob van DijkPeter PraamstraPeter PraamstraNick RoachNick RoachAlan WingAlan WingGuido NolteGuido NolteMichael BreakspearMichael BreakspearHermann HakenHermann Haken