the firing rate‐lfp relaon changes as a funcon of firing...

1
The Firing Rate‐LFP Rela1on Changes as a Func1on of Firing Rate in Humans Ashwin G. Ramayya 1 , Jeremy R. Manning 1 , Joshua Jacobs 3 , Michael J. Kahana 2 1 Neuroscience Graduate Group and 2 Department of Psychology, University of Pennsylvania, Philadelphia, PA 3 Dept. of Bioengineering, Drexel University, Philadelphia, PA Introduc1on Sliding window analysis reveals non‐linearity in FR‐LFP rela1on Broadband + Neurons (n=453) Gamma + Neurons (n=141) 1/f “pink noise”: calculated as mean of Robust‐fit line Regression framework separates broadband and gamma correla1ons Correlated with single neuron firing rate Broadband power (2‐150 Hz) is not an oscilla1on References Concave‐Up (n=285) Not Quadra1c (n=128) Concave‐Down (n=40) Concave‐Up (n=68) Not Quadra1c (n=47) Concave‐Down (n=26) 18% 33% 48% 63% 28% 8% Extracellular recordings 20 paVents Virtual NavigaVon Task 2,030 neurons idenVfied Spikes removed from LFP Analyzed in 500ms epochs Conclusions LFP= Summed ac1vity of all neurons Firing Rate= Spiking ac1vity of single neuron 2 Modes of a single neuron “Entrained” to Network Independent AcVvity Firing Rate LFP LFP Firing Rate High Firing Rates Low Firing Rates LFP Firing Rate 20 percenVle window Power Frequency Broadband power Narrowband power Narrowband OscillaVon Single neuron firing rate has been widely correlated with gamma power (30‐150 Hz) of the local field potenVal (LFP) 1 . Recent evidence suggests that broadband power (2‐150 Hz) is another electrophysiological correlate of single neuron firing rate in humans 2 . Firing rate Results Classifying + correlated neurons with a quadra1c fit Possible model of concave‐up neurons + CorrelaVon ‐ CorrelaVon Firing rate (Hz) Firing rate (Hz) Firing rate (Hz) Firing rate (Hz) Firing rate (Hz) Firing rate (Hz) Mean =2.7 Mean =3.0 Mean =3.2 Mean =2.9 Mean =2.3 Mean =2.9 3 subpopulaVons of BB+ and gamma+ neurons (fig. 4): 1. Concave‐up: Most common; FR‐LFP well correlated at high firing rates but not at low firing rates (see fig. 5 “possible model”) 2. Not Quadra1c: PosiVvely correlated neurons displaying either linear or higher order FR‐LFP relaVons. 3. Concave‐down: Least common; FR‐LFP well correlated at low firing rates but not at high firing rates. Previously described in monkey literature 3 . Frontal Hipp. ParaHipp. Amyg. Post. Cx Frontal Hipp. ParaHipp. Amyg. Post. Cx Frontal Hipp. ParaHipp. Amyg. Post. Cx Frontal Hipp. ParaHipp. Amyg. Post. Cx Frontal Hipp. ParaHipp. Amyg. Post. Cx Frontal Hipp. ParaHipp. Amyg. Post. Cx 1. Whijngstall, K. and LogotheVs, N. (2009). Frequency‐Band Coupling in Surface EEG Reflects Spiking AcVvity in Monkey Visual Cortex. Neuron, 64(2):281–289. 2. Manning, J., Jacobs, J., Fried, I., and Kahana, M. (2009). Broadband shims in LFP power spectra are correlated with single‐neuron spiking in humans. Journal of Neuroscience, 29(43):13613 – 13620. 3. Mazzoni A, Whijngstall K, Brunel N, LogotheVs NK and Panzeri S (2009). Understanding the relaVonships between spike rate and delta/gamma frequency bands of LFPs and EEGs using a local corVcal network model. Neuroimage, 52(3):956‐972

Upload: trandieu

Post on 13-Apr-2018

224 views

Category:

Documents


1 download

TRANSCRIPT

TheFiringRate‐LFPRela1onChangesasaFunc1onofFiringRateinHumansAshwinG.Ramayya1,JeremyR.Manning1,JoshuaJacobs3,MichaelJ.Kahana2

1NeuroscienceGraduateGroupand2DepartmentofPsychology,UniversityofPennsylvania,Philadelphia,PA3Dept.ofBioengineering,DrexelUniversity,Philadelphia,PA

Introduc1onSlidingwindowanalysisrevealsnon‐linearityinFR‐LFPrela1on

Broadband+Neurons(n=453) Gamma+Neurons(n=141)

1/f“pinknoise”:calculatedasmeanofRobust‐fitline

Regressionframeworkseparatesbroadbandandgammacorrela1ons

Correlatedwithsingleneuronfiringrate

Broadbandpower(2‐150Hz)isnotanoscilla1on

References

Concave‐Up(n=285)

NotQuadra1c(n=128)

Concave‐Down(n=40)

Concave‐Up(n=68)

NotQuadra1c(n=47)

Concave‐Down(n=26)

18%

33%

48%63%

28%

8%• Extracellularrecordings20paVents• VirtualNavigaVonTask• 2,030neuronsidenVfied• SpikesremovedfromLFP• Analyzedin500msepochs

Conclusions

LFP=Summedac1vityofallneurons

FiringRate=Spikingac1vityofsingleneuron

2Modesofasingleneuron “Entrained”to

NetworkIndependentAcVvity

FiringRate

LFP

LFP

FiringRate

HighFiringRates

LowFiringRates

NetworkatRest Networkattask

LFP

FiringRate

LFP

FiringRate

LFP

FiringRate

20percenVlewindow

Power

Frequency

Broadbandpower

Narrowbandpower

NarrowbandOscillaVon

Singleneuronfiringratehasbeenwidelycorrelatedwithgammapower(30‐150Hz)ofthelocalfieldpotenVal(LFP)1.Recentevidencesuggeststhatbroadbandpower(2‐150Hz)isanotherelectrophysiologicalcorrelateofsingleneuronfiringrateinhumans2.

Firingrate

Results

Classifying+correlatedneuronswithaquadra1cfit Possiblemodelofconcave‐upneurons

+CorrelaVon

‐CorrelaVon

Firingrate(Hz)

Firingrate(Hz)

Firingrate(Hz)

Firingrate(Hz)

Firingrate(Hz)

Firingrate(Hz)

Mean=2.7

Mean=3.0

Mean=3.2

Mean=2.9

Mean=2.3

Mean=2.9

3subpopulaVonsofBB+andgamma+neurons(fig.4):1.  Concave‐up:Mostcommon;FR‐LFPwellcorrelatedat

highfiringratesbutnotatlowfiringrates(seefig.5“possiblemodel”)

2.  NotQuadra1c:PosiVvelycorrelatedneuronsdisplayingeitherlinearorhigherorderFR‐LFPrelaVons.

3.  Concave‐down:Leastcommon;FR‐LFPwellcorrelatedatlowfiringratesbutnotathighfiringrates.Previouslydescribedinmonkeyliterature3.

Fron

tal

Hipp.

ParaHipp.

Amyg.

Post.Cx

Fron

tal

Hipp. ParaHipp.

Amyg. Po

st.Cx

Fron

tal H

ipp.

ParaHipp.

Amyg.

Post.Cx

Fron

tal

Hipp.

ParaHipp.

Amyg. Post.Cx

Fron

tal

Hipp.

ParaHipp.

Amyg.

Post.Cx

Fron

tal

Hipp.

ParaHipp.

Amyg.

Post.Cx

1.  Whijngstall,K.andLogotheVs,N.(2009).Frequency‐BandCouplinginSurfaceEEGReflectsSpikingAcVvityinMonkeyVisualCortex.Neuron,64(2):281–289.

2.Manning,J.,Jacobs,J.,Fried,I.,andKahana,M.(2009).BroadbandshimsinLFPpowerspectraarecorrelatedwithsingle‐neuronspikinginhumans.JournalofNeuroscience,29(43):13613–13620.3.MazzoniA,WhijngstallK,BrunelN,LogotheVsNKandPanzeriS(2009).UnderstandingtherelaVonshipsbetweenspikerateanddelta/gammafrequencybandsofLFPsandEEGsusingalocalcorVcalnetworkmodel.Neuroimage,52(3):956‐972