Bradley Voytek - Cognitive Neuroscience Society 2016
1. BradleyVoytek, Ph.D.
UC San Diego
Cognitive and Neural Dynamics Laboratory
Department of Cognitive Science
Neurosciences Graduate Program
The Institute for Neural Computation
bvoytek@ucsd.edu
@bradleyvoytek
2.
3. Support
• Alfred P. Sloan Research Fellowship in Neuroscience
• NIMH R01 MH095984
• UCSD Frontiers of Innovation Scholars Program (FISP)
• California Institute forTelecommunications and InformationTechnology (Calit2) Strategic
Research Opportunities (CSRO) program
6. Defining an oscillation
“In a power spectrum, brain oscillations
appear as bumps on top of this 1/f
slope…” - He B, Trends Cogn Sci 2014
“[when] particular oscillation frequencies
become dominant… a peak (bump)
appears in the respective frequency band.” -
Buzsáki el al., Neuron 2013
14. High gamma :: spiking
Source: Mukamel, et al., Science 2005 c.f.: Hermes, et al., Human Brain Mapp 2011
Note that population spiking
negatively correlates with < 30
Hz frequencies, has zero
correlation around 30 Hz, and
then positively correlates with >
40 Hz frequencies.
Very similar to “high gamma
reflects spiking/alpha inversely
reflects cortical potentiation”
ideas of past decades.
23. Simulating LFP
Source: Gao et al. (in preparation)
This means “background” 1/f may
carry information about E:I balance!
24. Spectral slope (1/f)
Source: Podvalny et al., J Neurophysiol 2015; Freeman & Zhai, Cogn Neurodynamics 2009
1/f varies with task and state!
25. What does narrowband power reflect?
Source:Voytek & Knight, Biol Psychiatry 2015
In each case above, narrowband filtering in oscillation range will show power
decreases, but each has very different physiological interpretation!
26. Alpha desync/gamma sync are same process!
Source:Voytek & Knight, Biol Psychiatry 2015; Mukamel, et al., Science 2005; Edwards et al., J Neurophysiol 2005
This “crossover” with low power desynchronization
with high power synchronization aren’t two
different processes, but rather reflect one unified 1/
f slope change!
27. 1/f noise and aging
Source:Voytek et al., J Neurosci 2015
32. Take aways:
• 1/f ~ E:I balance?
• Does your signal even have an oscillation, or is your
narrowband filtering just measuring background noise?
• Huge variability in oscillation center frequency,
bandwidth, and even presence with task, age, disease, etc.
•Take advantage of that!