e.g. bimanual coordination social coordination multimodal binding… etc e.g. imitation task 0-3...
TRANSCRIPT
Spectral peaks, spectral floor and EEG temporal scalesE. Tognoli, october 9th, 2008, HBBL meeting
Peaks~floor
floor
peak
Interim question 1: why are there more peaks in structured behavioral tasks?
Steady-State paradigms and structured behavioral tasks
e.g. bimanual coordinationsocial coordinationmultimodal binding… etc
e.g. imitation task
0-3 peaks1 or 2 typical
0-5 peaks3 or 4 typical
Diversity of 10Hz rhythms (not a comprehensive list)
Non Stationarity: temporal (de)composition of brain dynamics
Largest window for spectral analysis: the run (here 28.5 sec)
Largest window for spectral analysis: the trial (here 6 sec)
STEA
DY-S
TATE
, e.g
. m
ultim
odal
bin
ding
STRU
CTU
RED
TAS
K e.
g.
imita
tion
task
a plateau
an “event”
a quasi-stationary brain state
A predetermined temporal window (here centered at the grip)
a “homogenous” behavioral state
a quasi-stationary brain state
(Reminder: time-frequency tradeoff)
Fourier Wavelet
low
hightimetimetime
frequency
Steady-State paradigms and structured behavioral tasks
Longer paradigm Shorter paradigm
Topographical relations between phenomena of the short~long temporal scale:
bottom up approach
Topographical relation between instantaneous oscillations and spectral peaks: bottom-up
Temporal topography
Spectral topography
Spectral topography (cumulative in time)
Spectral distribution
Spectral distribution(cumulative)
Bottom-up (cnt’d): a drowsy subject
Bottom-up (cnt’d): a wakeful subject
Bottom-up (cnt’d): a wakeful subject with activities standing out
Bottom-up (cnt’d): another non-linearity
One peak, not task relatedNo peak, everything floor(or too many peaks)
Task related peaks
activation
Rest and task-related
Interim question 2: why don’t spectral peaks ever reveal distributed topographies?
Inphase locking
Spurious antiphase Antiphase-locking
“Out-of-phase” locking
Where are the distributed topographies?
Where are the distributed topographies?Re
pres
enta
tive
map
s of
sp
ectr
al p
eaks
: loc
aliz
ed
Expe
cted
map
s fo
rCe
rtai
n pa
tter
ns:
dist
ribut
ed
?
No distributed topographies: implications
≠ + + …
Long time scale (Fourier)
short time scale (pattern)
SOME REORGANIZATION
Topographical relations between phenomena of the short~long temporal scale:
top down approach
Temporal window scaling with task
Window scaling with cognitive/behavioral process
Brain microstate
CUMULANT ABOVE THE COMMON AREA
top down approach: handshake
Observation time scale and spectral content
Observation time scales and spectral content
infinite
Task-scale
process
REST
Task related (local area)
Brain state (local and network), transients
Should we be afraid of transients? (a.k.a. are brief events more likely to be noise than sustained ones?)
Sustained rhythms like alpha in drowsy subjects are often with poor task specificity.
They are not necessary to perform the taskThey may have good inter and intra-individual consistencyThey don’t tell us a lot about the task… yet they are quite systematically modulated by the task
Transients (dead zone of electro-physiologists) may be key to the behavioral or cognitive state under investigation. Especially, coordinated brain state are typically brief.
Strategies for evoked and induced paradigmsDYNAMICS
brain
behavior
brain
Floor: many peaks merged together in an indistinct soup
floor
peak
EEG oscillatory processesfigures of a temporal scale ~ ground of another
Take-home message