Scalp
topography and magnetic field are oriented perpendicularly:
EEG 
MEG 

Potential
/ Magnetfeld Topographie 

L2 –
Minimum Norm 
Radial sources do not generate a
measurable magnetic field:
EEG 
MEG 

Potential
/ Magnetfeld Topographie 

L2 – Minimum Norm 
The EEG Reference Dependency:
Inverse Methods are reference
independant:

EEG
Average Reference 
EEG
Linked Ears Reference 
Potential Topographie  
L2 – Minimum Norm 
EEG 
MEG 
Measures
surface potentials of extracellular currents generated by synchronous
excitatory postsynaptic potentials at apical dendrites of several thousands of adjacent and similarly oriented pyramidal cells. 
Measures
external magnetic fields of intracellular currents generated by
synchronous excitatory postsynaptic potentials at apical dendrites of
several thousands of adjacent and similarly oriented pyramidal cells. 
Measures
radial and tangential sources 
Measures
mainly (90%) tangential sources 
Is reference
dependent 
Is reference
independent 
Inverse
solutions, CSD, Average Reference are reference independent (not
uncritical) 
Is reference
independent 
Scalp
potentials do strongly depend on volume conductor properties. The
inverse solutions do strongly depend on the correct modeling of such
properties. 
External
magnetic fields depend to a smaller degree on volume conductor
properties. The inverse solutions do thus lesser depend on the correct
modeling of such properties. 
Independent
of sensor orientation 
Dependent on
sensor orientation 
In first
approximation independent of sensor positioning (standard positions).
For better approximation measurement of head shape and sensor
coordinates necessary. 
Always
dependent on sensor configuration (no standard positions) =>
Measurement of head shape and sensor coordinates always necessary. 
Group level
statistical analysis possible in sensor domain (in first approximation ) 
Group level
statistical analysis not possible in sensor domain (normalization on
standard sensor system necessary ) 
Smaller
impact of movements (electrode positions stay identical, noise
increases). 
Strong
influence of movements (correction not yet possible). 
Strong
superposition of multiple generators even if far away from each other. 
Smaller
superposition of multiple generators. 
Stronger
impact of ocular artifacts, lesser impact of cardiac artifacts 
Stronger
impact of cardiac artifacts, lesser impact of ocular artifacts 
Smaller
temporal stability due to changing impedancies 
High
temporal stability 
Smaller
impact of environmental noise (shielding chamber beneficial but not
necessary) 
Strong
impact of environmental noise (shielding chamber necessary) 
Small
attenuation of deep sources 
Stronger
attenuation of deep sources, especially in gradiometer systems with
small baseline 
High
dimensionality of the source space (radial & tangential,
superficial & deep sources) => High ambiguity of the inverse
solution 
Smaller
dimensionality of the source space (no radial, no deep sources) =>
Smaller ambiguity of the inverse solution 
Smaller
localization accuracy on single subject level due to volume conductor
dependencies. 
Higher
localization accuracy on single subject level. On group level
advantageous if based on big N. (structural and functional intersubject
variance asks for smoothing). 
Small head
size dependency 
Stronger
head size dependency. Smaller heads show worse S/N. 
up to 256
sensors 
up to 275
sensors 
Medium
investment and medium operating costs 
High
investment and high operating costs 