Neuroscience Letters 452 (2009) 119–123
Contents lists available at ScienceDirect
Neuroscience Letters
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / n e u l e t
Evidence that transduction of electromagnetic field is mediated by
a force receptor
Andrew A. Marino a,∗, Simona Carrubba a, Clifton Frilot b, Andrew L. Chesson Jr. c
a Department of Orthopaedic Surgery, LSU Health Sciences Center, P.O. Box 33932, 1501 Kings Hwy., Shreveport, LA 71130-3932, United States
b School of Allied Health, LSU Health Sciences Center, Shreveport, LA, United States
c Department of Neurology, LSU Health Sciences Center, Shreveport, LA, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Low-strength magnetic fields triggered onset and offset evoked potentials, indicating that the detection
Received 25 November 2008
process was a form of sensory transduction; whether the field interacted directly with an ion channel
Received in revised form 6 January 2009
or indirectly via a signaling cascade is unknown. By analogy with electrosensory transduction in lower
Accepted 21 January 2009
life forms, we hypothesized that the evoked potentials were initiated by a force exerted by the induced
electric field on an ion channel in the plasma membrane. We applied a rapid magnetic stimulus (0.2 ms)
Keywords:
and found that it produced evoked potentials indistinguishable in latency, magnitude, and frequency from
Magnetosensory evoked potentials
those found previously when the stimulus was 50 times slower. The ability of the field-detection system
Nonlinear analysis
in human subjects to respond to the rapid stimulus supported the theory that the receptor potentials
Sensory transduction
Patch-clamping
necessary for production of evoked potentials originated from a direct interaction between the field and
an ion channel in the plasma membrane that resulted in a change in the average probability of the channel
to be in the open state.
© 2009 Elsevier Ireland Ltd. All rights reserved.
The initial stages of sensory transduction include an interaction of
is unknown; animal studies suggest it is probably in the head [12],
the stimulus with plasma-membrane or intracellular structures in
possibly the cerebellum [8].
specialized cells, leading to changes in mean conductance of ion
Based on electrophysiological and modeling studies of the elec-
channels. For sound and touch, the ion channel is a force receptor
troreceptor in the catfish Kryptopterus bicirrhis, a species for which
that interacts directly with the stimulus; in other cases including
the neuroanatomy of the electrosensory system is well known, we
light and some kinds of chemicals, ion channels are the effectors of
proposed that the catfish detected EMFs by means of their inter-
a biochemical signaling cascade that results in a receptor potential
action with glyco groups attached to the gate of an ion channel,
(Fig. 1a–c). The electrical response of cells having force receptors in
resulting in a force tending to open the gate [11]. Reasoning by anal-
the membrane, hair cells for example, typically occurs 0.04–0.20 ms
ogy, it occurred to us that the electroreceptor responsible for the
following application of the force [7,16]. Vertebrate photorecep-
development of MEPs might also be a force detector responsive to
tors, in contrast, have latencies (delay between photon absorption
the induced electric field (Fig. 1d). Under this assumption, using the
and change in channel conductance) about 100 times longer, con-
patch-clamp technique [13], we measured single-channel proper-
sistent with the role of second-messengers in visual transduction
ties of the field-sensitive channel in the catfish to gain insight into
[14,16].
how rapidly we might expect the analogue human field-sensitive
The onset and offset of magnetic fields produced magnetosen-
channel to respond to a field. After establishing that the channel
sory evoked potentials (MEPs) in human subjects, consistent with
being measured was the field-responsive membrane ion channel
the view that the detection process was a form of sensory transduc-
(data not shown), we recorded single-channel currents (Fig. 2). The
tion [4,5]; the MEPs were observed using nonlinear analysis, but not
average open time of the channel, a measure of the switching time
by means of time averaging. The rise- and fall-times of the stimuli
between closed and open states, was about 0.2 ms. It could therefore
that produced the MEPs were about 10 ms, indicating that the sys-
be anticipated that 0.2-ms signals (and perhaps even more rapid,
tem that mediated transduction (assumed to be based on one of the
depending on signal intensity) might affect the probability of the
types of receptors shown in Fig. 1a–c) could respond to a stimulus
force receptor to be in the open state.
at least as fast as 10 ms. The location of the human electroreceptor
To further explore the idea that the EMF transduction pro-
cess responsible for MEPs involved a force receptor similar to that
in Kryptopterus, we applied 0.2-ms magnetic stimuli to human
∗
subjects with the intent of interpreting observations of MEPs as
Corresponding author. Tel.: +1 318 675 6180; fax: +1 318 675 6186.
E-mail address: amarino@lsuhsc.edu (A.A. Marino).
evidence that the ion channel involved in the EMF transduction
0304-3940/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.neulet.2009.01.051

---

120
A.A. Marino et al. / Neuroscience Letters 452 (2009) 119–123
Fig. 1. Signal transduction in sensory receptors. (a) Some stimuli (sound, touch as examples) mechanically induce conformational change of an ion channel. In other cases
such as detection of light (b) and chemicals (c), transduction is mediated by intracellular messengers released after the initial molecular events triggered by the stimulus. (d)
Model for detection of electric fields. The glycocalyx consists of negatively charged oligosaccharide side chains covalently bound to ion channels [11]. An applied electric field
E exerts a force F on the glycocalyx, thereby mechanically opening the channel gate.
process was a force receptor, recognizing that the evidence would
rise- and fall-times were 10 ms [4]. We reproduced the rise-time
not be direct proof.
as a positive control, and utilized a fall-time of 0.2 ms to assess
Ten clinically normal subjects were studied (5 males, 24–61
whether the more rapid stimulus (by a factor of 50) would also
years, and 5 females range 32–67 years). The subjects were
result in offset MEPs (Fig. 3b). To quickly change the fall-time of
informed of the goals, methods, and general design of the investiga-
the field, we used a NPN transistor (Fairchild TIP102) to switch off
tion, but were not told exactly when or for how long the field would
the coil current (the switching time was determined by the coil
be applied. Written informed consent was obtained for each sub-
current and inductance, and the capacitance across the transistor
ject prior to participation in the study. The review board for human
leads).
research at our institution approved all experimental procedures.
The subjects sat in a comfortable wooden chair with their eyes
To achieve precise control of the duration of the rise- and fall-
closed; care was taken to insure the equipment controlling the
times of the magnetic field we used dc (direct current) magnetic
coil current and recording the electroencephalogram (EEG) did not
fields. A uniaxial dc magnetic field having a strength of 1 G (0.1 mT)
provide sensory cues. None of the subjects consciously perceived
uniform to within 5% over the region of the head was applied to the
the field. Electroencephalograms (V(t)) were recorded continuously
subjects in the coronal plane using a pair of coils (Fig. 3a); details
from O1, O2, C3, C4, P3, and P4 (International 10–20 system) ref-
of the experimental system are given elsewhere [3]. The field was
erenced to linked ears, using gold-plated electrodes attached to
applied for 2-s intervals, each separated by a 5-s interval during
the scalp with conductive paste. Electrode impedances (measured
which there was no applied field. The geomagnetic field was 0.26 G,
before and after each experiment) were <10 k
in all subjects. The
59.9◦ below the horizontal (0.04 G along the direction of the applied
signals were amplified (Nihon Kohden, Irvine, CA), filtered to pass
field).
0.5–35 Hz, sampled at 300 Hz using a 12-bit analog-to-digital con-
We showed previously that 94% (16 of 17 subjects) exhibited an
verter (National Instruments, Austin, TX), and analyzed offline. The
onset MEP and 65% (11 of 17) exhibited an offset MEP when the
signals were divided into consecutive 7-s trials with field onset
Fig. 2. Single channel current from a voltage-sensitive channel in an electroreceptor cell in Kryptopterus bicirrhis.

---

A.A. Marino et al. / Neuroscience Letters 452 (2009) 119–123
121
vectors in the trajectory were near (defined as within 15% of the
maximum distance between any two states); distances were cal-
culated using the Euclidean norm. The plots were quantified using
two recurrence variables [18]: (1) percent recurrence (%R), defined
as the ratio of the number of points in the plot to the total num-
ber of points in the recurrence matrix; (2) percent determinism
(%D), defined as the fraction of points in the plot that formed diag-
onal lines consisting of at least 2 adjacent points. The process was
repeated using a sliding window of 1 point in V(t), yielding the time
series %R(t), which was smoothed using a 100-ms, step-1 averaging
window. The resulting time series, %R(t) and %D(t), were analyzed
for the presence of evoked potentials. All calculations were per-
formed using publicly available software [17], and verified using a
custom Matlab code (Mathworks, Natick, MA).
Onset and offset of a magnetic field each produced an evoked
potential [4]; we examined the same latency range to detect evoked
potentials in the present study. Each of the 60 points in %R(t) and
in %D(t) between 209 and 404 ms (which described the dynamical
activity in V(t) at 109–504 ms) were compared individually with
the corresponding points in the control epochs using the paired
t-test at a pair-wise significance level of p < 0.05 (identical results
were found using the Wilcoxon signed rank test). In preliminary
studies on baseline EEGs (no field) consisting of 2048 sets of 50
sham-field versus control comparisons, we found that the proba-
bility of observing ≥10 significant tests (out of 60) due to chance was
about 0.04. We therefore planned to regard a comparison of a set of
evoked-potential (onset or offset (Fig. 3a)) and control epochs from
any particular electrode as significant if ≥10 tests were pair-wise
significant at p < 0.05.
Filtering the EEG in the alpha band sometimes facilitated detec-
tion of an MEP, and the results depended on the nature of the
filtering (sometimes filtering 9–12 Hz but not 8–10 Hz was effec-
tive, and sometimes conversely) [6]. In addition, although use of
%R and %D often gave the same result, there were instances where
only one of the quantifiers detected a field-induced change in
the EEG [6]. Based on these prior observations, we systematically
Fig. 3. Application of magnetic fields. (a) Schematic diagram of the exposure and
considered all conditions of analysis previously shown capable of
EEG-detection systems (mid-sagittal view). (b) Onset and offset of the magnetic
revealing an MEP [6]. First, we analyzed %R(t) in all 6 electrodes.
field. (c) Procedures for nonlinear and linear analysis.
If we found an evoked potential (≥10 pair-wise significant tests
within the expected latency interval) in at least 3 electrodes, no
beginning at t = 0, field offset beginning at t = 2 s, and a field-free
further analyses were conducted. If fewer than 3 evoked poten-
period at 2 < t < 7 s (Fig. 3a).
tials were found, we analyzed %D(t). If a total of 3 evoked potentials
Each subject received two blocks of 80 trials. The magnetic field
were still not detected, we filtered V(t) prior to calculating %R(t) and
was applied in either the earlier or later block, as determined ran-
%D(t) and continued the analysis until either 3 evoked potentials
domly from subject to subject. In the block where the field was
were detected or all the 6 predetermined conditions (combinations
not applied, the data was analyzed as a negative control (sham
of recurrence variable and filtering conditions) were considered.
exposure). To help maintain alertness, 5 binaural 2-s 424-Hz tones
The overall results did not depend on the order; for presentation,
were presented prior to each of the field and sham sessions. Trials
we chose the sequence %R(t), %D(t), %R(t), after filtering the EEG
containing artifacts as assessed by visual inspection [10] were dis-
at 8–10 Hz, %D(t) after filtering at 8–10 Hz, %R(t) after filtering at
carded (<5% of all trials), and the artifact-free trials were digitally
9–12 Hz, %D(t) after filtering at 9–12 Hz. Whenever tests were done
filtered between 0.5 and 35 Hz. All results were based on data from
to compare evoked-potential and control epochs, the conditions
at least 50 trials.
being evaluated were also applied to the sham data (sham evoked
The rise and the fall of the field each produced a spike in V(t)
potential versus sham control). Thus, for example, when the exper-
that was broadened to 30 ms by the time constant of the EEG ampli-
imental data was filtered at 8–10 Hz, so was the sham data. At the
fier. In preliminary studies using electrical phantoms of the head,
conclusion of the study we calculated the a posteriori false-positive
we established that the spikes arose from Faraday-type induction,
rate (number of false-positive effects in the sham data divided by
and were unrelated to neuronal activity. Prior to analyzing V(t), the
the total number of tests performed on the sham data), and used
spikes were removed by deleting the first 30 ms of data (10 points,
that error rate to estimate the family-wise error (PFW) for the deci-
see below) after presentation of the stimulus.
sion that a subject had detected the stimulus.
Details of our nonlinear method (Fig. 3c) were given elsewhere
Prior to the study we were unaware of whether the probability
[3]. Briefly, the first 100 ms of each of the epochs of interest in V(t)
of detection of evoked potentials would depend on the electrode
(t = 0.03–1 s, 2.03–3 s, and 5.03–6 s, corresponding to onset, offset,
derivation. We therefore computed the contributions to PFW sepa-
and control intervals, respectively) (Fig. 3a) were embedded in five-
rately for the central, occipital, and parietal electrodes using the
dimensional phase space, using a time delay of 5 points (17 ms), and
binomial formula, and the overall family-wise error rate for the
the resulting trajectory was mapped to a two-dimensional recur-
occurrence of evoked potentials in each experiment was deter-
rence plot by placing a point at (i,j) whenever the ith and jth state
mined by the law of compound probability.

---

122
A.A. Marino et al. / Neuroscience Letters 452 (2009) 119–123
Table 1
Onset (a) and offset (b) evoked potentials in subjects exposed to a magnetic stimulus (1 G, DC) having a rise-time of 10 ms (onset potential) and a fall-time (offset potential)
of 0.2 ms.
Subject
%R
%D
%R (8–10 Hz)
%D (8–10 Hz)
%R (9–12 Hz)
%D (9–12 Hz)
All Effects
No. Tests
PFW
(a)
S1
C3
X
C4
C4
–
–
C3 C4 C4
23
0.074
S2
P3
O1 P3
–
–
–
–
O1 P3 P3
12
0.005
S3
C3 P3
C3 P3
–
–
–
–
C3 C3 P3 P3
12
0.000
S4
O1 O2
O1 O2
–
–
–
–
O1 O1 O2 O2
12
0.001
S5
C3 C4 P3 P4
–
–
–
–
–
C3 C4 P3 P4
6
0.000
S6*
O2
O2
X
C3
–
–
O2 O2 C3
22
0.024
S7
C4
X
P4
C4 P4
X
–
C4 C4 P4 P4
23
0.012
S8
P3
P3
X
C3
–
–
C3 P3 P3
22
0.024
S9
X
X
X
P3
O1 P4
–
O1 P3 P4
30
0.057
S10
C4
X
P4
P4
–
–
C4 P4 P4
23
0.030
(b)
S1
X
P4
P3
C4 P3
–
–
C4 P3 P3 P4
23
0.004
S2
O2 C3 P3
–
–
–
–
–
O2 C3 P3
6
0.001
S3
P3
P3
C3
–
–
–
C3 P3 P3
17
0.013
S4
P4
P4
C3
–
–
–
C3 P4 P4
17
0.013
S5
C4
X
X
X
X
X
C4
34
0.773
S6
X
X
X
X
O2 P3
O2 P3
O2 O2 P3 P3
36
0.025
S7
O2 C3
O2
–
–
–
–
O2 O2 C3
12
0.005
S8
X
X
O1
X
X
X
O1
35
0.785
S9
X
X
X
X
O1
O1
O1 O1
36
0.462
S10
P4
X
X
X
X
O1
O1 P4
34
0.234
Column heads indicate conditions of analysis. Effects in %D(t) are shown in bold. X, evoked potentials not detected. Bars indicate conditions not analyzed. PFW, family-wise
error for the decision that the subject detected the field. *False-positive result found in the sham-field analysis.
V(t) was also evaluated directly (no unfolding in phase space)
produced onset potentials in all subjects and offset potentials in
by time averaging to detect linear evoked potentials, should they
60% of the subjects (Table 1); the results were similar to those found
occur. The estimation of the a posteriori false-positive rate and the
when the rise- and fall-times were 10 ms (94%, 65%, respectively)
family-wise error for each of the two experiments in each subject
[4].
was identical to the analysis used to evaluate the recurrence time
For several reasons, the observed effects can be taken to have
series. We regarded a potential as nonlinear if it was detected by
been a result of true post-transduction changes in brain electri-
recurrence analysis but not by time averaging.
cal activity triggered by the magnetic stimuli. First, an alternate
Using %R(t), brain potentials were found in 9 subjects in response
explanation that the effects were unrelated to neuronal activity
to field onset, and in 6 subjects in response to field offset (Table 1,
but rather resulted from interactions between the field and the
first data column). In subject S2, for example, potentials occurred
scalp electrodes can be ruled out because we showed in prelim-
at P3 due to field onset, and at O2, C3, and P3 in response to field
inary experiments involving phantoms of the human head that
offset. Detailed results for P3 (Fig. 4) illustrate the appearance of
such interactions began instantaneously and lasted less than 30 ms
evoked potentials when assessed using recurrence variables; when
after stimulus onset or offset. In contrast, the observed potentials
the EEG signals were time averaged, evoked potentials were not
occurred several hundred ms after initiation of the stimulus, which
detected (data not shown). A total of 120 statistical tests involving
is a typical latency for evoked potentials. Second, the family-wise
the %R(t) time series were performed (2 stimuli × 6 derivations × 10
error rate for a decision that the subject detected each of the stimuli
subjects), resulting in 23 evoked potentials (Table 1, first data
was sufficient to rule out the possibility that the effects were due to
column). When a subject exhibited fewer than 3 evoked poten-
chance. Finally, the false-positive signal-detection rate as assessed
tials in response to either the onset or offset of the field, %D(t)
during sham exposure ruled out the possibility that the effect could
was computed and analyzed; onset potentials in S2 and offset
have arisen as a result of the analytical method used to analyze the
potentials in S1 were found that had not been detected with
EEG. For all these reasons, the observed changes in electrical activity
%R(t) (Table 1, second data column). Filtering the EEG to remove
were true MEPs.
8–10 Hz or 9–12 Hz prior to computing %R(t) or %D(t) revealed addi-
It might be argued that the difference between the onset and off-
tional potentials (Table 1, data columns 3–6); all subjects detected
set response rates (100% versus 60%) was partly due to differences
field onset, and 6 subjects detected field offset. The a posteriori
in the rise-times of the stimuli. However comparable response rate
comparison-wise error rate (computed from the sham data) was
differences were observed previously (94%, 65%) when the rise-
19 false-positive tests/439 tests = 0.043; thus PFW < 0.05 for each of
time of both stimuli was 10 ms. Further, onset evoked potentials
the 16 instances of field detection, except S1 onset (PFW = 0.074) and
due to auditory stimuli also occurred more frequently than offset
S9 onset (PFW = 0.057) (Table 1, last column). There was one instance
potentials [2,9,15]. The likelihood is, therefore, that the difference
of false-positive detection (Table 1, S6 onset).
in response rates observed here was not related to the difference in
Our main purpose was to test the hypothesis that the human
the rapidity of the stimuli.
magnetosensory system could respond to rapid stimuli (0.2 ms)
In the catfish, the three-dimensional orientation of the elec-
with efficiency comparable to that for relatively slow stimuli
troreceptor cells and their afferent innervation (4–30 cells synapsed
(10 ms). The underlying idea was that a sensory system capable
with a single neuron) help insure that the effect of the field on the
of responding to a 10-ms magnetic stimulus might be explainable
probability of the channels to be in the open state is not averaged
on the basis of a second-messenger signal system in the electrore-
away across the combined cellular ensemble. If the proposed model
ceptor cell, but that the quicker the field to which the system could
(Fig. 1d) were applicable to human transduction of EMFs, structural
respond, the more likely was the possibility that the field inter-
ordering of the electroreceptor cells and afferent innervation would
acted directly with the ion channel, as in force transduction. The
similarly be expected to insure that the response of the system was
magnetic field, which had a 10-ms rise-time and a 0.2-ms fall-time,
not averaged away. The geomagnetic field might also play a role,

---

A.A. Marino et al. / Neuroscience Letters 452 (2009) 119–123
123
Fig. 4. Evoked potentials detected from P3 in subject S2 using the nonlinear variable %R(t). (a) Magnetic-field and sham-field onset (left and right panels, respectively). (b)
Magnetic-field and sham-field offset (left and right panels, respectively). The curves at the tops of the panels show the average values of the stimulus (E) and control (C)
epochs (N ≥ 50 trials). The p(t) curves are the probability that the difference between the means of the onset and control epochs at time t was due to chance. Bar graphs
indicate the average value of %R over the latency interval for which p(t) < 0.05 (horizontal line); the standard deviations are not resolved at scale shown. The stippled regions
show the expected latency intervals.
possibly tipping ion-channel open-time probabilities one way or
[4] S. Carrubba, C. Frilot, A.L. Chesson Jr., A.A. Marino, Evidence of a nonlinear
the other [1].
human magnetic sense, Neuroscience 144 (2007) 356–367.
[5] S. Carrubba, C. Frilot, A.L. Chesson Jr., A.A. Marino, Nonlinear EEG activa-
Several considerations suggested that the mechanism by which
tion by low-strength low-frequency magnetic fields, Neurosci. Lett. 417 (2007)
the magnetic field triggered the response involved the induced elec-
212–216.
tric field. First, the strength of the induced field (determined from
[6] S. Carrubba, C. Frilot, A.L. Chesson Jr., C.L. Webber Jr., J.P. Zbilut, A.A. Marino, Mag-
netosensory evoked potentials: consistent nonlinear phenomena, Neurosci.
Fig. 1b, using Faraday’s law) was theoretically capable of modify-
Res. 60 (2008) 95–105.
ing the average open time of an ion channel [11], assuming the
[7] D.P. Corey, A.J. Hudspeth, Kinetics of the receptor current in bullfrog saccular
model depicted in Fig. 1d. Second, there is no published explana-
hair cells, J. Neurosci. 3 (1983) 962–976.
tion regarding how the magnetic field used in this study could be
[8] C. Frilot II, S. Carrubba, A.A. Marino, Localization of magnetosensory function
using positron emission tomography, Synapse 63 (2009) 421–428.
detected by a sensory cell, because the small energy of interac-
[9] J. He, Responses in the auditory thalamus of the guinea pig, J. Neurophysiol. 88
tion between the field and tissue (compared with thermal energy)
(2002) 2377–2386.
has thus far defeated all proposed models except those that postu-
[10] G.H. Klem, Artifacts, in: J.S. Ebersole, T.A. Pedley (Eds.), Current Practice of Clini-
cal Electroencephalography, Lippincott Williams & Wilkins, Philadelphia, 2003,
late the presence of ferromagnetic structures (which have not been
pp. 271–287.
observed in the human brain).
[11] O.V. Kolomytkin, S. Dunn, F.X. Hart, C. Frilot, D. Kolomytkin, A.A. Marino, Glyco-
We conclude that the human EMF transduction system is capa-
proteins bound to ion channels mediate detection of electric fields: a proposed
mechanism and supporting evidence, Bioelectromagnetics 28 (2007) 379–385.
ble of detecting signals that change at least as rapidly as about
[12] A.A. Marino, E. Nilsen, C. Frilot, Localization of electroreceptive function in
0.2 ms (on-to-off), possibly indicating that the signal transduction
rabbits, Phys. Behav. 79 (2003) 803–810.
process is directly initiated by a force receptor.
[13] E. Neher, B. Sakmann, J.H. Steinbach, The extracellular patch clamp: a method for
resolving currents through individual open channels in biological membranes,
Pflugers Arch. 375 (1978) 218–228.
References
[14] R.D. Penn, W.A. Hagins, Kinetics of the photocurrent of retinal rods, Biophys. J.
12 (1972) 1073–1094.
[1] C.F. Blackman, S.G. Benane, J.R. Rabinowitz, D.E. House, W.T. Joines, A role for
[15] H. Takahashi, M. Nakao, K. Kaga, Cortical mapping of auditory-evoked offset
the magnetic field in the radiation-induced efflux of calcium ions from brain
responses in rats, Neuroreport 15 (2004) 1565–1569.
tissue in vitro, Bioelectromagnetics 6 (1985) 327–337.
[16] R.G. Walker, A.T. Willingham, C.S. Zuker, A Drosophila mechanosensory trans-
[2] L.E.V. Campen, J.W.H. III, D.W. Grantham, Human offset auditory brainstem
duction channel, Science 287 (2000) 2229–2234.
response: effects of stimulus acoustic ringing and rise-fall time, Hearing Res.
[17] C.L. Webber, Jr., Recurrence Quantification Analysis, vol. 2008, 2007.
103 (1997) 35–46.
[18] J.P. Zbilut, J. Webber, C.L., Recurrence quantification analysis. In: M. Akay (Ed.),
[3] S. Carrubba, C. Frilot, A. Chesson, A. Marino, Detection of nonlinear event-related
Wiley Encyclopedia of Biomedical Engineering, John Wiley & Sons, Hoboken,
potentials, J. Neurosci. Methods 157 (2006) 39–47.
2006, pp. 2979–2986.

---

Document Outline

• Evidence that transduction of electromagnetic field is mediated by a force receptor
  • References

---