400 Electroencephalography and Clinical Neurophysiology, 1981, 52:400--404
Elsevier/North-Holland Scientific Publishers, Ltd.
F R E Q U E N C Y - F O L L O W I N G POTENTIALS IN MAN BY LOCK-IN TECHNIQUE
MANFRED EULER and JURGEN KIESSLING
Fb. 10, Physik-Technologie, Universita't Duisburg, D-4100 Duisburg, and University ENT-Clinic,
D-6300 Giessen (G.F.R.)
(Accepted for publication: June 25, 1981)
The early human auditory responses to low
frequency tones in the speech range are of
considerable interest for objective audiome-
try, because they probably test the integrity
of the low frequency auditory channels.
These responses are phase locked to the stim-
ulus and follow the stimulus wave form (fre-
quency-following response ( F F R ) , Marsh and
Worden 1968). It is possible to detect F F R by
scalp electrodes (Moushegian et a l . 1973).
Several workers have suggested the inferior
colliculus (IC) as a possible source of the
scalp-recorded response, mainly because F F R
latency (6 msec) equals the click stimulation
latency of the IC (Glaser et al. 1976). Further
evidence for the IC origin is based on depth
electrode measurements (Smith et al. 1975)
and on data from subjects with brain stem
lesions (Sohmer and Pratt 1977). Ablation
experiments, however, tend to question the
role of IC in F F R generation (Gardi et al.
1979).
Additionally, the question of h o w far F F R
is related to the apical turn of the cochlea
needs further clarification. Up to now this
problem could only be tackled by high-pass
masking noise (e.g., De Boer et al. 1977). By
using continuous tone stimulation and sensi-
tive frequency selective analysis techniques
the present investigation allows definite con-
clusions on these issues.
Conventional F F R investigations are based
on transient stimulation and c o m p u t e r averag-
ing procedures. Response latencies are some-
what arbitrarily estimated from the F F R on-
set. Instead of time domain averaging fre-
0013-4649/81/0000--0000/$02.50 © 1981 Elsevier/North-Holland Scientific Publishers, Ltd.
quency selective analysis methods appear to
be more appropriate because of the well-
defined response frequency. An off-line
spectrum analysis of the averaged F F R has
considerably improved sensitivity (De Boer et
al. 1977). This procedure does not preserve
time information.
As an alternative approach we use lock-in
amplification to obtain a rapid and sensitive
on-line F F R analysis. The lock-in amplifier
can be regarded as a very narrow tunable filter
designed to extract the amplitude of har-
monic signals from statistical background
noise. The analysis is phase sensitive, which is
an important feature, as the F F R phase con-
tains time delay information.
Method
Fig. 1 shows the experimental design. The
stimulus section consists of a voltage-con-
trolled oscillator (VCO) driving a high impe-
dance headphone (Sennheiser HD 415 X). To
avoid magnetic pick up the phone is shielded
by a /a-metal case and, additionally, the dis-
tance between acoustic stimulus generator
and recording site has been increased by a
plastic tube acting as an acoustic delay line
(70 cm minimum length). With these precau-
tions, tests with an equivalent circuit (10 kg2
wire loop) prove that the induced voltage is at
least 40 dB below the F F R voltage. Further
evidence, that no stimulus artefacts were mea-
sured, could be established by clamping the
plastic tube; thus the sound is prevented from
FREQUENCY-FOLLOWING POTENTIALS BY LOCK-IN TECHNIQUE 401

X Y recorder

clCF~ pLI t i e r
IO0
nV
VoH'oge
dela y 5O
t r~e

beac~
phorqe
I voitage
c or~ t r o t t e d
oscillator

,HJ - r'r'e t a t box

Fig. 1. Experimental arrangement for continuous 50

tone F F R measurements by lock-in technique.

reaching the subject's ear and the F F R van-

ishes. A ramp generator provides the voltage
to sweep the VCO frequency and to drive the
X-axis of the XY-recorder.
We use bipolar recording (Beckman mini
disc electrodes) between forehead and mas-
toid with the opposite mastoid as ground.
After amplification (g = 2 X 104) the lock-in
analysis yields two vector components (A"
cos ~ and A • sin ~) which are recorded simul-
taneously as a function of frequency. From
these two plots F F R amplitude A and phase
angle ~ in relation to the stimultis can be cal-
culated. Usually sweep rates are in the order
of 2 Hz/sec with analyser time constants of
4 sec (equivalent filter bandwidth 0.03 Hz).
So it takes about 4 min to obtain F F R
records in the frequency range from 100 to
600 Hz.
Ten normally hearing adults served as sub-
jects in the study. During the recording proce-
dure that t o o k place in a silent room they
were seated in a comfortable chair. As sub-
jects were cooperative elimination of move-
m e n t artifacts turned out to be unnecessary.
Results
To check the reliability of the m e t h o d and
to identify the measured F F R components
different acoustic delay lines and different
electrode configurations were used. Represen-
C
200 400 600 Hz 800 1000 7200
Fig. 2. F F R records to continuous tone stimulation as
a function of stimulus frequency. The records show
A • sin ~ output of the lock-in amplifier. A: ipsilateral
recording (approximately 60 dB HL), length of delay
tube 70 cm. B: ipsilateral recording (approximately
50 dB HL), length of delay tube 200 cm. C: contra-
lateral recording (approximately 50 dB HL), length of
delay tube 70 cm.
tative results for one person are given in
Fig. 2. Traces A and B show the A • sin ~ F F R
c o m p o n e n t from ipsilateral electrodes (differ-
ential amplification between forehead and
mastoid of the stimulated ear) using two dif-
ferent delay lines. The response extended up
to several kHz and exhibited a rather regular
variation of phase versus frequency with con-
stant spacings between maxima and minima,
depending on delay line length. Additional
white noise did n o t affect the response appre-
ciably.
Contralateral recording showed quite dif-
402
ferent A • sin ~ patterns, with response maxi-
ma below 500 Hz (Fig. 2C). These potentials
could be masked completely b y white noise at
intensities equal to those producing subjective
masking. The most remarkable feature of the
contralateral c o m p o n e n t is its frequency-
dependent phase change, indicated b y the dis-
tance between maxima and minima increasing
with frequency.
By computing the derivative of phase ver-
sus circular frequency ( d ~ / d ~ ) F F R group
delays could be obtained (i.e., the time delay
of a wave package composed of a group of
neighbouring frequencies). The ipsilateral
response phase was obviously a linear func-
tion of frequency. Its group delay, therefore,
was constant and turned o u t to be only 0.5
msec longer than the acoustic time delay of
the respective delay line.
In contralateral recording a dispersion
p h e n o m e n o n occurred, till n o w undiscovered
in F F R measurements. The group delay was
frequency dependent and increased for lower
frequencies. After subtraction of 2.78 msec
acoustic delay (70 cm tube), the delay of the
contralateral c o m p o n e n t amounted to 5.2
msec at 500 Hz and increased to 12 msec at
200 Hz.
Discussion
The results allow unequivocal conclusions
a b o u t the origins of the main scalp recorded
F F R components. In ipsilateral measurements
a cochlear microphonic c o m p o n e n t (CM) pre-
vails and supersedes neural components. The
latter can be isolated b y contralateral record-
ing. Such a view of CM components is gener-
ally accepted in far-field electrocochleography
(Terkildsen et al. 1973).
The far-field CM must originate in the basal
cochlear turn in consequence of its extremely
short latency of 0.5 msec in normal ears. This
conclusion agrees with findings b y conven-
tional recording technique (Sohmer and Pratt
1977). The absence of dispersion in the pres-
ent frequency-specific measurements gives
M. EULER, J. KIESSLING
additional evidence for a basal origin, as
responses from the best frequency region
would exhibit dispersion according to a travel-
ling wave phenomenon.
Obviously, the CM c o m p o n e n t is an indica-
tor of the integrity of the high frequency sen-
sitive basal cochlear turn. We expect the am-
plitude-phase characteristics in ipsilateral
records to shift considerably towards longer
time delays in recruiting ears.
The dispersion p h e n o m e n o n in contralat-
eral F F R c o m p o n e n t s is a definite p r o o f of its
apical specificity and demonstrates the
cochlear frequency dispersion mechanism.
The total group delay of this c o m p o n e n t con-
sists of the acoustic delay, the travelling wave
delay in the cochlea and neural delays. A
more sophisticated view may additionally
refer to the group delay due to the 'second
filter,' but, for simplicity, we shall include
this term in the total travelling wave delay.
Among these different delay mechanisms only
the travelling wave delay is frequency depen-
dent.
After subtracting 1 msec corresponding to
the supposed neural delay from the above cal-
culated group delays at 200 and 500 Hz, the
resulting delay times agree well with travelling
wave delays to the best frequency basilar
membrane region obtained in single nerve
studies (Anderson et al. 1971) and in derived
narrow band action potentials (Eggermont
1976). The data of Anderson et al. on squirrel
monkeys range from 3 msec delay at 500 Hz
to 6--7 msec at 200 Hz. The human-derived
action potential data come quite close to our
results (5 msec delay at 500 Hz, approxi-
mately 12 msec at 200 Hz as can be found by
extrapolating Fig. 5 of Eggermont 1976). The
present dispersion p h e n o m e n o n is consistent
with the assumption that, at moderate inten-
sities, F F R originate in neurones tuned to
respective frequency. As a consequence, the
6 msec argument for brain stem origin appears
questionable, as at 500 Hz the travelling wave
delay alone, already amounts to 4 msec at
least, and it increases considerably for lower
frequencies. The obvious contradiction to the
FREQUENCY-FOLLOWING POTENTIALS BY LOCK-IN TECHNIQUE
results of depth electrode measurements
(Smith et al. 1975) may arise from the differ-
ent latency determination procedures and
stimulus conditions. One possible explanation
is that transient stimulation involves neural
structures completely different from those
activated in the steady state, which are
probed b y the present method. The continu-
ous tone F F R is likely to reflect neural activ-
ity from a more peripheral origin, probably
related to N1 or N2 activity known from elec-
trocochleography. The study demonstrates
the feasibility of continuous tone lock-in F F R
analysis and shows the inherent advantages
over conventional techniques, owing to the
continuous frequency-specific information.
Apical audiometry (cf., Davis 1976) appears
to be practicable with the present detection
method, which may become a valuable tool
of electric response audiometry.
Summary
Frequency-following responses ( F F R ) from
continuous tone stimulation were investigated
b y lock-in analysis. The technique is superior
to conventional tone-burst stimulation,
because it allows continuous registration of
the response parameters with sweeping fre-
quency. The scalp recorded F F R consists of
at least t w o components separable from each
other by their phase/frequency relationship.
The microphonic c o m p o n e n t that dominates
in ipsilateral records shows latencies of a b o u t
0.5 msec with negligible dispersion and must
therefore originate in the basal cochlear turn.
The neural c o m p o n e n t exhibits considerable
dispersion, reflecting the travelling wave delay
up to the region of best frequency. These long
cochlear time delays are consistent with an
apical F F R origin, and brain stem sources of
the continuous tone response appear to be
questionable. The present F F R analysis tech-
nique opens up a very direct and comprehen-
sive way to assess the integrity of the entire
inner ear for high frequency basal and low fre-
quency apical cochlear regions as well.
403
R~sumd
Potentiels accord~s d la fr~quence ~tudi~s
chez l'homme par une m~thode 'lock-in'
La rdponse accordde ~ la frdquence ( F F R )
pour une stimulation sonore continue est
analys~e par 'lock-in'. Cette technique se
montre supdrieure ~ la stimulation conven-
tionnelle par bouffdes tonales, du fait que les
caract~ristiques de la F F R sont enregistr~es de
fa~on continue avec la fr~quence. La F F R
mesur~e par les dlectrodes de scalp est com-
posde de deux ~ldments principaux au moins
d'origine microphonique et neuronale, que
l'on peut distinguer par leur relation entre
phase et frdquence. La composante micro-
phonique, prddominante dans les enregistre-
ments ipsilatdraux, est produite dans la spirale
basale, vu sa courte latence (0,5 msec) et son
absence de dispersion. La grande dispersion de
la composante neuronale refl~te le d~lai de
progression le long de la membrane basilaire
jusqu'~ la rdgion de frdquence optimale. Ces
longs ddlais cochl~aires sont en accord avec
une origine apicale de la F F R , et jettent un
doute sur l'hypoth~se de leur origine dans le
tronc c~rdbral.
La prdsente mdthode d'analyse de la F F R
permet de vdrifier de faqon directe et com-
plete l'intdgrit~ de l'oreille interne tout
enti~re, aussi bien pour les rdgions cochldaires
basales (des hautes fr~quences) qu'apicales
(des basses frdquences).
References
Anderson, D.J., Rose, J.E., Hind, J.E. and Brugge,
J.F. Temporal position of discharge in single nerve
fibers within the cycle of a sine-wave stimulus: fre-
quency and intensity effects. J. acoust. Soc.
Amer., 1971, 49: 1131--1139.
Davis, H. Brain stem and other responses in electric
response audiometry. Ann. Otol. (St. Louis),
1976, 85: 3--14.
De Boer, E., Machiels, M.B. and Kruidenier, C. Low-
level frequency-following response. Audiology,
1977, 16: 229--240.
404
Eggermont, J.J. Analysis of compound action poten-
tial responses to tone bursts in the human and
guinea pig cochlea. J. acoust. Soc. Amer., 1976,
60: 1132--1139.
Gardi, J., Merzenich, M. and McKean, C. Origins of
the scalp-recorded frequency-following response in
the cat. Audiology, 1979, 18: 353--381.
Glaser, E.M., Suter, C.M., Dasheiff, R. and Goldberg,
A. The human frequency following response: its
behavior during continuous tone and tone burst
stimulation. Electroenceph. clin. Neurophysiol.,
1976, 40: 25--32.
Marsh, J.T. and Worden, F.G. Sound evoked fre-
quency-following responses in the central auditory
pathway. Laryngoscope (St. Louis), 1968, 78:
1149--1163.
M. EULER, J. KIESSLING
Moushegian, G., Rupert, A.L. and Stillman, R.D.
Scalp-recorded early responses in man to frequen-
cies in the speech range. Electroenceph. clin.
Neurophysiol., 1973, 35: 665--667.
Smith, J.C., Marsh, J.T. and Brown, W.S. Far-field
recorded frequency-following responses: evidence
for the locus of brain stem sources. Electroenceph.
clin. Neurophysiol., 1975, 39: 465--472.
Sohmer, H. and Pratt, H. Identification and separa-
tion of acoustic frequency following responses
( F F R s ) in man. Electroenceph. clin. Neurophys-
iol., 1977, 42: 493--500.
Terkildsen, K., Osterhammel, P. and Huis in 't Veld,
F. Electrocochleography with a far field tech-
nique. Scand. Audiol., 1973, 2" 141--148.