Exp Brain Res (2001) 139:454–464
DOI 10.1007/s002210100775
R E S E A R C H A RT I C L E
James R. Lackner · Ely Rabin · Paul DiZio
Stabilization of posture by precision touch
of the index finger with rigid and flexible filaments
Received: 18 July 2000 / Accepted: 18 April 2001 / Published online: 29 June 2001
© Springer-Verlag 2001
Abstract Light touch of the index finger with a stationary surface at non-mechanically supportive force levels
(<100 g) greatly attenuates the body sway of standing
subjects. In three experiments, we evaluated the properties of finger contact and of the contacted object necessary to produce postural stabilization in subjects standing
heel-to-toe with eyes closed, as well as how accurately
hand position can be controlled. Experiment 1 involved
finger contact with flexible filaments of different bending strengths, a flat surface, and an imagined spatial position. Contact with the flat surface was most effective in
attenuating sway; the flexible filaments were much less
effective but still significantly better than imagined contact. Experiment 2 compared the effectiveness of finger
contact with a flexible filament, a rigid filament of the
same diameter, a flat surface, and an imagined spatial
position. The rigid filament and flat surface conditions
were equally effective in attenuating body sway and
were greatly superior to contact with the flexible filament, which was superior to imagined contact. Experiment 3 included five conditions: arms by sides; finger
“contact” with an imagined spatial position; finger contact with a flat surface; finger contact with a flexible filament attempting to maintain it bent; and contact with the
flexible filament attempting not to bend it. The arms by
sides and finger “contact” with an imagined position
conditions did not differ significantly; all three conditions involving actual finger contact showed significantly less center of pressure and hand sway, but contact
with the flat surface was most effective in attenuating
both postural and hand displacement. In all three experiments, the level of force applied in fingertip contact conditions was far below that necessary to provide mechanical stabilization. Our findings indicate that: (1) stimulaJ.R. Lackner (✉) · E. Rabin · P. DiZio
Ashton Graybiel Spatial Orientation Laboratory
and Volen Center for Complex Systems,
MS033, Brandeis University, Waltham,
MA 02454-9110, USA
e-mail: lackner@brandeis.edu
Tel.: +1-781-7362033, Fax: +1-781-7362031
tion of a small number of receptors in the fingertip is adequate to allow stabilization of sway, (2) fingertip force
levels as low as 5–10 g provide some stabilization, (3)
contact with a stationary spatial referent is most effective, and (4) independent control of arm and torso occurs
when finger contact is allowed.
Keywords Posture · Stabilization · Somatosensation ·
Balance
Introduction
Contact of the index finger with a stationary surface can
greatly attenuate postural sway even when the level of
force applied is far below that necessary to provide mechanical support (Clapp and Wing 1999; Holden et al.
1987, 1994; Jeka and Lackner 1994, 1995; Reginella et
al. 1999; Riley et al. 1997). The fingertip signal, decoded
in relation to information about arm to torso and overall
body configuration, can specify the direction and velocity of body motion and guide implementation of compensatory innervations to attenuate body sway (cf. Jeka and
Lackner 1995; Rabin et al. 1999) well before vestibular
thresholds are exceeded (Fitzpatrick and McCloskey
1994).
Subjects in our experiments (e.g., Holden et al. 1994;
Jeka and Lackner 1994, 1995; Lackner et al. 1999) when
allowed contact with a rigid surface spontaneously adopt
force levels of approximately 40 g at their fingertips.
This is a very small value; for example, in studies of
Braille letter recognition, dot matrices are typically
scanned on the fingertip at a force of 60 g (Johnson and
Hsiao 1992). The usefulness of the sensory cues at the
fingertip for stabilizing balance depends on the ability to
monitor arm configuration in relation to the torso and the
contact surface (Rabin et al. 1999). Subjects adjust their
postural musculature so as to minimize sensory changes
at the fingertip and by so doing stabilize their body (Jeka
and Lackner 1995). Fingertip contact with rough and
slippery surfaces attenuates body sway equally although
455
in one case the finger remains in the same place and in
the other slides relative to the surface (Jeka and Lackner
1995). In both situations, the changes in horizontal and
vertical forces at the fingertip lead changes in body
sway, as reflected in center of pressure measures, by approximately 250–300 ms; the lateral force at the fingertip averages ≈0±7 g and the vertical force 40±7 g. EMG
activity in the leg muscles countering body sway lags the
fingertip signals by about 125–150 ms and leads the
change in body sway by about the same amount. This
overall pattern suggests that the force changes at the fingertip can be used to detect the direction and extent of
sway and to elicit compensations.
We report here studies concerning what types of fingertip cues are adequate to provide stabilization of posture and hand position. In particular, we examined: (1)
whether contact of a large surface area of the fingerpad
is necessary or whether a point contact indenting the fingertip suffices, (2) whether the spatial referent needs to
be rigid, and (3) what magnitude of fingertip force is stabilizing. Thus, we were concerned with the “integration
area” necessary to derive spatial directional information
with regard to the finger surface, the spatial range within
which the fingertip could displace and still attenuate
sway, and the smallest stabilizing force.
We carried out three experiments to address these
questions. In the first, we determined whether finger
contact with flexible filaments of different bending
strengths would stabilize posture relative to a control
condition in which subjects attempted to hold their finger in the same spatial position in the absence of contact.
The second experiment contrasted contact with a rigid
filament, a flexible filament, a flat surface, and attempting to hold the finger in the same spatial position without
physical contact. The final experiment involved finger
contact with a flat surface and with a flexible filament at
two attempted force levels: one adequate to buckle the
filament and the other not adequate. This experiment
also involved two control conditions: arms by the sides,
and attempting to hold the finger in a fixed spatial position without physical contact. The flexible filaments employed were nylon filaments, commonly referred to as
von Frey hairs, that are used to test sensory thresholds
for touch (e.g., Johansson and Vallbo 1980). Each hair
buckles at a vertical force level determined by its diameter. In our experiments, we mounted the filaments vertically on a horizontal bar located at the subject’s side so
that the subject could touch the top of a filament with his
or her index finger. This allowed us to limit the maximum vertical force the subject could apply with the fingertip and to have a non-rigid spatial reference that could
move within a relatively small area while finger contact
was maintained.
Fig. 1 Schematic illustration of the test situation and recording
equipment
Experiment 1:
fingertip contact with flexible filaments
Materials and methods
Subjects
Ten healthy young adults participated voluntarily for pay. They
were without sensory or motor impairments that could have influenced their balance. All signed informed consent forms approved
by the Brandeis University Committee for the Protection of Human Subjects.
Apparatus and instrumentation
Our test situation is illustrated in Fig. 1. The subject is shown with
right index finger touching a filament attached to a bar equipped
with strain gauges to measure the vertical (TV) and lateral (TL)
forces resulting from the contact. The strain gauges are dual element and temperature compensated [Kulite Semiconductor, Type
M (12) OGP-350–500]. Their signals were amplified and calibrated using known masses applied to the bar. Resolution accuracy
was approximately 1 g. A Kistler force platform (model 9261A)
and associated electronics were used to compute the coordinates
of foot pressure. The subject was surrounded in front and on the
sides by a padded safety railing. Its front width was 72 cm and its
height was 71 cm. The sides were 100 cm long and 81 cm high.
Conditions and procedure
The subject in stocking feet stood heel-to-toe along the central anterior-posterior axis of the force platform without the feet touching. The touch device was adjusted in height and lateral position
so the subject could touch the end of a filament projecting vertically from it using a comfortable, outstretched arm configuration.
Four different conditions were run. In a no-touch control condition (N), the subject held his or her finger just above the touch
456
device and attempted to maintain it in an imagined fixed spatial
position. The other conditions involved nylon filaments1 of different diameters (0.28 mm, 0.53 mm, 0.68 mm). Each filament was
3.8 cm long. We experimentally determined to the nearest 5 g the
vertical compressional force necessary to bend the hairs and found
them to be 10, 35, and 85 g according to increasing filament diameter. These values were obtained by applying the von Frey hairs
normal to a sensitive force plate until they bent. The buckling
force (FB) can also be computed using Euler’s buckling formula:
2
FB = nπ 2EI
L
4
where L=length, E=Young’s modulus, I = πr , and r=radius.
4
The factor n is determined by how the ends of the object are restrained. In our situation, with one end rigidly fixed, n=1/4. EI is
2
the flexural stiffness. Consequently, FB = π EI
.
4 L2
By contrast, if shear rather than compression force is applied
to the free end of a von Frey hair, it will deflect according to the
bending moment M.
M = EI = FT L
r
where FT is the tangential force. Rearranging:
FT EI
Lr
The distance d that the end of the filament will deflect is given by:
d=
FT L3
3EI
or
FT = 3dEI
L3
The ratio of the bending force to the compressive buckling force is
thus:
2
FT
= 3dEI
⋅ 4 L = 12d
FB
L3 π 2 EI π 2 L
To appreciate the significance of this relationship, consider a
3.8-cm-long von Frey filament, such as used in the present experiments, that is bent 0.5 cm by a tangential force, FT. That force will
be:
FT = 122 ⋅ 0.5cm ⋅ FB = 0.16 FB
π ⋅ 3.8cm
or about 1/6th of the maximum buckling force.
This relationship holds irrespective of fiber diameter. The important point is that the maximum force that can be generated by finger contact with the filament is the buckling force, but this depends on the fingertip only moving straight downward on the filament and not laterally. In our test situation, lateral movement will
always be occurring so the average force generated at the fingertip will necessarily be less than the ideal buckling force of the fibers and sometimes just the tangential shear force.
We will refer to the filaments as 10, 35, and 85 g according to
their buckling force values. The subject’s goals in the filament trials were: (1) to begin the trial by touching the end of the filament
with his or her index finger until it bent, (2) to maintain finger and
body as stationary as possible, and (3) not to let the filament tip
slip in relation to the fingertip. In the control condition, the subject
attempted to imagine a spatial position just above the touch device
and to keep his or her finger in “contact” with it.
At the beginning of a trial, a subject got into the test posture
and adjusted hand and finger as appropriate for the trial (in filament trials, bending the end of the filament), then closed his or her
eyes and, when ready, said “go,” and the experimenter initiated
data collection. All trials of all conditions were conducted with the
subject’s eyes closed so that visual feedback about performance
1
The filaments were obtained from Alimed Inc., Dedham, MA.
was never available. A single practice trial was given for each
condition prior to the start of the experiment proper.
The four conditions were run in four blocks of four trials with
each condition represented once, randomly, within each block. Trials were 25 s in duration. After each trial, the subject stepped off
the force platform and sat quietly for about 30 s before stepping
back on for the next trial. The entire experiment including practice
trials lasted approximately 1 h.
Analysis
Our earlier experiments (Holden et al. 1994; Jeka and Lackner
1994; Rabin et al. 1999) had shown that in the heel-to-toe stance,
anterior-posterior sway is minimal and not correlated with cues at
the fingertip; consequently, we report here only medial-lateral
sway of center of foot pressure (CPX).
For each trial, CPX and TL were sampled at 60 Hz and stored
digitally. Each time series of CPX was reduced to a mean sway
amplitude (MSA) as follows:
N
MSA = 1 ∑ xi − x
N i =1
where
N
x = 1 ∑ xi
N i =1
N=number of samples and x=CPX.
The average absolute deviations from zero of the vertical (TV)
and horizontal (TL) touch bar reaction forces were calculated.
Fourier analysis (1024 point FFT, 17.06 s of data at 60 Hz)
was performed on the CPX and TV signals and the centroid of the
frequency band was calculated according to:
N /2
CFB =
∑ x j fj
j =1
N /2
∑ xj
j =1
where fj is the center of the jth frequency bin of the power spectrum, xj is the power in that bin, and N is the number of bins.
Cross correlations were calculated between center of pressure
and fingertip contact forces at 16.07 ms/step over ±1500 ms to
identify where the maximum correlations occurred.
Data analysis
The statistical analysis included a 4×4 ANOVA to test for effects
of condition (N, 10 g, 35 g, 85 g) and trial order (1–4) for CPx
MSA. Condition was significant in the ANOVA (P<0.005) and trial order was not; consequently, we averaged across trials within a
condition for individual subjects and performed planned pairwise
comparisons to determine how touch conditions differed. The use
of multiple tests inflates the likelihood of a type I error; therefore,
we used a Bonferroni adjustment and only considered an effect
significant if its adjusted value was P<0.01.
Results
Center of pressure sway
Contact of the index finger with the von Frey filaments
reduced CPx MSA relative to the control condition (N)
not involving physical contact. Figure 2A presents for
each condition CPx MSA, averaged across subjects. CPx
MSA was significantly less in all filament conditions
than in the control (N) condition (P<0.01, all comparisons), but was not different across the filament conditions (P>0.05, all comparisons). The centroid of frequen-
457
Fig. 2 A Mean sway amplitude
of lateral center of pressure
(CPX) for the filament and control conditions of experiment 1.
Error bars represent standard
deviations (N=10). B Mean lateral (TL) and vertical (TV) forces applied by the fingertip in
the conditions of experiment 1
(N=10). C Cross-correlations
and time lags between lateral
center of pressure (CPX) and
lateral fingertip forces (TL) in
the conditions of experiment 1
(N=10)
cy band of CPx did not differ significantly across conditions, ranging between 0.52 and 0.53 Hz.
Fingertip contact forces
Figure 2B shows the mean TL and TV forces generated in
the filament conditions. Both TL and TV forces increased
with the increasing filament diameters. TL was always
below 0.05 N. TV was 0.03, 0.11, and 0.23 N for the filaments according to increasing diameters. One N equals
102 g.
Cross-correlations between CPx sway
and fingertip contact forces
The cross-correlations and temporal relationships between CPx sway and TL are shown in Fig. 2C. The largest correlations are negative and range from about –0.4
to –0.6 for CPx and TL. The time lags at which these correlations occur are in the 50-ms range. The maximum
correlations between CPx and Tv never exceeded 0.3 and
are not shown.
Subjective reports
All of the subjects found balance to be most difficult in
the control condition (N) not involving touch of the finger. They reported that touching the flexible filaments
made balancing easier but because the end of a filament
could move with the fingertip it could only serve as a regional spatial referent.
Summary: experiment 1
The experimental findings indicate a significant influence of the von Frey filaments in attenuating body sway.
In none of the test situations did the subjects actually apply enough vertical force to buckle the filaments. For example, subjects applied on average about 3 g of vertical
force to the thinnest filament (which has a 10 g buckling
force level) and about 2–3 g of lateral force, yet this still
attenuated their sway. When 3 g of lateral force is substituted into the equation presented above showing the relationship between lateral force, deflection amplitude, and
buckling force, it can be seen that the deflection amplitude must be about 0.95 cm. This means that the finger-
458
tip moved within a region of approximately 2 cm diameter. A lateral finger excursion of 2 cm during rigid body
sway would be associated with 0.6–0.7 cm of lateral CP
displacement. This corresponds closely with the actual
CPx lateral sway amplitude in the flexible filament conditions (see Fig. 2A). The filaments thus provided a spatial region within which sensory cues were physically
possible and CPx MSA was maintained within this range.
The correlations between CPX and TL were negative for
the filaments, because when the fingertip displaces a filament’s tip rightward the filament’s base exerts a leftward force on the touch bar, and vice versa for leftward
bending of a filament.
Experiment 2:
flexible versus rigid filament versus flat surface
Results
Center of pressure and head sway
CPx and HX MSAs were greatest in the no-contact condition, being about 0.9 and 2.2 cm respectively. The
flexible filament significantly reduced CPX and HX
MSAs relative to no-contact condition values by about
15% and 30%, respectively (P<0.01, both comparisons).
The rigid filament and flat plate conditions did not differ and had 70% lower CPX MSA and 90% lower HX
MSA than the control condition. Both had significantly
lower CPX and HX MSA than the flexible filament condition (P<0.0025, all comparisons). Figure 3 presents
typical trials from a single subject for the four conditions. Figure 4a summarizes the findings across subjects
for all conditions.
Materials and methods
Subjects
Nine undergraduates participated for pay after providing informed
consent. They were without sensory, motor, or skeletal anomalies
that could have affected their performance.
Apparatus and instrumentation
The same apparatus was used as illustrated in Fig. 1. The subject
also wore a headband with a light-emitting diode (LED) attached
at the midline of the head. The position of the LED was tracked by
an ISCAN video system to obtain medial/lateral (Hx) coordinates
of head sway.
Conditions and procedure
Subjects were tested in stocking feet in the heel-to-toe stance. The
touch bar was positioned lateral to the subject’s right side and adjusted in height to allow the subject to assume a comfortable arm
position.
Four conditions were run: (1) a no-contact condition in which
the subject maintained his or her finger above but not touching the
touch bar, attempting to maintain “contact” with an imagined spatial position, (2) a flexible filament condition involving fingertip
contact with the 85-g von Frey hair used in experiment 1 (the subject started the trial with the filament bent), (3) a rigid filament
condition involving finger contact with a rigid, steel filament,
1 mm in diameter, and (4) a flat surface contact condition in which
the subject touched a horizontal metal plate mounted on the touch
device. The rigid filament defected less than 0.5 mm (our measurement resolution) when a lateral force of 100 g was applied to
its free end. Conditions were run in four blocks of four trials with
each condition represented once, randomly, in each block. Trials
were conducted as in experiment 1 with the subject’s eyes closed.
Fingertip contact forces
TL forces were approximately 0.05, 0.15, and 0.22 N for
the flexible filament, rigid filament, and plate conditions,
respectively. TV forces were 0.2, 0.4, and 0.6 N, for the
same conditions. For both TL and TV, the plate condition
involved significantly greater force (P<0.001, both comparisons) than the rigid filament condition, which in
turn involved significantly greater force than the flexible
filament condition (P<0.005, both comparisons). See
Fig. 4B.
Cross-correlations between center of pressure
and head sway
CPX-HX correlations were about 0.75 in the no-contact
and flexible filament conditions. The rigid filament and
plate conditions had correlations on the order of
0.55–0.6, both being significantly less than the no-contact and flexible filament conditions (P<0.005, all comparisons). See Fig. 4C.
Cross-correlations between center of pressure
and lateral fingertip forces
CPX and TL correlations were virtually identical in the
rigid filament and plate conditions at about 0.65, and
were about –0.5 in the flexible filament condition (see
Fig. 4D).
Data analysis
A 4×4 ANOVA showed effects of condition (P<0.001) but not trial order on CPX and HX MSAs; therefore, we averaged trials of
each condition across subjects and performed pairwise comparisons with Bonferroni adjustments.
Time lags associated with correlations of center
of pressure and lateral fingertip forces
The CPX versus TL time lag was virtually zero for the
flexible filament condition; by contrast, the rigid filament and plate conditions had comparable time lags
about 250 ms of CPX to TL (Fig. 4D).
459
Fig. 3 Typical individual trials
of one subject from the four
conditions of experiment 2
showing the relationship between lateral center of pressure
(CPX) and the lateral (TL) and
vertical (TV) touch forces at the
fingertip for conditions involving finger contact
Summary: experiment 2
Fingertip contact with the rigid filament attenuated CPx
and head MSAs as effectively as contact with the flat
plate. The horizontal and vertical forces at the fingertip
were significantly less during contact with the rigid filament than with the plate. Both of these contact conditions were superior to the flexible filament condition in
attenuating sway, which was itself superior to the imagined-contact condition. Thus, only a small region of the
fingerpad is adequate to fully stabilize posture as long as
contact is made with a spatially fixed object.
Experiment 3:
buckled versus unbuckled flexible filaments
Materials and methods
Subjects
Eleven undergraduate students participated for pay after giving informed consent. They were without sensory or motor impairments
460
Fig. 4 A Mean sway amplitudes of lateral center of pressure (CPX) and lateral head position (HeadX) in the different
conditions of experiment 2.
Error bars indicate standard
deviations (N=9). B Mean lateral (TL) and vertical forces
(TV) applied by the fingertip in
the conditions of experiment 2
(N=9). C Maximum correlations and associated time lags
between lateral center of pressure (CPX) and lateral head position (HeadX) in experiment 2
(N=9). D Maximum correlations and associated time lags
of lateral center of pressure
(CPX) and lateral forces (TL)
applied by the fingertip for experiment 2 (N=9)
that could have influenced their performance in the experimental
conditions.
Apparatus and instrumentation
The basic test situation is illustrated in Fig. 1. An LED was attached to the nail of the subject’s right index finger. An ISCAN
video camera tracked the motion of the LED, allowing us to compute medial-lateral coordinates HDX of hand location.
Conditions and procedure
Subjects were tested in the heel-to-toe stance, in stocking feet. The
touch device was positioned to the subject’s right side and adjusted to a comfortable height. The five conditions included: (1) arms
by sides, (2) index finger above the touch bar trying to maintain
“contact” with an imagined spatial position, (3) index finger
touching and attempting to maintain the 85-g filament bent, (4) index finger touching the 85-g filament but attempting not to bend
it, and (5) finger touching a flat plate mounted on the touch bar.
The subject’s eyes were always closed during the experimental trials. The conditions were run in four blocks of five trials with each
condition represented once in each block, randomly.
Data analysis
An ANOVA showed that condition but not trial order affected CPX
and lateral hand (HDX) MSAs. The results of multiple planned
pairwise comparisons with Bonferroni corrections are reported below.
461
Fig. 5 A Mean sway amplitudes of lateral center of
pressure (CPX) and of hand
(HandX), experiment 3. Error
bars indicate standard deviations (N=11). B Mean lateral
(TL) and vertical (TV) forces applied by the fingertip, experiment 3 (N=11). C Maximum
correlations and associated
time lags of lateral center of
pressure (CPX) and of the hand
(HandX) with lateral forces (TL)
applied by the finger, experiment 3 (N=11). D Maximum
correlations and associated
time lags of lateral center of
pressure (CPX) and lateral hand
position (HandX) for the conditions of experiment 3 (N=11)
Results
Hand mean sway amplitude
Center of pressure mean sway amplitude
MSA of the hand was greatest in the condition involving
hand “contact” with an imagined spatial position. Both
the straight and bent filament conditions led to significantly lower HDX MSA (P<0.001 both comparisons);
HDX MSA was also significantly less in the bent than in
the straight filament condition (P<0.005). Hand MSA
was less than 1 mm in the condition involving finger
contact with the flat plate. See Fig. 5B.
CPX MSA was greatest in the arms by sides condition
and the imagined “contact” condition not involving finger contact with an object; these conditions did not differ
significantly (P>0.025). The bent and “straight” filament
conditions differed significantly from the two no-contact
conditions (P<0.025, all comparisons) having about 10%
lower MSA, but did not differ from each other. Contact
with the flat plate led to the lowest MSA, only about
one-third of that in the two no-contact conditions. It differed significantly from all other conditions (P<0.001,
all comparisons). See Fig. 5A.
Fingertip contact forces
These forces were similar and quite small for the bent
and straight filament conditions, with TL about 0.05 N
462
and TV about 0.06 N, or about 5 and 6 g. TL was about
0.13 N (13 g) and TV about 0.14 N (14 g) in the flat plate
condition. See Fig. 5C.
Cross-correlations of center of pressure with lateral
fingertip forces and associated time lags
CPX-TL correlations were about –0.45 in both the bent
and the straight filament conditions, and 0.55 in the plate
condition. The time lags associated with these correlations were about 30 ms, –40 ms, and 300 ms, respectively. See Fig. 5D.
Cross-correlations of center of pressure
and hand position and associated time lags
CPX-HDX correlations ranged between 0.6 and 0.75 for
the hand above the touch bar condition and the bent and
straight filament conditions, and was lowest (0.4) for the
touch bar condition. The CPX-HDX time lag was zero for
the hand above touch bar condition and about 150 ms for
the other three conditions.
Summary: experiment 3
The arms by sides and imagined contact conditions did
not differ, indicating that mental imagery does not enhance postural control in our test situation. The buckled
and straight filament conditions stabilized posture about
equally, but hand position was more stable in the bent
filament condition. Both conditions, although not involving contact with a rigid referent, led to greater stability of hand position than imagined contact, indicating
that 5–10 g of contact force on the fingertip enables
subjects to control the spatial position of their hand
more accurately than when contact is absent. In the plate
condition involving a rigid reference object, TL and TV
were only about 14 g but hand position varied less than
1 mm.
Discussion
We identified features of finger contact and of the contacted object that are important for enabling postural stabilization to take place. In all three experiments, the contact force levels at the fingertip were always too low to
provide significant mechanical stabilization, never exceeding 65 g, and often averaging 10 g or less. Sixty-five
grams of force at the fingertip can attenuate body sway
at most by 1.5% (see Holden et al. 1994; Rabin et al.
1999 for experimental assessment and physical modeling
of the sway attenuation produced by different force levels at the fingertip). In the rigid filament and flat plate
conditions of experiment 2, the only conditions where as
much as 65 g of force was exerted at the fingertip, the reduction in sway amplitude was almost 70%.
The three flexible filament conditions of experiment 1
all showed significant attenuation of sway relative to the
imagined contact condition; moreover, the attenuation
was comparable across the three conditions. There was
progressively less finger displacement in experiment 1
according to increasing fiber stiffness. Because the average lateral force levels were similar, this suggests that a
particular force (F) threshold at the fingertip had to be
achieved for stabilization to occur where F=kx, with k
being the spring constant of a filament for lateral deflection and x being distance.2 The average finger forces in
the thinnest filament condition were about 2 g vertical
and 1 g lateral. This means that forces at the fingertip
which are not greatly above perceptual detection thresholds (cf. Vallbo and Johansson 1984) are adequate to allow some postural stabilization to be elicited.
Rigidity or spatial stability of the contacted object
was important for eliciting maximum postural stabilization. The rigid filament condition of experiment 2 attenuated sway as effectively as the flat surface contact condition and more effectively than a flexible filament of
similar diameter. The diameters of the flexible filaments
used in the present experiments were 1 mm or less; the
rigid filament was 1 mm in diameter and contact with it
was equivalent for sway reduction to touching a flat surface. This means that “point contact” with the fingertip
is as effective as contact with a broad area in suppressing
sway so long as the contact point is spatially stable. Contact with a physical object was necessary in order for
sway to be attenuated. In experiment 3, finger “contact”
with an imagined spatial position had no stabilizing effect on sway whatsoever relative to the hands by the
sides control condition. This result contrasts with other
types of motor control tasks in which mental and motor
imagery has been shown to enhance performance (cf.
Jeannerod 1994). A “range effect” is also apparent in our
experiments. Experiment 1 involved either no physical
contact of the finger or contact with flexible filaments.
The least magnitudes of applied fingertip forces were
generated in this experiment even though more force
could have been used, e.g., for the “medium” and “thick”
filaments. In experiment 2 contact with the plate, the rigid filament, and the thick filament were involved. Considerably more force was applied to the thick filament in
this situation than in experiment 1, where subjects never
made finger contact with a rigid object.
The fingertip has a rich sensory innervation with four
types of receptors that are responsive to deformation of
the finger surface (Torebjörk et al. 1984; Vallbo and
Johansson 1984). They differ in their receptive field and
adaptation properties. Fast adapting (FA) receptors show
transient responses to stimulation onset and offset but
not a sustained response; by contrast, slow adapting (SA)
receptors have a maintained response. FAI and SAI receptors have relatively small, sharp receptive fields
(≈8 mm in diameter) whereas FAII and SAII receptors
have larger and less sharply defined receptive fields. The
2
We thank an anonymous reviewer for calling this relationship to
our attention.
463
respective cell types are Meissner, Merkel, Pacinian, and
Ruffini.
FAI and SAI receptors are most densely represented
in the fingertip, about 130/cm2 and 70/cm2 respectively;
FAII and SAII are more sparsely present, about 20/cm2
and 10/cm2 (Johansson and Vallbo 1979, 1980). Psychophysical thresholds for detection of contact depend on
the FAI units; even a single spike is perceptually detectable. It is likely that at least three of the receptor types
are activated in the touch conditions of the present experiments. The FAIIs may not contribute much toward
postural stabilization because they are relatively sparse,
deep in the fingerpad, and their receptive fields are large
and may not allow resolution of directional change at the
fingertip. They may not even be activated in the flexible
filament conditions in which less than 5 g of force was
applied at the fingertip (Johansson et al. 1980). The FAI
units would be activated and they are also very sensitive
to slip. The SAI units are known to be especially sensitive to skin indentation (Srinivasan et al. 1990). SAII
units also respond to indentation, and to skin stretching,
and have directional responses to skin displacement.
Their directional responses would be particularly useful
in signaling direction of finger shear in relation to the
contacted object.
One consequence of the dense innervation of the fingertip by SAI and FAI units is that there is considerable
overlap of their receptive fields, on the order of 15 SAIs
and 20 FAIs. SAII units have a less dense representation
in the fingertip and their receptive fields are directionally
polarized. The overall implication of these distributions
for the present experiments is that with “punctate” stimulation at force levels less than 5 g, a relatively small
number (10–20) of somatosensory receptors in the fingertip would have been activated and this was adequate
to lead to significant postural stabilization.
The precision touch achieved by our subjects depended on precise control of muscles of the arm and
hand (analogous to the precision grip studies by Johansson and his colleagues, reviewed by Johansson
1991 and Johansson and Cole 1994). In experiment 3,
the average absolute horizontal and vertical forces on
the fingertip were always less than 0.14 N or about
14 g. A typical subject’s arm weighs about 3.6 kg
(Bernstein 1967). When the subject is standing with
arm outstretched and index finger contacting the touch
bar surface at 15 g of force, only about 1/250th of the
weight of the arm is “supported” by the contact. The
rest has to be controlled by the shoulder musculature
supporting the long lever of the arm against the action
of gravity. The actual forces generated have to be huge
because the arm is controlled as a type 3 lever with
poor mechanical advantage, e.g., the attachments of the
shoulder muscles to the humerus bone are only a few
centimeters from the center of rotation of the shoulder
joint (Norkin and Levangie 1990). However, subjects
were consistently able to maintain as little as 10–20 g
of force on the touch bar surface plate in flat plate contact conditions.
This controlled level of low force contact is notable
because in studies of “fine force resolution” subjects are
typically seated with the test arm supported against gravity and the index finger is used to apply a force against a
surface (Henningsen et al. 1997). In this circumstance,
when asked (starting from a 20 g baseline force achieved
with visual feedback) to make the smallest possible increase in finger force that they can without visual feedback, they exert about 25 g of additional force. In our experiments, subjects are standing blindfolded in an unstable posture with arm unsupported but they are still able
to control finger force level at least as well as the subjects in the fine force resolution experiments. In experiment 3, where we measured finger position, the average
amplitude of finger displacement was always significantly less than that of the center of pressure. The greatest
stabilization of posture was achieved with finger contact
with the rigid filament or a flat surface. In these conditions, there was about a 250-ms time lead of the lateral
fingertip force in relation to center of pressure so that
force direction reversals at the fingertip were occurring
prior to center of pressure reversals. This means the fingertip is not functioning as a passive probe to detect a
shift in body posture but is being actively and independently controlled. The ability to keep the finger spatially
stable is diminished when the contacted object is flexible, and the attenuation of sway is also lessened.
In our test situation, the subject is essentially swaying
as a single link inverted pendulum and center of pressure
measures reflect center of mass position (cf. Jeka and
Lackner 1995, for quantitative measurements). Precision
contact with a rigid filament or rigid surface coupled
with brachial proprioceptive and efferent information
about finger position in relation to the torso enables subjects to control their fingertip independently of body
sway. This allows the forces at the fingertip to be kept at
low levels and modulated to keep the finger from slipping. By dynamically stabilizing the fingertip in this way
and by initiating reversals of postural sway direction
when finger force modulation occurs, the subject also
stabilizes his or her body. The time lag between changes
in finger force signals and EMG activity in the leg muscles is about 125 ms and changes in center of pressure
follow after another 125 ms (Jeka and Lackner 1995). In
precision grip studies involving grasp of “active objects,” latencies of about 125 ms occur before grip force
adjusts to object motion; this is also the latency of pursuit eye movements in response to motion of a visual
target (Johansson and Cole 1994). The prevalence of
125-ms latencies suggests common principles may be
underlying postural, oculomotor and haptic control. Because the masses of the finger and eyeball are tiny in relation to that of the legs (grams vs kilograms), it takes
considerably longer for adequate forces to be generated
to affect the center of mass of the body. Our findings imply detailed monitoring of arm and finger position and
comparisons of expected with ongoing patterns of sensory feedback based on internal models of the arm and
body (cf. Cohn et al. 2001; Flannagan and Wing 1993;
464
Miall and Wolpert 1996) that allow appropriate postural
compensations to be initiated before thresholds for either
ankle proprioception or vestibular contributions are exceeded. Recent neurophysiological studies of cortex are
beginning to provide insights into the central somatosensory and motor components of precision grip and precision touch control and how such monitoring may occur
(e.g., Bennett and Lemon 1996; Lemon et al. 1995;
Salimi et al. 1999a, 1999b, 1999c).
Haptic information about postural sway derived from
contact with other parts of the body can also attenuate
sway. In ongoing experiments, we have found that if the
forehead or nose or leg is allowed contact at non-supportive force levels with a stable surface or rigid filament, body sway will be attenuated about half as much
as with contact of the hand. In fact, contact with virtually
any part of the body surface can influence the control
and perception of body orientation (cf. Lackner 1981;
Lackner and Graybiel 1978, 1979, 1983). However, the
hand and arm are especially flexible in providing spatial
contact information while also allowing a considerable
range of body displacement.
Acknowledgements This research was supported by NASA
grants NAG9-1037 and NAG9-1038. We thank Dr. Joel Ventura
for technical advice and assistance.
References
Bennett KMB, Lemon RN (1996) Corticomotoneuronal contribution to the fractionation of muscle activity during precision
grip in the monkey. J Neurophysiol 75:1826–1843
Bernstein N (1967) The coordination and regulation of movements. Pergamon Press, Oxford
Clapp S, Wing AM (1999) Light touch contribution to balance in
normal bipedal stance. Exp Brain Res 125:521–524
Cohn JV, DiZio P, Lackner JR (2001) Reaching during virtual rotation: context specific compensations for expected Coriolis
forces. J Neurophysiol (in press)
Fitzpatrick R, McCloskey DI (1994) Proprioceptive visual and
vestibular thresholds for the perception of sway during standing in humans. J Physiol 478:173–186
Flannagan JR, Wing AM (1993) Modulations of grip force with
load force during point-to-point arm movements. Exp Brain
Res 95:131–143
Henningsen H, Knecht S, Ende-Henningsen B (1997) Influence of
afferent feedback on isometric fine force resolution in humans.
Exp Brain Res 113:207–213
Holden M, Ventura J, Lackner JR (1987) Influence of light touch
from the hand on postural sway. Soc Neurosci Abstr 13:348
Holden M, Ventura J, Lackner JR (1994) Stabilization of posture
by precision contact of the index finger. J Vestib Res 4:285–
301
Jeannerod M (1994) The representing brain: neural correlates of
motor intention and imagery. Behav Brain Sci 17:187–245
Jeka JJ, Lackner JR (1994) Fingertip contact influences human
postural control. Exp Brain Res 100:495–502
Jeka JJ, Lackner JR (1995) The role of haptic cues from rough and
slippery surfaces on human postural control. Exp Brain Res
103:267–276
Johansson RS (1991) How is grasping modified by somatosensory
input? In: Humphrey DR, Freund H-J (eds) Motor control:
concepts and issues. John Wiley, Chichester
Johansson RS, Cole KS (1994) Grasp stability during manipulative actions. Can J Physiol Pharmacol 72:511–524
Johansson RS, Vallbo AB (1979) Tactile sensibility in the human
hand: relative and absolute densities of four types of mechanoreceptive units in glabrous skin. J Physiol (Lond) 286:283–300
Johansson RS, Vallbo AB (1980) Spatial properties of the population of mechano-receptive units in the glabrous skin of the human hand. Brain Res 184:353–366
Johansson RS, Vallbo AB, Westling G (1980) Thresholds of mechanoreceptive afferents in the human hand as measured with
von Frey hairs. Brain Res 184:343–351
Johnson KO, Hsiao SS (1992) Neural mechanisms of tactual form
and texture perception. Ann Rev Neurosci 15:227–250
Lackner JR (1981) Some contributions of touch pressure and kinesthesis to human spatial orientation and oculomotor control.
Acta Astronautica 8:825–830
Lackner JR, Graybiel A (1978) Some influences of touch and
pressure cues on human spatial orientation. Aviat Space Environ Med 49:798–804
Lackner JR, Graybiel A (1979) Parabolic flight: loss of sense of
orientation. Science 206:1105–1108
Lackner JR, Graybiel A (1983) Perceived orientation in free fall
depends on visual postural and architectural factors. Aviat
Space Environ Med 54:47–51
Lackner JR, DiZio P, Jeka JJ, Horak F, Krebs D, Rabin E (1999)
Precision contact of the fingertip reduces postural sway of individuals with bilateral vestibular loss. Exp Brain Res 126:
459–466
Lemon RN, Johansson RS, Westling R (1995) Corticospinal control during reach grasp and precision lift in man. J Neurosci
15:6145–6156
Miall C, Wolpert D (1996) Forward models for physiological motor control. Neural Networks 9:1265–1278
Norkin C, Levangie P (1990) Joint structure and function. FA
Davis, Philadelphia
Rabin E, Bortolami S, DiZio P, Lackner JR (1999) Haptic stabilization of posture: changes in arm proprioception and cutaneous feedback for different arm orientations. J Neurophysiol
82:3550–3562
Reginella RL, Redfern MS, Furman JM (1999) Postural sway with
earth-fixed and body referenced finger contact in young and
older adults. J Vestib Res 9:103–109
Riley MA, Wong S, Mitra S, Turvey MT (1997) Common effects
of touch and vision on postural parameters. Exp Brain Res
117:165–170
Salimi I, Brochier T, Smith AM (1999a) Neuronal activity in somatosensory cortex of monkeys using a precision grip. I. Receptive fields and discharge patterns. J Neurophysiol 81:825–
834
Salimi I, Brochier T, Smith AM (1999b) Neuronal activity in somatosensory cortex of monkeys using a precision grip. II. Responses to object texture and weights. J Neurophysiol 81:835–
844
Salimi I, Brochier T, Smith AM (1999c) Neuronal activity in somatosensory cortex of monkeys using a precision grip. III. Responses to altered friction perturbations. J Neurophysiol 81:
845–857
Srinivasan MA, Whitehouse JM, Lamotte RH (1990) Tactile detection of slip: surface microgeometry and peripheral neural
codes. J Neurophysiol 63:1323–1332
Torebjörk HE, Schady W, Ochoa J (1984) Sensory correlates of
somatic afferent fibre activation. Hum Neurobiol 3:15–20
Vallbo AB, Johansson RS (1984) Properties of cutaneous mechanoreceptors in the human hand related to touch sensation.
Hum Neurobiol 3:3–14