Perception &Psychophysics
1974, Vol. 15,No. 3,529·532
Visual proprioceptive control of
standing in human infants·
DAYID N. LEEt and ERIC ARONSON
University ofEdinburgh, Edinburgh, EBB 9TJ, Scotland
Human infants learning to stand use visual proprioceptive infonnation about body sway in order to
maintain stable posture. Moreover, the visual proprioceptive infonnation is more potent than the
nonvisual. This is shown by an experiment in which infants were caused to sway and even fall forward 01"
backward in response to appropriate visual stimulation.
According to the classical view formulated by
Sherrington (1906), the receptor systems of the body
may be classified as exteroceptors, proprioceptors, or
interoceptors. It is assumed that each receptor system
subserves a unique function, exteroceptive,
proprioceptive, or interoceptive.
J. J. Gibson (1966) has questioned this assumption.
He argues that if exteroception is defined to be the
obtaining of information about events extrinsic to the
organism, and proprioception the obtaining of
information by the organism about its own actions, then
vision, in particular, is not only exteroceptive, as is
classically assumed, but is also proprioceptive.
The proprioceptive function of vision seems apparent
enough in driving a vehicle; the driver clearly uses visual
information about his and the vehicle's movement to
guide the vehicle. Does vision also function
proprioceptively in the control of more basic activities?
The activity we chose to study is standing.
Standing involves continuous compensatory
adjustments of the musculature. It is a process of
feedback control. Any sway of the body away from the
vertical has to be registered and compensatory muscular
adjustments made else balance is lost.
The classical view is that the information about body
sway that is used in standing comes from receptors
("proprioceptors," in classical terminology) in the
vestibular canals and in the joints and muscles,
particularly of the ankles and hips (Eldred, 1960). We
may refer to these as mechanoreceptors, since they are
responsive to mechanical force.
The question of whether vision too functions
proprioceptively in standing does not seem to have been
considered. There is, however, some suggestive evidence
in the literature. For example, Wood (1895) reported
that people standing on a stationary "haunted swing"
while the surrounding room was rotated about a
*The work was supported by S.R.C. Research Grant
B/RG/631 aDd M.R.C. Research Grant G/969/559C. We tbaDk
Jane Dunke1d aDd J. R. Lishman for their assistance in running
the experiment.
tRequests for reprints should be addressed to D. N. Lee,
Department of psychology, University of Edinburgh, Edinburgh
EH8 9TJ, Scotland.
horizontal axis tended to be unsteady. Also, when
normal vision is impaired by imposing an impoverished
or unstable visual field or by blindfolding a person, body
sway tends to increase (Edwards, 1946; Witkin &
Wapner, 1950). While the available evidence is
suggestive, it is not conclusive. For example, the increase
in body sway with impaired vision may simply be due to
increased tension in the person, as Edwards (1946) in
fact suggested. However, if it could be shown that
balance can be disturbed by visual stimulation in a
direction specific to that stimulation, this would furnish
direct evidence that vision functions proprioceptively in
standing.
Visual information about movement of the body
relative to the environment is available in the optic array
at the eye. When the head is moving, as when the body is
swaying, there are certain properties of the optic flow
pattern at the eye that specify that movement (Gibson,
1958; Lee, 1973). The question is whether or not this
visual proprioceptive information is actually used in
maintaining balance.
We chose to use as Ss human infants with limited
experience in standing and walking. There were several
reasons for this. First, the standing posture in humans is
manifestly less stable in the early stages of its
development; we therefore expected that any effect on
balance that our experimental procedure might produce
would be more pronounced, and so more easily
measured, in infants than in adults. Second, the naivete
of infants should militate against their inadvertently
complying with the E's expectations. Finally, the
present study is part of an ongoing series using both
infant and adult Ss, and here we were particularly
interested in determining whether or not, and if so to
what extent, visual proprioception is an integral part of
the postural control system in the early stages of
development.
METHOD
The standing S was presented, for short periods, either with an
opti: flow pattern similar to what normaUy would accompany
backward body sway or with the opposite. This was done by
529
530 LEE AND ARONSON
Fig. 1. The moving room and the experimental situation just prior to moving the room past the infant S.
moving an experimental "room," comprising three walls and a
ceiling, forward or backward past the S standing on a stationary
floor (see Fig. 1). The "room," 3.6 m long x 1.8 m wide x 2 m
high, was constructed of rigid white polystyrene on a metal
frame and was open at one end. It was suspended in a horizontal
position by four ropes, one to each upper corner, from the
7.2-m-high ceiling of a lecture theater, so that the bottoms of the
walls were about 1 em above the floor of the theater. The room
could be swung noiselessly along its length from one lockable
position 47 em to one side of its stable hanging position to
another lockable position the same distance the other side. The
time for such a swing was about 2.5 sec, and the motion was
approximately simple harmonic, with a maximum velocity, in
the middle, of about 40 em/sec. During its swing, the room
remained horizontal and the variation in its height above the
floor was only about 4 em. The room was lit by two 60-W bulbs
rigidly attached to its ceiling 30 em from the closed end. Pictures
were stuck on the walls to make the room more interesting.
Consider a S standing still in the room facing the closed end.
Motion of the room forward, in the direction he is facing, will
produce optic flow patterns at his eyes that are similar, except
for the lower sector corresponding to the stationary floor, to
what would normally accompany backward sway of his body
(see Fig. 2). If he uses visual proprioception in maintaining
posture, the visual information for backward body sway should
induce muscular action to produce a compensatory forward
torque about his feet. If this torque were maintained, he would
sway forward until he lost balance. However, once he was
swaying forward, a conflict between visual and mechanical
proprioception should arise; vision would be still specifying
backward body sway and so calling for the maintenance of the
forward torque, while his mechanoreceptors should be
registering the actual forward sway of his body and so calling for
a compensatory backward torque. The converse holds for
backward motion of the room. Thus, any abnormal sway or loss
of balance in the direction the room is moved would indicate
dominance of the visual proprioceptive information over the
mechanical proprioceptive information.
Subjects
Seven normal healthy human infants served as Ss in the
experiment. They ranged in age from 13 to 16 months, and in
walking experience, according to their mothers' accounts, from 1
to 22 weeks (see Table 2).
Procedure
At the start of a session, the experimental room was
stationary at one of its two lockable positions. (Four of the Ss
started the experiment with the room in one position, and the
other three with it in the other position.) The infant S was then
brought into the room by its mother and an E, who remained
there throughout the experiment. They sat on the carpeted floor
at the closed end of the room, facing the open end, while the
infant was free to move around in the room and play with toys
for about 3 min. If necessary, the infant was encouraged to stand
and/or walk during this 3-min pretest period, and any falls were
recorded, providing a measure of the infant's normal stability. As
throughout the whole experiment, three independent records
were taken: one by the E inside the room, one by an E outside
the room, and one by a TV camera directed through the open
end of the room, the video record subsequently being analyzed
by a third person. On the great majority of trials there was
consensus among the three indepenent records, and when there
was not consensus, a majority criterion was applied.
The main experiment followed immediately after the pretest
and comprised up to 20 trials. Before starting a trial, the S was
required to be standing still, unsupported, facing the closed end
of the room. Normally, the E simply waited until the S was in
this position before starting a trial; on occasion, the S was
manoeuvred into the position. On a few trials, the S started to
VISUAL PROPRIOCEPTIVE CONTROL 531
Table 1
'(he Response Categories Used in the Experiment
Behavior
+++
++
+
-+
+-
Positive
Negative
Initial sway WR and fall WR
Initial sway WR and stagger WR
Initial sway WR
Initial sway OR and stagger or fall WR
Initial sway WR and stagger or fall OR
Initial sway OR and fall OR
Initial sway OR and stagger OR
Initial sway OR
Zero 0 No abnormal sway
Mixed
Category Code
RESULTS
walk just as the trial was starting, and when this lead to
uncertainty in categorizing his behavioral response, a "zero"
response was recorded (see Table 1).
On each trial, the room was swung by an E behind the S from
one stationary locked position to the other. It was swung back
again on the next trial, and so on. At the start of a trial, the S
was usually standing near the middle of the room, and during no
trial was the closed end of the room ever closer to him than 1 m.
The interval between trials was usually 30-60 sec. Whenever
possible, a 3-min posttest, similar in form to the pretest, was
given immediately after the main experiment.
The behavioral response to a swing of the room was
categorized according to the scheme shown in Table 1.
Unless there was a clear change in posture during the
2.5 sec that the room was moving, a "zero" response was
recorded. The experimental results and S data are
presented in Table 2. Three of the Ss (2, 4, and 5)
became very distressed during the main part of the
experiment, and the experiment had to be curtailed.
Since these Ss had appeared quite content during the
pretest period, their distress was presumably due to their
instability when the room was moved. The remaining
four Ss, however, showed no undue distress throughout
the experiment, though their instability during trials was
,
e:">,
8
Note- "WR"/"OR" (with room/opposite room) mean in the
same/opposite direction as the room was moving.
just as marked. The experiment had to be stopped after
the lOth trial for one of these Ss (No.1) due to the S's
becoming sleepy.
As argued above, a "positive" response (i.e., a sway,
stagger, or fall in the direction the room was moved)
would indicate that the S was making compensatory
adjustment to his posture in accord with the visual
proprioceptive information about body sway produced
by the motion of the room. The results (Table 2)
indicate that such occurred on the great majority of
trials, frequently resulting in loss of balance. Overall,
82% of responses were positive-26% sway, 23% stagger,
and 33% fall.
We discount two other possible explanations of the
results. First, it might be argued that a backward positive
response may simply be an avoidance response to the
"looming" of the room. However, this would leave
unexplained the forward positive responses that
occurred when the visible end of the room was moved
away from the S. Secondly, the draught produced by the
motion of the room, while consistent in direction with a
positive response, was so slight' as to make it a most
unlikely cause of such a response, particularly one
involving loss of balance.
DISCUSSION
Fig. 2. Motion of the experimental room forward (A)
produces optic flow patterns at the S's eyes that are similar to
what would normally accompany backward body sway (B).
Postural adjustment to counteract this apparent backward body
sway would result in the S's swaying forward and possibly losing
balance (C). The opposite holds for backward motion of the
room.
c
It is, therefore, evident that an infant learning to
stand uses visual proprioception in maintaining his
posture. Furthermore, since the conflict created in our
experiment between mechanical and visual
proprioception was, in the majority of cases, dominated
by the visual, it may be concluded that, for infants at
least, visual proprioceptive information is more potent
than mechanical proprioceptive information. If one
compares the mechanical and visual proprioceptive
information likely to be available to an infant, it is,
indeed, not surprising that visual proprioception should
playa dominant role in the maintenance of standing.
The most sensitive mechanical proprioceptive
information for standing is probably that from the
mechanoreceptors in the ankle joints and associated
muscles, and those in the soles of the feet. The vestibular
system appears to be insensitive to fine body movements
532 LEE AND ARONSON
Table 2
Subject Data and Experimental Results
81 82 83 84 S5 86 87
Weeks walking 1 4 8 8 9 10 22
Age (months: weeks) 16: 0 15 : 3 14: 1 15 : 2 14: 3 13 : 1 16 : 2
Sex F M F F F F M
Number falls in 3-min pretest 0 0 0 If 0 0 0
Number falls in 3-min posttest Ib 0 0 0
Main Experiment
Number trials 10 12 20 6 4 20 20
Number (+++) responses forward: backward 3:3 0:5 4:4 0: 1 1 : 1 5 : 1 0:2
Number (++) responses forward: backward 0:0 2: 1 1 : 1 1 : 1 0:0 4: 1 3:6
Number (+) responses forward: backward 1:2 3:0 3:4 0:0 1 : 0 0:5 3:2
Total number positive responses 9 11 17 3 3 16 16
Number mixed responses 0 0 0 0 0 0 0
Number negative responses 0 0 1 0 1 0 0
Number zero responses 1 1 2 3 0 4 4
(Birren, 1945). However, for an infant, these
mechanoreceptors are necessarily sensitive to the growth
changes in lengths and weights of his body parts, and so
probably can be kept finely calibrated only by means of
continual practice in some controlled activity that
utilizes them. But few, if any, of the activities that an
infant engages in before standing are of this type, and so
when he begins to stand, his mechanical proprioceptive
system will afford only rudimentary and imprecise
information, being essentially still in embryonic
condition (see Trevarthen, 1973, for a discussion of
behavioral embryology).
The infant's visual proprioceptive system, on the
other hand, should be more developed and considerably
more reliable. There seems no reason why its calibration
should be affected by skeletal growth, and the eye is
quite close to its final form. Moreover, the infant will
have had much visual proprioceptive feedback from his
prestanding activities, e.g., sitting, rocking, crawling, and
being carried.
In some more recent experiments, to be reported in
detail elsewhere, it has been found that adults show
similar behavior to that of infants in the experimental
room. Body sway is generally greater when the room is
moving forward and back than when it is stationary or
when the eyes are closed, the body sway tending to be in
phase with the motion of the room. However, the effect
on balance of the motion of the room is not as
pronounced as in infants, indicating that an adult has a
more finely tuned mechanical proprioceptive system.
We would suggest, therefore, that when an infant is
learning to stand he relies heavily on visual
proprioception, and only later, as a result of practice,
does his mechanical proprioceptive system approach the
same degree of efficiency as his visual proprioceptive
system.
REFERENCES
Birren, J. E. Static equilibrium and vestibular function. Journal
of Experimental Psychology, 1945, 35,127-133.
Edwards, A. S. Body sway and vision. Journal of Experimental
Psychology, 1946,36,526-535.
Eldred, E. Posture and locomotion. In Handbook of physiology.
Sec. 1, Vol. II. Washington, D.C: American Physiological
Society. 1960.P. 1067.
Gibson, J. J. Visually controlled locomotion and visual
orientation in animals. British Journal of Psychology. 1958,
49,182·194.
Gibson, J. J. The senses considered as perceptual systems.
Boston: Houghton Mifflin, 1966. Pp. 31-46.
Lee, D. N. Visual information during locomotion. In Studies in
perception: Essays in honor of J. J. Gibson. Ithaca, N.Y:
Cornell University Press, in press.
Sherrington, C. S. The integrative action of the nervous system.
Cambridge: University Press, 1906.
Trevarthen, C. Behavioural embryology. In E. C. Carterette and
M. P. Friedman (Eds.), The handbook of perception. New
York: Academic Press, 1973.
Witkin, H. A., & Wapner, S. Visual factors in the maintenance of
upright posture. American Journal of Psychology, 1950, 63,
31·50.
Wood, R. W. The "haunted-swing" illusion. Psychological
Review, 1895, 2,277-278.
(Received for publication September 24, 1973;
revision received December 24, 1973.)