Neuroscience Letters 419 (2007) 5–9
Haptic selective attention by foot and by hand
Alen Hajnal a,∗ , Sergio Fonseca b , Jeffrey M. Kinsella-Shaw a,c , Paula Silva a ,
Claudia Carello a , M.T. Turvey a
a
Center for the Ecological Study of Perception and Action, University of Connecticut, Storrs, CT, USA
b Department of Physical Therapy, Federal University of Minas Gerais, Belo Horizonte, Brazil
c Department of Physical Therapy, University of Connecticut, Storrs, CT, USA
Received 24 January 2007; received in revised form 15 March 2007; accepted 20 March 2007
Abstract
Nonvisual perceptions of a wielded object’s spatial properties are based on the quantities expressing the object’s mass distribution, quantities that
are invariant during the wielding. The mechanoreceptors underlying the kind of haptic perception involved in wielding – referred to as effortful,
kinesthetic, or dynamic touch – are those embedded in the muscles, tendons, and ligaments. The present experiment’s focus was the selectivity of
this muscle-based form of haptic perception. For an occluded rod grasped by the hand at some intermediate position along its length, participants
can attend to and report selectively the rod’s full length, its partial lengths (fore or aft of the hand), and the position of the grip. The present
experiment evaluated whether participants could similarly attend selectively when wielding by foot. For a given rod attached to and wielded by
foot or attached to (i.e. grasped) and wielded by hand, participants reported (by magnitude production) the rod’s whole length or fractional length
leftward of the point of attachment. On measures of mean perceived length, accuracy, and reliability, the degree of differentiation of partial from full
extent achieved by means of the foot matched that achieved by means of the hand. Despite their neural, anatomical, and experiential differences,
the lower and upper limbs seem to abide by the same principles of selective muscle-based perception and seem to express this perceptual function
with equal facility.
© 2007 Elsevier Ireland Ltd. All rights reserved.
Keywords: Dynamic touch; Foot; Hand; Perception
A large body of research has examined the nonvisual perceptions
of spatial properties of firmly grasped and manually wielded
rigid objects. The experimental objects have varied widely in
shape, size and material composition, and they have been of
either uniform or nonuniform density. The research has shown
that the nonvisual perceptions of a wielded object’s spatial properties are tied to the moments (mass, static moment, moment of
inertia) of the object’s mass distribution, physical quantities that
remain unchanged during the wielding [6].
The subsystem of the haptic perceptual system involved in
wielding a handheld object is referred to as dynamic or effortful
touch [9,22] or kinesthetic touch [16]. Unlike the most commonly understood form of touch involving the skin receptors,
the receptor basis for dynamic touch comprises the spindles and
∗ Corresponding author at: Department of Psychology, U-1020, 406 Babbidge
Rd., University of Connecticut, Storrs, CT 06269-1020, USA.
Tel.: +1 860 486 0998; fax: +1 860 486 2760.
E-mail address: [email protected] (A. Hajnal).
0304-3940/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.neulet.2007.03.042
Golgi tendon organs embedded in muscle and muscle’s attachment to the skeleton. Dynamic touch is Bell’s muscle sense [1]
broadened to include muscle’s exteroceptive capability as well
as its proprioceptive capability.
Like seeing, hearing, and the cutaneous form of haptic perception (e.g., [20]), dynamic touching can be selective. For
example, when wielding an occluded rod grasped at some intermediate position along its length, a person can make distinct
judgments about the rod’s full length, its partial lengths (e.g.,
fore or aft of the hand), and the position of the grip. Each of
the preceding selective perceptions seems to be constrained in
a specific way by the time-invariant moments and each can be
shown, through formal procedures of experimental design and
analysis, to be independent of the others [7].
Although one is inclined to link dynamic touching to manual abilities, the involvement of the hand is not necessary.
Within limits, any part of the muscular-skeletal system can move
(wield) an attached object, engender activity in muscle-related
mechanoreceptors, and, therefore, express the capabilities of
dynamic touching. Recent research has compared dynamic
6
A. Hajnal et al. / Neuroscience Letters 419 (2007) 5–9
touching by lower and upper limbs [10]. On a trial, a rod was
attached at one end to the right foot or held at one end by the right
hand. The limb and rod in both cases were occluded. The participant’s task was to wield the rod and to estimate (by magnitude
production) its full length. Despite the seemingly considerable
differences between foot and hand in the experience of wielding
and perceiving object properties nonvisually, full-length perceptions by foot and hand were similar in accuracy and dependence
on the mass moments of the wielded rods.
Understanding the muscle-based perceptual capabilities of
the lower limbs is of potential importance to addressing the
sensory neuropathy presented by diabetic and stroke patients.
Carello et al. [4] reported a clinical case in which a person with
loss of discriminative touch in the left arm but no comparable sensory deficits in the right arm could, nonetheless, express
dynamic touch capability with her insensate left arm (rod lengths
were perceptible) that was comparable to that of her unaffected
right arm. If muscle-based exteroceptive perceptual capabilities
are preserved in the absence of discriminative touch then the possibility arises that they might be deployable in remedial ways to
offset the loss of discriminative touch.
In the present research, we investigated the dynamic touch
ability of the lower limbs in respect to selective attention. In the
experimental arrangement depicted in Fig. 1, we asked whether
participants could attend selectively during wielding to (a) the
full length of a rod attached to the foot and (b) the partial length
leftward of the point of attachment, where the foot’s attachment
was at the rod’s midpoint. We asked this question in relation
to the participants’ selective ability to perceive full and partial
lengths when wielding a rod by hand with the grasp at the rod’s
midpoint.
The selective perception capability was evaluated in terms
of the experimental manipulations schematized in Fig. 2a. Consider whole length. If whole length perception is a function of
the rod’s second moment (moment of inertia), then stimulus A
should be perceived as shorter than stimuli B and C which, in
turn, should be perceived to be equal in length. When considering whole length, the attached mass makes the resistances to
rotational acceleration of stimuli B and C equally greater than
that of stimulus A. Consider partial length. If partial length leftward of midpoint is perceptually separable from whole length in
a way that is functionally equivalent to cutting off the rightward
portion of the rod, then stimuli A and B should be perceived to
be equal in length (same moment of inertia about the midpoint
qua endpoint) and shorter than stimulus C (greater moment of
inertia about midpoint qua endpoint).
The key question was whether the predicted pattern of whole
length and partial length reports would be reproduced for both
limbs. A matching of the highlighted data pattern of Fig. 2
by the data of both foot and hand would lend support to the
understanding that muscle-based perception by the lower limbs
operates according to the same principles as muscle-based perception by the upper limbs. Countering a possible sameness of
principles are differences between lower and upper limbs (a) in
degree of experience with dynamic touching, (b) in sensitivity to
basic sensory variables [14,17], and (c) in neural and anatomical
components and functions [13,21].
Fig. 1. A hollow handle was either grasped in the hand or taped firmly to the
bottom of the shoe and foot sheathed in a plastic material. The wielding limb was
extended under an occluding curtain. On a given trial, the experimenter inserted
a rod through the handle and secured it at the midpoint with thumbscrews. The
occluded rod was wielded freely by foot (a) or by hand (not shown) as two
visible markers were adjusted to indicate the length of the entire rod (b) or a
single marker was adjusted to indicate the location of the left tip of the rod (c).
Twenty participants 18–22 years of age were recruited from
University of Connecticut undergraduates. All gave their consent in accordance with the University of Connecticut’s internal
review board’s regulations for studies with human participants.
The stimulus materials were three pairs of wood rods of length
(L) .6, .8, and 1 m each, with a plastic handle at the midpoint.
To produce the stimulus B and C versions depicted in Fig. 2b,
a 150 g weight was attached to one member of each pair at a
distance 1/8 L from either the left or right end. For each trial, the
participant was seated such that his or her frontal plane was parallel to a curtain that occluded the limb (either the right arm or the
right leg) and the rod (either grasped by hand or attached to foot).
The latter arrangement, focusing upon the foot, is depicted in
Fig. 1 together with the magnitude production task by which the
participant reported his or her impressions of the whole length
or the partial length.
In total, there were nine stimulus conditions: three rodlengths by three mass manipulations (no mass, mass rightward,
mass leftward). Each participant received three trials at each
A. Hajnal et al. / Neuroscience Letters 419 (2007) 5–9
7
cation from comparison of the panels – that selective perception
by foot and by hand abided by a common principle – was confirmed by the non-significance of the three-way interaction of
limb × stimuli × report, F(2, 36) = 1.5, p = .23.
Corroboration of the latter implication was sought through
an examination of the accuracy and reliability of full and partial
reports: Were they likewise patterned in the same way for foot
and hand? Comparison of the consistency or reliability of partial
and whole length judgments across limbs was conducted by
means of the average deviation (AD) measure computed for
each participant:
No Nrep
i=1 j=1 |Lpij − Lpi |/Lpi
AD =
× 100,
No × Nrep
Fig. 2. With wielding from the center of the rod for both hand and foot, conditions A, B, and C were created, respectively, with no attached mass, mass to the
right of the limb, or mass to the left of the limb (a). The expected data pattern
for A, B, and C (each averaged over three lengths) – highlighted in the center
gray panel – was obtained for both hand and foot (b).
stimulus condition for both kinds of wielding (by foot, by hand)
resulting in 54 trials. The order of presentation of hand trials
and foot trials was blocked and counterbalanced over participants. The order of the rods was randomized within each block.
Before each block of trials, participants performed three practice
trials to familiarize themselves with the task. The whole experiment lasted approximately 1 h. There was a 2-min break between
blocks. The participants received no feedback during the experiment and never once saw the rods used in the experiment.
Analysis of variance (ANOVA) was conducted on the magnitude production measure with factors of report (partial versus
whole), rod length, stimulus (A, B, C), and limb. Four fundamental expectations from the highlighted data pattern of Fig. 2,
those preliminary to the primary hypothesis, were confirmed.
First, partial report was less than whole report, F(1, 18) = 12.92,
p < .01, length perceptions increased with rod length, F(1,
18) = 23.81, p < .0001, the stimuli (A, B, and C) were perceived differently, F(1, 18) = 19.84, p < .0001, and the partial
report–whole report contrast depended on the stimuli, F(2,
36) = 13.97, p < .0001.
Other results preliminary to the primary hypothesis were
those involving limb but not stimuli. Magnitude productions
were numerically smaller for foot (0.39 m) than for hand
(0.44 m) but not significantly smaller (p = .07); the size of the difference between perception by foot and perception by hand was
dependent on rod length, F(2, 36) = 5.29, p < .01, an interaction
that was indifferent (p = .5) to report (partial versus whole).
Turning now to the primary hypothesis, the question of significance was whether the patterning of magnitude productions
over stimuli (A, B, and C) and report (partial versus whole)
were identical for wielding by foot and by hand. The patterns
are shown in the bottom left and right panels of Fig. 2. The impli-
where Lpij is the perceived length (partial or whole) of object
i on the jth trial, Lpi is the mean perceived length (partial or
whole) for object i, No is the number of lengths (partial or whole),
and Nrep is the number of repetitions. A comparable measure of
accuracy, expressed as a percentage of actual extent (La, partial
or whole), is provided by the Weber-like quantity of mean-rootsquare (MRS) error:1
No Nrep (Lpij − La)2 /La
i=1 j=1
MRS =
× 100
No × Nrep
Four AD measures and four MRS measures were obtained
for each participant. The four measures were: perceived partial
and whole length by foot and perceived partial and whole length
by hand. A three way ANOVA was conducted with factors of
measure (MRS, AD), limb (foot, hand) and report (whole length,
partial length). In concert with the primary hypothesis, the threeway interaction was not significant, F < 1. The relation between
measure and report was the same for both foot and hand meaning that AD and MRS patterned in the same way for partial and
whole length judgments regardless of the limb doing the wielding. Overall, foot total “error” (MRS plus AD) exceeded hand
total “error” (29% versus 22%), F(1, 18) = 34.29, p < .0001.
The present results corroborate the finding of Hajnal et al.
[10] that the foot is comparable to the hand in the perception of
object length by dynamic touch. They go beyond that finding,
however, in showing that the foot compares favorably to the hand
in its ability to perceive selectively. The reported experiment has
shown that for a rod attached to the foot or grasped by the hand, a
person can differentiate the rod’s extent to one side of the foot or
hand from the rod’s full extent. Most importantly, the experiment
has shown that the degree of differentiation of partial from full
extent achieved by means of the foot matched that achieved
by means of the hand. The various differences between lower
and upper extremities numerated above do not translate into
differences in selective perception by dynamic touch.
1 In contrast with the standard root mean square (RMS), in which the deviations of perceived from actual length are summed prior to taking the root, the
benefit of MRS is that it scales the error as a dimensionless Weber fraction
(and functions, therefore, like the reliability measure). The standard RMS is
not invariant over different units of measurement and, in consequence, is more
difficult to interpret. RMS and MRS are linearly correlated.
8
A. Hajnal et al. / Neuroscience Letters 419 (2007) 5–9
The remarkable feature of this selective ability shared by
foot and hand is summarized by the following: when the mass
moments of the involved rods are different (e.g., stimulus A and
B), identical partial lengths are perceived to be the same and
identical whole rod lengths are perceived to be different. As
noted in preceding remarks, partial length perception could be
likened to dividing the rod at the foot or hand’s point of attachment leaving only the leftward part in the “grasp.” In previous
experiments, directed at this selective attention phenomenon, it
has been shown that a comparison between a stimulus B with
a 50 g added mass and a stimulus B with a 100 g added mass
reveal that the partial length of the non-weighted side is perceived the same in the two cases whereas the partial length
of the weighted side is perceived as different in the two cases
with longer perception for the 100 g case [23]. At issue is how
the perceiver achieves the perceptual isolation. One hypothesis ties the ability to a spinor-like (quaternion-like) quantity in
conjunction with the conventional mass moments [23]. The key
feature of this latter quantity is that it simultaneously represents
the attached object’s mass-based reference frame in two oppositely oriented orientations of θ and 2π − θ relative to the hand
(or foot). The observed ability to attend selectively – above or
below, right or left – is attributed to a selection of one of the
two oriented rotations. Multiplying the object’s inertia tensor
by θ for attend above or left and by 2π − θ for attend below or
right yields an independent measure that captures the orderliness
of the observed data [23,24]. The general idea is that selective
attention within dynamic touch is realized as tuning into physical
variables specific to the to-be-attended-to property.
The neurophysiology that might support selective dynamic
touch is unclear. Somewhat analogous achievements by the
somatosensory system reveal activity (measured by ERP and
PET) modulated by the act of attention [8] but without distinctions as a function of the particular attended property [2].
Although the activity is at somatotopically appropriate sites (S1
in the precentral gyrus and S2 within parietal opercular regions),
additional foci have been observed in non-sensory regions of
frontal cortex (i.e., at the fundus of the postcentral sulcus and
cingulate gyrus; [2]). Moreover, a role for the anterior cingulate
in selective attention is suggested by the literature on patients
with strokes in this region [15,18] or those who have undergone anterior cingulotomies [12,25]. In combination with data
showing increased activation in S2 when the tactile stimuli are
behaviorally relevant [2] and in S2 when the limb is in motion
[11], these data challenge any simple account of a hierarchical or
serial “flow” of processing from S1 to S2 (cf. [19]) when selectively attending to a multi-dimensional haptic stimulus array.
Minimally, theses findings suggest that the neurophysiology
underwriting haptic selective attention during wielding will be
at least as complex, interactive, widely distributed over parietal
cortex, as in previously reported paradigms.
Whatever the underlying mechanism for selective dynamic
touch, its equivalent manifestation in foot and hand is perhaps
unexpected given the neuromuscular differences and the different functional roles typically assumed by foot and hand in
everyday life. It is the case, however, that other capabilities of
the muscle-based perception under study in the present article
are known to be widely manifest across segments of the body
(e.g., [3]) and robust against the sensory declines that accompany aging [5] and peripheral neuropathy [4]. The present results
should be seen as lending further support for the interpretation
of muscle organizations implementing dynamic touch as perceptual instruments that are softly assembled [3]. The meaning
of soft is two-fold. First, it means that the instrument is temporary, assembled specifically for the given measurement task
(e.g., partial versus whole length). Second, it means that a given
instrument can be assembled over different anatomical structures or, similarly, that different instruments can be assembled
over the same anatomical structure. Most generally, the qualifier soft serves to emphasize that haptic perceptual instruments
(such as those studied in the present research) are assembled
from general dynamic properties, not from specific anatomical
components. The equivalence of foot and hand in the present
research encourages further examination of the soft assembly
hypothesis of haptic perceptual instruments.
Acknowledgments
Preparation of this manuscript was supported by grants
from the National Science Foundation (SBR00-04097) and the
Provost’s Office at the University of Connecticut.
References
[1] C. Bell, On the nervous circle which connects the voluntary muscles with
the brain, Philos. Trans. R. Soc. 116 (1826) 163–173.
[2] H. Burton, N.S. Abend, A.M.K. MacLeod, R.J. Sinclair, A.Z. Snyder, M.E.
Raichle, Tactile attention tasks enhance activation in somatosensory regions
of parietal cortex: a positron emission tomography study, Cereb. Cortex 9
(1999) 662–674.
[3] C. Carello, P. Fitzpatrick, I. Domaniewicz, T.-C. Chan, M.T. Turvey, Effortful touch with minimal movement, J. Exp. Psychol.: Hum. Percept. Perf.
18 (1992) 290–302.
[4] C. Carello, J.M. Kinsella-Shaw, E. Amazeen, M.T. Turvey, Peripheral neuropathy and object length perception by effortful (dynamic) touch: A case
study, Neurosci. Lett. 405 (2006) 159–163.
[5] C. Carello, S. Thuot, M.T. Turvey, Aging and the perception of a racket’s
sweet spot, Hum. Mov. Sci. 19 (2000) 1–20.
[6] C. Carello, M.T. Turvey, Physics and psychology of the muscle sense, Curr.
Dir. Psychol. Sci. 13 (2004) 25–28.
[7] M. Cooper, C. Carello, M.T. Turvey, Perceptual independence of whole
length, partial length, and hand position in wielding a rod, J. Exp. Psychol.:
Hum. Percept. Perf. 26 (2000) 74–85.
[8] M. Eimer, B. Forster, The spatial distribution of attentional selectivity in
touch: evidence from somatosensory ERP components, Clin. Neurophysiol.
114 (2003) 1298–1306.
[9] J.J. Gibson, The senses considered as perceptual systems. Wavelan,
Prospect Heights, IL, 1983, Original work published in 1966.
[10] A. Hajnal, S. Fonseca, S. Harrison, J.M. Kinsella-Shaw, C. Carello, Comparison of dynamic (effortful) touch by hand and foot, J. Motor Behav. 39
(2007) 82–88.
[11] J. Huttunen, H. Wikstrom, A. Korvenoja, A.M. Seppalainen, H. Aronen,
R.J. Ilmoniemi, Significance of the second somatosensory cortex in sensorimotor integration: enhancement of sensory responses during finger
movement, Neuroreport 7 (1996) 1009–1012.
[12] K.W. Janer, J.V. Pardo, Deficits in attention following bilateral anterior
cingulotomy, J. Cogn. Neurosci. 3 (1991) 231–241.
[13] P.M. Kennedy, J.T. Inglis, Distribution and behaviour of glabrous cutaneous receptors in the human foot sole, J. Physiol. 538 (2002) 995–
1002.
A. Hajnal et al. / Neuroscience Letters 419 (2007) 5–9
[14] C.M. Kets, M.E. Van Leerdam, W.H. Van Brakel, W. Deville, F.W. Bertelsmann, Reference values for touch sensibility thresholds in healthy Nepalese
volunteers, Lepr. Rev. 67 (1996) 28–38.
[15] D. Laplane, J.D. Degos, M. Baulac, F. Gray, Bilateral infarction of the
cingulate gyri and of the fornices. Report of a case, J. Neurol. Sci. 51
(1981) 289–300.
[16] J.M. Loomis, S.J. Lederman, Tactile perception, in: L.K.K. Boff, J. Thomas
(Eds.), Handbook of Human Perception and Performance, Wiley, New
York, 1986, pp. 31.01–31.41.
[17] P.M. Meier, C.B. Berde, J. DiCanzio, D. Zurakowski, N.F. Sethna, Quantitative assessment of cutaneous thermal and vibration sensation and thermal
pain detection thresholds in healthy children and adolescents, Muscle Nerve
24 (2001) 1339–1345.
[18] M.-M. Mesulam, A cortical network for directed attention and unilateral
neglect, Ann. Neurol. 10 (1981) 309–325.
[19] J.T. Pons, P.E. Garraghty, D.P. Friedman, M. Mishkin, Physiological evidence for serial processing in somatosensory cortex, Science 237 (1987)
417–420.
9
[20] S. Soto-Faraco, A. Ronald, C. Spence, Tactile selective attention and body
posture: assessing the multisensory contributions of vision and proprioception, Percept. Psychophys. 66 (2004) 1077–1094.
[21] M. Torgen, C. Swerup, Individual factors and physical work load in relation
to sensory thresholds in a middle-aged general population sample, Eur. J.
Appl. Physiol. 86 (2002) 418–427.
[22] M.T. Turvey, Dynamic touch, Am. Psychol. 51 (1996) 1134–1152.
[23] M.T. Turvey, C. Carello, P. Fitzpatrick, C.C. Pagano, E.E. Kadar, Spinors
and selective dynamic touch, J. Exp. Psychol.: Hum. Percept. Perf. 22
(1996) 1113–1126.
[24] M.T. Turvey, S. Park, S.P. Dumais, C. Carello, Nonvisible perception of
segments of a handheld object and the attitude spinor, J. Motor Behav. 30
(1998) 3–19.
[25] R.T. Watson, K.M. Heilman, J.C. Cauthen, F.A. King, Neglect after cingulectomy, Neurology 23 (1973) 1003–1007.