J Comp Physiol A (1999) 185: 367±372
Ó Springer-Verlag 1999
REVIEW
K. C. Catania
A nose that looks like a hand and acts like an eye:
the unusual mechanosensory system of the star-nosed mole
Accepted: 31 May 1999
Abstract The star-nosed mole (Condylura cristata) has a
snout surrounded by 22 ¯eshy and mobile appendages.
This unusual structure is not an olfactory organ, as
might be assumed from its location, nor is it used to
manipulate objects as might be guessed from its appearance. Rather, the star is devoted to the sense of
touch, and for this purpose the appendages are covered
with thousands of small mechanoreceptive Eimer's organs. Recent behavioral studies ®nd that the star acts
much like a tactile eye, having a small behavioral focus,
or ``fovea'' at the center ± used for detailed explorations
of objects of interest. The peripheral and central nervous
systems of the mole re¯ect these behavioral specializations, such that the small behavioral focus on the nose is
more densely innervated in the periphery, and has a
greatly enlarged representation in the somatosensory
cortex. This somatosensory representation of the tactile
fovea is not correlated with anatomical parameters (innervation density) as found in other species, but rather is
highly correlated with patterns of behavior. The many
surprising parallels between the somatosensory system
of the mole, and the visual systems of other mammals,
suggest a convergent and perhaps common organization
for highly developed sensory systems.
Key words Cortical magni®cation á Somatosensory
cortex á Development á Evolution á Behavior
Introduction
The star-nosed mole is renowned for its unusual nose,
which consists of 22 mobile, ¯eshy appendages that
surround the nostrils. This peculiar nasal specialization
K.C. Catania
Department of Psychology, Vanderbilt University,
301 Wilson Hall, 111 21st Ave South,
Nashville, TN 37240, USA
Tel.: +1-615-322-7491; Fax: +1-615-343-8449
is unique among mammals and raises several basic
questions about star-nosed mole biology. What is the
star and how is it used by the mole to explore its dark
underground environment? What kinds of sensory receptors are associated with the star? How is information
from the star represented in the neocortex? Such questions have been the subject of a number of recent studies
exploring the sensory biology of this species and its
relatives (Catania et al. 1993; Catania 1995a, b, c, 1996;
Catania and Kaas 1995, 1996, 1997a, b). Each aspect of
the sensory system, including the mole's behavior, the
structure of the nose and sensory receptors, and the
physiology and anatomy of the cortex, has provided
some important insight into these basic questions. But
the greatest insights come from combining these studies.
Unusual specializations in the cortex are explained by
behavioral observations. Anatomical comparisons
between the star and its cortical representation reveal
unexpected parallels between the mole's somatosensory
system and the primate visual system. These results
provide an example of how a specialized species can
reveal general principles of brain organization.
Star-nosed mole behavior reveals a ``tactile fovea''
Star-nosed moles live in wetlands in the eastern United
States and Canada. They build extensive tunnel systems
with their heavily clawed forelimbs, and seldom come to
the surface. Moles are insectivores with a high rate of
metabolism, and they must ®nd and eat a large number
of invertebrate prey each day. As would be predicted
from their fossorial lifestyle, they have small eyes and a
tiny optic nerve. Thus a major challenge to their survival
is ®nding sucient quantities of often small prey in their
dark underground tunnels. The star seems to have
evolved mainly for this purpose, and may be the most
sensitive touch organ among mammals.
When moles search for food or explore their environment, the star is in constant motion and is repeatedly
touched to the substrate or objects of interest. Each
368
touch consists of raising the nose upward while the nasal
rays swing backward, then swinging the rays forward
while the star is brought into contact with the substrate.
This behavior is very rapid; star-nosed moles may touch
ten or more di€erent places each second. Similar behavior in other moles without a ``star'' has been described as tapping the ground with the nose (Nagorsen
1996). When a prey item ± such as an earthworm ± is
encountered, the rays are repeatedly touched to the prey
as it is bitten, torn up, and eaten. When prey is missed by
the star, by even a millimeter or two, the mole proceeds
past without apparent notice (whether searching in water or soil), arguing against the hypothesis that the star
functions to detect prey from a distance through electroreception (Gould et al. 1993).
With the naked eye, the details of foraging and prey
encounters are dicult to discern, but slow motion
videotaped behavior of moles locating small prey reveal
a stereotyped sequence of nose movements and a behavioral focus or ``fovea'' on the star (Fig. 1). Ray 11,
the center-most ventral ray (Fig. 2) is preferentially used
to explore prey items, acting in a manner analogous to
the fovea in the visual system of other mammals. For
example, whenever a prey item is ®rst contacted with the
peripheral rays (rays 1 through 10) the nose is then
shifted so that further detailed explorations are made
with ray number 11 (Fig. 1A). No food item is eaten
without ®rst being explored with ray 11. But since ray 11
is one of the smallest rays on the nose, food is usually
Fig. 1A±C The use of the nose while foraging for food. A A
schematic representation of the nose with the rays numbered from 1 to
11, shown in relation to a prey item (gray oval). The ®rst touch made
contact with rays 4 and 5 as the mole discovered the food item. The
second touch was centered on the 11th central pair of rays, after which
the food was eaten. For brevity, the prey is shown changing position
relative to the star ± but in reality, the star is moved relative to the
prey. This sequence of movements is characteristic of prey encounters,
which often include multiple touches with the 11th pair of rays. B A
¯ow diagram illustrating the progression of touches from the lateral
rays to the central 11th ray before the food is eaten. C An example of
the distribution of touches across the nose for ten consecutive preys
encounters, showing the preferential use of the 11th and surrounding
rays. B and C from Catania and Kaas 1997b
®rst encountered by the larger peripheral rays which
together make up most of the surface area of the star.
This stereotyped sequence of movements is very rapid.
Moles can touch a small prey item with the peripheral
rays, shift the star for several additional touches with ray
11, and take the prey into the mouth, all in about 400 ms
(Catania and Kaas 1997b).
The division of the star into peripheral touch and
central touch seems analogous to the retina in the visual
system of other mammals. In primates, for example,
photoreceptors in the peripheral retina typically detect
potentially important stimuli ®rst, and then a saccade
brings the image onto the higher resolution, denser region of photoreceptors of the retinal fovea. Does the
anatomy of the nasal rays also re¯ect the di€erent roles
they play in behavior?
The peripheral anatomy and sensory receptors of the star
Although the star is a specialization of the distal portion
of the nose, it is obviously not an olfactory structure. At
®rst glance, one might imagine the nose acting as an
extra hand. But the rays do not contain muscles or bones
and are not used to manipulate objects or capture prey.
They are controlled through tendons by a complex series
of muscles that attach to the skull, and their role seems
to be purely mechanosensory. For this purpose, they are
composed almost entirely of a series of small mechanosensory organs called ``Eimer's organs''. Under the
scanning electron microscope, these organs are visible in
a honeycomb pattern of epidermal domes on the surface
of each nasal ray (Fig. 2). Eimer's organs are very sensitive to light touch (Catania and Kaas 1995, 1997a) and
they are found on the snout of every member of the
family Talpidae that has been examined (Catania
1995b). However, the star-nosed mole has many more
Eimer's organs than any other species and their distribution across the nasal rays is unique: while most moles
have at most a few thousand Eimer's organs surrounding their nostrils, the star-nosed mole has over 25 000 on
the star. Each organ is supplied by a number of primary
a€erents, thus the star is very densely innervated. There
369
A.
2
1
B.
Terminal Swellings of
Free Nerve Endings
C.
3
4
Epidermis
5
N
6
Dermis
11
Merkel Cell
Neurite Complex
Encapsulated
Corpuscle
7
10
9
8
Fig. 2A±C The nose and sensory organs of the star-nosed mole. A
The nose is composed of 22 appendages, or ``rays'' that surround the
nostrils (N). There are 11 symmetric pairs, numbered from 1 to 11 on
each side. The rays are used by the mole to explore its dark
underground environment through touch. The rays can be moved
through tendons that connect to facial muscles, and they are
repeatedly touched to the substrate as the mole moves through its
system of tunnels. B Each ray is covered with many hundreds of
domed sensory organs called ``Eimer's organs''. These are apparent in
a honeycomb distribution on the skin surface at high magni®cation.
The rays are essentially composed of these sensory organs which are
densely innervated by a branch of the infraorbital nerve. C Each
Eimer's organ is a specialization of the epidermis that contains an
array of sensory receptors, including an encapsulated corpuscle, a
merkel cell-neurite complex, and a series of intra-epidermal nerve
endings that terminate in swellings at the apex of the central cell
column. Plate A from Catania and Kaas 1995
are over 100 000 myelinated ®bers innervating the relatively small star, and a substantial cross section of each
ray is taken up by a large nerve branch (Catania 1995a).
The internal structure of the star-nosed mole's Eimer's organs is illustrated in Fig. 2C. Each organ is a
roughly 40-lm swelling of the epidermis that contains a
central column of keratinocytes. At the bottom of each
organ, in the connective tissue of the dermis, there is a
single encapsulated corpuscle. Just above this, among
the keratinocytes that form the base of the central cell
column, there is a single merkel cell-neurite complex.
Within the cell column, a series of nerve ®bers branch
from three myelinated ®bers in the dermis and ascend
along the margins of the column to terminate in a circular arrangement of swellings just below the outer layer
of epidermis. This concise geometric arrangement of
terminal nerve swellings is entirely enclosed and encapsulated by a single circular keratinocyte at the apex of
the cell column and the overlying outer layer of epidermis is tightly sealed by many desmosomal adhesions
(Catania 1996).
While the response properties of the separate nerve
terminals at the apex of Eimer's organ have not been
Myelinated
Fibers
10µm
Eimer's Organ
characterized, recordings from the cortex show the star to
be highly responsive to several di€erent features of ®ne
tactile stimulation (Catania and Kaas 1995). The super®cial nerve terminals in each Eimer's organs are ideally
positioned to be stimulated by pressure to the top of the
cell column when the rays are touched against an object.
The Eimer's organs have the same basic structure
across the di€erent rays on the star. Ray 11, the behavioral focus of the nose, does not have a higher density of organs, or more organs than other rays, as might
be guessed from its important role in search behaviors.
In fact, because it is relatively short and has little caudal
surface, it has fewer Eimer's organs than almost any
other ray. Ray 11 has about 900 Eimer's organs on its
surface while some of the lateral rays have well over
1500 (Catania and Kaas 1997b).
Rather than having more sensory organs, a di€erent
way in which a skin surface may be more sensitive to
mechanoreceptive input is by increasing its innervation
density and decreasing each a€erent's receptive ®eld. To
test for this possibility, the number of Eimer's organs
and the number of myelinated ®bers innervating each
ray were counted and compared (Fig. 3). Rays 1
through 9 each had about 4 ®bers per Eimer's organ,
while rays 10 and 11 had signi®cantly higher innervation
densities of 5.6 and 7.1 ®bers per organ, respectively.
Thus there is a peripheral specialization of the rays most
used for exploratory behaviors, such that ray 11 has a
much higher innervation density than more lateral rays.
Previous studies of the organization of the somatosensory cortex, where touch information in mammals is
processed, have found a correlation between peripheral
innervation density and the size of the corresponding
representation or map of the sensory surface in cortex.
To investigate this relationship in moles, we examined
the organization of the mole's somatosensory cortex and
related these areas to the anatomy of the star (Catania
and Kaas 1997b).
Average Fibers per Eimer's Organ
370
8
Fibers per Organ
Anatomical proportions
7
6
5
4
3
2
1
0
1 2 3 4 5 6 7 8 9 10 11
Ray Number
Fig. 3 The ratio of myelinated ®bers to Eimer's organs from the 11
rays of the nose (from four moles). While rays 1 through 9 have an
average of roughly 4 ®bers per Eimer's organ, the 10th and 11th have
signi®cantly higher innervation densities of 5.6 and 7.1 ®bers per
organ, respectively (from Catania and Kaas 1997b)
The representation of the star in cerebral cortex
To explore the organization of the somatosensory cortex
in star-nosed moles, neuronal responses were recorded
using microelectrodes in anesthetized moles, while the
nose and body were gently stimulated with small probes
(Catania et al. 1993; Catania and Kaas 1995). A large
area of cortex responded to stimulation of the star, while
smaller areas responded to the rest of the body and the
limbs (Fig. 4). After mapping the locations of receptive
®elds from the star and body of the mole onto the somatosensory areas, the moles were perfused and their
cortex was ¯attened, sectioned parallel to the cortical
surface, and stained for the metabolic enzyme cytochrome oxidase, which often reveals cortical subdivisions.
In the area of cortex that contained an electrophysiological map of the star, a set of dark stripes was found
in brain sections processed for cytochrome oxidase
(Fig. 5). Each stripe corresponded precisely in location
to an area that responded to a single nasal ray in the
microelectrode recordings. Thus the representation of
the star in somatosensory cortex is visible in brain sections, much as the representation of whiskers in rodents
is visible as a set of cortical barrels in their cortex
(Woolsey and Van der Loos 1970). However, di€erences
between the barrel cortex in rodents and the star representation in moles were immediately apparent.
First, multiple maps of the star were seen in sections
of the mole's cortex. As in all mammals, each half of the
body is represented in the opposite cerebral hemisphere,
but in each mole hemisphere there were at least two
clearly visible sets of 11 stripes representing the contralateral star. In some favorable cases, a smaller third set
of stripes was also apparent (Catania and Kaas 1995,
1996). The most prominent set of stripes corresponded
to the primary somatosensory cortex, S1. The second set
Cortical proportions
Fig. 4 Cortical magni®cation in star-nosed moles. The upper drawing
shows the anatomical proportions of the various body parts. The
lower drawing of a ``moleunculus'' shows the relative size of each body
part as it is represented in somatosensory cortex. As would be
predicted from its high innervation density and behavioral importance, the nose dominates the cortical representation. The forelimb
also has a large cortical representation, probably re¯ecting its
important sensory role in the excavation of tunnels, rather than in
locating prey (from Catania and Kaas 1996)
of stripes was a mirror image of the ®rst, and corresponded to the second somatosensory area, S2 (Catania
and Kaas 1995). S1 and S2 are somatosensory areas
found in all mammals (Kaas 1987, 1995), including rodents, but there is not a visible representation of each
whisker representation in S2 of rodents. The third set of
stripes also responds to tactile stimulation of the nose,
but it has not yet been fully mapped because of its small
size and lateral location in the brain. The presence of
two and perhaps three visible maps of the star in different somatosensory areas of the mole is unique, and
may be a specialization related to the very high innervation density and large number of mechanoreceptors in
the star.
Another unusual but not entirely unexpected ®nding
was a hugely enlarged cortical representation of the 11th
nasal ray. This was seen in electrophysiological recordings (Catania and Kaas 1995, 1996), and is also strikingly clear in the cytochrome oxidase pattern of stripes
visible in brain sections (Fig. 5). This ray often takes up
over 25% of the entire S1 cortical representation of the
nose, despite its relatively small size on the star. Behaviorally important sensory surfaces often have greatly
enlarged cortical representations relative to their anatomical size. But these enlargements are generally attributed to their higher innervation densities, rather than
speci®c in¯uences of behavior. For example, a direct
linear correlation has been found between the size of
cortical barrels in rodent somatosensory cortex and the
number of a€erents innervating the corresponding
whisker on the face (Welker and Van der Loos 1986). To
371
A.
b
4
3
5
6
7
2
8
1
10
9
11
B.
8
4
5
6 7
9
10
3
2
1
Average Area of Cortex
Per Afferent (µm2)
C
150
11
S1 Cortex per Fiber
(Afferent Magnification)
120
90
60
30
0
1 2 3 4 5 6 7 8 9 10 11
Ray Number
explore this relationship in moles, the area of each ray's
cortical representation was measured and compared to
its corresponding peripheral innervation (Fig. 5C).
Cortical representations re¯ect behavior,
rather than innervation density
While ray 11 was found to have a higher innervation
density than the lateral rays (Fig. 3), this increased
Fig. 5A±C The cortical representation of the star. A A rotated view
of the left side of the nose (dorsal is to the left, lateral is up) oriented to
match the nose representation in the primary somatosensory cortex
(S1) of the right hemisphere (below). B A tangential section through
layer 4 of the primary somatosensory nose representation in cortex,
processed for the metabolic enzyme cytochrome oxidase. The pattern
of 11 rays on the nose is mirrored in the brain as a series of 11 dark
stripes. Note, however, the 11th subdivision in the brain is greatly
enlarged relative to the size of the 11th ray on the nose. The
disproportionate representation of the 11th ray, and to a lesser extent
the adjacent rays, re¯ects their behavioral importance (Fig. 1). C The
average area of cortex in S1 per primary a€erent for each nasal ray.
The representations of the rays in cortex are not proportional to their
respective innervation densities. Rather, the 11th and surrounding
rays have greater areas of cortex per a€erent. The pattern of ``a€erent
magni®cation'' is very similar to the distribution of touches across the
nose during feeding behaviors (Fig. 1C), suggesting a possible role for
behavior in shaping these cortical areas during development. Plate C
from Catania and Kaas 1997b
innervation density did not account for the size of the
11th subdivision in somatosensory cortex (Catania
1995c; Catania and Kaas 1997b). Ray 11 contained
about 7% of the Eimer's organs on the star, received
about 11% of the nerve ®bers innervating the star
(accounting for the higher innervation density per organ), but took up about 25% of the cortical representation of the star. Thus ray 11 is specialized both in the
periphery and in the cortex, by a higher innervation
density and a larger cortical representation per a€erent,
respectively.
These ®ndings are di€erent from the results of similar
investigations in rodents (Welker and Van der Loos
1986), but nevertheless may be a common condition in
mammalian somatosensory cortex that is simply dicult
to quantify in most species where the areas of cortical
representations and corresponding innervation densities
of skin surfaces are dicult to measure. This preferential
magni®cation of primary a€erents has been termed
``a€erent magni®cation'', to distinguish it from the traditional term ``cortical magni®cation'' (Catania and
Kaas, 1997b). The latter is used to describe enlarged
representations of a skin surface in the corresponding
cortical representation, but does not take into account
innervation density. The pattern of a€erent magni®cation of the rays (Fig. 5C) is strikingly similar to the
distribution of touches across the mole's nose when exploring prey (Fig. 1C).
The relationship between peripheral innervation
density and cortical representational area has long been
debated for the visual system of primates (Malpeli and
Baker 1975; Drasdo 1977; Myerson et al. 1977; Perry
and Cowey 1985; Silveira et al. 1989; WaÈssle et al. 1989,
1990). The ®ndings in moles support the most recent
®ndings in primates (Azzopardi and Cowey 1993) which
report a disproportionately large cortical representation
of ganglion cells from the retinal fovea. The relative
degree of a€erent magni®cation is similar in the two
sensory systems (Catania 1995c). Thus there seem to be
parallels between visual system organization in primates
and somatosensory organization in moles, perhaps re-
372
vealing a general aspect of the cortical representations of
important inputs across di€erent sensory systems.
All of these ®ndings raise a number of additional
questions about star-nosed mole sensory biology. For
example, why have multiple cortical representations of
the nose, rather than one large representation? Does
adding new representations of the nose increase computational eciency through parallel processing of different aspects of touch? These representations are also
topographically interconnected both in the same hemisphere and across hemispheres to the representation of
the contralateral star. The ability to identify each respective ray representation in multiple maps will allow
future studies to determine the circuitry of this system in
detail (with the aid of anatomical tracers) to determine
the degree of parallel or hierarchical organization of the
cortical representation.
Another area of future research includes the functioning of Eimer's organs. The concise geometric arrangement of nerve terminals at the apex of each organ
may be used to detect the surface features of objects in the
10- to 100-lm size range. If so, the sensory world of the
star-nosed mole could include a new realm of perceptions
not usually considered salient to mechanosensation, such
as microscopic textures and surface features previously
considered impossible to detect by mammalian touch.
However, the most striking result of these studies is the
similarity between the patterns of behavior in moles
(Fig. 1C) and the pattern of a€erent magni®cation of the
rays (Fig. 5C). The sizes of the cortical modules representing the rays are not correlated with the anatomical
parameters traditionally assumed to be driving these
specializations, but rather are highly correlated with the
patterns of mole behavior (see Catania and Kaas 1997b
for more details). Here we may be able to examine the role
of behavior in brain development to see if it is in fact
driving the enlargement of cortical areas, or whether these
behavioral and anatomical parameters are somehow independently matched to one another during development.
Acknowledgements Thanks to Jon Kaas and Glenn Northcutt for
their support, guidance, and many fruitful collaborations in the
course of these studies. Neeraj Jain, Christine Collins and Melanie
Catania provided helpful comments on the manuscript. Special
thanks to Hrathkus and Bill Catania for their help capturing starnosed moles.
All experiments and animal care procedures were approved by
the Vanderbilt University Animal Care Committee and follow the
National Institute of Health guide for the care and use of laboratory animals.
References
Azzopardi P, Cowey A (1993) Preferential representation of the
fovea in primary visual cortex. Nature (Lond) 361: 719±721
Catania KC (1995a) The structure and innervation of the sensory
organs on the snout of the star-nosed mole. J Comp Neurol 351:
549±567
Catania KC (1995b) A comparison of the Eimer's organs of three
North American moles: the hairy-tailed mole (Parascalops
breweri), the star-nosed mole (Condylura cristata) and the
eastern mole (Scalopus aquaticus). J Comp Neurol 354: 150±160
Catania KC (1995c) Magni®ed cortex in star-nosed moles. Nature
(Lond) 375: 453±454
Catania KC (1996) Ultrastructure of the Eimer's organs of the starnosed mole. J Comp Neurol 365: 343±354
Catania KC, Kaas JH (1995) The organization of the somatosensory cortex of the star-nosed mole. J Comp Neurol 351:
536±548
Catania KC, Kaas JH (1996) The unusual nose and brain of the
star-nosed mole. Bio Science Rep 46: 578±586
Catania KC, Kaas JH (1997a) The organization of somatosensory
cortex and distribution of corticospinal neurons in the eastern
mole (Scalopus aquaticus). J Comp Neurol 378: 337±353
Catania KC, Kaas JH (1997b) Somatosensory fovea in the starnosed mole: behavioral use of the star in relation to innervation
patterns and cortical representation. J Comp Neurol 387:
215±233
Catania KC, Kaas JH, Northcutt RG, Beck PD (1993) Nose stars
and brain stripes. Nature (Lond) 364: 493
Drasdo N (1977) The neural representation of visual space. Nature
(Lond) 266: 554±556
Gould E, McShea W, Grand T (1993) Function of the star in the
star-nosed mole, Condylura cristata. J Mammal 74: 108±106
Kaas JH (1987) The organization and evolution of neocortex. In:
Wise SP (ed) Higher brain functions. Wiley, New York, pp 347±
378
Kaas JH (1995) The evolution of isocortex. Brain Behav Evol 46:
187±196
Malpeli JG, Baker FH (1975) The representation of the visual ®eld
in the lateral geniculate nucleus of Macaca mulatta. J Comp
Neurol 161: 569±594
Myerson J, Manis PB, Miezin FM, Allman JM (1977) Magni®cation in striate cortex and retinal ganglion cell layer of owl
monkey: a quantitative comparison. Science 198: 855±857
Nagorsen DW (1996) Opossums shrews and moles of British
Columbia. UBC Press, Vancouver, British Columbia
Perry VH, Cowey A (1985) The ganglion cell and cone distributions
in the monkey's retina: implications for central magni®cation
factors. Vision Res 25: 1795±1810
Silveira LCL, Picanco-Dinz CW, Sampaio LFS, Oswaldo-Cruz
E (1989) Retinal ganglion cell distribution in the cebus monkey:
a comparison with the cortical magni®cation factors. Vision
Res 29: 1471±1483
WaÈssle H, Grunert U, Rohrenbeck J, Boycott BB (1989) Cortical
magni®cation factor and the ganglion cell density of the primate
retina. Nature (Lond) 341: 643±646
WaÈssle H, Grunert U, Rohrenbeck J, Boycott BB (1990) Retinal
ganglion cell density and cortical magni®cation factor in the
primate. Vision Res 30: 1897±1911
Welker E, Van der Loos H (1986) Quantitative correlation between
barrel-®eld size and the sensory innervation of the whiskerpad:
a comparative study in six strains of mice bred for di€erent
patterns of mystacial vibrissae. J Neurosci 6: 3355±3373
Woolsey TA, Van der Loos H (1970) The structural organization
of layer IV in the somatosensory region (SI) of mouse cerebral
cortex: the description of a cortical ®eld composed of discrete
cytoarchitectonic units. Brain Res 17: 205±242