Original Paper
Brain Behav Evol 2004;63:1–12 Received: May 28, 2003 Returned for revision: July 11, 2003 DOI: 10.1159/000073755 Accepted after revision: July 24, 2003
Tactile Foveation in the Star-Nosed Mole
Kenneth C. Catania Fiona E. Remple
Department of Biological Sciences, Vanderbilt University, Nashville, Tenn., USA
Key Words and both distributions were closely correlated with the Somatosensory W Mechanosensory W Saccade W Touch W degree of cortical magnification of the appendage repre- Cortex W Mammals W Star-nosed mole sentations in primary somatosensory cortex (S1). Copyright © 2004 S. Karger AG, Basel
Abstract Star-nosed moles have a specialized somatosensory sys- Introduction tem with 22 mechanosensory appendages surrounding the nostrils. A pair of appendages (the 11th pair on the Many mammals with well-developed visual systems ventral midline) acts as the tactile fovea and is used for have a retinal fovea characterized by a high density of detailed investigations. Here we used a high speed video photoreceptors and correspondingly enlarged areas of the camera to document movements of the star while moles central nervous system for processing information from searched for small prey items. Mole foraging behavior this important area of central vision. In humans it has was very fast; the star, which is just over a centimeter in been estimated that only one ten-thousandth of the visual diameter, was touched to different areas of the environ- field is seen with full clarity [see Carpenter, 1988 for ment approximately 13 times per second. This suggests review] and therefore the eyes must be constantly shifted that a mole foraging without interruption could potential- in a saccadic, or jerky, manner in order to analyze a visual ly investigate 46,000 cm2 of surface area per hour. In 100% scene. Some echolocating bats have developed an analo- of 526 trials in which prey was identified and eaten, star- gous organization for their auditory system complete with nosed moles made rapid, saccadic movements of the star an acoustic version of a saccade. For example mustached to investigate the contacted prey with the foveal appen- bats (Pteronotus parnellii) have an acoustic fovea corre- dages. The movements of the star were similar to visual sponding to the 2nd harmonic of their echolocation pulses saccades in other species. Maximum velocity of the star [Suga and Jen, 1976; Suga, 1989] but echos returning during saccades was approximately 40 cm/s, and most from flying prey are often Doppler shifted to a frequency saccades were between 30 and 60 ms in duration. As in outside of the fovea’s narrow range. By constantly shifting the primate visual system, small corrective saccades were the frequency of their outgoing pulses (a behavior called often needed to accurately foveate. We quantified the Doppler shift compensation) bats can focus returning number of contacts different appendages made with prey echos on their acoustic fovea [Schnitzler, 1968] much like items of various sizes during each encounter and com- an eye movement centers important visual information pared this distribution to a previously proposed simula- on the retinal fovea. tion of star movements during prey encounters. The The more recent discovery of a mechanosensory fovea behavior pattern and the simulation produced similar dis- in the star-nosed mole [Catania and Kaas, 1997] provides tributions of contact between the appendages and prey, further evidence for a remarkable degree of convergence
© 2004 S. Karger AG, Basel Kenneth C. Catania, PhD ABC Vanderbilt University, Department of Biological Sciences Fax + 41 61 306 12 34 VU Station B 351634 E-Mail [email protected] Accessible online at: Nashville, TN 37235 (USA) www.karger.com www.karger.com/bbe Tel. +1 615 343 1079, Fax +1 615 343 0336, E-Mail [email protected] Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM in the design of sensory systems that handle large volumes A of complex information. Visual, auditory, and somatosen- sory specialists have all hit upon the same solution of div- iding the sensory system into a small, high resolution area for detailed analysis and a larger, low resolution area for scanning a sensory scene. An obvious advantage of this design is the conservation of processing territory in the central nervous system – a large proportion of which is devoted to the representation of the fovea. Although primate eye movements and the behavior of echolocating bats have been the subjects of numerous B C 1 cm investigations [eyes – Carpenter, 1988; Goldberg et al., Nose 2 Cortex 1990; Gaymard and Pierrot-Deseilligny, 1999; Schall and 3 1 Thompson, 1999; bats – Asanuma et al., 1983; Kujirai S1 Nose and Suga, 1983; Simmons, 1989; Suga, 1989], far less is 4 known about tactile foveation in the star-nosed mole. The star consists of 11 pairs of bilaterally symmetric appen- 5 dages surrounding each nostril (numbered 1–11 from the 6 9 8 7 6 5 dorsal most appendage). The tactile fovea is the pair of 7 11 10 4 3 2 relatively small, 11th appendages just above the mouth 10 11 (fig. 1). Here we have documented the movements of the 8 9 1 Fovea star during foraging behavior in the laboratory using a 2 mm 1mm S1 Nose high-speed camera. Our main goal was to provide a Representation description of the star movements in response to tactile stimuli (prey items) of various sizes to assess the proposed Fig. 1. The star-nosed mole and its sensory specializations. A A star- analogy (in the behavioral dimension) between the star- nosed mole emerges from its underground tunnel system showing the 22 appendages that ring the nostrils and the large clawed forelimbs nosed mole’s somatosensory fovea and the visual and used for digging tunnels. B Schematic of the star with the numbering auditory foveas described for other species. We also test system for the separate mechanosensory appendages. The 11th the predictions of a previously proposed simulation, or appendage acts as the tactile fovea. C The primary somatosensory model, of tactile foveation in the star-nosed mole [Catania cortex of the star-nosed mole contains a series of 11 stripes visible in and Kaas, 1997]. This simulation predicts the distribu- cytochrome oxidase processed tissue. These stripes are the areas where information from each tactile appendage is processed [see tion of contacts between the appendages and prey items of Catania and Kaas, 1995]. The anatomical reflection of each appen- different sizes that are encountered during foraging bouts. dage in the cortical map allows the area of each appendage represen- Finally, we revisit the organization of somatosensory cor- tation to be accurately measured and compared to star-nosed mole tex in star-nosed moles in relation to observed behavior behavior (see text). patterns.
Materials and Methods worm 1–6 mm in diameter) were placed. A high speed video camera (S-Series MotionScope, Redlake Imaging Corporation) was posi- For our behavioral recordings, four star-nosed moles were tioned below the glass plate and illumination was provided from trapped under PA scientific collecting permit number COL00087. below with two fiber-optic light sources. Behavior was generally The moles were housed separately in 38 ! 55 cm containers filled to filmed at a sampling rate of 250 or 500 frames per second and a a depth of 20 cm with moist peat moss. These home cages were con- shutter speed of 1/2,500 s. Data frames were stored digitally in a vid- nected by a 5 cm diameter plastic tube to a 26 ! 40 cm container eo buffer and then transferred to S-VHS videotape for archiving and filled with 6 cm of water. Fresh water was provided daily. The mole’s scoring. Digital images from high speed video were captured for fig- captive diet consisted mainly of commercially raised nightcrawlers ures from the VHS tapes using a Macintosh G4 computer and iMovie supplemented occasionally with small crayfish. Plastic tubes (5cm2.0 software (Apple Computer). All procedures conformed to Nation- inside diameter) connected the containers to a variety of Plexiglas al Institutes of Health standards concerning the use and welfare of chambers in which the behavioral observations and filming took experimental animals and were approved by the Vanderbilt Univer- place (fig. 2). The bottom floor of the Plexiglas container was a sity Animal Care and Use Committee. removable glass plate on which small prey items (pieces of an earth-
2 Brain Behav Evol 2004;63:1–12 Catania/Remple Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM Results basic pattern of search behaviors and the frequency of tac- tile foveation. The contact of different appendages to prey We filmed trials from the 4 moles as they encountered items was also quantified in 526 behavior trials for com- and ingested prey items (small segments of an earthworm) parison with previous measures of cortical magnification ranging in size from 1.0 to 6.0 mm. Our first goal in this in star-nosed moles [Catania and Kaas, 1997]. study was to examine nose movements to determine the Search Behavior Foraging behavior depends on the star, which is in con- stant motion as these moles explore their underground Prey habitat in search of food. Prey items in the field consist of a variety of small invertebrates found in the wetland hab- Entrance Item itat [Hamilton, 1931]. In the laboratory, we used small pieces of earthworms as prey items in order to document Filming Chamber the movements of the star during foraging. When explor- ing their enclosures and searching for food, moles made a series of high speed touches with the star (13 per second) to different areas of their environment. These movements VCR were too rapid to be accurately observed with the naked eye, but analysis of high speed video allowed us to docu- ment and measure the details of these behaviors. High Speed Digital Buffer The different components of star-nosed mole search Camera behaviors were stereotyped and showed little variation. Each contact of the star to the substrate or to a prey item consisted of brief but firm compression of the star against the object being touched. These contacts were very short, Fig. 2. The experimental set-up for filming movements of the star. lasting an average of only 26 ms (SD 5 ms, see table 1). Moles entered the filming chamber from a habit-trail tube connected Compression of the star against the surfaces being to their home cage. A high speed video camera with a macro lens was touched was clear from the movements of the star, the positioned below a glass plate that formed the floor of the chamber. Prey items were placed at various locations on the glass plate and the deformation of the appendages during contact, and from plate was cleaned between trials. Video frames were stored in a dig- the observation that blood was pushed out of the vascular ital buffer and then transferred to VHS tapes for later scoring. skin surface during contact.
Table 1. The average duration of different components of star-nosed mole foveation No prey contacted, ms 1st prey contact, ms Fovea touch touch saccade touch + touch saccade touch + saccade to fovea saccade
Mole 1 26 51 78 24 38 62 27 Mole 2 27 47 74 28 38 66 30 Mole 3 18 52 70 21 39 60 29 Mole 4 31 50 81 32 34 66 32 Mean 26 50 76 26 37 63 30 SD 5.4 2.2 4.8 4.8 2.2 3.0 2.1
Data are from 10 trials for each of moles 1 through 4 (n = 40). No prey contact shows data for saccades during search behavior without contact to prey. The touch duration, saccade duration, and combined touch and saccade duration are shown. 1st prey contact shows the same data for an encounter with a prey item (touch) followed by a star saccade to place the fovea in contact with the prey. The last column indicates the mean contact time made by the 11th, foveal appendages to prey.
Tactile Foveation in the Star-Nosed Mole Brain Behav Evol 2004;63:1–12 3 Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM Each brief touch with the star was typically followed by eye coil allows similar measurements to be made for a saccadic movement of the nose that repositioned the visual saccades [Robinson, 1963]. Star-nosed mole tactile mechanosensory appendages onto a different area of the saccades to foveate on prey items were surprisingly simi- environment. The movement portion of the behavior lar to primate eye movements, having a time course that lasted approximately 50 ms (SD 2 ms) but could be short- was typically between 20 and 50 ms (mean duration was er or longer depending on the distance the nose was 37 ms) – e.g., most human visual saccades have a duration shifted. Thus a single ‘unit’ of search behavior, consisting of 30–60 ms [see Carpenter, 1988]. The average velocity of a touch followed by a saccadic star movement, lasted an of the star was approximately 18 cm/s, but this was average of 76 ms (SD 5 ms) when no object or prey item divided into slower acceleration and deceleration compo- was being explored. This allowed for approximately 13 nents at the beginning and end of most saccades, with a touches per second if a mole was searching without inter- higher, peak velocity in the middle of the saccade ruption, although rates as high as 17 touches per second (fig. 3, 4). Saccades to the 11th appendages were not were observed. Saccadic movements of the star were always accurate, and thus sometimes an additional, cor- made almost entirely as the result of head movements, rective saccade was required (fig. 4). These shorter sac- however small movements of the star and groups of cades, as might be expected, were generally of shorter appendages are possible through muscles on the side of duration. the head that connect to the star through a series of ten- dons [Grand et al., 1998]. Distribution of Contacts to Different Appendages Exploratory behavior appeared to be random, as food The obligatory saccades described above resulted in a that was not contacted with the appendages of the star was higher number of contacts between prey items and the invariably missed and passed over, whereas contact with 11th appendages of the star than occurred for the lateral any part of the star was usually sufficient for detection of appendages that form the more peripheral, low resolution the prey item. The distribution of contacts to prey items sensory surface. This was also true for contacts with other with different appendages of the star (see below) was also objects of interest to the mole, such as distractors that consistent with random initial encounters, but once the resemble prey items [Catania, unpublished observations]. prey was contacted, subsequent movements of the star We were particularly interested in the distribution of con- consisted of predictable saccadic foveation to the 11th tacts that different appendages made to prey items be- appendages. cause a number of studies suggest that the cortical repre- sentation of tactile sensory surfaces might expand as a Foveation result of high levels of mechanosensory stimulation [Mer- In 100% of the trials in which food was detected and zenich and Jenkins, 1993; Xerri et al., 1996; Coq and Xer- eaten, star-nosed moles made mechanosensory saccades ri, 1998; Benuskova et al., 2001], and previous investiga- that positioned the 11th appendages onto the prey item tions suggested this might be true for star-nosed moles as for one or more discrete touches before the food was eaten well [Catania and Kaas, 1997]. In addition, the distribu- (figs. 3, 4). Interestingly, this stereotyped foveation for tion of touches that different appendages make to an further tactile exploration of prey consisted of the same object of interest has been predicted from a relatively sim- basic unit of behavior that was used during random ple simulation of star-nosed mole foveation movements searching. That is, contact with the 11th appendages of (fig. 5). This simulation assumed that the star would be the star during foveation consisted of brief but firm con- shifted such that the 11th appendages made contact with tact of the skin surface against the prey item (mean con- the perceived center of each prey item as detected from an tact time was 30 ms). This was then followed either by initially random encounter [see Catania and Kaas, 1997]. additional exploratory contacts, or by repositioning of the Given this assumption it was possible to predict how a mouth over the food item which was then grasped with schematized star would contact idealized, circular prey the front teeth. If the initial, random contact with the food items of different sizes. However the accuracy of this item happened to be with the 11th appendages, a second simulation has never been assessed. exploratory contact with the 11th appendages invariably Here we were able to quantify the number of times dif- followed before the item was taken into the mouth. ferent appendages made contact with actual prey items of By tracking the location of the star every 2 ms, we were different sizes for all 526 trials from the 4 subjects. This is able to measure the trajectory and velocity of the star dur- shown in graphical form in figure 6, which illustrates the ing foveation (for example, fig. 3 right panels) much as an average number of contacts made by the different ap-
4 Brain Behav Evol 2004;63:1–12 Catania/Remple Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM 5 28 ms 20 ms
2 mm
10 ms Vel max = 55 cm/s Mean Vel = 25 cm/s
2 mm 0 ms 5 60 ms
40 ms
32 ms Vel max = 24 ms 31 cm/s Mean Vel = 18 cm/s 0 ms 16 ms 2 mm 2 mm 12 ms 5
24 ms Mean Vel = 17 cm/s 20 ms
Vel max = 10 ms 34 cm/s 0 ms 2 mm 2 mm 5 Mean Vel = 0 ms 14 cm/s Vel max = 20 ms 28 cm/s
40 ms
52 ms 2 mm 2 mm
5 38 ms
30 ms
Mean Vel = 20 ms 15 cm/s Vel max = 40 cm/s
10 ms 0 ms 2 mm 2 mm
Fig. 3. Examples of saccadic movements of the star during prey cated at the end-point. Note the acceleration and deceleration compo- encounters. A–E show selected video frames captured at 500 frames nents at the beginning and end of each track and the period of maxi- per second with high shutter speeds (1/2,500 second). Each numbered mum velocity in the middle. These tracks resemble primate eye move- set designates a separate trial. The panels on the right illustrate the ments in many respects. Dots on the far right panels indicate the loca- entire track of the star during the saccade shown on the left, with time tion of the star every 2 ms (trials A, C, and E) or 4 ms (trials B and D). points labeled in milliseconds. The total time of the saccade is indi- Arrows mark the locations of small prey items in A, C, and E.
Tactile Foveation in the Star-Nosed Mole Brain Behav Evol 2004;63:1–12 5 Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM A 1 EN 580 2 3
42 ms 0 ms
30 ms 20 ms 10 ms
Mean Vel = 22 cm/s Vel max = 33 cm/s
2 mm
Touch 1 Touch 2 2 mm 4 5 6 H1
12 ms
Mean Vel = 17 cm/s Vel max = 21 cm/s 0 ms
2 mm Touch 3 2 mm EN 524 2 3 B 1 52 ms 40 ms
30 ms
Mean Vel = 15 cm/s Vel max = 38 cm/s 20 ms 0 ms
Touch 1 Touch 2 2 mm 2 mm 10 ms 4 5 6 H2
14 ms
0 ms Mean Vel = 22 cm/s Vel max = 33 cm/s
2 mm 2 mm Touch 3
Fig. 4. Two examples of saccadic movements of the star that required a second, corrective saccade to accurately foveate. A1–3. Selected video frames from one trial, showing the initial contact to the prey item and subsequent saccade. Right Panel: Illustration of the path of the star at 2 ms intervals. A4–6. Selected video frames showing the second short saccade that more accurately places the 11th appendages on the prey item. Right Panel: The path of the star at 2 ms intervals for the second saccade. B1–6. Conventions the same as above for a second trial, also including a short corrective saccade. pendages to a prey item during each encounter. The most they were ingested. In contrast, the average number of obvious result observed for all prey sizes is the greater contacts to small prey items for the lateral appendages average number of contacts made by the 11th appendage was between 0.04 and 0.07 (fig. 6A) – close to what would to the prey items. For the 11th appendage (from one side be expected from random encounters with prey. There are of the star) the average number of contacts per prey item 22 appendages making up the star, and thus each would was greater than 1.0 for each prey size, and this follows be expected to have a chance of approximately 1 in 22, or naturally from the observation that one or both of the a 4.5% chance of making the initial random contact. Of 11th appendages always contacted the food items before course a number of other variables would be expected to
6 Brain Behav Evol 2004;63:1–12 Catania/Remple Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM A B 2 1 3A
4
B 5 C 4.5 4 3.5 3 2.5 2 1.5 1 6 Relative size in mm 11 7 10
9 8
C 800 800 800 800 700 1.0 mm Prey 700 1.5 mm Prey 700 2.0 mm Prey 700 2.5 mm Prey 600 600 600 600 500 500 500 500 D Sum of Touches For All Prey Sizes 400 400 400 400 6000
Touches 300 300 300 300 200 200 200 200 100 100 100 100 5000 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Appendage Number Appendage Number Appendage Number Appendage Number 4000
Touches 3000
800 800 800 800 4.0 mm Prey 4.5 mm Prey 700 3.0 mm Prey 700 3.5 mm Prey 700 700 2000 600 600 600 600 500 500 500 500 400 400 400 400 1000 300 300 300 300 Touches 200 200 200 200 0 100 100 100 100 1 2 3 4 5 6 7 8 9 10 11 0 0 0 0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Appendage Number Appendage Number Appendage Number Appendage Number Appendage Number
Fig. 5. A simple simulation of the movements of the star relative to center of the prey during the first touch, the prey is often not centered prey items of various sizes [reproduced with permission from Cata- precisely on the 11th appendages for the second touch. The appen- nia and Kaas, 1997]. Using the areas of overlap between prey and dages that made contact with the prey item in each encounter were nasal appendages, and assuming the star is always shifted to position counted (see B). Any overlap of the prey item and an appendage was the 11th appendages on the perceived center of the object, the distri- scored as a touch. B The size of the simulated prey items used in the bution of contact to the appendages was determined for a series of model relative to the nose. C The sum of contacts across the appen- different sized prey. A Example of an encounter with a prey item dages for prey items of different sizes, weighted to be equivalent to a 2 mm in diameter. After first contact the star is shifted in location so total of 166 prey encounters for each prey size. The number of that subsequent touches are with the 11th appendage pair. The first touches was always highest for appendage 11, but as prey size touch was assumed to be at a random location on the nose, thus all increased, the appendages close to appendage 11 were heavily areas of the nose were equally weighted in the model. For this, a grid recruited. D The sum of touches across the appendages for all of the of points was superimposed on the schematic of the nose, and a single simulated prey items shown in B and C. This pattern of touches prey encounter was initiated at each point. This was considered across the nose is similar to the actual behavior pattern and the pat- the first touch. Because this area of overlap was often not the true tern of cortical magnification (fig. 6).
affect this outcome, such as the size of each appendage, ples see fig. 3B, D). Second, there should be a marked the area of the star first contacted (e.g., overlap of the prey increase in the contacts to prey items made by the areas item with adjacent appendages) and the generally forward adjacent to the 11th appendages (i.e., nearby appendages motion of the mole while searching. Given the potential 1, 9, and 10) as foveation to large prey items will necessar- effects of these other variables, the results seem consistent ily involve larger areas of the star surrounding the fovea. with a random search pattern. This was also evident in the simulation. Very large prey As prey size increases, two additional trends could be items would obviously be contacted by most of the star reasonably predicted and were evident in the simulation during foveation. (fig. 5). First, more frequent (average) contact between all Both of these trends were observed in the actual forag- of the appendages and the prey should occur as larger prey ing behavior of star-nosed moles (fig. 6). The average are contacted, as the initially random encounters are more number of contacts to the lateral appendages was lowest likely to involve multiple appendages (for actually exam- for the smallest prey items and steadily increased as the
Tactile Foveation in the Star-Nosed Mole Brain Behav Evol 2004;63:1–12 7 Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM A B C D 1.5 1.0 to 1.5 mm Prey 1.5 1.6 to 2.0 mm Prey 1.5 2.1 to 2.5 mm Prey 1.5 2.6 to 3.0 mm Prey
1.2 1.2 1.2 1.2
0.9 0.9 0.9 0.9
0.6 0.6 0.6 0.6
0.3 0.3 0.3 0.3
0.0 0.0 0.0 0.0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11
E F G H 1.5 3.1 to 3.5 mm Prey 1.5 3.6 to 4.0 mm Prey 1.5 4.1 to 5.0 mm Prey 1.5 5.1 to 6.0 mm Prey 1.2 1.2 1.2 1.2
0.9 0.9 0.9 0.9 Average Contacts Per Prey Encounter Average 0.6 0.6 0.6 0.6
0.3 0.3 0.3 0.3
0.0 0.0 0.0 0.0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Appendage Number Appendage Number Appendage Number Appendage Number
J
I ) 2 1.5 Average Contacts per trial 1200 S1 Cortex per Organ All Prey sizes Combined (Cortical Magnification) 1.2 1000
800 0.9 600 0.6 400 0.3 200
0.0 S1 Cortex Per Eimer's Organ (µm 0 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 11 Appendage Number Appendage Number
Fig. 6. The distribution of contacts that different appendages made ters (1/22 or 4.5%). However, as prey size increased there was more to prey items of different sizes in the behavior trials (reflecting con- frequent contact to the lateral appendages as there was greater over- tacts to appendages from 1/2 of the star). Data is summed from the 4 lap of prey items with multiple appendages during each touch. This animals. Panels A–H show results for different prey sizes, as labeled trend was also observed in the simulation. A second trend was the above the histograms. Several trends are apparent from the data, increasing number of contacts to appendages close to the somatosen- which is similar to data derived from the simulation in a number of sory fovea, as larger prey items overlap appendages close to the fovea respects. For all prey sizes, the 11th appendage most frequently con- after each saccade. The bottom left panel (I) shows the average num- tacted prey items, reflecting its role as the tactile fovea. The disparity ber of contacts to the different appendages for all prey items summed between contacts to the foveal appendages and the most lateral, for all moles. This distribution of actual use of the appendages to peripheral appendages was most prominent for the smallest prey contact prey is similar to the degree of cortical magnification of the items (A). For small prey items the average number of contacts to the different appendages (J). Number of trials: A 176; B 62; C 105; D 72; lateral appendages was similar to that predicted for random encoun- E 46; F 23; G 25; H 17 ; J 25. Bars = standard error of the mean.
8 Brain Behav Evol 2004;63:1–12 Catania/Remple Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM sizes of the prey items increased (see fig. 6A–H). Similar- particularly appropriate. In addition to this general simi- ly, the number of contacts to appendages 1, 9, and 10 larity between the star-nosed mole’s somatosensory sys- increased steadily as prey size increased and foveation tem and visual systems of other species, the details of star contacts began to involve the areas of the star surrounding movements are similar to eye movements in other mam- the 11th appendage. As predicted, at the largest prey sizes mals. Thus it seems appropriate to extend the analogy most of the appendages contacted each prey item during with the visual system to include the term ‘saccade’ when every encounter (fig. 6H). referring to star foveation, as these movements resemble Finally, the sum of contacts the appendages made to all visual saccades in their rapid, jerky motion (the term sac- of the different prey sizes was similar in both the foraging cade is derived from middle French for ‘jerk’ or ‘twitch’) data and in the simulation of star movements (fig. 5D, 6I). and they have a time course of 30–70 ms, very similar to Thus both the predicted results and the actual behavior that of most human eye movements [30–60 ms – see Car- pattern of the star-nosed mole show an overall trend in penter, 1988 for review]. Star-nosed mole saccades are use of the different appendages that is centered around often inaccurate, requiring short corrective saccades for the mechanosensory fovea (the 11th appendages) and ta- proper foveation (fig. 4) and similar corrective saccades pers to a minimum at the most peripheral appendages are characteristic of human eye movements of large am- (2–7) that are furthest from the fovea. When the results for plitude (140 degrees). Finally, star-nosed mole saccades all prey sizes are summed (fig. 6I) these trends in patterns of typically have an acceleration and deceleration compo- use for the different appendages closely mirror the pattern nent at each end, flanking a period of maximum velocity of cortical magnification observed in the primary somato- (fig. 3, 4). sensory representation of the star (fig. 6J) as determined in Similar components are typical of saccadic eye move- previous investigations [Catania and Kaas, 1997]. ments and it is interesting to compare some specific parameters between these two disparate sensory systems. The human eye moves at a maximum velocity of approxi- Discussion mately 700 degrees/s [Carpenter, 1988]. This parameter is difficult to compare directly to star movements, however The results reveal parallels between the mechanosenso- the velocity of the surface of the human eye can be esti- ry system of the star-nosed mole and the visual systems of mated from its typical 25 mm diameter. Thus given a cor- other mammals and these findings raise a number of responding circumference of about 8 cm rotating at a peak issues for discussion. For example, many of the details of velocity of 2 rotations per second (720 degrees/s) the sur- star-nosed mole foveation movements resemble the eye face of the human eye moves at a maximum velocity of movements of more visual species and this raises the roughly 16 cm/s. The peak velocity of the star during a questions of why such a system might have evolved, how tactile foveation was approximately 40 cm/s or about 2.5 star-movements are guided, and what other specializa- times the maximum velocity of the surface of the human tions may be found in the mole’s somatosensory system eye. Despite the many obvious differences between the that parallel visual system organization. In addition, the primate visual system and the star-nosed mole somato- distinctive pattern of stimulation for the central appen- sensory system, they share many features in common. dages compared to the more lateral appendages closely These results also raise the question of how star move- matches the previously described degree of respective cor- ments are guided. The superior colliculus (or optic tectum tical magnification for each part of the star [Catania and in non-mammals) plays a critical role integrating sensory Kaas, 1997] and this raises the question of how different inputs to guide eye and head movements in more visual sensory surfaces become magnified in cortical representa- species [see Stein and Meredith, 1993 for review] but this tions. Finally, the observations of star-nosed mole forag- structure is usually dominated by visual inputs. However ing allow some extrapolations to be made regarding the recent investigations in the star-nosed mole reveal a supe- ability of star-nosed moles to explore their environment rior colliculus dominated by somatosensory inputs and and this might help explain the utility of the unusual star. containing a greatly magnified representation of the star [Crish et al., 2003]. Thus rather than degenerating in the Star Movements versus Eye Movements absence of significant visual inputs, the superior colliculus The obligatory foveation movements made by star- of the star-nosed mole has been retained and might play a nosed moles during tactile discriminations suggest that major role in guiding saccadic movements using somato- the designation of the 11th appendages as a tactile fovea is sensory inputs. Receptive fields for neurons in the collicu-
Tactile Foveation in the Star-Nosed Mole Brain Behav Evol 2004;63:1–12 9 Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM lar representation of the star are large [Crish et al., 2003] of their primary (S1) cortical star representation to the compared to neurons in similar representations at the lev- fovea [Catania and Kaas, 1997], devoting the same el of the cortex [Sachdev and Catania, 2002]. Similar find- amount of central processing territory to the entire star ings have been reported for the visual system [McIlwain would result in a doubling of cortical area needed for the and Buser, 1968; Sterling and Wickelgren, 1969; Schiller star in S1. Of course other cortical (e.g., S2) and subcorti- and Koerner, 1971; Goldberg and Wurtz, 1972; and see cal areas would likely have to increase in size as well. The Stein and Meredith, 1993] and this is thought to reflect a net result would be a significantly larger brain and higher widely distributed, population code of sensory representa- overall metabolic costs. Clearly foveas provide an impor- tions in the colliculus, such that positional information tant conservation of CNS space, with the trade-off that can be accurately extracted from the population of neu- behavioral and sensory specializations for orienting the rons with large, overlapping receptive fields despite the fovea must be adopted. relatively poor coding of space available from individual neurons [McIlwain, 1976, 1991; Findlay, 1982; Findlay Function of the Star and Walker, 1999]. Although further investigation is Several conclusions can be drawn from the behavioral needed, the large receptive fields of neurons in the superi- data regarding the sensory abilities of star-nosed moles. It or colliculus of the star-nosed mole might reflect a similar- is clear, for example, that star-nosed moles can detect very ly distributed coding of position based on tactile inputs. small prey items (e.g., fig. 3A, C). This was true despite a The discovery of a tactile fovea in the star-nosed mole feeding regimen in the lab that consisted of large night is evidence of convergent evolution in the design of com- crawlers supplemented with occasional crayfish. Yet no plex, high-resolution sensory systems suggesting there are training was required for moles to locate and consume important benefits for this configuration. Visual, audito- very small prey items. It seems reasonable to conclude ry, and somatosensory foveas share a number of features from these observations that star-nosed moles include in common, most obviously the division of the sensory small prey in their normal diet, and indeed the remark- system into a large, low resolution area for scanning sen- able efficiency with which they can identify and ingest sory inputs and a small, high resolution area used for small prey suggests they specialize to some extent on these detailed analysis of a sensory scene. The retinal fovea and items. This conclusion is also supported by gut content the star-nosed mole’s mechanosensory fovea act in an analysis of wild-caught specimens [Hamilton, 1931]. analogous manner – through physical movement of the Thus one obvious function of the star is to make rapid receptor surface (saccades) during foveation. The audito- and very accurate sensory discriminations in a complex ry fovea of echolocating bats is less intuitively similar; it is tactile environment. The usefulness of this ability is clear only when Doppler shift compensation is considered that from the individual behavior trials documented here, as it surprising parallels with visual and somatosensory sys- was rare for a mole to miss potential prey once it had been tems are apparent. Foveation in bats is necessary because contacted with the star. In addition to these impressive echos returning during flight are invariably Doppler discriminatory abilities, the foraging speed of star-nosed shifted to different degrees depending on the relative moles allows us to make some additional observations. speed of the bat and the target being probed. Bats ‘foveate’ For example, while moving through their tunnel systems by altering the frequency of their outgoing echolocation star-nosed moles can contact approximately 13 different pulses such that the Doppler shifted returning echo pre- areas per second. The utility of this rapid behavior cisely matches the narrow frequency range of their acous- becomes clear when one considers the amount of territory tic fovea [Schnitzler, 1968; Suga et al., 1987; Metzner, that a hungry mole can cover over time. 1989; Gaioni et al., 1990]. The star has a surface area of approximately 1 cm2 and A major advantage of subdividing sensory systems into therefore a mole can investigate roughly 13 cm2 of tunnel high resolution and low resolution areas is the conserva- surface per second. Extrapolating to 1 h of foraging, a tion of processing territory in the central nervous system. mole could in theory explore over 46,000 cm2 of tunnel Given that only about 1/10,000th of the human retina has surface. Of course in reality foraging behavior is inter- maximal resolution and the representation of this area rupted by other behaviors such as grooming, consuming takes up a large proportion of many cortical and subcorti- captured prey, tunnel construction, rest, and other activi- cal visual areas, it would require a manifold increase in ties. Nevertheless, if a star-nosed mole foraged continual- human brain size to convert the entire retina to the resolu- ly for only one fifth of its day it could in theory cover over tion of the fovea. In star-nosed moles, which devote 25% 200,000 cm2. This is an impressive surface area that could
10 Brain Behav Evol 2004;63:1–12 Catania/Remple Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM potentially encompass over 300 m of tunnel floor each sistent with a number of studies of primate brain organi- day [assuming prey is primarily located on the lower half zation that suggest levels of peripheral tactile stimulation of a tunnel 4 cm in diameter – Hickman, 1982]. These can alter the size of representations in S1 [Merzenich and considerations highlight the utility of the star, particularly Jenkins, 1993; Xerri et al., 1996; Coq and Xerri, 1998; for a species that lives in wetlands rich in small inverte- Benuskova et al., 2001]. However a recent investigation brates [Hamilton, 1931]. of developing star-nosed moles revealed the early devel- opment of the 11th appendages in embryos, and the corre- Appendage Use during Foraging Bouts and Cortical sponding early maturation of sensory receptors (Eimer’s Magnification organs) on these appendages [Catania, 2001]. This could A previous investigation of star-nosed mole foraging provide a competitive advantage for the foveal area of the suggested that contact with the appendages to prey items star in capturing cortical territory during early critical is concentrated on the 11th appendages and tapers off to a periods of development, before the adult patterns of minimum at the lateral appendages [Catania and Kaas, behavior described here are exhibited. Thus it is possible 1997]. These observations were based on relatively few that intrinsic developmental mechanisms play the major behavior trials. Nevertheless, these data suggested a simu- role in determining the relative sizes of cortical represen- lation, or model, of star movements that could predict tations, largely independent of behavior patterns. Alter- how the different appendages contact prey (fig. 5). Here natively, a combination of intrinsic developmental pro- we have extended these preliminary studies and con- grams and early behavior patterns could contribute to the firmed the basic premise of the simulation. Star-nosed development and maintenance of patterns of cortical mole invariably foveate to the 11th appendages and this magnification. Star-nosed moles might help determine results in a predictable pattern of preferential contact to the relative roles of these two influences on cortical devel- prey items with the fovea and nearby areas of the star opment. (fig. 6). When a range of prey items is considered (fig. 6I) these patterns of contact across the star appendages closely mir- Acknowledgements ror the pattern of cortical magnification of the star in pri- Special thanks to Sandra Schurman for help observing and mary somatosensory cortex (fig. 6J). This correlation be- recording behavioral data. This research was supported by NIH grant tween cortical representational size and behavioral pa- R01 MH58909 and the Searle Scholars Program (to K.C.C). rameters suggests a causative link, and this would be con-
References
Asanuma A, Wong D, Suga N (1983) Frequency Coq JO, Xerri C (1998) Environmental enrichment Gaymard B, Pierrot-Deseilligny C (1999) Neurolo- and amplitude representations in anterior pri- alters organizational features of the forepaw gy of saccades and smooth pursuit. Curr Opin mary auditory cortex of the mustached bat. J representation in the primary somatosensory Neurol 12:13–19. Neurophysiol 50:1182–1196. cortex of adult rats. Exp Brain Res 121:191– Goldberg ME, Wurtz RH (1972) Activity of superi- Benuskova L, Rema V, Armstrong-James M, Ebner 204. or colliculus in behaving monkey, I. Visual FF (2001) Theory for normal and impaired Crish SD, Comer C, Marasco PD, Catania KC receptive fields of single neurons. J Neurophy- experience-dependent plasticity in neocortex of (2003) Somatosensation in the superior collicu- siol 35:542–559. adult rats. Proc Natl Acad Sci USA 98:2797– lus of the star-nosed mole. J Comp Neurol: In Goldberg ME, Colby CL, Duhamel JR (1990) Re- 2802. Press presentation of visuomotor space in the pari- Carpenter RHS (1988) Movements of the Eyes. Findlay JM (1982) Global visual processing for sac- etal lobe of the monkey. Cold Spring Harb London: Pion Limited. cadic eye movements. Vision Res 22:1033– Symp Quant Biol 55:729–739. Catania KC (2001) Early development of a somato- 1045. Grand TE, Gould E, Montali R (1998) Structure of sensory fovea: a head start in the cortical space Findlay JM, Walker R (1999) A model of saccade the proboscis and rays of the star-nosed mole, race? Nat Neurosci 4:353–354. generation based on parallel processing and Condylura cristata. J Mammal 79:492–501. Catania KC, Kaas JH (1995) Organization of the competitive inhibition. Behav Brain Sci 22: Hamilton WJ (1931) Habits of the star-nosed mole, somato-sensory cortex of the start-nosed mole. 661–721. Condylura cristata. J Mammal 12:345–355. J Comp Neurol 351:549–567. Gaioni SJ, Riquimaroux H, Suga N (1990) Bioson- Hickman GC (1982) Influence of the semiaquatic Catania KC, Kaas JH (1997) Somatosensory fovea ar behavior of mustached bats swung on a pen- habitat in determining burrow structure of the in the star-nosed mole: behavioral use of the dulum prior to cortical ablation. J Neurophy- star-nosed mole (Condylura cristata). Can J star in relation to innervation patterns and cor- siol 64:1801–1817. Zool 61:1688–1692. tical representation. J Comp Neurol 387:215– 233.
Tactile Foveation in the Star-Nosed Mole Brain Behav Evol 2004;63:1–12 11 Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM Kujirai K, Suga N (1983) Tonotopic representation Robinson D (1963) A method of measuring eye Stein BE, Meredith MA (1993) The Merging of the and space map in the non-primary auditory movement using a scleral search coil in a mag- Senses. Cambridge, MA: MIT Press. cortex of the mustached bat. Auris Nasus La- netic field. IEEE Trans Biomed Eng 10:137– Sterling P, Wickelgren BG (1969) Visual receptive rynx 10:9–24. 145 fields in the superior colliculus of the cat. J McIlwain JT (1976) Large receptive fields and spa- Sachdev RN, Catania KC (2002) Receptive fields Neurophysiol 32:1–15. tial transformations in the visual system. In: and response properties of neurons in the star- Suga N (1989) Principles of auditory information- Neurophysiol II, Vol 10, pp 223–248 (Porter R, nosed mole’s somatosensory fovea. J Neuro- processing derived from neuroethology. J Exp ed). Baltimore: University Park Press. physiol 87:2602–2611. Biol 146:277–286. McIlwain JT (1991) Distributed spatial coding in Schall JD, Thompson KG (1999) Neural selection Suga N, Jen PH (1976) Disproportionate tonotopic the superior colliculus: a review. Vis Neurosci and control of visually guided eye movements. representation for processing CF-FM sonar sig- 6:3–13. Ann Rev Neurosci 22:241–259. nals in the mustache bat auditory cortex. McIlwain JT, Buser P (1968) Receptive fields of Schiller PH, Koerner F (1971) Discharge character- Science 194:542–544. single cells in the cat’s superior colliculus. Exp istics of single units in superior colliculus of the Suga N, Niwa H, Taniguchi I, Margoliash D (1987) Brain Res 5:314–325. alert rhesus monkey. J Neurophysiol 34:920– The personalized auditory cortex of the mus- Merzenich MM, Jenkins WM (1993) Reorganiza- 936. tached bat: adaptation for echolocation. J Neu- tion of cortical representations of the hand fol- Schnitzler HU (1968) Die Ultraschallortungslaute rophysiol 58:643–654. lowing alterations of skin inputs induced by der Hufeisennasen-Fledermäuse (Chiroptera, Xerri C, Coq JO, Merzenich MM, Jenkins WM nerve injury, skin island transfers, and experi- Rhinolophidae) in verschiedenen Orientie- (1996) Experience-induced plasticity of cuta- ence. J Hand Ther 6:89–104. rungssituationen. Z Vergl Physiol 57:376–408. neous maps in the primary somatosensory cor- Metzner W (1989) A possible neuronal basis for Simmons JA (1989) A view of the world through tex of adult monkeys and rats. J Physiol Paris Doppler-shift compensation in echo-locating the bat’s ear: the formation of acoustic images 90:277–287. horseshoe bats. Nature 34:529–532. in echolocation. Cognition 33:155–199.
12 Brain Behav Evol 2004;63:1–12 Catania/Remple Downloaded by: Weizmann Inst. of Science 149.126.78.33 - 1/26/2016 10:32:20 PM