THE JOURNAL OF COMPARATIVE NEUROLOGY 258:281-296 (1987)

Variability in Hand Surface Representations in Areas 3b and 1 in Adult

Owl and Squirrel Monkeys

MICHAEL M. MERZENICH, RANDALL J. NELSON, JON H. KAAS, MICHAEL P. STRYKER, WILLIAM M. JENKINS, JOHN M. ZOOK,

MAX S. CYNADER, AND AXEL SCHOPPMANN Coleman Laboratory and Departments of Otolaryngology and Physiology, University of

California at San Francisco, San Francisco, California 94143; Department of Psychology, Vanderbilt University, Nashville, Tennessee 37240

ABSTRACT Detailed microelectrode maps of the hand representation were derived

in cortical areas 3b and 1 from a series of normal adult owl and squirrel monkeys. While overlap relationships were maintained, and all maps were internally topographic, many map features varied significantly when exam­ined in detail. Variable features of the hand representations among different monkeys included a) the overall shapes and sizes of hand surface represen­tations; b) the actual and proportional areas of representations of different skin surfaces and the cortical magnifications of representations of specific skin surfaces, which commonly varied severalfold in area 3b and manyfold in area 1; c) the topographic relationships among skin surface representa­tions, with skin surfaces that were represented adjacently in some monkeys represented in locations many hundreds of microns apart in others; d) the internal orderliness of representations; e) the completeness of representa­tions of the dorsal hand surfaces; and f) the skin surfaces represented along the borders of the hand representation.

Owl monkey maps were, in general, internally more strictly topographic than squirrel monkey maps. In both species, area 3b was more strictly topographic and less variable than was area 1.

The degree of individual variability revealed in these experiments is difficult to reconcile with the hypothesis that details of cortical maps are ontogenetically specified during a period in early life. Instead, we propose that differences in the details of cortical map structure are the consequence of individual differences in lifelong use of the hands. This conclusion is consistent with earlier studies of the consequences of peripheral nerve tran­section and digital amputation, which revealed that cortical maps are dy­namically maintained and are alterable as a function of use or nerve injury in these monkeys CMerzenich et al., '83a,b, '84a; Merzenich, '86; Jenkins et al., '84; Jenkins and Merzenich, '87).

Key words: monkey, hand, somatosensory representations, cortex, area 3b, area 1, representational variability

Accepted September 17, 1986.

R.J. Nelson's present address is the Department of Anatomy, University of Tennessee, Memphis, TN 38163.

J.M. Zook's present address is the Department of Zoological and Biomedical Science, Ohio University, Athens, OH 75701.

© 1987 ALAN R. LISS, INC.

M.S. Cynader's present address is the Department of Psychol· ogy, Dalhousie University, Halifax, Nova Scotia.

A. Schoppmann's present address is the Abteilung Verglei· chende Neurobiologie, Oberer Eselsberg, Universitat Ulm, 7900 Ulm, Germany.

282

It has long been known that cortical sensory and motor representations are not identical among all individuals of a species. Some of the variability observed may arise from variation in the placement of cortical fields with respect to sulcation patterns rather than from differences in the inter­nal topography of cortical maps, but some variability exists within cortical maps as well (Tusa et aI., '78, '79; Albus and Beckmann, '79). While not all the sources of variability are known, it is likely that some of the variability is due to differences in peripheral innervation (Lee and Woolsey, '75): for an extreme example, mice that lack a particular vi­brissa (Van der Loos and Dorfl, '78). From the literature it is difficult, however, to determine the extent of variability of the internal features of cortical maps. Knowledge of the extent of this variability is prerequisite to understanding its sources, and of the factors that give rise to cortical maps.

We have measured this variability by taking advantage of the large number of detailed maps of the normal repre­sentation ofthe hands in areas 3b and 1 that were obtained in the course of plasticity studies on owl and squirrel mon­keys (Merzenich et aI., '83a,b; Wall et aI., '83, '86). In this report, we describe the internal organizations of these cor­tical fields, with special emphasis on the variability of the map structures among individual adult animals.

MATERIALS AND METHODS Surgery and recording procedures

The cortical mapping procedures employed in these stud­ies have been described in several earlier reports (Merzen­ich et aI., '78, '81; Nelson et aI., '80a). Experimental animals were neurologically intact owl and squirrel monkeys with intact hands. The monkeys were obtained from a variety of sources. Animals were anesthetized either with ketamine hydrochloride i.m. followed by a small single dose of sodium pentobarbital i.p. immediately before surgery, and then maintained with supplemental ketamine i.m. or with so­dium pentobarbital i.v. The hand representations in areas 3b and 1 were exposed by a wide parietal craniotomy, and a photograph of the brain surface vasculature was taken and printed at 25-50 x for siting microelectrode penetra­tions. The exposed cortex was bathed in a well of liquid silicone (dimethyl poly siloxane) throughout the course of the experiment. The monkeys' temperatures were main­tained at 38°C.

Glass-coated platinum-iridium and glass-, parylene-, and lacquer-coated tungsten electrodes were used in different experiments. Electrode penetrations in each experiment were approximately normal to the cortical surface and were parallel to each other. "Minimal receptive fields" were de­fined for neurons or for small clusters of neurons in the middle cortical layers using fine hand-held blue glass probes to produce just-visible indentations of the skin. Receptive fields were carefully drawn on enlarged illustrations con­structed from photographs of hand surfaces. Earlier exper­iments have shown that defined receptive fields change little across the strongly driven middle cortical layers (ap­proximately 500-1,200 jlm in depth), so that a single de­fined receptive field is representative for neurons throughout any vertical penetration (Powell and Mountcas­tle, '59; Paul et aI., '72; Merzenich et aI., '78, '81). The same criteria for delineating receptive fields were used through­out these studies.

M.M. MERZENICH ET AL.

Penetration density and map construction In the earliest experiments, maps of areas 3b and 1 were

derived from the receptive fields defined in 150-250 pene­trations into these two fields in each individual monkey. Cytoarchitectonically defined boundaries between areas 3a and 3b, 3b and 1, and 1 and 2 were determined, and were found to correspond approximately with functional borders, defined either between cortex devoted predominantly to deep-receptor and predominantly to cutaneous-receptor-in­puts (for 3a-3b and 2-1) or by a reversal in the sequence of cutaneous representation (for 3b-l). In more recent studies, maps were derived in more detail. Between 250 and 450 penetrations were used to map the area 3b or area 1 hand representation. The majority of these more recent normal maps were limited to the zones of representation of the fingers in area 3b, and the borders of the area 3b represen­tation were usually defined electrophysiologically but not cytoarchitectonically.

Data from a representative experiment are shown in Fig­ures 1 and 2. The penetration grid is illustrated in Figure 1 (top). The sequences of receptive field centers are illus­trated in Figure 1 (bottom), and many ofthe receptive fields for this case are shown in Figure 2. The penetration density in this map is representative of all of the owl monkey maps illustrated in this report. Squirrel monkey maps were de­rived at 1.2-1.8 x this grain. The construction of topo­graphic maps from this data has been described previously (Merzenich et aI., '78, '81; Nelson et aI., '80a). Regions with receptive fields centered on a given body part were outlined (e.g., see Fig. 1). The positions of the borders between body parts on the map are biased to reflect the relative distance of each receptive field defining the border from the bound­ary of the body part represented. Thus, a border between D2 and D3 is drawn closer to the penetration site yielding a receptive field on D2 when one of the receptive fields that define this border is on the extreme ulnar aspect of D2 and the other is midway along the radial face of D3. All mea­surements of representational areas were made with a dig­ital planimeter.

Reproducibility and accuracy of cortical maps We have attempted in two ways to determine the ade­

quacy ofthe microelectrode sampling methods used in these experiments for determination of map areas. First, the re­producibility of our receptive field determinations for a given cortical locus was evaluated. In experiments reported in the accompanying paper (Stryker et aI., '87), one part of area 3b was mapped 2 or 3 times over the course of a single day in individual monkeys. Different experimenters de­fined most receptive fields in at least one different map repetition in each experiment. Each of these repeated maps was made without knowledge of the previous findings. Re­ceptive fields defined at nearly identical map locations un­der different conditions of anesthesia by different experimenters were closely overlapping. These results in­dicate that the peripheral receptive field defined for given map points on the cortex are reproducible.

Second, given this evidence that receptive fields are rela­tively stable and are accurately defined, we considered the precision with which representational territories can be determined from microelectrode sampling points spaced in a grid that is coarse (usually 50-300 jlm) compared to the size of cortical neurons. Mathematical analyses and com-

CORTI CAL MAP VARIABILITY

3

11 26

,!2 102

126 154 ,-

...-184

216 ;;;; 2

26 9

29 5"-

47

Verti ca l Rows

1~

" , , \

1 2 3

2 L-../

,'t: --, -- "[

" , , -r ,

, 'T , , , ,

, ~ -- - --t

4 5 6

v02 / 03 10 ,/

25 -r- - -- -<- ~- ,

, , , , ;;.

.-, , '- " , ,

" , , , , ~ ~ , 180 , ,

~ , -, --- - - '

7 8 9 10 11

46 ;{04

, 71 , 101 , 125 ,

183

~

12 13

153 ~ :179

215

245

268

f 294

320

14

283

Fig. 1, Penetration grid for II repreBentali\"1I map oflh .. &"'. oI'represen. datum oft!>is report. In the hand drawings al the bottom, thoU;!eII ofcenwno t.anon oIlhe hand within aNa 3b. Each I)'mool reprel!enta !.he site of II ofn!OOpth-e fields are indicated by the filled circle.. Filled lIquarC8 represent microelectrodc penetration. The '-crtical burs are 300 I'm apart. Diamond, penetration .it" in which neurons were drh'en from deep inpllt$. Open lila .. , and large dot:. ",!)resent penetration. with neurons having rea:ptive IIQUlln!S represent penetrations in which realpliW! fields were centered over field!; limilPd 1.0 the ,labl"OUII !fUrface ofdigitll2. 3. or". I'f!Speo;liwly. OQhed donal ikln Mlrface an'!1lIl, Dotted "'luare:& repraent pen.etrations in .... hich liMII around c1 ..... ~ of e~rode penelraliOlUl for eao:h dlll'l ,lIu!ll.nlte the refilptive field. were on lhe face. ....".Lulilln of boundoriel IO' ilhin the nul", lhat ~htute the principal

puter simulat ions demonstrated that only small errors in senting, for example, a pal'ticular d il{it clln be determined estimations of rcpl'csentational areas are likely, given sam· from OUt' maps by placing the border of the digit represen· piing !:.'Tids of the dens iLy used in these studies. The differ· tation through a U t.he recording sites e it.her just within t he enccs in area a mong the hand representations of the representation (for a minimum area), or just outside the monkeys of this study rar exceed the possible sampling representation (for a maximum area), The true border must errors. The maximum error in determining the area repre- lie between these two extreme borders. In the digit repre·

284 M.M. MEnZENICH ET AL.

'4

5 '

Fig. 2. All reo:epti"e fielda defined in penelratiollB within the 3()O."m· wide "vertical rowI" demarcated in Figure 1. Numbers in thelle drawings COITeI!pOnd to the vertital row numbers in Figure I. Thi, typical map was limited to the area 3b hand .urface repreMmtation; the mOfll extreme PI'OJ<' imal alI~ oI'the palm waa inoomple~ly mapped ,

COHTICAL J\tAP VARlABILITY

sentations shown in Figure 1, these maximum a nd mini­mum arens d iffer from the borders drawn by about 20%_

It can be shown geometrically that the maximum error in the determination of a compact a rea for a n evenly spaced grid of sample sites nearly equals 2 x n-'\ where n is t he number of siles within t he representation whose area is to be measured. This maximum error, however, would occur only extremely rarely, as it is statistically unlikely t ha t t he true border wou ld lie a t t he same extreme position relative to all of the points t hat determine it . The precision of the area determina tion is measu red by the standard deviation of repeated deterrninations of t he same lrue area . This a rea measure nltmL is influenced, of course. only by t he points along t he perimeter of the area. The number of t hese perim. ete r points is proportional ro area lit . From information the­ory, we know that the precision of a determination based on repeated imprecise measurements increases with the square root of the number of measureme nts. Hence, t he standard deviation is proportional to (area"')'" 0 1' area v.. The coefficient of variat ion, which expresses the variation as a fraction of the mean area, is the standard deviation divided by the mean. Hence, the coefficient of variat ion of area equals Vo divided by a rea, which equals area"'. 'rh is relation is ,brj"en by sampling theory and holds t rue irrespective of whether t he sampling grid is evenly spaced.

This predicted statistical distribution of measured areas was confirmed by a Monte Carlo method. In this computer simulation of the microelectrode mapping process, the coef­ficient of variation of the measured area was equal to 0.64 X (mean area)"'" (r "" 0.996). In actual practice. the situa. tion is somewhat better for two reasons: 1) the experimen. ters someti mes biased the sampling grid for defining the locations of interesting borders, a nd 2) horder definitions marked by receptive field overlap were biased at each loca· tion to renect the degr'Cc of t ha t overlnp. Thus, these bor. ders are determined morc precisely t han they would be by the random sampling of L"quivalenL density assumed in the foregoing ana lysis .

Given the more conservative equivalent.-.density model. the consequence of this degree of variability for our exper i­mental measurements of area illustrated in Fibrure I are as follows: 1) For the digit 2 representation, consisting of 36 sites, the coefficient of var iation of t he area measurements would be 4 .3%, yield ing95% confidence limits that the arca as drawn was L'OI'I'eCt to within ± 8,4%. 2) For the digit 3 representa~ion, with 35 sites, the variation wou ld be 4,4%, giving 95% confide nce limits of ±9,4 % of the measured area. We conclude t hat errors in the meas urements of rep­resentational areas associated with the spacing of these microelect.rode gr ids are smaU compa red to the map varia· fiollS recorded in this study.

RESULTS The major findings of this paper are drawn from IIlSpec.

tion and measurement of t he maps ill ustrated in Figures 3, 4, 6, and 7. The paragraphs below poi nt out the constant

Fig. 3. Pa.ttern. of ~pre8e ntlt tion of the hand in cortical area 3b in () adult owl monkcya. Outlined ZQnl.'!l a.re the cortical area.! in whicb all recepth-e lielclll an! centered on the skin surfScetl dl.'3ignllted in the hand drowing at the ri ght. Arell!l of repTe8entation of dorsa l hand .urfltCJe!l are o:roas.hBtehl!d in each drawing. The tytoar'ehilectonic area 3b---.re& 1 bonier ... 1llI delined in lIM< mape . ho,,·n in I) and }:, but not in A. C.

285

AREA 38 OWL MONKEY

286

AREA 38

, L.m

OWL MONKEY (cont.)

w

Fig. 4. HlI.od represenlatiQnJI in 3b in 4 adult owl monkeya. 001">181 repre­!lentationlll zones e r&~ros&hDtclled. The arell 3b-1 horder was anatomically d~fined ail shown. for the mllj)JI illUlitroted in F and G. Abbreviations <lTlj

marked on Lh .. hand drawing Inset in Figure 3.

a nd variable features of these maps, tabu.late their quanti­tativeaspects, and display th~ regularities in receptive field sequences that underlie t hese maps.

Common features of area 3b maps of the hand among adult owl a nd squirrel monkeys

Figures 3, 4, and 6 show nearly complete maps of area 3b derived in a series of adult owl and squirrel monkeys. Several featul'es were common to all maps: a) There was a

M.M. MERZENICH ET AL.

DIGITAL REPRESENTATION ZONES

Ow, ""O"k., ••• , • • 3b

<!: 1.5 1-o ~ • • z w • .., 1 g: E 1.0 w • " " o • w

" •

h oL~~~~~~~~~ __ 12345 12345

GLABROUS DORSA L

Fig. 5. Mean OreaS of repre".mtation of digital glabrous and doraa l skin 8urfa<*S in the nine adu lt owl monkeys whO&! maps are shown in Figul"elO:I and 4. The rll llgll!l of cortical represen tation areas ate illdieatN by thin ~ertical1ineB.

single and complete or nearly complete representation of the glabl"Ous hand surface in area 3b. b) Tips of digi ts 2-4 were, with one exception, represented along the extreme rostral margi n of the field. c) The digits were always repre­sented in orderly sequence. d) An overall internal topo­graphic order was maintained th,'oughout the field, with very few receptive fields out of oopographic sequence. e) The dorsal surfaces of digit 1 (and somet imes 2 and sometimes 3) were represented continuously in the most lateral aspect of most of these maps, as was the dorsum of digit 5 (and sometimes digits 4 and 3) in t he most medial aspect. 0 Dorsal skin surfaces were represented in a number offrag­mentary, small , discontinuous patches amid a predomi­nantly glabrous digita l surface representation.

Varia ble features of a rea 3b hand maps Maps of hand surfaces were a lso highly individual with

numerous variable features. r ile IJ/ltlined configu ration of fhe hand represenlatiQnai

area. Note the idiosyncratic shapes of the hand represen· tational zone in a rea 3b in both species.

The total cortical ana of representation of the hand. The hand representation in Figure 3E. for example, is approxi­mately .55 x the area of representation of the hand in the case illustrated in 3D. In the 9 owl monkeys illustrated in Figures 3 and 4, the glabrous digits occupied from about 3.2 mm2 (in E) to about 5. 1 mm:! (in D). Similarly, t he hand a nd digital representations shown in t he t h ird squirrel monkey map in Figure 6 were only about 50-60% of the size of the representations in the first monkey illustrated..

Tile overall tern'tories of representatio" of different hand su rfaces, Note, for example, the differences in the area of representation of t he individual digi ts in area 3b in differ­ent monkeys. T he territory of representatioll of the thumb is only 0.5 mm2 in owl monkey C (Fig . 3); it is L4 mm2 in owl monkey B. The ranges of areas of digit surfaces in a rea

CORTI CAL MAP VARIABILITY

AREA 38 Squirrel Monkey

w

287

3b in both owl monkeys and squirrel monkeys, and for palmar sufaces in area 3b in squirrel monkeys are i,ndicated by t he thin vertical bars in Figures 5, 8, and 10. Roughly twofold differences have been seen in comparing the corti· cal areas of representation of most glabrous surfaces.

ProportilllJaI areas of r llpreselliation of g iven halld sur­faces, When the di("rit representations of diffe,'ent mon· keys were normalized by exp,'essing them as a fr action of the total area of all the 3b digit representations in each animal , strikingly different proportional differences in rep· resentation are still present. This can be a ppreciated by a comparison of cortical maps A, B, D, F, a nd H in Figures 3 and 4, which are all nearly the same overall size. The magnitude of size differences in areas of representation in this group of monkeys is nearly as great as the full range of absolute a real var iation noted above. Similarly great proportional differences were seen in area 3b in squirrel monkeys. Note, for example, the large differences in ten'i­tories of representation of different hand surfaces between the first and fifth or second and third monkeys' maps in Figure 6. The two in each pair are about the same overall size.

Cortical maglli/ica1ifm fIf representation of differellt skin sIIrfaces. Repl'esentationa l magnification is defined as the area of cortical territory of representation divided by that of the ski.n surface represented. Magnifications fol' all digits of owl and squirrel monkeys illustrated are summarized in Table f. Of course variations in magnification directly re­flect variations in actual cortical representational area, discussed above.

The exlellts of pate li es representing different dorsal digi, tal surfaces, alld the extclits of skin surfnCelJ rep resented i ll those patches. One of the most striking features of these maps is the idiosyncratic a ppearance of the small patches of cortex representing dorsal skin surfaces (cross· hatched zones in Figs. 3, 4, 6). This feature was not noted in our inilial descri ptions of the hand surface representa· tion in area 3b in t he owl monkey (Merzenich et aI., '78), which were based on maps del'ived with definHion on only one-third to one-fifth the grain of present maps. We then described the relatively large, conti nuous dorsal hand rep· resentations along the lateral and medial margins of the hand representational zone, but missed the small but inter­nally continuous islands of dorsal skin representat ion that lie amid the larger glabrous digital ski n surfaces. No such patches have ever been seen within the palmar represen, tationa l zone.

These patches of digit dorsum rep"esentation varied markedly in size among t he different digits of the same monkey. as well as among tbe representations of t he same digit in different monkeys. For example, note the territory of representation of dorsal digit 2 (0.600 mm2) in owl mon· key I (Fig. 4), as compared with the nonexistent represen· tation of the dorswn of digit 3 in the same monkey and the ni nefold diffe rence in the territory of representation of dol" sal digit 1 vs. dorsal digit 5 in monkey G (Fig. 4). Note also the wide range of variabi lity of the territory of the dorsum

Fig. 6. Mapa of the cortical representatiOn!! of the hand in area 3b of 5 adult !!quirrel monkeys. Abbreviations as in Figure 3. hand inset. DoI'SRI skin ropresentational ZOneft are cros.a·halched; the territ(lrieft of repre!lenta· lion of the glabrouo """aces of the digit;; are marked by diagonal lines. The Ill't'll 3b-~rell 1 borders were defined On llnIIt(lnticlll baJ!/'ll in all iIlUlltrated ell""",

AREA 1 Squi r re l Monkey

w

L.,

Fig. 7. Map' or the corti",,1 areal of rtp~ntDtion of hand .,,",teII in corticalllnl8 1 in.five adult Bquil'l'tll monkeys. Abbrevintions 111 in Figure 3. Jk,1'>Il,l1 n:llrt!llentational ~U)1'"Ij ore IIhllded: glabrous digi tal n::pr~nla· lionu! art"" IiTII cr ....... ·lull.chcd.

of the same digits in different monkeys, shown ill Fibrure 5 . There, about 13·fold variations in the territories of repre­sentation of dorsal digi ts 2-5 wel'e evident. In t he squirrel monkeys studied, these dorau! patches were even less con­stant, and dorsal representations for some fingers seemed to be absent.

M.M. lUERZENICH ET AL.

DIGITAL REPRESENTATIONAL ZONES SquIrre l Monkeys

Z Q .... '" .... Z W "'::' WE a: E "-W 0 a:.:.:;. u. 0

'" W a:

'"

3.

2.

2.

I.

1.

.5

AREA 38

1 2 3 -4 5 DIGI TS

AREA 1

1 2 3 4 5 DIGITS

Fig. 8. Mean a reS$ of rep._nlation of the glabrom digital surfaces in 1I"'9.lI Sb and I in 5 adull squirrel monkeys. The range. of 8"'83 are i"di~ated by thin vertitallines.

Even in maps from monkeys in which representations of a ll dorsal digi ts were found, those representations were h ighly individual and /l ever complete. Examples fro m sev­eral owl monkeys are shown in Figure 9. There, dotted zones illustrate the entire skin surface represenl.ed in the Isq,'Cr, conti nuous dorsal representational zones on the lat,. eral and medial margins of the hand representation. All of the re<:eptive fields defined in penetrations within islands of dorsal skin representation a mid the glabrous digital zone are outlined. The representation of the hand dorsum in the monkey shown in B is one of the most compieU! seen in this series. Those shown in A and C are more ty pical. Note as well the marked local differences in the extents of represen· talion of dorsal skin surfaces. Some were represented strongly, oLhers weakly, others not at all. Note also that in some instances skin surfaces were represented both in patches and in t he lateral s nd medial zones.

Topological neighoorhood reiutiOlUlhipi! i" tile maps. Many variants of map topology were evident. Most striking is the split representation of the glabrous thumb in owl monkeys B, 0 , F, H, a nd I (Figs. 3, 4) in compa rison with the continuous representation in other c8SCS. Discontinui­ties in A and I were especially st.riking, because interven· ing distances were Isl1.'C, and the representation of digit 2 intervened. In both cases, the more caudal zone represented only proximal phalangeal sunaces, whiJe the more rostral zone represented distal pha langeal. sUlfaces. Similar split glabrous sunace representations were seen on digits 2 and 3 on monkeys A, G, and I and on digit 5 in F' (Figs. 3, 41.

CORTICAL MAP VARIABILITY

TABLE 1A. Area 3b: Owl Monkeys'

Glabrous digits Hairy digits

Experiment No. 2 3 4 5 2 3 4 5

A .67 1.30 1.22 1.12 .88 .56 .22 .20 .12 INC2

B 1.41 1.08 1.15 .93 .66 .23 .33 .08 .07 INC C .56 1.11 .86 .88 .61 INC .35 .16 .14 INC D .88 1.00 1.32 1.44 .75 .26 .42 .31 .05 INC E .74 .79 .72 .74 .52 .33 .22 .19 .13 INC F .80 .92 1.32 1.39 .68 .28 .44 .14 .04 .33 G .53 .77 .97 .82 .49 .64 .28 .09 03 .36 H .71 1.36 1.16 1.11 .88 .22 .13 .16 0 .10 I .72 1.06 1.22 .99 .48 INC .60 0 INC INC Mean .78 1.04 1.10 1.05 .66 .36 .33 .15 .07 .26 SD .26 .20 .21 .24 .15 .17 .14 .09 .06 .14 Skin surface 1.00 1.84 2.09 2.01 1.43 NV5 NV NV NV NV

areas4

Cortical .78 .57 .53 .52 .46 NV NV NV NV NV magnification6

IAll cortical representational area measurements are in mm2. 2Representational territories were incompletely mapped. These values not included in means. 3No area of representation identified at this grain of mapping. 4Mean of three normal adult monkeys. Values are cm2. 5No value obtained or not measured. 6Cortical magnification = cortical representation -.:- skin surface area * 100.

TABLE IE. Area 3b: Squirrel Monkeys

Glabrous digits p,! Palmar pads

Experiment No. 2 3 4 5 PTH P2 P3 p. PIN PHY

79-35 1.53 1.82 1.88 1.25 .77 .94 .39 .21 .32 .44 .13 79-30 .90 1.40 1.30 .82 .84

, .38 .28 .14 .17

, 79-34 1.29 1.44 1.45 1.27 .69 .14 .14 .38 .16 .13 79-11 2.29 2.43 2.80 1.52 INC 1.32 .67 INC INC .15 79-73 1.68 1.90 1.74 1.06 INC 1.10 .34 .19 .23 .15 Mean 1.54 1.80 1.83 1.18 .77 1.12 .38 .21 .27 .21 .13 SD .51 .42 .59 .26 .08 .19 .19 .06 .11 .13 0 Skin surface .57 .98 1.15 1.12 .81 1.30 .44 .43 .53 .61 1.42

areas Cortical 2.70 1.84 1.59 1.05 .95 .86 .86 .49 .51 .34 INC

magnification

IPTH and PI fields sometimes broadly covered both pads, i.e., sites of centers of receptive fields could sometimes not be resolved. In other cases, PTH-P1 and PHY representational territories were incompletely mapped. These values were not included in means.

TABLE 1C. Area 1: Squirrel Monkeys'

Glabrous digits P,' Palmar pads

Experiment No. 2 3 4 5 PTH P2 P3 p. PIN PHY

79-35 .17 .41 .63 .70 .83 1.82 .37 .48 .48 .25 INC 79-30 .30 .41 .72 1.05 .98 .59 .17 .17 .06 .36 79-34 .32 .17 .27 .48 .37 .57 .33 .77 .21 .33 79-11 1.24 .94 1.33 1.30 .96 1.96 INC INC INC INC INC 79-73 .10 .26 .39 .49 NV 2.48 .50 .57 INC .90 .24 Mean .43 .44 .67 .80 .79 2.09 .51 .39 .47 .36 .06 SD .46 .30 .41 .36 .28 .35 .10 .18 .30 .37 .06 Skin surface .57 .98 1.15 1.12 .81 1.30 .44 .43 .53 .61 1.42

areas Cortical .75 .45 .58 .71 .98 1.61 1.16 .91 1.09 .57 .22

magnification

'See footnote to Table IE.

289

Split representations of digit 5 have also been seen in other normal monkeys that were mapped less completely.

The skin surfaces represented across borders between digits or digits and palm varied considerably. The base of the glabrous digit 2 representation in owl monkeys, for example, in some cases bordered a hairy surface represen-

tation of the digit (cases B, F); or the first pad, PI (I); P2 (C, D, G, F); or, more equally, P2 and PI (A, E, H) (Figs. 3,4). Similar variability was seen in this feature in the squirrel monkey. This can be readily appreciated by scanning the posterior (in the drawing, lower) border of the digit 2 rep­resentation in the five examples illustrated in Figure 6.

290 M.M. MERZENICH ET AL.

Dorsal Skin Represented Area 3b (owl monkey)

Fig. 9. All dorsal skin surfaces represented in three relatively completely mapped hand surface representations. The dotted zones indicate the skin surfaces represented in continuous dorsal represen­tational wnes along the lateral and medial aspects of the hand representations. Outlined receptive fields are all fields defined in discontinuous islands of dorsal skin representation within the larger, predominantly glabrous representational zone.

The relationship ofthe hypothenar and fourth palmar pads with digits 4 and 5 was highly variable and virtually idio­syncratic in both species.

A number of other examples could be cited; many such topological features, when considered in detail, varied as much.

The skin surface represented along the borders of the hand surface representation

The hand-face border. The glabrous digit 1 margin may be predominant along this border (e.g., owl monkeys B, H) or on the other extreme, not approach the border at all (A). Digit 2 may be represented along part of the border (owl monkeys B, C, G, I), or may not approach within hundreds of microns of it (A, D, E, F, H). Similarly, dorsal digit 2 representations may lie along a lengthy border sector (owl monkey I; squirrel monkeys A and E); or be nowhere near the border (owl monkeys A, G, H; squirrel monkey D).

The area 3b--area 3A border. This border, along the rostral margin of area 3b, was occupied by representations of digit tips in most monkeys, but dorsal representations (often limited to nail bed fields) were also seen along this line (e.g., owl monkey I; squirrel monkeys, B and E). In a few cases (squirrel monkey A, and in several less com­pletely mapped normal cases) palmar fields were also found along a limited sector of the rostral area 3b border.

The area 3b--area 1 border. The exact sequence of skin surfaces along the same glabrous digital border was, again, almost idiosyncratic in detail. In some monkeys, digit sur­face representations extended down to the borders (e.g., digit 4 in owl monkey E, and several less completely mapped owl monkeys; digits 3 and/or 4 and/or 5 in the squirrel monkeys A, C, E). In such instances, the palmar

PALMAR REPRESENTATIONAL ZONES

Squirrel Monkeys

z o I­<C

1.51- AREA 1

~ 1.Ot- AREA 38 w w_ ~ E a.. E w c: c:: .-u. o <C w c:: <C

-.51-

+ o~~~~~~--~~~~~~-

P, P, P, p. ~ P,N P, P, P, p. ~ P,N

PALMAR PADS PALMAR PADS

Fig. 10. Mean areas of representation of the palmar pads in areas 3b and 1 in five adult squirrel monkeys. Abbreviations as in the inset in Figure 3. Thin vertical lines indicate the ranges of measured areas in these 5 individuals.

CORT ICAL MAP VARIABIL ITY

INTERNAL MAP ORDER

2 3

,5 7

" ,,-'orv6

B '>0 •

" •

Fill . 11 . Inll!rJUlllOpogTaphy of hand aurfaee rnapi' in lhrH intena;"ely mllpped adult owl monkf:ya. Line!! tnlck the centen of ~Ive field!; rerorded for .. U pt!nelnotiona ,,~thin 300.,.m.wide •• lpl! acroD the hand ~p'--nlat}on. 1'h_ litOpi' are 81ig!led in the aagitta! ph.". in Ih .. draw. inp at the Jell, and in the. orthogonal fron .... 1 plllJ>e in the dru"';nga at lhe ri¥hl.

INTERNAL MAP ORDER (CONT.)

5 3

"

, 5 7

.,

,

,

291

Fill· 12. Inlernll l t.opogrophy or hand lI .. rfaoo maps (continued ), All akin surfaces represenlA.>d in all penetrations in 3OO'l'm.wide cortiCAl strips ~ ing area 3b liN! indicated by the "utlined lind dotted lire .... Fie ld>; from alternating IJlrips are ~hown. that is, outlined or dOlUK! woe!! represent .very fourth conkel strip. Drawings at the left R1'e from I'OII1rocaudal1y (lTiented bands: drawings at the right lin! from med iolaleral1y oriented bends.

representation was spli t, wit h t he most ul nar aspect rep­resented in a separate zone medial to the digital surface representation in such monkeys.

The halld-wrist border. Similar indi vidual var iability a ppeared to exist at the ha nd-wrist border, al though th is was not completely mapped in all cases of Lhis experimen­tal series.

Thus, nearly a ll details of a rea 3b map structure vary in individual monkeys. As will be seen below, a n overall conti nuous interna1 topography is (with some exceptions) preserved, but regional territorial distri butions are hjghly variable.

Map variability in area l Variabili ty in area 1 maps of the hand in adult squirrel

monkeys is illust rated in Figure 7. Severa l maps of area 1 in norma l owl monkeys (not illustrated) have also been defi ned in a pproximately equal detail. As shown in Figure 8, area 1 maps were far more var iable t han were area 3b maps, a nd t he variable features described fo r maps of area 3b applied even more emphatically to area 1 maps. For example, note t he d iffe rence in the overall areas of repre­sentation of the hand between t he first and third monkeys

292

of Figure 7. Both absolute and proportional areas of repre­sentation and cortical magnifications of specific hand sur­faces were dramatically different in different monkeys. Compare, for example, the territories of representations of the glabrous surfaces of digit 1 in the second and third with the fourth monkey in Figure 7; the manyfold difference in representation of the insular pad between the third and fifth; the large territories of representation of the radial and hand dorsum seen in some monkeys (e.g., first and fifth) but not others (e.g., the fourth); the overall territories of representation of the glabrous digits (compare first and third monkeys with fourth and fifth); or of the dorsal digits (compare second and fourth monkeys). There were also highly idiosyncratic patterns of representation of the dorsal hand.

Major differences in topographic neighborhood relation­ships were evident in the area 1 maps. Note the separation (by hundreds of microns) of the representations of glabrous digits 1 and 2 in the first three cases; while these digits were adjacent in the last two. Note as well the inverse relationship between pads 1 and 2 in the second and third monkeys. Many more such examples could be cited. Skin surfaces represented along the borders of the area 1 hand representation were also highly individual.

In area 3b, representational magnifications were rela­tively similar across the glabrous digits (usually not vary­ing by more than a factor of 2). In area 1, however, there was in some cases a disproportionately greater magnifica­tion of the representation of the more ulnarward digits.

Internal order in cortical maps While there were many differences in the area and topog­

raphy of the representation of different skin surfaces, most such variations occurred nonetheless with preservation of topographic neighborhood relationships. That is, glabrous skin representations were almost invariably internally con­tinuous, even when their order was examined in finest ~etai.l. This feature is illustrated by three examples, shown III FIgure 11. They were selected for illustration simply because they were the three most intensively studied nor­mal owl monkeys, with maps of area 3b defined by 340-420 penetrations. In this figure, dotted lines trace the axis of the centers of receptive fields plotted for electrode penetra­tions within contiguous 300-JLm-wide strips crossing the mapped sector of area 3b. Note a) the overall orderliness of receptive field sequences; b) the continuous shifting of re­ceptive field centers all across these maps in both axes· and c) the relatively few anomalies in rec~ptive field se: quence locations (usually on the representation of thumb). In most regions, glabrous finger representational zones were continuous along these lines. However, in some instances (described above) dorsal patches completely separated rep­resentations of more proximal from more distal sectors.

In addition to overall topographic continuity, overlap of receptive fields was relatively constant within each of areas 3b and 1. Examples are shown in Figures 12 and 2. There, outlines of all receptive fields from alternate 300-JLm-wide strips of cortex are shown. Thus, the nearest receptive fields illustrated in adjacent strips were from penetrations 300-600 JLm apart. Note the similarity in the separation of different vertical and different horizontal rows consistent with a rule of constant receptive field overlap.' (Sur et aI., '80a). The proportion of overlap is roughly equivalent across both mediolateral and rostrocaudal cortical axes.

M.M. MERZENICH ET AL.

DISCUSSION The basic organization of the representations of the body

surface has been defined in the parietal somatosensory strip in a number of primate species (see Woolsey et aI., '42; Powell and Mountcastle, '59; Werner and Whitsel, '68; Pu­boIs and Pubols, '72; Paul et aI., '72; Dreyer et aI., '74, '75; Krishnamurti et aI., '76; Merzenich et aI., '78; Sur et aI., '80b, '81, '82; Nelson et aI., '80b; Iwamura et aI., '81, '83; C~rlson and Welt, '81; Carlson and Fitzpatrick, '82; Fitzpa­trIck et aI., '82; McKenna et aI., '82). Although individual differences have been noted, none of these studies have dealt directly with map variation. Such comparisons have been difficult because, with only a few recent exceptions, these maps were derived with use of a relatively coarse sampling grid. Further, in most mapping investigations there has been the inherent assumption that the map in anyone individual applies for all individuals of the same species. Commonly, data from several animals were com­bined within a single homuncular or territorial outline representation. A main goal of these studies has been to produce such a composite for each studied species.

This study reveals that when maps are defined in detail, they have striking idiosyncratic features. What might be the meaning of such variation? Variation of this degree constitutes a serious challenge to some conventional views of nervous system organization. In this discussion we first review what varies and what does not vary in these maps; relate these results to those of earlier studies; consider the possible sources of cortical map variability; consider the implications of the partial representations that are evident for different submodal inputs and for the dorsum of digits; and finally, review the implications of observed map varia­bility for concepts of map generation in development, and for functional cortical operations in adults.

What is constant in cortical maps? What is variable?

When considered on a large grain, the cortical represen­tation of the hand in areas 3b and 1 in adult monkeys has constant features. Thus, for example, the digits and palmar pads are represented in an orderly radial-to-ulnar lateral­to-medial sequence, with the digit tips directed toward the rostral border of area 3b and toward the caudal border of area 1. However, considered on a finer grain, all map fea­tures varied substantially. Thus, the actual and propor­tional areas of representation varied severalfold in area 3b and sometimes manyfold in area 1. Cortical magnification~ varied similarly. Internal topographies also differed, with representational zones that bordered each other in one map sometimes separated by hundreds of microns in other mon­keys. There was especially great variability in dorsal hand skin surface representation in these fields.

What, besides an overall general topography, is constant across these individuals? All representations were locally topographic. As a rule, overlapping but shifted receptive fields were recorded at neighboring cortical position. The possible significance of these constancies coupled with rep­resentational territorial variations is discussed below.

Relationship to earlier studies of primate cortical map organization

Our initial description of the organization of the represen­tation of the hand in area 3b of the owl monkey was based

CORTICAL MAP VARIABILITY

largely on two relatively complete maps of the hand, de­rived at about one-third to one-fifth ofthe grain of definition of the present maps (Merzenich et aI., '78)_ This description of the internal organization of area 3b (and of area 1) has been confirmed in the present more complete study, with three significant exceptions_ First, in all more detailed maps, islands of representation of dorsal digital skin sur­faces have been found amid the large glabrous digital rep­resentations, in both owl and squirrel monkeys. None were noted in initial maps. Second, there is substantial individ­ual variation in the shapes, areas, and, to a lesser extent, the neighborhood relationships of different skin surfaces, not fully appreciated in initial studies. Some of these indi­vidual differences are actually greater than typical inter­specific differences between owl and squirrel monkeys. Third, the dorsal digital surface representation does not appear to be complete in either area 3b or 1, in at least a number of intensively studied monkeys.

Surprisingly, since our initial maps were derived in these and other primate species, investigators using other proce­dures have failed to confirm this basic organization (Iwa­mura et aI., '81, '83; McKenna et aI., '82; Carlson et aI., '82). Some investigators have hypothesized that the maps derived in our studies have been substantially biased by anesthetic artifacts (Duncan et aI., '82; McKenna et aI., '82; but see accompanying paper, Stryker et aI., on this issue), with our earlier results being interpreted as being other­wise compatible with a single map of the body extending across areas 3a, 3b, 1 and 2. Recently, Carlson et aI. ('82) concluded that area 3b contains a representation of the glabrous surfaces of the digits, while area 1 provides the dorsal digital representation. We have now mapped the representation of the hand in area 3b in over 20 normal owl and squirrel monkeys, and, in equal detail, in more than. 30 addi~ional owl and squirrel monkeys following restrIcted perIpheral or central lesions. Everyone of these hand representation maps has been derived on a grain ~ev~r~l to many ti~es finer than in maps generated in any mdividual monkey m the contrary studies. Moreover, there are great technical advantages in defining map structures in a cortical sector in which the representation is spread entirely across a flat cortical surface, rather than partially buried within a deep central sulcus. Both the owl and squirrel monkey hand representations offer these advan­tages. The variation in internal map structure also demon­strates that any combination of data from different monkeys to produce a common map of the representation (e.g., as in the studies of Whitsel et aI., '71; Iwamura et aI., '81, '83; McKenna et aI., '82) will necessarily result in an obfusca­tion of map detail. The problem of map definition is also ~ot solved com~le~el! by deriving relatively complete maps m one or two mdividual monkeys, because of substantial individual variation.

These problems may be aggravated when a map is de­rived from recordings made in an individual monkey over a long period of time. Our own plasticity studies indicate that two further difficulties may be encountered in such experiments: 1) Inaccuracies are introduced in precisely locating the neurons studied with respect to the cortical surface. 2) The assumption that the receptive fields of mon­keys are static over long periods of time is questionable (see Jenkins et aI., '84; Merzenich et aI., '84b; Jenkins and Merzenich, '87; Merzenich, '87).

293

What are the sources of cortical map variability? Why are somatosensory cortical maps so variable? Inher­

ent errors in mapping procedures are significant, but they cannot account for most aspects of the variability noted in this study (see Methods).

The two absolute constancies in the hand maps are 1) at the largest scale, a constant overall orientation, and 2) at the very finest scale, a strict internal topography. The vari­able t~polo~es of digit representation in the maps can be reconclle~ wIth these constancies, at least on a formal level, by ass~mmg that, after the coarse orientation of the map is estabhshed, the process that gives rise to the fine structure of maps acts only on the very local scale and does not include a mechanism for maintaining topology. The next section presents one proposal for such a process, but any of a ~eat many. mechanisms that maintain the peripheral neIghbor relatlOns among inputs to the cortex could account for these findings equally well (for example Fraser '80 '85). ' "

The variable areas of digit representation present more of a problem. One is reluctant to believe that the consider­able variation in area of cortical representations is a mere reflection of variation in peripheral innervation. This is espec~any the case for the dorsal surfaces of the digits, some of whIch appear to lack an area 3b representation entirelv. It is, however, difficult to construct a model that maintai;;'s l~cal topography accurately but gives rise to representa­tional areas that do not accurately reflect innervation den­sity of the input, because the forces that drive such models toward ~arge-scale o~ganization also tend strongly to spread out the mputs to umform density (e.g., Fraser, '80, '85). The proposal made in the next section is the only model we kn~w that can easi.ly account for variable area of represen­tatlOns together wIth the constancies that are observed in normal maps, and it has the virtue of accounting as well for the findings of plasticity experiments.

On the other hand, we know nothing about the genetic or acquired variability in cutaneous innervation among these new world monkeys. So it remains possible that the differ­ences in cortical map area among normal animals do reflect differences in the periphery. Under this hypothesis the normal variability would not result from the mechan'isms that are active in plasticity studies.

Another possibility is that the variable maps we see in the adult animal reflect the fixation of a possibly variable functional map in early development. Such mechanisms ha~e been pr~posed for visual and vibrissal systems, in whIch dramatic alterations in the cortex can be produced by limiting the early pattern of peripheral input (Woolsey and Wann, '76; Rubel et aI., '77; Killackey and Belford '79' LeVay et aI., '80; Woolsey, '80; Movshon and Van Slu;ters' '81; Sherman and Spear, '82; Simons et aI., '84). There is n~ r~~so~, however, to expect any particular degree of varia­bIhty m the early functional map, and this hypothesis does not explain findings of adult plasticity studies.

A proposal to account for variability and plasticity of cortical maps

We propose that there are two major, interlocked sources of this map variability. First, other studies that we have conducted suggest that these adult cortical maps are shaped by experience, and subject to alteration by use throughout

294

life (Merzenich et aI., '84b; Merzenich, '87). Thus, map structure is substantially altered following peripheral nerve transection (Merzenich et aI., '83a,b), digital amputations (Merzenich et aI., '84a), or restricted cortical lesions (Jen­kins et aI., '82; Jenkins and Merzenich, '87). Recent experi­ments have directly revealed changes in map detail resulting from different, special uses of the hand by an adult monkey (Jenkins et aI., '84; Jenkins and Merzenich, '87). Thus, we hypothesize that details of functional map structure are largely created and are alterable by experience (Merzenich et aI., '83b, '84a,b) and that the forms of these adult maps at least to a significant extent reflect the con­sequences of different lifelong patterns of hand use in these monkeys. Note that, with few exceptions, although map details are substantially variable, internal topographic re­lationships and shifting overlap relationships are pre­served. We have hypothesized elsewhere that this active maintenance of overlap is probably controlled by temporal synchronization or sequencing of inputs (Merzenich et aI., '83b, 84a,b; Merzenich, '87). The preservation of overlap relationships in the face of substantial map variability it­self constitutes evidence that differences in map detail are largely determined by organizing activities through and/or after a critical developmental period.

Second, developmental studies indicate that cortical con­nections are almost certainly not precisely genetically pre­determined, as had been earlier hypothesized (e.g., see Sperry, '63). To the contrary, early addressing of connec­tions may be guided by cell-adhesion molecules that can effect the generation of only relatively crude local topogra­phies (see Edelman, '83, for review). Numerous studies in­dicate that these initially crude connections are refined during an initial period of development (see Harris, '81; Sherman and Spear, '82). Both the initial relatively crude specification of connections and variable early experience might contribute to variable features of cortical maps re­corded in adults.

We have hypothesized that even after this developmental shaping there is still a substantial neuroanatomical spread of inputs in somatosensory cortex, i.e., that the "neuroana­tomical map" in the adult monkey is crude. Therefore, the detail of cortical maps is a product of a process of selection of effective driving inputs by experience (Merzenich et aI., '84b; Merzenich, '87). By this process, there are very many possible detailed forms of functional representation that can be created over time. The potential for variability is limited by the degree of divergence and convergence of anatomical inputs (Merzenich et aI., '84a,b; Merzenich, '87), but within such limits considerable variation is possible. This view is consistent with the theory of neuronal group selection, postulated by Edelman (,82).

In models in which maps are believed to organize along a gradient (as opposed to their being specified genetically point by point), it has been argued that one margin of gradient is likely genetically specified (see Horder and Mar­tin, '78; Dykes, '78). It is interesting to note, then, that skin surfaces represented along all borders of the hand surface representation are highly variable.

What is the significance of the incomplete representation of dorsal hand surfaces in areas 3b and I? How can evidence for slowly adapting and quickly adapting columns in area 3b be consistent with evidence for a single, simple representation of

the glabrous hand within this field? One perhaps unexpected feature of these maps is the

fragmented and often incomplete representations of the

M.M. MERZENICH ET AL.

dorsal digits in areas 3b and in area 1 in many individual monkeys. Their disposition suggests that a topographically continuous or complete overt representation of these sur­faces is not prerequisite for what area 3b contributes to aspects of perception attributable to this field.

Studies of plasticity suggest that the dorsal digital skin surface representation is special, in several respects. Thus a) a complete and very highly detailed map of the dorsal digital representation arises in the map after removal of glabrous digital inputs (Merzenich et aI., '83a,b); b) dorsal digital representational zones have been seen to be occu­pied by glabrous digital inputs, following restricted cortical lesions (Jenkins et aI., '82; Jenkins and Merzenich, '87); c) in a few instances, neurons at one recording site have appeared to switch their receptive fields from glabrous to dorsal skin following heavy stimulation of the dorsal skin (unpublished observations).

These findings suggest that both dorsal and glabrous inputs are present and closely overlapping in area 3b. Both are capable of creating large, detailed maps over this corti­cal zone; but in the monkeys that we have studied, with their individual histories of hand use, glabrous digital rep­resentations predominate.

In studies of the distributions of neurons driven by slowly adapting inputs, several investigators have presented evi­dence that there are separate, complete representations of the skin for each submodal class (Sur et aI., '81; Dykes and Gabor, '81). On the other hand, these detailed maps indicate that there is a single, simple representation of the glabrous hand surfaces in these fields. In area 3b of monkeys, Sur and colleagues ('81) have proposed that in layer IV, slowly adapting (SA) and rapidly adapting (RA) inputs are segre­gated into narrow alternating bands of partially shifted somatotopic overlap so that overall, SA and RA bands sep­arately represent the same skin surfaces. However, possibly consistent with the existence of a single overall map, they believed that these inputs merged into a mixed representa­tion in the superficial and deep layers. This proposal is similar to that made for striate cortex in monkeys where ocular dominance bands represent the same visual space individually for the two eyes in layer IV and combine to form a single binocular representation in other layers. In the present highly detailed maps, only a single highly to­pographic map was evident; separate representations were not apparent. Map reorganization after peripheral lesions suggest that the boundaries of SA and RA zones may be subject to change. Thus, we hypothesize, SA and RA bound­aries may be functionally derived distributions, alterable by stimulus aspects of predominant hand uses.

Evidence for map variability in other cortical regions

If map variability is primarily a consequence of differ­ences in hand use in individual monkeys, then map varia­bility in other cortical fields might constitute evidence for the possible extents of use-dependent alterability in those zones. Again, variability in map structure has not been the primary objective of any earlier study in cortex. However, differences have been described in several regions.

In auditory cortical fields, there is substantial variation in the details of best frequency organization in individual cats, although the overall pattern of cochleotopic or "tono­topic" representations can be remarkably similar (Merzen­ich et aI., '75, '84b; Reale and Imig, '80). In the isofrequency dimension of the primary auditory cortex, functional orga­nization is very highly idiosyncratic (see Merzenich and Brugge, '73; Imig and Adrian, '77; Imig et aI., '77; Schrei-

CORTICAL MAP VARIABILITY

ner and Cynader, 'S4; Merzenich et al., 'S4b). In the cat, for example, neurons at given sites along the isofrequency axis of Al have highly specific, complex response properties which differ from cat to cat. Neuroanatomical projections from the auditory thalamus are massively convergent and divergent along this field axis (Colwell, '77; Merzenich et al., 'S2, 'S4b; Middlebrooks and Zook, 'S3).

Other somatosensory zones are likely more variable in representational order than are areas 3b and 1. Thus, for example, raw data from different SII field maps are sub­stantially different in the relatively detailed studies of Ro­binson, Burton and Freidman, conducted in macaques and cats (Robinson and Burton, 'SOa; Freidman, 'Sl; Burton et al., 'S2). Very great individual differences were noted, in an important study of the somatosensory representation of area 7b and other outlying fields in the macaque (Robinson and Burton, 'SOb; Burton and Robinson, 'Sl).

We have earlier hypothesized that the limits of cortical map alterability are determined by the extents of spread of anatomical inputs (Merzenich et al., 'S3b, 'S4a,b; Merzen­ich, 'S7). In this vein, the very substantial variation re­corded along the isofrequency axis of auditory fields and within SII and area 7 somatosensory fields are consistent with this view, as inputs to all of those cortical zones are more widely convergent-divergent than are those to areas 3b and 1.

Variability in visual field organization has been less well studied. There, maps have, as a rule, been defined in coarser grain, as mapping has again been largely directed toward defining visual representational entities, and toward deter­mining the overall grand pattern of retinotopy (e.g., see Zeki, '69, '74; Allman and Kaas, '71a, '71b, '74, '75, '76; Tusa et al., '7S, '79, 'Sl; Palmer et al., '7S). Indeed, individ­ual comparisons are more difficult because of a limitation in the frames of reference for comparing map details. The representation of the visual field periphery as a variable number of "islands" within areas lS and 19 in the cat (e.g., Donaldson and Whitteridge, '77) would appear to be equiv­alent to those described herein (also see Tusa et al., '7S, '79, 'Sl; Albus and Beckmann, '79). It would be surprising ifthe sources of substantial map variability operating within so­matosensory and auditory cortical fields are not also pres­ent within visual cortical areas.

ACKNOWLEDGMENTS This research was supported by Nlli grants NS-10414,

NS-16446, The Coleman Fund, and Hearing Research, Inc. It was written, in part, while Dr. Merzenich was a visiting fellow at the Neuroscience Institute, 1230 York Avenue, New York, NY 10021.

LITERATURE CITED

295

gatus). Science 191 :572-575.

Burton, H., and C.J. Robinson (1981) Organization of the SII parietal cortex. Multiple sensory representations within and near the second somatic area of cynomologous monkeys. In C.N. Woolsey (ed): Cortical Sensory Organization, Vol 1: Multiple Somatic Areas. Clifton, NJ: Humana Press, pp. 67-119.

Burton, H., G. Mitchell, and D. Brent (1982) Second somatic sensory area in the cerebral cortex of cats: Somatotopic organization and cytoarchitec­ture. J. Compo Neurol. 210:109-135.

Carlson, M., and C. Welt (1981) The somatic sensory cortex: Sm1 in pro­simian primates. In C.N. Woolsey (ed): Cortical Sensory Organization, Vol. 1: Multiple Somatic Areas. Clifton, NJ: Humana Press, pp. 1-27.

Carlson, M., and K.A. Fitzpatrick (1982) Organization of the hand area in the primary somatic sensory cortex (Sm!) of the prosimian primate, Nycticebus coucang. J. Compo Neurol. 204:280-295.

Carlson, M., S. Dianhua, and X. Bingxuan (1982) Significance of topography and submodality in the organization of Brodmann's area 1 in Sm1 of Macaca. Soc. Neurosci. Abstr. 8:36

Colwell, S.A. (1977). Reciprocal Structure in the Medial Geniculate. Ph.D. Thesis, University of California San Francisco, San Francisco.

Donaldson, LM.K., and D. Whitteridge (1977) The nature of the boundary between cortical visual areas II and III in the cat. Proc. Roy. Soc. Lond. [BioI] 199:445-462.

Dreyer, D.A., P.R Loe, C.B. Metz, and B.L. Whitsel (1974) Differential contributions of spinal pathways to body representations in postcentral gyrus of Macaca mulatta. J. Neurophysiol. 37:119-145.

Dreyer, D.A., P.R Loe, C.B. Metz, and B.L. Whitsel (1975) Representation of head and face in postcentral gyrus of the macaque. J. N europhysiol. 38:714-733.

Duncan, G.H., D.A. Dreyer, T.M. McKenna, and B.L. Whitsel (1982) Dose and time-dependent effects of ketamine on SI neurons with cutaneous receptive fields. J. Neurophysiol. 47:655-699.

Dykes, RW. (1978) The anatomy and physiology of the somatic sensory cortical regions. Prog. Neurobiol. 10:33-89.

Dykes, RW., and A. Gabor (1981) Magnification functions and receptive field sequences for submodality-specific bands in SI cortex of cats. J. Compo Neurol. 202:597-620.

Edelman, G.M. (1982) Group selection as a basis for higher brain function. In F.P. Schmidt, F.G. Worden, G. Edelman, and S.G. Dennis (eds): The Organization of Cerebral Cortex. Cambridge: MIT Press, pp. 535-563.

Edelman, G.M. (1983) Cell adhesion molecules. Science 219:450-457.

Fitzpatrick, K., M. Carlson, and J. Charlton (1982) Topography, cytoarchi­tecture, and sulcal patterns in the prosimian primate, Perodictius potto. J. Compo Neurol. 204:296-310.

Fraser, S.E. (1980) A differential adhesion approach to the patterning of nerve connections. Dev. BioI. 79:453-464.

Fraser, S.E. (1985) Cell interactions involved in neuronal paterning: An experimental and theoretical approach. In G.M. Edelman, W.E. Gall, and W.M. Cowan (eds): Molecular Bases of Neural Development. New York:Neurosciences Research Foundation, pp. 481-507.

Friedman, D.P. (1981) Body topography in second somatic sensory area. Monkey SII somatotopy. In C.N. Woolsey (ed): Cortical Sensory Organi­zation, Vol. 1: Multiple Somatic Areas. Clifton, NJ: Humana Press, pp. 121-165.

Harris, W.A.(1981) Neural activity and development. Ann. Rev. Physiol. 43:689-710.

Horder, T.J., and K.A.C. Martin (1978) Morphogenetics as an alternative to chemospecificity in the formation of nerve connections. In A.S.G. Curtis (ed): Cell-Cell Recognition. Cambridge: Cambridge U. Press, pp. 275-358.

Albus, K., and R Beckmann (1979) Second and third visual areas of the cat: Hubel, D.H., T.N. Wiesel, and S. LeVay (1977) Plasticity of ocular dominance Interindividual variability in retinotopic arrangement and cortical 10- columns in monkey striate cortex. Proc. R Soc. Lond. [BioI] 278:377-cation. J. Physiol. (Lond.) 299:247-276. 409.

Allman, J.M., and J.H. Kaas (1971a) A representation of the visual field in caudal third of the middle temporal gyrus of the owl monkey (Aotus triuirgatus). Brain Res. 31:85-105.

Allman, J.M., and J.H. Kaas (1971b) Representation of the visual field in striate and adjoining cortex ofthe owl monkey (Aotus triuirgatus). Brain Res. 35:89-106.

Allman, J.M., and J.H. Kaas (1974) The organization of the second visual area (VII) in the owl monkey: A second order transformation of the visual hemifield. Brain Res. 76:247-265.

Allman, J.M., and J.H. Kaas (1975) The dorsomedial cortical visual area: A third tier area in the occipital lobe of the owl monkey (Aotus triuirgatus). Brain Res. 100:473-487.

Imig, T.T., and H.O. Adrian (1977) Binaural columns in the primary field (A1) of cat auditory cortex. Brain Res. 138:241-257.

Imig, T.T., M.H. Ruggero, L.M. Kitzes, E. Javel, and J.F. Brugge (1977) Organization of auditory cortex in the owl monkey (Aotus triuirgatus). J. Compo Neurol. 171:111-128.

Iwamura, Y, M. Tanaka, and O. Hikosaka (1981) Cortical neuronal mecha­nisms. In Y Katsuki, R Norgren, and M. Sato (eds): Brain Mechanisms of Sensation. New York: J. Wiley & Sons, pp. 61-70.

Iwamura, Y., M. Tanaka, M. Sakamoto, and O. Hikosaka (1983) Functional subdivisions representing different finger regions in area 3 of the first somatosensory cortex of the conscious monkey. Exptl. Brain Res. 51 :315-326.

Allman, J.M., and J.H. Kaas (1976) Representation of the visual field on the Jenkins, W.M., M.M. Merzenich, J.M. Zook, B.C. Fowler, and M.P. Stryker medial wall of occipital-parietal cortex of the owl monkey (Aotus triuir- (1982) The area 3b representation of the hand in owl monkeys reorga-

296

nizes after induction of restricted cortical lesions. Soc. Neurosci. Abstr. 8:141.

Jenkins, W.M., M.M. Merzenich, and M.T. Ochs (1984) Behaviorally con­trolled differential use of restricted hand surfaces induce changes in the cortical representation of the hand in area 3b of adult owl monkeys. Soc. Neurosci. Abstr. 10:665.

Jenkins, W.M., and M.M. Merzenich (1987) Reorganization of neocortical representations after brain injury: A neurophysiological model of the bases of recovery from stroke. In F.J. Seil, E. Herbert, and B.M. Carlson (eds): Progress in Brain Research. Amsterdam: Elsevier (in press).

Killackey, H.P., and G.R Belford (1979) The formation of afferent patterns in the somatosensory cortex of the neonatal rat. J. Compo Neurol. 183:285-304.

Krishnamurti, A.F., F. Sanides, and W.I. Welker (1976) Microelectrode map­ping of modality-specific somatic sensory cerebral neocortex in slow loris. Brain Behav. Evo!. 13:267-283.

Lee, K.J., and T.A. Woolsey (1975) A proportional relationship between peripheral innervation density and cortical neuron number in the so­matosensory system ofthe mouse. Brain Res. 99:349-353.

LeVay, S., T.N. Wiesel, and D.H. Hubel (1980) The development of ocular dominance columns in normal and visually deprived monkeys. J. Compo Neurol. 191:1-51.

McKenna, T.M., B.L. Whitsel, and D.A. Dreyer (1982) Anterior parietal cortical topographic organization in macaque monkey: A reevaluation. J. Neurophysiol. 48:289-318.

Merzenich, M.M., and J.F. Brugge (1973) Representation of the cochlear partition on the superior temporal plane of the macaque monkey. Brain

·Res.50:275-296. Merzenich, M.M., P.L. Knight, and G.L Roth (1975) Representation of the

cochlea within primary auditory cortex in the cat. J. Neurophysiol. 38:231-249.

Merzenich, M.M., J.H. Kaas, M. Sur, and C.S. Lin (1978) Double represen­tation of the body surface within cytoarchitectonic areas 3b and 2 in SI in the owl monkey (Aotus trivirgatus). J. Compo Neurol. 181:41-73.

Merzenich, M.M., M. Sur, RJ. Nelson, and J.H. Kaas (1981) The organiza­tion of the SI cortex: Multiple representations of the body in primates. In C.N. Woolsey (ed): Cortical Sensory Organization, Vol. 1: Multiple Somatic Areas. Clifton, NJ: Humana Press, pp. 36-48.

Merzenich, M.M., S.A. Colewell, and RA. Andersen (1982) Auditory fore­brain organization: Thalamocortical and corticothalamic connections in the cat. In C.N. Woolsey (ed): Cortical Sensory Organization, Vol. III: Multiple Auditory Areas. Clifton, NJ: Humana Press, pp. 43-58.

Merzenich, M.M., J.H. Kaas, J.T. Wall, RJ. Nelson, M. Sur, and D.J. Felle­man (1983a) Topographic reorganization of somatosensory cortical areas 3b and 1 in adult monkeys following restricted deafferentation. Neuro· science 8:33-55.

Merzenich, M.M., J.H. Kaas, J.T. Wall, M. Sur, RJ. Nelson, and D.J. Felle­man (1983b) Progression of change following median nerve section in the cortical representation of the hand in areas 3b and 1 in adult owl and squirrel monkeys. Neuroscience 10:639-665.

Merzenich, M.M., RJ. Nelson, M.P. Stryker, M.S. Cynader, A. Schoppmann, and J.M. Zook (1984a) Somatosensory cortical map changes following digit amputation in adult monkeys. J. Compo Neurol. 224:591-605.

Merzenich, M.M., W.M. Jenkins, and J.C. Middlebrooks (1984b) Observa­tions and hypotheses on special organizational features of the central auditory nervous system. In G. Edelman, M. Cowan, and E. Gall (eds): Dynamic Aspects of Neocortical Function. New York: J. Wiley & Sons, pp.397-424.

Merzenich, M.M. (1987) Dynamic neocortical processes and the origins of higher brain functions. In S. Bernhard (ed): The Neural and Molecular Bases of Learning. Berlin: Dahlem Konferenzen (in press).

Middlebrooks, J.C., and J.M. Zook (1983) Intrinsic organization of the cat's medial geniculate body identified by projections to binaural response­specific bands in the primary auditory cortex. J. Neurosci. 3:203-224.

Movshon, J.A., and RC. Van Sluyters (1981) Visual neural development. Ann. Rev. Psychol. 32:477-522.

Nelson, RJ., M. Sur, D.J. Felleman, and J.H. Kaas (1980a) Representations of the body surface in postcentral parietal cortex of Macaca fascicularis. J. Compo Neurol. 192:611-643.

Nelson, RJ., M.M. Merzenich, J.T. Wall, M. Sur, D.J. Felleman, and J.H. Kaas (1980b) Variability in the proportional representations of the hand in somatosensory cortex of primates. Soc. Neurosci. Abstr. 6:651.

Palmer, L.A., A.C. Rosenquist, and RJ. Tusa (1978) The retinotopic organi­zation oflateral suprasylvian visual areas in the cat. J. Compo Neurol. 177:237-256.

Paul, RL., M.M. Merzenich, and H. Goodman (1972) Representation of slowly and rapidly adapting cutaneous mechanoreceptors of the hand in

M.M. MERZENICH ET AL.

Brodmann's areas 3 and 1 of Macaca mulatta. Brain Res. 36:229-249.

Powell, T.P.S., and v.B. Mountcastle (1959) Some aspects of the functional organization of the cortex of the postcentral gyrus of the monkey: A correlation of findings obtained in a single unit analysis with cytoarchi­tecture. Bull. Johns Hopkins Hosp. 105:133-162.

Pubols, B.H., and L.M. Pubols (1972) Neural organization of somatic sensory representation in the spider monkey. Brain Behav. Evol. 5:342-366.

Reale, RA., and T.J. Imig (1980) Tonotopic maps of auditory cortex in the cat. J. Compo Neurol. 192:265-292.

Robinson, C.J., and H. Burton (1980a) Somatotopographic organization in the second somatosensory area of M. fascicularis. J. Compo Neurol. 192:43-67.

Robinson, C.J., and H. Burton (1980b) Organization of somatosensory recep­tive fields in cortical areas 7b, retroinsula, postauditory and granular insula of M. fascicularis. J. Compo Neurol. 192:69-92.

Schreiner, C.E., and M.S. Cynader (1984) Basic functional organization of the second auditory cortical field (AIl) of the cat. J. Neurophysio!. 51: 1284-1305.

Sherman, S.M., and P.D. Spear (1982) Organization of visual pathways in normal and visually deprived cats. Physiol. Rev. 62:738-855.

Simons, P.J., D. Durham, and T.A. Woolsey (1984) Functional organization of mouse and rat Sml barrel cortex following vibrissal damage on different postnatal days. Somatosen. Res. 1:207-245.

Sperry, RW. (1963) Chemoaffinity in the orderly growth of nerve fiber patterns and connections. PNAS 50:703-710.

Stryker, M.P., W.M. Jenkins, and M.M. Merzenich (1987) Anesthetic state does not affect the map of the hand representation within area 3b somatosensory cortex in owl monkey. J. Compo Neurol. 258:297-303.

Sur, M., M.M. Merzenich, and J.H. Kaas (1980a) Magnification, receptive. field area, and hypercolumn size in areas 3b and 1 of somatosensory cortex in owl monkeys. J. Neurophysiol. 44:295-311.

Sur, M., RJ. Nelson, and J.H. Kaas (1980b) Representation of the body surface in somatic koniocortex in the prosimian galago. J. Compo Neu­rol. 189:381-402.

Sur, M., J.T. Wall, and J.H. Kaas (1981) Modular segregation of functional cell classes within the postcentral somatosensory cortex of monkeys. Science 212:1059-1061.

Sur, M., RJ. Nelson, and J.H. Kaas (1982) Representations of the body surface in cortical areas 3b and 1 of squirrel monkeys: Comparisons with other primates. J. Compo Neurol. 211:177-192.

Tusa, RJ., L.A. Palmer, and A.C. Rosenquist (1978) The retinotopic organi­zation of area 17 (stri"te cortex) in the cat. J. Compo Neurol. 177:213-236.

Tusa, RJ., A.C. Rosenquist, and L.A. Palmer (1979) Retinotopic organiza­tion of areas 18 and 19 in the cat. J. Compo Neurol. 185:657-678.

Tusa, RJ., L.A. Palmer, and A.C. Rosenquist (1981) Multiple cortical visual areas: Visual field topography in the cat. In C.N. Woolsey (ed): Cortical Sensory Organization, Vol. II: Multiple Visual Areas. Clifton, NJ: Hu· mana Press, pp. 1-32.

Van der Loos, H., and J. Diirfl (1978) Does the skin tell the somatosensory cortex how to construct a map of the periphery? Neurosci. Lett. 7:23-30.

Wall, J.T., D.J. Felleman, and J.H. Kaas (1983) Recovery of normal topog­raphy in somatosensory cortex after nerve crush and regeneration. Science 221 :771-773.

Wall, J.T., J.H. Kaas, M. Sur, RJ. Nelson, D.J. Felleman, and M.M. Merzen­ich (1986) Functional reorganization in somatosensory cortical areas 3b and 1 of adult monkeys after median nerve repair: Possible relation· ships to sensory recovery in humans. J. Neurosci. 6:218-233.

Werner, G., and B.L. Whitsel (1968) Topology of the body representation in somatosensory area 1 of primates. J. Neurophysiol. 31 :856-869.

Whitsel, B.L., D.A. Dreyer, and J.R Roppolo (1971) Determinants of body representation in post-central gyrus of macaques. J. Neurophysiol. 34:1018-1034.

Woolsey, C.N., W.H. Marshall, and P. Bard (1942) Representation of cuta­neous tactile sensibility in the cerebral cortex of the monkey as indi­cated by evoked potentials. Bul!. Johns Hopkins Hosp. 70:399-441.

Woolsey, T.A., and J.R Wann (1976) Areal changes in mouse cortical barrels following vibrissal damage at different postnatal ages. J. Compo Neurol. 170:53-66.

Woolsey, T.A. (1980) Some anatomical bases of cortical somatotopic organi· zation. Brain Behav. Evo!. 15:325-371.

Zeki, S.M. (1969) Representation of central visual fields in pre striate cortex of monkeys. Brain Res. 14:271-291.

Zeki, S.M. (1974) Functional organization of a visual area in the posterior bank of the superior temporal sulcus of the rhesus monkey. J. Physiol. (Lond.) 236:549-573.