•

I. INTRODUCTION, DO MA INS IN THE INTERMEDIATE LEVElS

We are concerned in this chapter with the transactions that occur in organized subsystems of receptors, neurons, and effectors. T heir mechan isms wi ll be cen­tral to the issues of how the nervoussysttm works as it communications machine that recognizes, decides, and commands. Of Ihe va rious domai ns of operations thai are ava ilable for our scrutiny. we include here fi ve, and these form the headings of sections II to VI.

T his is the rirsl encQunter, between these covers, wit h the phys iology of useful arrays of neurons. It wil l help to introduce two dassrs of function that emerge from and pervddc the activi ties of such arrays.

Each pari of th e nervous system-but especia lly the receiv ing s ide, from Ihe battery of receptors to the networks of higher-order neurons of the aHerent systems-ca n be thought of as a filter . The sense organs send a pa tt erned stre.1m of impulses, in space and t ime, to the afferent centers, and these represent

to the afferent centers a coded Form of the sense organs' selection, or fil trate, of the act ual sti muli. The aHerent net­works-meaning the interconnected second-, th ird- and higher-order neurons that receive thei r input primarily fro m the receptors, though also from ot her sources-do not merely pass on this information. They use convergence of separa te cha nnels (comparison), divergence of each channel (pa rallel processing), later.ll inh ibi tion (enhancing contrast), and other processes to modify the signal. Special Attribu tes of the original input are passed on (recogni­tion), and much in form.l lion is dis­carded . The structure and coupling Functions of the network determine what gets through .

Compara ble networks exist on the out put side. They are also fil te rs, but since they Formulate and send to the eHectors commands thil l are crucia lly patterned in sp.1Ce and ti me, they are oft en thought of as pallern genera tors. T hey convert triggeri ng inpu t or steering input from receptors or their own spon­taneous discharge or a mixture of these

242

Chapter 7 Integrdlion at the iniermediat(' Levels

Figure 7.1 Rf(rrlilm(ll/ ill /I muscle wilh IIlru molor III1i/S.

Silent mOlor

Muscle fibers '1. inactive

Low ~sjon Axons with

impuls.!;. trains

~-v--High tension

into ad.lplively patterned ("coordil1<1ted") sire.llTIS of impulses in the output chan­nels. Again the connectivity and dynamic properties of the network determine its outpul, but not merely by passive filter ­ing. Often there are intrinsic rhythms of spontaneous impulse bursts, and these result in the generation of specified con­stellations and sequences of activity in populations of units and finally in the effectors. Let us look first at the final link.

II. NERVOUS CONTROL IN EFFECTORS: DIVERSITY OF PERIPHERAL INTEGRATION

The best known vertebrate skeletal muscles consist of muscle fibers whose cell membrane is capable of producing propagated all-or-none impulses like those of nerve fibers. Each muscle fiber is innervated, in general, by only one axon with one terminal at the speci"lized end-plate. A motor nerve impulse arriv­ing there gives rise to an end-plate or junctional potential, like an e.p.s.p., which is usually suprathreshold for the initiation of a propagated muscle action potential. Each motor neuron innervates many muscle fibers. The motor neuron and its muscle fibers comprise a motor unit. Each muscle contains from one to several thousands of motor units. An impulse in one unit, or a synchronous volley in many, causes a brief contraction to summate if they overlap, or to fuse into a smooth contraction called a teta­nus if the frequency is high. When there are many motor units per muscle, a means of grading the strength of con­traction of the muscle is to vary the number of active un its. This method of control is called recruitment (Fig. 7.1). A

second means is to vary the frequency of impulses in each unit, since either the aver,'ge tension in a series of twitches or the tetanic tension is a function of fre ­quency. You can observe frequency and recruitment;control phenomena by plac­ing a stethoscope over your eyelid and listen ing to the twitches of its muscle while you control them willfully. Re­cruitment may be important in verte­brate skeletal muscle only at low ten­sions.

Since there are many motor units in most vertebrate skeletal muscles, oper­ating these unit s out of phase with each other can give a smooth overall contrac­tion even when the frequency in each is so low that it contracts in a series of twitches. Another possible utility of r having many units is that there could be a rotation of activity during low or medium work loads, sa..that units could rest. But this old notion is apparently not confirmed in the best studied materials.

Other muscle fibers in vertebrates arel incapable of producing propagated Mtion potentials. They are innervated not by a single end-plate, but by numer­ous spatially distributed motor nerve terminals. Such multiterminally in­nervated muscle fibers (Fig. 7.2) respond electrically with junctional potentials only, but since these occur at many sites, eJectrotonically spreading depolarization can activate the whole fiber. Muscle fibers of this type m"y be mixed with others, as in some frog muscles. They are usually innervated by small diameter j motor axons.

Since individual vertebrate muscles are excited by many axons, we think of them as being driven by pools of motor neurons. There is a diversity of s ize within the pool of motor neurons. Gen­er<llly, the smaller ones have lower

Figure 1.2 T!lI'''s of s1 riR /fd /H 1/S(lf illlUrvRlio ,l,

""""""""""""""" £;"""""""'""""',: Unilerminal innervation

Mul1i ' lIrminal innllrval ion

";::Z""""""\C""",,z',,,n,,;:;$';, Dinou ronal innerva,ion

!!I ,~II"" I " :m , , ,:::25tii, Polynooronal innervation

thresholds for natural stimulation and are more tonic in their discharge. They innervate relatively fewer mu scle fibers and produce wea ker contractions, but are probably principals in th e ord inary load of muscle work. The larger phasic fibe rs are called forth by stronger stim­ulat ion and produce vigorous action. T his relation among threshold, fi ber s ize, and tonic versus phasic mode of dis­cha rge is probably of gener,J\ s ignifi c.mce in neurophysiology. 11 is easily demon­strated in sensory systems as well.

Many whole arthropod muscles, and some in an nelids, are in nervated by only one or a very few motor axons. [n muscles innervated by only one axon, recruitment cannot be li ke that in verte­brates, which add motor units by centra l enl istment of motor neurons. Bul a s ingle axon can recru it incre"sing Ilu m-

bers of the muscle fi bers it supplies because diffe rences in the facili talion properties of the synapses allow trans­mission to be effecti ve "I different fre­q uencies of arriving nerve impu lses in different fibers. There are also differ­ences in excita ti on-cont raction coupling threshold in some muscles.

Innerva tion of all th ese muscles is of the mul ti lerminal type. At norma l fre ­quencies of "rriv ing nerve impulses, the junctions commonly ex hibit the t ime­dependent properlies of facilitation or antifacili tation. Contract ion strength can be controlled not only by average fre ­quency of moior axon impu lses, but, in some muscles, also by deta iled tempora l structu re of the impu lse train. For example, the opener muscle of the cray­fi sh claw is innerva ted by only one excitatory neuron, and the myoneural junction shows sl rong faci litation. A steady rhythmic Irai n of impulses causes a contraction of a given strength. But if the same lotal number of impulses in a given time is delivered grouped in pairs, the contraction is stronger. The junction is considered to be pattern sensitive in that it responds to delai ls of innervating tempora l p,1 I1ern, nol just ave rage fre­quency (Fig. 7.3,A). Recordings from intact moving animals show Ihat the eNS sometimes issues molar comm.1 nds in impulse doublets, but il is not yel certa in how this potential code is em­ployed, wheth er independently of or correlated wit h mea n frequency.

Muscles may be polyneuronally innervated in that si ngle fibers can receive input from more than one motor axon (Fig. 7.2). Each innervating axon di ffers in the properties of its endings on the same muscle fi ber. O ne of the axons is usually d so-c<l lled "("sl" or phasic axon that normally car ries short bursts of

243

Se<:tiol'l II Nervous Con'rol In Effec,ors:

O ivers ily of Peripheral In.egraUon

244

A SLOW

B FAST

20 mV I 0.25 9 I ltm\\\~~I~ II'lI~\ I'I\I\I\I\\"\I'I. - ~

~

C 2 sec

ISO

2 • ~ 0 _ 0 0-o· 100 . , •• "-_0 0 ·-," -= oE

'~a 50

• ~ 0 U

o ~O--:2C--:.--:,C--~8-:-:I~O--:I2;--:I';'--:'~6 ~.3 Excita tory spacing interval (msec)

Fig ure 7.3 Differences .llllong muscles in response 10 stimulus inlerv"L A. A muS(le fiber in the c!.1\V. cl05cr muscle o f " cTab responds to repetit ive sl imul,l! ion

of " "slow" ~ )(on with 1.ITge detion potentials (upper Ir",e) hav ing slow (,lci lit,ltio n, and wi th gr,u(u,llly SUlllllldling smooth tetanic contTdc! ;o n (lower I rd CC shows tenSion). B. T he Sdme Illuscle fib er exc ited vi .. a " foist" ,,"on shows 5111.111, qllickly pl,l\cduing action potentia ls and small, clonic . ( :::; nonfused) conlrdclions. IAtwood, 1967. ] C. Pd ttern-sensilive (fi lled cir­cles) ,1lId p,lttcrn-insensi livc (open c ircles) muscle of (rayfish. At a con­st,lnt ,,,~n,, freque 'lcy o f 30 shocks/ sec (== 33.3 msec {ntervals), trains o f ~hock s Me delivered ei ther ,1 t uniform intervals (extreme r ight) or at alter­ndtely short (d b5dss~ v~ l ucs) ~nd long illter\'~ls; response plo tted as a per_ centdge of the e\'enly sp~C(!d cOllt raction. Refrdctoriness red uces responses at the shortest interv.ll s. IRipley .Ind Wiersmd, 1953. 1

impulses at high frequency for quick movements and produces la rge junc­tional potent ia ls in the muscle fi ber (Fig. 7.3,8), .md these elicit spike-shaped loca l response potentials and rapid cont rac­tions. Anotller axon is a "slow." or tonic, nerve fiber that normally carries long trains of impulses at low frequency; these impulses decrement severely as they approach the nerve ending, and prod uce smaller and more facilitating junct ional potentials that cause slowly developing con tractions. Commonly there is also an inhi bitory axon. Each innervating axon may end upon a large fraction of all th e fibe rs of the muscle, the fast axon preferentially reaching the fast (short sa rcome re) and intermediate muscle fibe rs; the slow axon, the s low (long sarcomere) and intermediate mus­cle fibe rs (Fig. 7.4). Th e presynaptic endings of the same axon upon muscle fibe rs of d ifferent sarcomere length and hence contraction properties differ in effectiveness, in arousi ng post junctional potentia ls, and in degree of faci litation. An extr.lord inary heterogeneity of mus­cles and actions and thei r smooth grada ­tion ca n now be understood, even in such animals as arthropods, in which only a few axons supply a whole muscle. Th is heteroge neity is the resu lt of the di fferent combinat ions of phasic, tonic, and inhib itory axons, th e different de­grees of polyneu ronal innerva tion, th e d iversity of type of terminal of each axon upon dirfel'ent muscle fi bers, the diver­s ity of type of muscle nbe r, and the diversity of rise and fa ll times of facili ta­ti on.

Inhi bitor axons exert two distinct effect s, presynapti c and postsynaptic (Fig. 7.5). Postsynaptic inhibition causes hyperpolariz..ll ion and increased con­ductance of the muscle fi ber membrane;

•

"Fast"" a~on (phasic)

Large " fas!" ' e.p.s,p. + spike

\LliLJ:LLLL1.LLLL.'.LLJ..LlllLLllJJ.J Fast conlraction

Slow contraction

Large "slow" e.p .s,p.

"Slow" a~on (tonic)

to msec

Figure 7,4

More 1 facilitat ion

Less j fac ilitalion

Crustacean neuromuscular innervation. The diagram illustrates the matching of phdsic ("f.1St axon"') a nd tonic ("slow axon") innervation with different ca tegories of muscle fiber (indicdted by sarcomere length ) in a crab muscle. [Atwood, 1973.]

245

Section 11 Nervous Control in Ef(ectors:

Diversity of Peripheral Integration

Figure 7.5 Di~grtHlwmli( re"restllialioll of III ~

Hcitatory a"d i"/Iihitory imlervaliml of a (msla(e~tI mllsdt fiber.

Nerve /terminaIS~

Excitatory Inhibitory

Presyn. inhib.

Postsyn. inhib. m ) ) ) ) ) ) )6

Muscle

ExdlMory urn'e It l'ltliuui, form rl eIH·O· mlisO/Ia. ' )I>1apse, . Inlribilory la mi· Iluis forlll 'Ifllrorllllmdar sy'W/'SfS (/wslsyrlapli( illhihilioll) I1Ild nxo· axo· /lal syrla/lSf, H/wrl fxril~lory lerll/illais (llre,y>1aplic illllibiliOlI). [LI1IIS IlUd Alwood,1973· 1

246

Chapter 7 Integration at Ihe Intermediate levels

,

c

, ~

D F

Fig ure 7 .6

The ,tn"tomical distribution of ,1XOnS to the dista l thor"cic limb muscles of several groups of Mabcostrac ... BI.lCk lines represent motor excitor "xons; colored lines represent inhibitory axons. Br,Kke!s indicate tht: distribution ofaxons behveen nerve blllldies. A. Br,lChyu,,,.,.u. AIlDmur" . C. Stom"lopod,,- D. Palinu,a. E. Astaell'''. F. Stenopodide,' (N"I",,!i.,). The boxes rep,esent the muscles in the three di stal segments o f the legs, .IS indicated by thei r COllllllOn IMllle5: Il. ,1Ccessory nexor; b, bender; (, closer; f, extensor; f. nexor; 0, opener; r, rol,l\or; ~. stre tcher. IWiersll1.l and Ripley, 1952.)

presynaptic inhibition redu ces mi niature e.p.s.p.'s and the amount of excitatory transmitter released per impulse. Both depend on the arrival time of the inhibi­tory impulse relative to the excitatory, but in different ways. They are widely different in importance. Inhibition in some muscles is mainly postsynaptic; in others, presynaptic. In some mu scles the response to a slow excitor is inhi bited presytl<1ptically, and the response to a fast excitor in the same muscle is 'Iess inh ibited . In the crayfi sh claw opener muscle, inh ibit ion caused by 10 impulses per second is 90% presynaptic, but a t 40 impulses per second it is only 50% pre­synaptic. Inhibitory transmitter is re-

leased with facilitation, which varies among junctions.

An additional degree of complexity of muscle innervation in arthropods can be appreciated if one examines the innerva­tion of the muscle of a whole limb, such as the walking leg of a crayfish. Several muscles may be innervated by a single motor axon (Fig. 7.6). The overlapping but nonidentical innervation patterns still .1110w independent action. For example, the opener and stretcher mus­cles are innervated by the same excitor axon but by different inh ibitors. Never­theless, the whole limb is innervated by only 12 excitatory and J inhibitory motor fibers, and this fact of small numbers

•

makes it seem possi ble that we will be ab le to understand fully ,l t the level of s ingle nervous units the normal control of movement in arthropods.

The important lesson from the exa mples chosen is that even ,l t th e last lin k, the nervous control of effectors, th ere is diversity in the principles em­ployed, some cases ex hi biti ng a sub­stantial degree of in tegration in the periphery. Wags have sa id crabs can think (evaluate) in their legs. We h,we not exha usted the diversity: fast insect flight muscles, a va riety of smooth mus­cles (Box 7.1), electric organs, glands, cilia, chromatophores, and other effec­tors manifest additional principl es of se­ries and parallel control (Fig. 7.7).

III. ANALYSIS OF SENSORY INPUT, PARALLEL AND SERIES PROCESSING

The general principles of sensory recep­tion are common to systems as different as vision and taste and to inputs that rarely re,lch consciousness, such as those from pressure receptors and che mo­receptors in the grea t Mteries, as well as those that do. Foremost is the principle of parallel cha nnels: many receptors in­dependently sample the sti mulu s world and send their imp ulse-coded reports to the centra l nervous system in parallel afferent fiber s. The basic independence may in some organs be abridged by superimposed in fluence fro m the centra l nervous system (centrifugal or efferent infl uence) or from neighboring receptors (lateral inhibition).

Inti mate ly rela ted is the second principle, that of overlapping receptive fields: neighboring receptors are gener-

a ll y stimulated by a fract ion of the im­pi ngi ng world that is large but circum­scribed (the excita tory receptive fi eld, ERF) and which overlaps substantially with that of cont iguous ch,lllneis. T his applies in some systems to topographic oV('flap and field s, such as areas of the skin, retina, and basilar membrane (which distributes sound frequencies to aud itory afferents). In some systems the overlap involves ambiguity in modality (Chapter 6, p. 214).

The third principle relevan t to the present section states that sensory input typically exhibits divergence and con­vergence in th e central nervous system, the former d istributing information to analyzers concerned with different as­pects of the sti mu lus world, the la tter permitt ing resolution of ambigu ities, sha rpening of fields, and abstraction of special features. Generally, divergence predominates in sensory pathways, so that even though each neuron still re­ce ives synapses from many different presynaptic cells, there are increasing nu mbers of cells at each higher neu ral level enroute to the sensory cortex . One characteristic example is the first- to second-order neuron stage in the audi­tory system, already mentioned (p. 108). The value of such wide divergence is that different postsynaptic cells, receiving in put from subtly different, overlapping, presynapti c populations, can perform d ifferent types of information-process ing operations in parallel, more fully ex­tracting all available in for m'ltion. Thus a visua l signal ca n be analyzed by different cell populations for brigh tness, shape, color, distance, and movement; an audi­tory signal, for frequency, rate and di­rection of change of frequency, duration, in tens ity, presence of overtones and temporal patterns, localization of the

247 Figure ??

Various Iyl"'s of ,!f,·c/ors.

, ,

~: :, - , -, , / '\ ... , , ,

- - , Fly wing muscles

Snail heart

Bladder. ureter. u rethra

0 'VIJ\iIfWWVlM!l!\ilflfV\f\']

rlJlit/lJlJlflJVWW\IWIJV\I) ~'1NlfWV\N\I\I\fV\.I\i\iUvt,'j

e""""'VWINVlIVWWVW] Electric organ

c~ , Salivary gland

Chromatophore

248

,.------Box 7.1 The Nervous Control o~ We will briefly revicw the sl.lte of knowledge with Unitary muscles ,lI1d some o thers ,ue typically respect to vcrtebr<ltcs. Prosser (1 973), from whom we arrJngcd in bund les of fibers glnncctcd to each other by take much of this .lecount, divides 5moolh muscles of " nexuses," areas of dose ilpposi tion .1Ild e lecirica l low vertebrates into two kinds, unit<ll'y and multiunitary, resisl,mce. The motor nerve termina ls release trans-with some intermedia tes. Unitary muscles include Iho~ mitter from large numbers of varicosities (swellings). of the ma'or viscera - uterus ureter, and gastro- Only some muscle cells receive direct inncrvdtion (see lnies!in .. ! IT"'t. These s ow spontdllcOUS rhythmicil , figure). O the rs are excited by their d irect electrotonic and distr ibution of exci tation IS tned rom muscle coupling 10 these. Still others aTe excited indirectly, fi ber to fibe r. They ca n be stimulated by st retch and pcrll<lps by a combinat ion of e lectronic coupling .Iud mo(lii1;ueaby nerves. Multiunitd ry muscles ii1c"lude diffu se transmitter. In di fferent viscer.l the smooth nicl ildlJ ng membrane an pi omolor, ci lia!), •. <tnd iris muscles vary widely in the proportions of these three nruscles. These are norma lly aclivdted not spont.lne- types,.lS well as in cable properties, ion dependence. o~y or by stretch, b'"Ut by nerves or hormones; d is - the effects of transm itte rs and drugs, sym p.lthetic ilnd tribulion of excitation within the muscle is normally by pa r.lsym p.lthetic innervation, and Spont.lIlCOUS activity nerves. I here may be 1 ,lcilitation of p.s.p.'s; severdl pattern . nerve- fibers may influence one muscle fiber.

~ "Direclly innervated "" cell with close (200 A) neuromuscular junc tions

C iii,' . :J "Coupled"" cell exhibits junct ion potent ials Carried by e tectrotonic coupl ing

c ~ "" tndirectly coupled·' ce ll exhibits only ac tion potentials l ow-resistance pathway

Varicose nerve fiber

Schelnatic represenldtion of the types of autonomic innervat ion of smooth muscle. All the smooth­muscle fibers are In terconnected by l ow-re5 i 5 t~ n ce electrotonic junc tions, shown here as bridges. Some receive direct nerve endings (ddrk shading): other.; (light shading) do not, but Me clo~

enough to the foregoing to show junction potentidls th~t h,we spread elect rotonical1y. The most remote (unsh .. ded) show no junct ion potentials; they n",y be elicited by electrotonic sprNd of aClion potent ials from the preceding d"ss and by tr .. nsmitter rele .. sed from the varicose nerve fibers at some disl"nce. [BurnstO(k and IWJ~·ama, 1971.1

•

source in space, etc. After bu ilding up sharply specillc response requirements, these elements can again converge to combine their specificities.

Ani mals abstract from their total sen· sory input specia l qua lities of that input before formulating a motor command. A frog jumps and snaps at any sma ll, dark, moving object wit hi n range if it is in the mood. A male stick leback fi sh attempts to CQmt with any oval, silvery, red­bottomed object of su itable s ize, whether it be a female stickleback or a crude mode l. A few ChM.lclers of Ihe whole constellation of visual inputs associated with the female stickleback seem to be the only relevant ones to the ma le (sec Chapter 8). How might fil tering of this quality (recogni tion) go on in the nervous system?

We could discuss mtering networks for any sensory modality, but space does not permit a survey. Instead we will concent rate on visua l filtering. Since notions of moda lity, submodality, labeled lines, and tempora l coding have been dealt with in Chapter 6, we can go directly to a consideration of relatively complex network functions.

The histological structure, cell types, and connectivity of the vertebrate retina have already been descri bed (pp. 12·1-126), together with some of the electro­physiological properties expressi ng the coupling functions of the connections.

At least thirteen functiona lly distinct types of optic nerve IIbers carry infor­mation to Ihe brain from the eye in the ca t. Ta ble 7.1 gives ,I recent classificat ion. Note that 92% have exci tatory receptive fie lds (ERF's) that are concentrically organized. These arc either brisk or sluggish, referring to the promptness and vigor of their responses, and either tr.lnsienl or sus tai ned, referri ng to thei r

mainly phasic or mainly tonic char.lcter. For each of the four combinations of these properties there are two types, distingu ished by the s ign of their COIl­

centric orga niza tion. There are ON­center-OfF-surround units and Ihe con­verse, OFF-center-ON-surround units. Units of the former sorl are excited by the ON of a sma ll spot of light in the centra l zone of the ERF or by the OFF of an annular illu mi nation of the surrou nd­ing zone. Units of the latter sort arc converse in character. The table lists all types of uni ts in the order of their axon diameter, the largest first. The largest and fastest un its, those in the brisk­transient group, are also ca ll ed V-neuro ns; the next la rgest un its, those in the brisk-s ustained group, arc also called X-neurons. (All the rest, both concentric sluggish and nonconcentric, are coIiectively ca lled W.neurons, but this term embrilces too heterogeneous a set to be useful. )

For ma ny years the cat was descri bed as havi ng ess·entially only two types of optic nerve fibers, ON-center and OFF­center. In contrast, the rabbit, ground squirrel, and gray sq uirrel were de­scribed as more frog-l ike in having several com mon nonconcenl ric and more complex types-for example, those pre­ferring movement, some specillc for direction of movement, some even for orientation. It now appea rs that the cal has about the same v.uiely of Iloncon­centric units, including th ese complex types, but in much small er propo rtion, 8% compared 1034% in the rllbbi t. In the cal, "loca l edge detectors" are fi ve ti mes as com mon as "direction selective" units; in the rabbit the lalter predominate, except in the viSual st reak (the equiva­lent of the are.' centralis, for high resolu ­tion) .

249

5«lIon III AnJiY5iJ of ~n'iOry Inpul:

I'Molliel ~nd Series ProcesSi ng

Overlap 01 receptive lields

Convergence

250

Clupler 1 Intcgr.ltion at the I nter l11ed l~ 11!' levels

All of these optic nerve fi be r Iypes bespeak transa ctions tha t process the information encoded by rods and cones. La tcral interact ions, convergence, and highly specified connect ivity (see Chapler 3) among receptor, horizontal, bipola r, amacrine, and ga nglion cells are a ll indicated. This is usua lly referred to as early extraction of fea tures to dis­tinguish it from further changes in the meaning of cell discharge at laler stages in the visual pathways. It also evidences paralle l processing for centra l destina­tions of quite different fu nction. The c£> l1 lr.11 targets of these o ptic nerve fiber

Tabte 7. 1

types are known only in parI. In the ca l, X- and V-fibers go ma inly to the dors.l l late ral geniculate, e.lch to its priVate class of geniculate neurons; these in turn project 10 the vi sual cortex, X-fibers to the so callea simple cell s and Y-fibers to the complex cells. W-fi bcrs go to the midbr.l in, largely to the tectum. It should be pointed ou t here that the re are a t least s ix cent ral targets of optic nerve fibers, of whi ch the dorsa l later.ll genicu la te and th e tectum are on ly two. Evidence sug­gests that they have d ifferent functions.

T he retinas of frogs. lizards. and pigeons have even fewer concentrica lly

Receptive fi eld types of 960 cal re lina l ganglion cell s, Latencies are the ranges of anti­dromic conduction ti mes (rom the optic tract stimulus si te, and can be t,lken as pro­portional 10 Ihe reciproca l of axon diameter.

Types

Concentr ically organized Brisk

Transient ON-center OFF-center

Sustained ON·center OFF-cenler

Sluggish Sustained ·

ON-cenler OFF-cenler

Transient" ON-center OFF-cenler

Nonconcentrically organized local edge detector Direction·selecti ..... e Color-coded Unilormily detector Edge.lnhibitory OFF-center Unclassilied J

Sou"e. Cleland and Levl&k. 1914

Number

887 m 243 115 128 531 271 260

113 44 22 22 27 13 14

73 45 11 6 5 3 3

Pereen/age

92 BO 25

55

12

B 5 1

<1 <1 < 1 <,

-Larency (msee)

1.0-2.4 ) y."I1,

2.5-5.9 ] X-cells

4.6- 24.0

6.1-18.7

6.6-15.9 6.1- 12.4 3.8-14.2 8.7-13.9 3.9-6.6

W·cells

' In e lurtl'lOr 42. Ihere wera insuiliciont obserYlllLorlS to distinguish whelher sustained Or transient (21 ON-center. 22 OFF·center)

I Insu1liCIIlnt ObServalions 10 reach e conclusion.

organized ganglion cells and many more movement-specific cells. This is prob­ably not a phylogenetic trend; it may have some relation to habit of life and the roles of vision. frogs are known best from ~oneering work by ~et vi n, Maturana, NrccU11iJcI~s, who

SeCI-Aim:u! h5f more Itah.lf<frk1ilOs than flashes of diffuse fields or focused beams, and from the quantitative work of the Griissers (Fig. 7.8). In contrast to the cat, in which all optic nerve fibers are myelinated, the vast majority in frogs are very fine and unmyelinated. The five types of optic nerve afferenls may be summarized in the following way. (a) Type 1 fibers, called "susta ined-edge detectors," respond to a sharply focused l edge of an object, light or dark, moving I or recently having moved into the 1 ° -30 l ERF. (b) Type 2 fibers, called "convex­edge detectors" (Fig. 7.9) respond only to a sma ll object, darker than the back­ground, that moves into or has recently moved into the field; they must detect not only the change of position in the ca. 3° ERF but also that there is little or no change of position in the ca. 15° sur­rounding inhibitory receptive field (IRF). (c) Types 3 fibers, called "changing con-I trast detectors:' respond to any edges in motion, large 01' small, dark-an-light, or light-an-dark, if the contrast and rate are adequate and the edge is not too fuzzy. These fibers give a weak response to nonmoving tempor.l[ changes in light; they aTe classical ON-OFF units, but much prefer motion. (d) Type 4 fibers, c.llled "dimming detectors" respond to any dimming or darkening, whether caused by motion or not; these are OFF fibers but not like the concentric OFF­center units of the C.1t. (e) Type 5 fibers respond to any brightening; these are ON fibers, but not concentrically organized,

with an OFF-sensitive surrounding ex­citatory field, as in the cat; they are more sensitive to blue light than to white or other colors.

Types 1 and 2 are fine, unmyelinated axons in the optic nerve and by far the most numerous. Types 3, 4, and 5 are myelinated. Types Ito 4 go to the optic tectum of the mesencephalon, type 5 to the dorsal [ateral geniculate of the diencephalon. Histo[ogica[ types of ganglion cells with distinctive dendrite branching patterns are probably asso­ciated with these fiber types (see Fig. 3.4).

Types 1 and 2 do not respond to gen­eral illumination, and their responses to movement are not influenced by the [evel of ill um ination over a range of intensity of mOTe than a hundredfold. Given their req uirements they di scharge at a rate that encodes contrast of the moving object (but not light [evel) and rate of motion of the object minus that of background objects in the IRF (but not relative motion). The discharge is ambiguous for certain severely con­stra ined combinations of object size, contrast, speed, and amplitude of move­ment.

Change of position is as good a stim­ulus .15 visible movement; th us a good response is elicited by briefly illuminat­ing a sta tic scene (no response) and th en, during the dark period between flashes, moving an object within the ERF. The "memory" of the position of objects during the first flash lasts through at least one second of darkness (Fig. 7.10). An additional and remarkable property of type 2 units is erasability. When such a unit fires upon motion of a sufficiently convex edge toward the ERF center, it continues to fire at a [ower rate for some seconds after the motion stops; but it

251

Seclion III Analysis of Sensory Input:

I'arallel and Series I'rocessi ng

252

Oitfuse light

---I 1 sec I 00 Of<

Class t

Class 2

Class 3 I Class 4 I I I 111 11111111

Horizontal movement of a 2 x 10' vertical

black bar

Location of optic nerve fiber endings

at tacta l cell ________ dendrites

./- ~

Class! 11 11111 11 ~~~ Class 2 ....:::::::::

. Class 3 11111 ~ ,.,." .. g05. Class 4 11111 I I II ----"'- .

_ Excitatory synapses

----l Postsynaptic inhibitory synapses

~I Presynaptic inhibitory synapses

Movement of a 2' black spot

J f" ERFsize

Class 1 --tttHtIHHlHtttttHtttttttt--~ 2- 4'

Class 2

Class 3 11111 1111 I II 11111

Class 4 --------

Horizontal movement of a 2 x 10' horizontal

black bar

./ Class 1 61111 111111 11 I I I I II Class 2 111111 1 I I I Class 3 1111

Class 4 111 1 I I I I I

R

I I

25-4' 6_8'

~ 10'

Fig ure 7.S

Types of ganglion cells and their optic nerve fibers in the frog. Above. The four types found in the optk tectum, distinguished by thei r responses to four kinds of tests. ERf, excitatory receptive field; s.g.r., stra tum griscum centrdle of the tectum; •. g.s. , stratum griseum superficiale. Below. Diagram of the connections between the elements of the retina converging on a ganglion cell; presumably differ­ences in the details of these connect ions and their transfer functions account for the types. II. ama­crine cells; B, bipolar cclls; G, ganglion cells; fi , horizontal cells; R, receptor cells. [Grilsser and GUsseT-Cornehls, 1972.]

Type 1 tlbor

No spikes in response 10 room light ON or OFF or to e large moving bar

or to i lluminating or moving a natural-looking image.

Good re5ponse to small, dark mOving object In II 3" field,

even if the illumination is very taint. but

not ilthe edges ere too fUllY ,

or to reversed contrast.

Figure 1.11

8

Abstrdctlon e .. rly in ,m .1(ferellt pJthway. 1m· pulses recorded from .. frog oplk nerve fiber in the roof of the midbr .. in. [Based on d .. ... of M"tur"nil al aI. , 1960.)

Light inion sily

0

-'

light inten sily

0

Figure 7.10

Flash Flash

• ~

Stationary obloct

Flash Flash

'~. 1p::..J

Change in position during dark

No response

8

Response

8

"Movement neurons" mdY respond to chJngc of position, with ~ forgetting time.

253

254

Chapter 7 Integr~lion at the Intermediate levels

promptly (eases if the light is turned off briefly, and does not resume after the light is back on.

Notice that units of type 2 respond to any sma ll, dark, moving object, espe­(i,llly if the motion is jerky. A hungry frog will jump at .my such object. Normally any object having these char­acteristics in the environment of a frog will be a bug, an edible object. Since there are many detectors of type 2 spread over the visual field, the activity of one or morc fibers of that type can both signal the presence of a bug and give its spatia l coordinates, hence both activate and steer a jump and a tongue flick. Tadpoles may lack some of the Iypes more important for the metamorphosed, hunting frog. Type 4 units in the frog's optic nerve signal general dimming, perhaps the approach of a predator or any large object.

The known visually oriented behavior of frogs in a natural environment in~

cludes finding food and avoiding ob­stacles and large, threatening objects. The fi ltering processes that occur in the retina in frogs can abstract the relevant aspects of the whole visual input signal for those behaviors. Studies by Ingle and by Ewert, us ing single-unit recording, ablation, and stimulation suggest that central processing builds on the retinal filtering by separately analyzing optic input for different behavioral meanings in different structures. The frog's attrac­tion to blue depends on the dorsal geniculate; its avoidance of large ob­stacles in jumping and its retreat from large threatening objects may depend on distinct, overlapping parts of the pretec­tum, and its approach to food upon the

tectu m. We may speculate that the phase locking of its circadi,'11 rhythm with the environment.,1 photoperiod depends on a hypoth.1lamic center, the suprach ias­matic nucleus, .'s has been shown in rats. Each of these four central structures re­ceives its own optic input, consisting la rgely of a distinct mix of the ganglion cell types.

As activity proceeds through first-, second-, third-, and nth-order neurons in a sensory pathway, the me.1ning of the impulse activity changes. We have been considering differences in meaning in parallel neurona l pathways as a result of divergence; we should note the differ­ences in meaning in successive neurons as a result of convergence. When the criteria for firing include .1 significant fraction of the complex features of ., stimulus that releases norma l behavior, we m.,y speak of recognition cells; th is is a matter of degree. Higher-order ce ll s may add dimensions to the criteria, such as novelty or familiarity. A novelty un it deep in the frog tectum may fire in response to a small, dark, moving object anywhere in a large (30°) field, but will soon cease, only to resume if th e same or another object wiggles in a fresh part of the field. A familiarity unit fires in response to a similar object, but, instead of ceasing, continues, even maintaining a low rate of discharge ("muttering") for many seconds if the object stops moving. It now ignores fresh, moving objects within its la rge (30°) field. It will fla re up if "its" object s lowly moves about within Ihe field, but will lose it and go silent if the object jumps too far-to a fresh part of the field!

In the auditory sphere too, we know of

•

units of a wide r,lllge of co mplexity of criteri;!. , In bats, for ins tance, a series is found leading to ce lls that do not fire in rcsponse to any pure tone at any intensity but on ly to frcq ucncy­modulated tones with a certain range, rate, a l.ld direct ion of modulation, like the bat's echo-ranging cry. In squ irrel monkeys, cells arc found that s trongly prefer a certain one out of some 20 tape-recorded sounds chosen from the 35 or so natural vocalizations in the species' repertoire.

Some workers distinguish between recognition of natural sti muli and feature extraction; thc form er is potentially more complex and may result from con­vergence of cells of th e laller type. The term "fea ture" in this context means lim ited aspects of a complete natural stimulus, such as duration or frequency modulation . Our information is still too lim ited to decide whcthcr some sub­systems in some animals work differ­ently from others in a fundamental sense. It is oft en supposed, for example, that some subsystems fu nnel s uccessive featu re detectors down to a s ingle recog­nition unit, like a " bug detector," whereas o thers never quite converge the set of re leva nt feat ure detcctors. We dis­cussed the samc problcm from another di rection on pp. 238-240. By whatever means, complex natura l stimulus recog­niti on mu st occur widely in nervous sys· tems-sometim es early in the pathw.1YS, sometimes late.

Modulation of input by ce ntral in­fluence via centrifugal (efferent) fibers is a potentia lly important part of the active fil tering in many sense orga ns, as was noted on pages 170 and 225. Inh ibitory

, d,

f/ " ., ga

Figure 1.11 Efferent fibers ' 0 <l sense orgdn. Axons from the brain to the ret in~ in the pigeon. Golgi prepdfa­tion. ig, displaced ganglion cell; f, n~t ~macr ine cell; gil, ganglion celt I~yer; ir, inner nurleM I~ yer; ip, inner plexiform I~ yer; 5, Snl,l 11 pM,Isol amacrine cell. [Maturana and Frenk, 1965.J

fibers to muscle receptor orga ns in cray­fish are illustrated in Figure 2.75. The y-effcrents to muscl e spindles a rc trea ted below (p. 267 el seq. ). T he mammalian coch lea (Fig. 2.80,C) and the avian retina (Fig. 7.11) arc also well-known exam ples. but the fu nctional Significance of the effercnts is nol yel adeq uately under­stood.

IV. ElEMENTARY NEURONAL NETWORKS, EMERGENT PROPERTIES OF CIRCU ITRY

In Chapter 3 we introduced some well­stud ied examples of connectivity and ccrtain general prin ciples of circuits of neuron-like units. Here we extend the di scussion to emphas ize th e physiolog­ica l consequences. Three kinds of ,Hrays will be chosen; these may be c.lllcd "networks," fo llowing th e usage in th e literature on neural modeling, me,m ing any assemblage of con nected neurons .

255

Secllon IV Elemen' MY Neuronal Nelworks: Emergent P,o!,(, rties of Circuitry

I

256

Ch3pter 7 Integration at the Intermediate levels

A. Mutually Exdtatory or Positive-feedback Networks

If two or morc neurons are capable of exciti ng each other, then input that ex­ceeds th e threshold of one will likewise excite the others (Fig. 7.12A), Should the others exceed threshold as it result, they feed excitation back to th e first, and a runaway process ensues. The whole

/

Inetwork may come into a state of max­imum activity and, without lim iting processes, might stay in that condition. Most neurons have relatively long-term sel f-inhibitory processes, such as adapta­tion, accommodation, or fatigue. As the neurons in the positive-feed back net­work begin to fatigue, one or more of them will decrease in frequency. They then excite the others less, and hence

I I Figure 7.12 Simple networks wilh (AI mulua1 excitMion and (B) muhIJ I inhibition. lnpul (omes from presyn­dplic axons. Excitatory synapses shown by forked axon terminals, inhibilory by te rminal balls.

also receive less excitatory Feed back. Just as the whole network was able to rUIl

away to maximum activity, it now ru ns away negatively to a min imum state. Once the adaptation or fa tigue has worn away, a n~ cycle C,lll begin. Networks of cells, each of which may not be capa­ble of rhyth mic bursts of activi ty, can produce bursts of more-or-less syn­chronous activity in all members.

Networks of this kind have been demonst rated in severa l cases and may be widespread. Inspiratory interneurons in the medulla are probably so con­nected, perhaps leading to their rhythm ic bursting. Certain cells in the brai n in several gastropod molluscs have been found to be positively coupled. They produce synchronous rhythm ic bursts ') that act as triggers in the control of feeding and other activities. Deca pod crustacean hearts al'e controlled by t,ri\ ganglia containing only nine cells. ®-; ~ ~ baweerr- some of these is known at least to aid in the build up of the heart-beat.

B, Mutually Inhibitory or Negative-feedback Networks

Two cells, or two clusters of cells that inh ibit each other, may produce alter­nating single impulses or rhythmiC bursts. This arrangement, ca ll ed reciprocal inhibition, is a common fea­ture of the control of antagonistic effec­tors or actions. The activation of motor neurons of one group of muscles is often coupled, both by feedback and feed for ­ward, to inhib ition of the motor neu­rons controlli ng antagon istic muscles (see Section V, p. 266) . This ki nd of reciprocal inhibition is di,lgrammatically simple in

,

Vest>bular l'Iuclel VIii n(trve roolS

Figure 7.13 A reciproedlly inhibi ting pair of neurons. The gi.1l1 t cells of Mduthner in the medull~ 0' nsh, sche· 'll~ticdlly shown, with the inhibitory colldlerdl (/I) Indicdtcd only on the left, the indirect VIII nerve afferenls (8) only on the lefl, ~rrd the dired VJJJ netve dffercrrts (e) CO.lllrlg orrly from the right, ~lthough .III the5e components ~re rully bildleul. I Rct~l.lff dnd Fontdlne, 1960. 1

the Cdse of the paired giant Mauthner's fibers of teleost fish and tailed amphibia (Fig. 7.13). The cell bodies and large dendrites ofl hese neurons are situated in the medulla, where they collect input, including particu larly Ihat from Ihe eighth crania l nerve (sec also Fig. 2.17), The axons cross over before descending the spinal cord to synapse on the motor neurons of the longi tudinal musculature. Each axon has a bra nch that ends in an inhili,ilQry: 5 na se on til contra I ral Maulhner's cell axon hil , nca r the spike- initiating site. (This is an electricill synapse. ) An impulse in one cell pre­vents a simu ltaneous one in the other. Each descending impulse activates, nearly synchronously, an extensive lon­gitudinal body wa ll musculature on one side, causing a twi tch-li ke curvature of the posterior body region or tail. This startle reaction begins the familiar, sud-

den "jump" of some fish when the aquari um glass is struck. The vibrations excite the sensory endings in the vest ib­ular apparatus of the ear. Although these cells have the largest axons in the body, and although each has thousands of input terminals and must be often bom­barded with very many impulses per second, they produce on ly the occasional output necessary for start le reactions.

Another system of one-to-one alterna­tion is found in the neurons driving the tymbal muscles in some cicad.l s. A pace­maker interneuron firing about 200 im­pul ses per second drives two mola l' neurons, which each fire at halF this rale in exact alternation and precisely phased with respect 10 the pacemaker. The alternating clicks of the two tymba ls during song double the sound frequency poss ible if they were synchronous.

Reciprocal inhibition Cdn lead to

257

Secllon tV Eleillelltuy Nrurorrdl Networks: Emergent Properlie5 of CIrcuit ry

258

C hapter 1 [tlt cgr~tlon at the [nlermedl .t!e le ... el~

Threshold 20 mV I

Rebound "'lI f table (Iormal)

II II I Stimuli Inhibitory spikes in

I I I I I I I Spikes out

A

1111 1 1111+1--111111 1 11111 11 1I1IIH-lH+1I I 1-1 --+�m���------~II I 11111 1 111111 1111111 1111111 III III I 3 iIIl ~III IIUIII!llIIllllllyllmllYilll Briel I.p.s .p. barrage starlS long barrage

alternat ing bursts slops

8

O>---~ '--IlIII----!IIIUIIIIII-I -IIIllllIIIIlIlIlIlIllI--IIa8111f--IIIII-~IIII'BIIIIlHlIl Il IlI----III! IIIiIIIIIII--II~1 1l1li '-.' - --+1111111111 1I 111111 111l---111111111 lI ;mllll- 11IIlI1I1I 1I1"1111U1l 11 11111 11 1111 RIll--3 I I I I I I Single I.p.s.p.

triggers paUern

c Figure 1.14

l a tephsse intensifies

Laler phase resets

Rccip rOC.t1 il1hibilion .. nd pastinhibitory rebound provide ncxible mechanisms (or ~cnerJling bursts, A. POSlinhibitory rebound in a model neuron. A neuron 11t.11 is not spont.lneously ,lcl i"e receives dn inhibitory input and produces spike o utput by rebound . lJ. Two such neurons, reciprocally inhibi . tory. c;on giW! a long 5eries o f alternating bursts to ~ brief input bArr~ge from ~ third neuron. C. A single inp ut impulse has d iffe rent effects .according to the phase of the .alternat ion when It url\"!'s. [perke l and M ulloney, 1974.1

alternating bursts of act ivi ty in otherwise nonbursting cells. The networks shown in Figure 7.14,B and C consist of only two interconnected cells, but they could represen t two populat ions. Input to the network is inhibitory in this exam ple, .1nd reaches only one of the cells. Th e

com mon neurona l property of post­inhibitory rebound (Fig. 7.14,A) is invoked in this model, and causes bursts that suppress the ot her cell. As the re­bound burst in the first cell s lows down, the second is dis inhibited. Released and rebound ing in its tum, the second fires

, 1

1

,

and inhi bits th e first. As the seco nd ce ll s lows, th e fir st recovers, and the whole cycle repc.l ts.

Whether dependent on rebound or not, some such si mple mechanism for alternating burst acli vity, though difficult to demonstrate, see ms to operate in f.wo rable materials, like the lobsler stomatog.lstric ganglion. We believe it to be an important mechanism for many kinds of rhythmic and alternati ng be­havior. Inspiratory and expi ratory inter­neurons in the mam malia n medulla probably inhibit each other. Locomotory systems in ma ny animals may involve r~i procal inhibition between pace­makers for antagon istic muscle sets. Ikciproca ll y inhi biting networks can

perform a variety of fu nct ions, according to the pa rt icular transfer functions of the

' synapses and the input and output con­nections. T he following examples are theoretica l and q uali ta tive; proof that reciprocal inhi bition is the mech.1 nism that operates in living an imals is incom­plete. Given certain properties, recipro­cally inhibiting networks can act as gates, switching rapid ly fro m control of the output by one input line to another. This can recur at a steady r.lIe, for ,1

steady-s tate input, providing a pace­maker in which no s ingle cell is the essential clement, as discussed above. With cert ain dynam ic properties the same simple circuit C,ln act as an intensity-lo-Ii me con verier that m,lY be useful in comparing the strength of two inputs by their relative dura tion of con­trol of some downstre,lm system. Slight changes could provide sensory sca nning by periodically or irregularly sampling e,lch of a number of input li nes and giving them control, in turn, of some later elements. In a cell with a large number of input li nes converging on it, a

drastic increase of acti vity in one li ne would ord ina rily be drowned in the background activity of the ot hers; but reciprocal inh ibi tion in the same cell might a ll ow one line to dom inate if its activity were to rise above some level, and thus serve to d irect allention or to switch control. Still another possible use of such circuits, with only slight changes in the coupling functions, might be as an alarm system. Time- and load-sharing /,1 delay lines. null detection, and filt ering ! are other theo retically availa ble con- I seq uences of such networks.

An array of inhibitory cross-con­nections can be thought of as an arelM in which there is competiti on be­tween par.ll1el streams of impulses. Both the nonlinearity of dependence on act ivity levels and the cri tical inflections in the input -output functions wou ld a ll ow one strea m to win control. A spectrum of properties is possible from democratic to oligarchical to dictatorial.

The reliability of performance of networks is considered on p. 233.

C. Lateral Inhibition Networks

Th is term differs from the precedi ng head ing in directing a ttention more to l'l yers or arrays than to alternate cells or groups. Many neural tissues consist of layers of similar neurons that inh ibit e.lch other either direct ly or ind irect ly. The inhibitory connections spread from any part icular cell to make conta ct wit h neighboring and more distant members of the layer, but wilh decreas ing dens ity, aSwe saw in Chapter 3 (p. 11 1). Hence the effect iveness of inhibition decreases with distance.

Some effects of latera l inhibition are ill ustrated schematically in Figure 7.15.

259

Seellon IV Etrmt'nluy Ne uron,,] Ndworks: EIlIt'rgt'nl t'rop"'rlies of C ircuitry

260

-

A

Stimuh.l! in tensity

B

Inhibitory sensory nOl,l rons Interneu rons

5 /0 + $ O..;'-------~,,?--~J:0"''----

0..,/

0' ,s'---------<<E;"-"~" s ~ - (0,:'-----a.. __ '"

o,:s'--------~C~"-'-<'~-(o,:S'--C+~1 "Yo"v/'"

O,S'-"+"E'--____ ~O"-'-'-"J S + E + I ~ - (0':'-'--"-''-' ;<,',).

0;s'-+:':.!E'--____ ~<::~"~,-,-:'~,1os + E + 21

O"S'-'+"E"--____ ~cr:c"-'-'-'~(OS + E + 21 0;;;;:0..,,/"

"S'-+:':.!E'---____ ~C:"-'-''---', 5 + E + 21 0- ~ (O~'-"'-'--" Receptors 0 Second-order

Inhibitory sensory neurons Interneurons

c

-

" Darker" "Brighter"

~----. Physiological

ellect

o

1

,

Flgu.~ 1. IS I/llriHg r~SO') l .ltl' r.ll inhibillon. A. A p.lUl'rn 0' unifolmly gr.y ,"r.lS with shup f>dgn. rrpr~rnlinR .I "isu.1 fit'ld _n by .In eye. (M.lch b.nds • .In illusion in ",hlch .Ipp.onmtly d.lrker .nd lightl'r lunds .Irt' _ n .I round t'.tch t'dgt', Mt' not evidl'nt to us in this conrrguration.) B. The sllmulus ploul'd <IS inlensity (horizontlll) .Ivinsl lpo1 lial e>:tt'n t (,-ert ic", IL emphllsizing the unifonnily of the pnysiclll intt'Mily within ('IIC"h IIrt'll. C. A network of rtc~ptOfS ",nd S('cond~rd('r neurons with rt'ciprocal Inhibitory C"onntelions vi. inll'rneurons (brokr n linrs). Spont.lnrous act i\~ l y in the re«plors (5) is .IugmenlOO by ucildtion (E ) due 10 light. Thl' network CdU$e5 Ihe sl"<ond~rder nl'u.ons 10 show 5 dC"ti vity dUg-1111'n tl'd by E . nd/ or rt'ducl'd by inhibition (I ) in s ingle or doubll' (11) dose. D. Th(' output, ('(jUi"d' lenl to our $ensdlion, piatti'd .IS dMker or ligh ter th.n tht' bolCkground dut' to S.

First consider Ihe behavior of one spon­taneously active cell in the network while a stimu lus is moved about its input field . As the stimulus approa ches, it first excites neighbors of the recorded ce ll. Si nce they inh ibit Ihat cell, it responds by a decrease in firing frequency. If the stimulus passes directly over the re­corded cell, it is excited 10 fire above its normal ra te, but as the stimulus moves on il is again depressed by ils neighbors. Any cell in Ihe network may be char­acterized as having a recept ive fi eld that has an excita tory center and an annular inhibitory surround. Th is is compa r.lble to the ON-center ganglion cells of the cat ret ina.

If, ins te.ld or looking at the response made by a s ingle cell in the spatial array or Figure 7.15 to a moving sti mulus, we examine the ou tput of a whole line of cells in the array while one hal f of the array is stimulated more than the other half, lVe see an abstracting function of the network. Cells in either uniformly stimula ted half of the fi eld all in hi bit each other symmetrically, but th ose at the edge do not. Ce lls on the strongly stim ulated side of the edge are inhibited wcak ly by their nCighbors across the edge whi le they strongly inhibit those !lame neighbors. The result is especially high .llld low firi ng frequencies CI t the

stimulus edge. In terms of the fir ing frequencies or Ihe cells in the network, the stimulus edge has been enhanced; there has been a spatial di ffere ntiation of the input signal. A second layer of neurons cou ld be so constructed tha t it lVould detect the edge only. Latera l in­hibition probably explains our psycho­physical illusion known as "Mach bands" (Fig. 7.15,0), and perhaps oper­ates widely to en hance the sensitivity to cont rasts.

Since the lateral spredd and ex trd synapse take time, there is a delay in the inhibi tion. Th is confers a tem poral prop­erty of freq uency se lect ion or fil tering that acts to attenu.l te ra pid changes as a function of dista nce rrom the center and to prolong the exaggerat ion of the pri­mary response at the center, again fa vor­ing low frequencies.

It Cdn also be seen that the inhibition exerted on a given cell by those sur­rounding it can be redu ccd by stilllulation of a slightly more distant popu lation. The di stant popu latio n red uces the lcvel of activity of the nearer one, wh ich "d isi nhibils" the center.

These phenomena were first described by Hartline and his .lssocia tes in a se ries of elegant ex periments on the compound eye of Lil/IIIIIIs. The same latera l inter­action exist, however, in virtua lly "II

261

51"<1Ion IV El tm~nl.lry Neuronal Ntlworks: Emt'rgtnt Propt'Tl ies of Circu itry

262

Chapter 1 Integr.lHon at the Inlerm~d l~te l evels

sensory syslems in both vertebrates and invertebrates and may o peT,l le at the earliest stage in the pathw,lY and / or at later stages, even in the cortex (Fig. 7.16).

We h,wc seen in Chapter 3 how lateral inh ibition might result from recurrent collalera ls of the axon ending on neigh­bori ng cells (Fig. 7.17). We saw holV such inhibition ca n be complicated in the cerebellar cortex by th e inhibitory influ­ence of recurrent coJlalerals of Purkinje cell axons, not only on neighboring Purkin je cells bu t on basket cells, thereby disinhi bit ing Purki njes in a cer­tain geometric pattern. Renshaw cell laler,l] inhibition via molor neu ron re­curren t collaleraJs is trea ted on page 269 (Fig. 7.24).

D. Mixed Networks

If both excitatory and in hibitory connec­tions exist in homogeneously in a set of cell s, the variety of possible outputs expands to the degree that it is useless to make (l priori or genera lized, uncon­strained models. A recently studied real exa mple is, however, worth describing (Fig. 7.18). T he stomatogastric ganglion of crustacea ns controls the stomach, one part of which contains the gastri c mill used to grind food. The gangli on con­ta ins 30 cells, almost a ll of which are identifiable ,md consta nt in con nections and influence. Ten are motor neurons to the g.lstric mill part and 14 to the pyloric part of the stomach, but both groups also have direct infl uence upon each other. Furthermore, two interneurons are pres­ent with connections to both sets of motor neurons. There are 123 known inhibitory connections and only 6 ex­ci tatory junctions. Twenty-nine of the junctions are electrotonic, the rest un-

known <l nd presu mably main ly chemica l. All the fu nctiona lly established connec­tions can be an.l torn ica ll y justified by fibers vi sualized by inj ection of Procion yell ow. How mu ch spontaneity there is cannot be sf'ated, but it must be con­s iderable. Among these junctions, ., v,uiety of integrative input-output prop­erties are found. It seems like ly that if we could unrave l the web of influences, there wou ld be bot h excitatory and in­hibitory reciprocity, latera l effects, and mixtures of spontaneous rhythmicity with imposed bu rst-sha ping effects. The norlllal activity is largely a rhythm ic se ries of bu rsts of repeatable but labile spike pattern; the known connections ex plain much of the deta il of the bursts . Alth ough many of the connections of this network are known (nearly a ll; hence far more in proportion than for any ot her known system of some com­plexit y), it is difficult to assign causes to th e bursting phenomenon itself. Does th e pos itive feedback, by itself, cause one group of cell s to burst, or is the reci proc.ll inhibition relationship with anot her group necessary, or even suffi­cient? In theory, ei ther mechanism could prod uce bursting, but physiological evi­dence suggests that both kinds of con­nect ions exisl. Perha ps both mechanisms oper.lle synergistica lly to make th e whole system more stable.

E. Connections Ensuring Synchrony of Activity

Although exactly synchronous activity of neu rons is not known to be req ui red in m.1 ny insta nces, it is important in a few, and these are interesti ng in showing the fl ex ibility of design with whi ch cells ca n be connected, among th emselves .l nd to

•

",gill :g 20 .,.30 '~ 40 ·

~ SO 20 30 50 100

i ~~1iI 30 ~ 40

50 20 30 50 100 A Vib rat ion frequency (Hz)

Figu re 7.16 Sh~rjX'n i ng by suppressing sensitivity on e.lch side of the best frequency, with converging input. A. Thresholds for sensMion as the end­point; vibr<ltion felt by 2 or 3 fi ngers. 8 . Thresh­olds for single neuron firi ng as the endpoint;

0

- 20

-" - 60 , - 80 ,

m • .~ - '00 oE ." ,-£~ 0> 0 -0 0_ < • - 20 • < ~

-" - 60

- 80

- 100

• "'.

Lower aUditory center

0.2 0.5 2 5

V ! • t • ..

• • • • \ • Higher .. auditory H cen ter \1

0.2 0.5 2 5

unit responses to sound at lower and higher ~ udi ­

tory centers. It should be noted t h ~ 1 such narrow curves at higher centers are no t common; units wi th m.my types of curves are found. {Von Ilckesy, 1967.) B Sound Irequency (kHz)

~i

f igure 7.11 Recun en! (011,,11'. ,,15, The line fibers are the array of colliller .. ls of three pyramiddl cells in the corte .. of the killen. IScheibel and Scheibel. 1970d.)

263

10 20

10 20

264

""

" GASTRIC MILL '" REGION

PYLORIC REGION

A

• E lPGN al (2)

c

B

ALN '1 $ " hO' !· .. ·· t! ! ::; 'J ' Io\;",IHI I!

LGN "1' j ' " . ," "' " . ~~~ ifjirg ii . k':: :: •. -".,If:;'; 'I:;:: }ff PNI ~ I • , I I , I , , • I I I ! , , I I I I I I I I I I I , I

d-l VN :' .••. ' r !.' ~,. .:!",': . •.. · ... - .~ · ..... I.' •••

~ ____ r, : E I I i ibe.s I : (2) leG's of 0 o 0 .... n ___ J

Inl 1

MGN LGN

gm l .2,3a gm4 c7

,,' cpvl a. lb. 2b ,2. p2b- 13

•

the periphery, to ,lccomplish precise ti ming. These can be rcg,lrded as specia l cases of the gene r,\l problem of ass uring precise lim ing of sequences of activ ity. Among the best exa mples of systems requ iring nearly exact synchrony are the electric organ discharges of electric fi sh. (Ot her exa mples are the oculomotor neurons and the neurons innervating sound-producing muscles and wings in certa in ani ma ls.) The electric organ celis (electrocytes) are in most cases com­posed of modified muscle cells thai are oriented in seri es so th,ll Iheir depola ri­zalions ca n be sum med. It is necessary in Ihese orga ns Iha t a ll the elect rocytes be activated nearly sim uli,lneously, and in some species wit hin « 0:1 Insec. Th e

Figure 7.18 URdHS P~8t)

com n1.1nd or p.lccmaker neurons ,1rC usually located in the medull 'l 01' mid­brain and act th rough interneurons and motor neurons on the electrocytes. In all C,lSes, these neurons are electrica lly coupled to each other via gap junctions (see p. 333). The electrotonic connections may be between celi bod ies, or den­drites, or via presyn,l ptic fibers that form electrical synapses on Ill.lny of the cells. These synapses, by equa lizing the level of depolarization between different celi s, are both excitatory and inh ibitory, but ensure synchronous acti vi ty of all cou­pled cells. The simplest exampl e of such coupling is in th e electric catfish, Mn/npierllrJIs, in which there are two electromotor neurons, one on ei th er s ide

A si mple sys tem of some th irty neurons; the stomdtog<lslric gdnglion of .I lobster. A. Side view 01 lobster stom­.. ch. The two principal funct ion~ 1 divisions, the g.1strk mill region .Ind the pyloric region, are $Cpdr .. ted by .. broken line. PMt of the st Ont.l t og~ st ric nervous system is shown together with. few of the stom .. ch muscles th .. t it innerva tes. The stomdlog .. stric g.1ngl ion (SIG) ",n be seen on the dOrs.l1 surf .. ce 01 the stom .. ch just dbove g.>stric mill muscle I (gm l). Other lettering identifies muscles .. nd nerves. B. The two bdsic rhythms pro­duced by the isolJ ted g~ngl ion Cdn be $Cen in the extr .. cellul .. r recordings from neryes supplying the hvo different regions of the stomdch. The top thru nen'es (/U N, LGN, DGN), not a ll shown in part /I.. supply muscles th .. t oper .. te the g.lstric mill . Note thdt the bUr5ts of dct ivity be .. r a particuldr ..se rela tionship wi th e .. ch other . nd th .. t the dura tion of cadi burst is ~lseConds. -The th ru lowe"i=tr .. ces cont~ in axons of mot~ns (MVN, PN, dol VN) supplying p.1!2ric musclt!"s. The bunts are much shorter in dur ... ion, and the overa ll frequency is •• bout seven times t ha~U!!.e g_~st~1. Note ~ Iso t hat the bunts of ~ctivity ,;;afnt ~in d pJTticular phJse rel ... ionship. The d.LVN t r~ce cont .. ins dxons innervating both regions (sec 'lbo\'e) and the long bun ts sun in this tr,lce ,I re from .. xons to muscle 8111 3 ~ . WoIrts A ~nd H, Selverston and Mulloney, 1974. 1 C. Neurond l con nect ivity diagrdm for the lobster stom .. tog.ilstric ganglion. All the cells except interneurons 1 and 2.

Me motor neurons. The top ten neu rons control the g .. st ric mill cycle, and the bottom fOllrtee n ce lls cO'ltrol the pyloric rhythm. The dxonal pathw"ys, ~s well as the muscles innervated by the celis, Me known. Somol, neuro­pile, and ~xon .. 1 pMtS of the celis Me Indicated on the left. Broken lines Mound some of the neuropile Meas in_ dicate that ce lls of tl1.l1 group Me electrotonically conilCc ted and can be considered together. Known connec­tions with the centrdl nervous system Me shOWIl at the top. Round dots represell l chemic.l l inhibitory syll.lpse5; tridngles represent chemicoll excitatory synapses and resistors (= electrotonic junctions): F, functional synapse wi th strong effect but no clear uni t"ry po5t syn~pt ic potential; f, exci tatory fiber input from commissuroll 8<ln-gli ... LPGN, Iolterol l post('rior g.st ric neuron; MGN, medi .. n gas tric neuron; LGN, latera! gastr ic neuron; '"' I ,md 2. interneuron neuron I .. nd 2; GM, g .. str ic mill neuron; DGN. dorS<l1 golstric ne\l ron; AMN. ,Interior me­di.lI1 neUTOn; IC, inferior cMdiolc; VD, ventr iculM dilator; PD. pyloric dil .. tor; A8, .Interior burster; LP, l"'er .. 1 pyloric; PY. pyloric; eG, conlmissur .. 1 8"ngli.; STGN. stomdtog.lstric nerve. ICourtesy of A. Seiv('r5ton. j

265

Section IV Elfnlent.lrY Neuronal Networks: Emergent Properties of Circuitry

266

Cluplt'r 7 Integrdtlon .at Ihe Inlerp.edi.alt' uvels

of the /lrst spina l segment, lightly coupled to each other. In ot her fi sh, there may be as many as 20- 50 p.lccmahr cells coupled to each other, a ll of which electrically excite internuncial relay cells that are also electrotonica lly coupled.

Given a synchron ized command, how aTC electrocytes at di fferen t distances from the brain activated synchronously? Thi s is accomplished, in most cases, either by systematically shortening the lengt h of branches to the progressively mOTe distant electrocyles, or by altering their diameter so that the s lower con­duction velocities ofaxons innerva ting the nearer electrocytes compensate for Ihe shorter distance. In several species the gra ded delay is built into th e electro­cytes. Dista nt ones are innervated nea r their principal surfa ce, nearer ones at the end of long, slow-conducti ng stalks of the elect rocyte.

The ex istence of these mechanisms for building compensating delays into neu­ra l ci rcuits opens the poss ibility that similar refinements of st ructure may be involved in other systems that are sensi­tive to the precise tim ing of inputs, such as the parts of the auditory system re­sponsible for sound localization or the motor systems controll ing speech, eye movements, midd le ear muscles, and the li ke.

V. STIMULUS-TRIGGERED REACTIONS, THE ORGANIZATION OF REFLEX ES

With some of the principles of effector control. of sensory input, and of ele­mentary circuits in mind, we can now turn to the lowest levels of motor re-

sponse to IMturaJ stimuli . What can we say is the s implest nervously mediated response?

The verlebrate stretch reflex, while specialized for simplicity and speed rather thaI!, be ing primitive, is a good starting poi nt because it is mono­synaptic-that is, no interneurons arc interpolated between afferent fibers and motor neurons (Fig. 7.19). Skeletal mus­cles contain numerous sense organs, called muscle spind les, that are sensitive to s tre tch in the ax is of the muscle fibers (Fig. 7.20). Passive stretch, as by the action of gravity or of other muscles, causes a train of im pulses to arise in the term inals of a sensory axon in a muscle spindle, and these are conducted via the dorsal root to the spinal cord, there 10 be distributed in axon collaterals to the dorsal and ventral horns of the same segment on both s ides of the cord, 10 nearby segments up a';;'d down the cord, and to the dorsa l colum ns ending in nuclei in Ihe medulla. Of these destina­tions one is the large alpha (a ) motor neurons of Ih e sa me muscle, wh ich are excited. This excitation is distribut ed by the axon branches of the motor neuron to a group of mu scle fibers, usually between WO and 1000, called a molor unit. Conlrdction of the molar unit tends to ca ncel the s tretch. This proprioceptive refl ex acts as a tonic muscle-length servo (like a gyrocompass), tending to main­tain the length aga inst any change in load either way. Gravity is an import.lnt norma l stimulu s; th is re Aex arc is the principal ant igravity circui t.

AI th e sa me ti me, coll alcrals of the spindle afferents fil'c interneurons thai in turn exci te synergistic motor neurons and inhibit motor neurons of antag­onistic muscles on the sa me side whi le

•

FIgure 7.20 267

Til, mll"multiaH /IlI/SlI, sp;/Jdl,.

Figure! 7.19 The monosYn.lpl i<: slrttch rrAu p.ilhlV"Y in IllAmm .. ts, simplifi~ by oml .. ing Ihr olher dtslindlions of Ihr s.am(' .. ffrrr ni n('U ron, Ihr oIher inputs 10 Ihe s.amt molor nturon, ils oulpul rt.'Currenl coll .. ler~ ls, .md Ihe molor conlrol of Ihe spindle!. IGlrduer, 1963.[

Nuclear chain fiber

Annulo· splrlll

endings

motor neurons of the homologous mus­cle and its synergists on the contralater.ll s ide are inhi bited and antagonists ex­cited. The phenomenon of reciproca l innervation contributes to coordination by prevent ing antagonists from working against e.lcn other.

Given this iength-maint.lining re Aex. how ca n the organ ism walk or volun­tari ly ch.l nge muscle length? If higher cen ters were s imply to comma nd "" ­motor neurons to greater or less activity. the stretch renex wou ld quickly cancel the effect. Somehow. the set point (SoH wert) must be changed. This c.lIl be done by adjusti ng tension in the muscle fi bers within the spi ndle. ca lled in tra­fusal fibers.

There is another set of motor neurons in th e spinal eOI'd that also se nd th eir e(ferents to th e mu scle. These are the gamma (y) effcrents. emana ting fro m small motor neurons (Fig. 7.21). They innerv.lte the muscle fi bers in the muscle spindle and alter its sensitivity. In the centr.ll region of the spindle fi bers is a

Flower -'.-,...,~ spray

endings

..... SillS' orgjHl ,x(;I,d by j1fl$S illt 51ft/ell , 0' IIdivII/ioli 01 ;/$ 01011 ;III,i,Jsir ml/Sllt

fibtr5 by ga"",111 m% r axo/IS.

noncontractile enlclrgement filled with nucle i, and it is to this zone that the stretch receptor fibe rs come (Fig. 7.20). The end regions of the spind le fibers are contractile and are innervated by the y-efferents. When the y-effe rents in­crease their activity. the spindle fibers shorten, the cen tral region is st retched, and the stretch receptor fi bers are ex­ci ted. It is as if the muscle had been passively s tretched. The opposite effects obtain if the efferents have reduced activity or the main muscle has increased activity.

The set point of th e postur.ll renex ca n be affected by tens ion changes in the spi ndle muscle fi bers. When these con­tract they stretch the muscle receptors but do not change the lengt h of the muscle. The resulting increase in stretch

1111 '-H -t-7 r Molor J axons

Type Aa afferent

axon

Type All al/eren!

aKon

Nuclear bag fiber

Exl ralusal skelelal

muscle fiber

>68

Chapter 7 Integr~tion at the Interm~dlale levels Spinal

cord

axon

Muscle tiber

Figure 7.21

From brain 2

I

The y·loop servomechanism. Commands (rom the brain may operate by contract ing the mu~le fibers of the spindle via the y-efferents in the venlral root (2). This excites the stretch reflex (3) Jnd hence the main muscle. There are also direct conllections (not shown) from the brain to the Ill,,;n motor neurons. [MNton, 1972. J

receptor firing f<lle excites a -effercnts, which elicit greater tension and hence muscle shortening. Shortening continues until stretch receptor firing rate is re­duced again to normal values. The cen­tral nervous system can command a long-lasting new length by varying fre­quency in the ),-efferents and thus changing the sensitivity of the stretch receptors with respect to muscle length. The setting of muscle length by way of the y-efferent is called y-loop activation.

The motor neurons (both y and n) for one muscle are loosely grouped in the ventral horn of the spinal cord. T hey receive many inputs, usu.1lly in parallel. The stretch receptor afferents excite only the n-efferents. If they excited y-efferents the reflex would comprise a positive feedback loop, which would run away to

maximum or minimum tension . Descend­ing excitation or inhibition from the brain impinges on both y- and Ct,-fibers. We have already encountered the gen­eral rule- that small units are more tonic and have lower thresholds for n.1tural stimulation. The y-neurons are smaller and more sensitive than the Ct"s. A weak command from the brain excites the y's, which in turn change the set point of the muscle-receptor-Ct'-efferent reflex, and muscle length changes to compensate for this. Strong descending-movement com­mands excite both y's and n's. The faster-conducting a's initiate a move­ment that wou ld be later cancelled by reflex function were it not for the more slowly developing effect of y-excitation, which changes the set point of the reflex. Thus volunt.wy and other brain control

•

Goigi tendon

receptor

Figure 7.22 The tendon receptor for st retch. Another sense org~n, besides the spindle, is in the tendon, which differs by being stretched (excited) when the muscle conlr~cts as well .1$ when it is lo .. ded. The sign of its innucnce is such as to preven t overconlr.l(lion.

of movement oper.l tes largely via the y.loop balanced in vc1rious degrees with coactiv<l tiol1 of a's.

Anoth er set of receptors is on the muscle tendons, and these respond to stretch of the tendon (Fig. 7.22). In can· tr<lst to spind les, therefore, they respond in the same d irection to im posed load and to contr.1Ction of the mu scle (Fig. 7.23) . These receptors (Colgi tendon organ receptors) have no monosynaptic endings, but, via inlerneurons, they send (a) inh ibitory input to motor neurons of the muscle they innerv.1 tc and to its synergists, (b) excita tory input to molar neurons of antagon ist muscles, and (c) opposite inputs to motor ncurons of the correspond ing muscles on the opposite side of the body. These inputs might be said to perform a tension- regulating

function, restraining th e motor neuron from causi ng the muscle fibers to COll­

tract too violently. O ne more automatic subsystem is

important in hel ping to determine motor neuron activity loca ll y; this is thc Renshaw cell nega tive-feedback loop, acting on motor neurons in the spi nal cord (Fig. 7.24). Situated in the ventral roots there are branches of a-motor axons called recurrent collalera ls because they turn back and reenler the ventra l horn to synapse with small interneurolls, na med for Renshaw, who discovered them physiologically. These cells fire at high frequency when excited by motor neuron output and have the effect of inhibiting the same and neighboring motor neurons. The roles of this inh ibi­tion may include preventing motor neu­rons from excessive activity, focusing activity upon certain cells and perhaps su ppressing phasic responses more than ton ic ones.

Many reflexes are elicited by cxtra­muscu lar stimuli. For exa mple, painful cutaneous s timulation often gives rise to fl ex ion of a limb (Fig. 7.25). The fl exor refl ex to nocicept ive stim uli is not sharply loca lized; it may affecl all the muscles of a li mb. Both strength of re­sponse and degree of spread of response are related to stimulus strength. Input fibers do not impinge direct ly upon motor neurons, but on interneurons; the reflex pathw.1Y is polysynaptic. The input excites flexor action and at the same tim e inhibits ext ensor motor neu­rons. If the stimulu s is quite strong its effects may spread even to the conlra­lateral limb, where they are opposite in sign. The crossed extcnsion reflex pre­pares one limb 10 bear the extra weight shifted to it when the painfully sti m­ulated one flexes . Similar refl ex con-

269

Seello ll V Stimulus-Triggered RNction5: The Organization of HeOu t'S

figure 7.23 Spimflt IIfrsrl5 I"rdorr 't(flrlor~.

j Pll llorr , 1965.J

Muscle con tracted

SPINDLE RECEPTORS " IN PARALLEL"'

Muscte stre tched

Muscte contrac' ed

~

Discharge

Acceterated discharge

TENDON RECEPTORS ""IN SERIES·'

270

+eo Motor neuron

+<0

0 •

- 40

- eo ,-.s , '~

+ 80

~ +<0 n

• 0 0 • • D E -40 • ,

- eo

- 120

Motor

• -60~ -70 ___ :_p_s_p,

a! -80 I

~"'"W" 0 5 " 15 20 25 30

msec Figure 7.24 The Renshaw type of inhibitory interneuron. Axon collaterals from motor neurons activate Renshaw cells to high frequency discharge, which set~ up summating inhibitory postsyfl<lp­tic potentials in neighboring motor neurons of the same pool. [Eccles, 1964.)

ncctions are seen for cutaneous touch, pressure, and temperature receptors.

We have seen now some qui te genera l contrasts between flexion ,lnd ex tension reflexes, nociceptive .1nd proprioceptive, pain and postural, cutaneous and muscle-afferent, phasic and Ionic re­flexes . These are overlapping but not synonymous dichotomies. Anothet use­ful one, referring to their roles in causing normal behavior is the distinction be­tween elementa l and tuning reflexes; the fonner cause the basic sequences, th e laller adjust to momentary conditions.

Sherrington sta ted the concept of the reflex, one of the truly great and fruitful abstractions in biology, with these words: "The Ullil reae/iOIl hi l1erVOU5 integra­liOlI is the reflex. because every refl ex is an integrative reaction and no nervous action short of a reflex is a complete act of integration" (1906, emphasis his). Many authors have pointed out the limitations of the concept or criticized it as art ificia l, and Sherri ngton, as clearly as anyone else, emphasized the non­existence of a discrete circuit insu lated from others. Nevertheless, the funda-

,

Nociceptor / in skin

Figure 7.25 Nociceptors .. nd the polysyn .. ptic p .. thw .. y. An· other !OCt of inputs impinging on the motor neu­ron comc from pain and sirn ila r receptors. The sign of their aclion is generally excitatory to the ipsilateral Aexor and contra lateral extensor mus· cles ~nd inhibitory to the ips ilateral extensors .md contralateral nexors.

mental usefu lness of recognizing this category of responses has been amply proved by th e ins ight that experi ments based upon it have provided. The refl ex is a useful abstraction, but we qualify the bro.."ld sta tement that it is the unit of all nervous integration: (a) the arousal func­tions of the reticular activa ting system in ma mmals cannot be resolved into re­fl exes, nor can (b) mere sensing, (c) autochthonous action (arising from within), or (d ) many instincts.

Familiar examples of the phenomena under consideration are the stretch refl ex, the fl exion, crossed extension, sha ke and scratch (dog) refl exes, the refl exes of micturition and defecat ion,

the pi nna, swallowing, stepping, sneez­ing, sal iv.ltion, blink, accommodation, and tonic neck refl exes, placi ng, hop­ping, and righting reflexes. They arc re.ldy- mad e, un leamed, adapti ve move­ments, prompt and coordinated . The coordination is not observed just withi n each refl ex, which we might ex plai n as a fi xed pattern, but is observed even when conflicting refl exes are sti mulated simul­taneously, for th ere is al most in va riably a resolution of the pote nt ia lly maladaptive conflict or intermed iate action in fa vor of adaptive selection .lnlOng them. Cooper­ative interplay is also marked between sim ple segmental refl exes, long spinal intersegmental refl exes of foreli mbs and hind limbs, and coord ination of body, li mbs, neck, hea d, and eye refl exes, as in visually guided wa lk ing.

Though the present account of re­flexology necessa rily draws mainly from ma mmalian literature because lower forms have been less studied in this respect, we believe the princi ples en unciated are probably gene ra I. J m pl ici t in the conce pt of reflexes-si nce there is almost invariably more than one refl ex util izing a given muscle, and hence more than one cent ra l mechanism converging on the sa me motor neurons-is the con­cept of the fin al common path. The very ex press ion em phasizes the int egrative function.

The propert ies of refl exes (see Box 7.2) and the ru les by which they are used ma ke the best case for their reali ty and importance. We have already learned ma ny of these rules in the exam ples detailed above. Let liS look further at the ways they com bine.

Separate reflexes may be either com­patible or incom patible. The fortner combine adaptively into compound re~

flexes. For exa mple, a grav ity refl ex that

271

Section V Stimulus-Triggered RCdcllons, The O rganlution or Hrncxe5

272

Chapter 7 In tegration at the Intermediate Levels

keeps a fish uprigh t adds to a ligh t reflex that keeps the dorsal side toward the light, so that if light comes from the side, cert.lin fish tilt to a degree graded according to the ligh t intensity, the strength of gravity, and a central evalua­tion thai multiplies each in put by a weight ing faclor. The weighting depends on time of day, temperature, hunger, and other inputs, such as chemical signal s associated with food, mecha nica l inputs that ind icate a 5ubslr,liul1l, and visual input su fficiently imaged and "under­stood" to represent a substratum.

the nervous system d oes not release them both, One o r the other is sup­pressed, at each moment. Incom pa tible reflexes do not add algebraically or com­bine li nearly; instead, switches or pat­terned contro l ins ure normally adaptive interaction. This may show either in­hibition or facilitat ion . It is as though the system were preorganized for useful movements.

Incompatible reflexes are those that cannot be accomplished at the same time. If the hind leg reflex to scratch the back is elicited on one side in a dog and at the same time the sti m ulus for an

The tonic neck reflexes and a number of related postural reflexes d ue to vestibular and proprioceptive input interact with each other and with phasic movements as though add ing a bias o r "tuning" the response to the conditions o f th e moment. For example, if a load is li fted by wrist flex ion, more work c.1I\ b e done (stretch reflexes faci li t.1ted) with

extensor thrust of the same sid e is given, the head bent down or turned away from

Box 7.2 Properties of Reflexes

WIMt are the properties of reflexes that mani fest integration? (a) The threshold st imulus is very much dependent on conditions. (b) Above the threshold, gradation of response does not closely correspond with grJd,ltion of st imulus, (c) If the stimulus is repetit ive, there is USUJlly a poor correspondence between its rh ythm and that of the reflex response. (d) Single afferen t impulses are U511,11ly not adequate; temporJl summation is usually necessary to el icit a response. (e) A depressed excitability typically follows a reflex and is oft en qui te long. (f) Afterdisch.uge, or the prolong.dion of the motor neuron activity after the cessation of the stimulus, is ,1 prom inent fea ture of nl.lny reflexes, as though the mecha nism were org,lIlized to com plete a certai n movement in a controlled W<1y. (g) Spa ti .l lly and temporally patterned con trol of severa l muscles is probably involved in .111 reflexes.

The timing of inhibition is coord inated with that of excit<1tion, as is cleMly seen in the alternating reflexes, shaking, stepping, .lnd scratching. Even sim ple flexion and crossed extension reflexes show ch.lfacteri stic temporal patterning. (h) Irradiation with increasing intensity of stimulus occurs in some reflexes-for example, the protective flexion reflex. At threshold <1 response lll<1y involve a lim ited part of a synergic muscle group <1cross one joint, but with irrJd i,ltion it may spre.ld to other joints of the same append,lge, to other appendages and segmcnt.llievels, to the he,ld and neck. The spre<1d is generally salt,ltory and is confi ned strictly to certain Jines or muscle groups. However, apparently uninvolved muscles m,lY in fact be involved as objects of inhibition. The possible movements Me thus cirnllTIscribed in a chM<1cteristic p<1ltern. These properties help to define reflexes, to bring out thei r integrative nature, and to emphasize th e centr.ll deter­mination of detai ls of form and timing.

•

tha t Mill, because of the tonic neck re­flexes; if the Io.ld is met by wrist ex­tension, the opposite head movements enh'lIlce ou tput.

Refl exes Me normally \'/Oven into an integra ted f" brie, without sharp li nes. They Me gr"ded in amplitude by influ ­ences descend ing from higher centers and comb ined under the ru les of ,l

hierarchy. We ca nnot help wondering whelher the sa me ci rcuits Ihal are re­fl ex ly t riggered from Ihe periphery might be centrally triggered in patterned sequences.

Hav ing erected this edifice of plausible ass umptions and conclus ions from studies of reflexes, we should now raise the questi on as to wh.,t evidence th ere is for the centr"l origin of patterned im­pul se discharge.

VI. CENTI!AllY SCORED BEHAVIOR: PATTERNING IN SPACE AND TIME

One way to sta te Ihe fu nction of the n('rvous system is that it formulales appropriately pa tterned messages to drive the effectors. A core qu('stion in the study of intermedia te level integrAtion is: " How is this doner' T he patterni ng in tim(' may be treated as a more serious issue than tha i in space. Theoret ica lly it could arise in either or both of two W.1YS: by followi ng (a) timing cues from peripheral sensc organs or (b) liming cues frOIll central pacemakers or patt el'll generators. We may call the second mecha nism ,I central scor(' to empha size its potential complexity of detail, its mod ifiability from outs ide on any given occas ion, and its reali ty as a slored progr.un dPdft from external inputs. A score hdS ti mi ng built in, though subject

to exterlld l influen ce; a progr.1Il1 might have the same, but the term ,'pplies also to cases tha t merely ca ll for reactions, leaving the tim ing to effectors, trdns­ducers, loops, .lnd largely peripher.l l events.

Instead of a s imple d ichotomy we may disti nguish several poss ible mechanisms (Fig. 7.26). In A, we have a s imple reflex, . such as an eye blink, a sWdllow, or a cough, triggered by a stimulus. Slartle responses media ted by gia nt ribers belong in this category (see Box 7.3). Even here the t('mpora l pa Hern 0 I messages to various muscles is deter­mined by central pathways and integra­tive junctions. In B, there is sensory feedback from proprioce ptors ea rly enough to determin e a rh yt hm of recur­rence; a chain refl ex aCCOli nts for fre­quency, phasing, and amplitude. The cxtreme case, in E, is purely centra lly timed, without any immediate feedback. C and 0 are combi ned mechanisms in which the central spontaneity can deter­mine the bas ic rhythm but feed back may alter either the rhythm (C) or only the deta ils of the expression of the rhythm (D).

Probably all five mechanisms are com­ilion, though usually there has not been enough analysis to be sure which class an activity belongs to. In this section we are concerned particul.uly with some examples of C, 0 and E. In each exa mple, there is permiss ive or essentia l input from sense org,lIlS that may start or stop the whole pattern or innu ence the overall "central exci tatory state," to lise ShNrington's phrase. This is what lVe mean by spontaneit y, not that there is independence of the envi ronment for permissive conditions, bu t only for trig­gering the succession of ,1(tions. Spon­taneous rhyth ms ca n be in nuenced in

273

5~dlon VI C~nt'~ lI y 5c:ored 8~h~y lor:

I'.! tln ning in 5p~c~ .and Time

A B c o E

Figure 7.26 Five mechanisms of pattern fOTmulat ion. The three levels of neurons are understood to represent brJnching chains in whose functions integrat ive properties may a lter the actua l impulses ,md distr ib_ ute them spat ially as well as temporilily to the effectors (bollo"I). A and B are shown with receptors; C, 0, and E, with spontaneous p<lcema kers giving simple or group dischMges. Band C have proprio­cept ive feedback acting on the trigger neuron; D, only on the shaping of the pattern. [Bullock, 1961a.)

Box 7.3 Giant Fi bers and Startl e Responses

Striking among the behaviors tha t are merely triggered by environmenta l st imuli, but not further guided by ei ther environmental or propriocept ive inputs, arc the responses med iated by giant fibers . Giant fibers are Found in many invertebrate .m imals and fish (see ChJpter 10). Their phyletiC distri bution is scattered Jnd their st ructures are diverse; hence they a rc likely correl<1 ted only in func tion, not through evolutionJry or developmental relationships. BecJuse of this diversity, the follOWing generalizat ions l1lily have exceptions. Beciluse of their size they conduct impulses TJpid ly. Perhaps related to size is the fact that they can have ,1

large divergence rJtio-that is, one giant fiber can excite mJny other neu rons or muscle fibers . Moreover, the curren t available for electricJI trJnsmission of the act ion potenti<l l to downstre,ml neurons is large; elec­trica l transmission may be common in these systems. In all of the adequ,ltely studied C<1ses g i<1nt fiber ac tion

results in rapid, neJrly synchronous, widespread mus­cle act ivity. Behaviora lly, giant fibers ordina ri ly fire only with rather special input requirements; these often have the ch'Hacteristics we call startle.

The giant axons of squid arc the best known of all nerve fibers. We may briefly review their funct ional ana tomy (see <1lso Ch,lpter 10, p. 434) and behaviora l role. The muscles of the mantle of the squ id a re doubly innerv,lted. Many small motor axons From the stellate g,l1lglion excite rela tively few muscle fibers each, and these slnJIl neurons control the slower movements involved in respir,l\ion and ordinary swim mi ng. The single giant fibe r in each of the 6 or 8 stellar nerves on e,1(h s ide together innervate most or all of the muscle fibers of the m'11l tle. EJch giant fiber has In,my cell bodies in a lobe of the g,lnglion and is therefore a fused syncytium of many Illotor neurons. "The" giant fiber of the squid is the I<lst, longest, ,md I,lrgest of the 6 or 8 on

•

Box 7.3 (wlI/'-mud )

c.leh side; these fibers arc properly thc third~ordcr

gianis. Their input is from two second-order giant fibers Ih.lI arise in the viscera l lobe of the bra in, enter the stell.ltc g.lnglion, and synapse all each third-order giant fib er in tha t gilnglion. The 8 i.1nl synapses are one­to-one re lays; every input impulse causes an all-or­none twitch throughou t the mantic, nearly simulta­neously, .md a vigorous ejection of water through the funnel . When a squid is slilfl led. it aet iv.lles Ihe giant fiber system probolbly Viol the 5ingle bilaterally fused firs t-order giant unit in the ped.d lobe of Ihe brain. This unit is .J cornm,llld unit. as ddlncd on p,lge 279. There is probably no immediate feedback influcndng Ihe oper­a tion of the giant system.

The gia nt fiber system of earthworms consists of two pd rallel cha ins of electric~!ly coupled segmental COl11-mand units: a median fiber and a latera l eledrotonically coupled pair of fibe rs (see p. 405). These premotor inle reurons control largely overlapping musculature, the longitudina l or shortening muscles, plus separate muscles for the setae of anterior .lnd poste rior seg­ments. The inputs h.we some labi le overlap. The median giant is activated by stilrtling stimul i, such as a vibration or a tap anywhere on the anterior third of the worm, and causes anchoring by protrus ion of setae in the tail, plus shorten ing, which therefore retracts the head. The lateral giant pair is activ<l ted by mecha nica l stim uli in the posterior two-th irds, and causes anchor­ing of the head end, hence pulling up the tail. Each giant is therefore a unique, consistent chain of neurons acting as a decision unit. The adeqUAte input is from m,my receptors, dnd redches threshold only when some subtle criterion is met thdt involves ra te of rise, recent history, sp.ltial pattern, and a centrill excitatory st.lte.

The crayfish giant syste m hols been more completely studied, and its circuitry is diagr,ul1l11ed in the sketch.

Ma uthner's fibers in teleosts and .1Cj uatic amphibia are likewise premotor COllltll<lnd units receiving a large input from many sources, especially vibrdt ion receptors (see pp. 30, 106,257, 440). They normally fire only once or twice, firsl one side and Ihen the olher, causing the

" Lateral giant fiber /

en and iJ responses)

Fast lIexol motor neulons

Tactile rBceptors

Anlilacili tating

'h~'r".p""

~ , / B ","W", ,,,"" 1 L '"""'''00'

LG ;---electncal synapses

/ Rectifying synapse

MG MOlor gian t cell

FleJ;or musculature

Zucker's dwit for the Tdpid tail nexion of the crayfish. Schem~ti(' diagrolm of the known elements and ronnl"Ctions for ph.sic mechanical 51imuli to abdomen. [Zucker, 1972.1

initial body bend of a s tdrile response. The wea lth and v<1rie ty of synapses known-on the so ma, axon hillock, ,lI1d large dendrites-sugges t tens of thousands of im­pulses may arrive per second, during the reaction time of a single afferent impu lse, representing a high degree (but perhaps not dtypical for neurons) of integrative filtering, recognizing. decid ing, .1l1d com manding .

275

276

occurrence and Frequency both by tonic input and by higher central levels (e.g., by changes of mood).

A. Central Rhythms and Reflex Modulation

The peripheral or reflex hypothesis fails to account adequately for respiratory control of vertebrates. A crucial test, total deafferentation, leaves a functioning central system. Similar tests have shown that many other rhythmic control sys­tems have built-in central scores. The motor neuron discharge seq uence, which withdraws the mantle and closes the valves in a clam, Mya, occurs after de­afferentation. The copu latory move­ments of a praying mantis, the flight rhythm of insects, the walki ng patterns of insects and amph ibia, respiratory movements in insects, beating of the swimmerets in decapod crustacea, strid­ulatory singing in crickets and grass­hoppers, side-la-side alternation of longitud inal muscle activity in sharks, heartbe.1t and gastric mill contro l in crustacea, and the swimmi ng beal of jellyfish have been shown with varying degrees of rigor to be centrally con­trolled.

A few of these examples deserve more detailed discussion. But first we should poi nt out that proprioceptive feedback functions have also been demonstrated in nearly a ll of th em, and in the dis­cussion of centrally patterned motor output we should attend to the role of reflexes as well. In studying cases of oscillatory behavior, as in all considera­tions of oscillatory phenomena, three measures always merit notice: frequency (or the reciprocal of frequency, th c pe­riod) of the osci lla tion, amplitude, and

phase of one element of the oscillatory pattern relative to another. In severa l of the known centra lly driven beh.wiors, we can relate the functions of peripheral feedback to one or more of these pM­ticular parameters.

Lomsl Flight . One of the most decisive studies on central rhythms is that of Wilson on wingbeat control in grass­hoppers. These ani ma ls will sometimes fly when the g.1nglia of the head and abdomen are removed, so that the pat­tern generator must be present within the thoracic segments. Th e normal motor score has been analyzed in such detail that the temporal sequence and phasi ng of every motor neuron impulse during Aight is known. This pattern of motor neuron discharge can be recognized in the central stumps of the thoracic nerves, even after all those"'11erves have been severed. Isolated thoracic nerve cord preparations are not spontaneously ac­tive in the flight rhythm, but random electrical st im ulation of the nerve cord ca n elicit nearly the same pattern as is found in intact animals during Aight. The output pattern of deafferented prepara­tions is defi cient in one major respect; it is low in frequency. The decre.1sed fre­quency can be ascribed to lack of input from four stretch receptor cells, one in the hi nge of each wing. These stretch receptors discharge when the wing is elevated. During flight they fi re one to a few impulses toward the end of the upstroke in each wingbeat cycle (Fig. 7.27). The number of impulses in the burst is correlated with wingbeat ampli­tude, the timing of the burst with wing­beat phase, and the burst repetition rate with wingbeat frequency-all the meas­ures of an oscillation. This information,

•

I I I I I I I I I o

, I ' '00

I I I I I I I I I I I I

'00 Tima (msec)

I I I I I , I ' 300

I I I I I I I , I 400

Figure 7.27 Propr ioceptive feedbac k in a fly ing insect. ~nsory dischMges in nerves from the wing and wing h inge in a locus t, recorded with wires manipul,l ted into the largely eviscerated tho­racic cavity of a locust. The top record is of downstroke muscle potentials, which are re­pea ting ,It the wing-beat frequency. The bottom record is of a sensory (stretch) receptor from one wing, firing one or two times per wing beat. [Wilson, 1968.J

upon enteri ng the eNS, would be ade­q uate to trigger and control the events of th e next cycle. Apparently, however, it does not even signi fi can tly affect the next cycle. In spite of th is rich detail of in­formation about wing position, there is li tt le input/output correlation except a relatively long-termed one relati ng aver­age input frequency in th e stretch re­ceptofs to average wingbeat freq uency

and ampli tude. Th e ganglion integrates and smooths the input over several wingbeat cycles, thereby almost bu t not quite los ing the phasic informati on. The fil tering process is analogous to th e smoothing and integration in a re­sistance-capacitance electrica l net work. In su m, locust fli ght reaffercnce primar­ily plays a role in con trolling the average excitation of the motor p.l t1ern genera tor,

277

Sectlo n VI Cenerally Scored Behavior:

l'.IUerning in Space and T ime

278

Chapter 7 integrdtiOI1 at the intennedidte levels

thus affecting wing beat frequency and power. It has a very we<lk effect on wingbeat phase.

An impo rt.l ll t lesson to be Je.1!"ned from this exam ple is thai, even though one can demonstrate that proprioceptive input in a rhythm ic system fits the re­qu irements for a refl ex feedback model, that input may not be necessary for th e normal pattern, and significant infor­mation parameters for the peripheral hypothesis may not even be used in the normal operation of intact anima ls. They may be discarded in a filtering process.

Swimming ill Slzarks. When many fi shes swim, the longitudinal musculature of the two sides contracts in alternate metachronal waves. If a shark is curar­ized to the point of total paralysis, motor output in the segmental nerves may still result when the anima l is stimulated, but since no movement occurs there can be no correlated proprioceptive feedback . In th is circumstance, bursts of impulses still issue alternately through contralateral nerves, but at unusually low frequency. Proprioceptive input may, as in locusts, be necessary for ton ic excitation of cen­tral state. But in sharks, compMison of the outputs on the same side in different segments shows them all to be syn­chronous. The metachronicity is lost. On the basis of presently available evidence, it appears that proprioceptive feed back is necessary for the phasi ng of outputs in proper segmental seq uence as well as for the ma intenance of normal frequency (see Fig. 10.63).

Crayfisl! find Lobster Swimmerei Beal. The abdomina l appendages of decapod crustaceans also have a metachronal rhythmic beat. The completely isola ted abdominal nerve cord can be stim ulated

to produce th e swimmeret beat com­mand, and the output of a deafferented or isolated preparation is normal in both frequency and segmental phasing (Fig. 7.29). The known proprioceptive reflexes cannot even modulate these parameters. They do, however, modulate th e force of each st roke, affecting the velocity and amplitude by influencing the number and repetition rate of motor impulses during each cycle-in other words, the magnitude of the motor discharge.

B. Command Cells

A superficially quite different category of centra lly scored pattern is that called up by com nl.1nd cells, actually just the end of a spectrum of mechanisms that occur in all degrees. Command units, first discovered in crayfish, are now known in many arthropods, annelids, molluscs, and vertebrates (see Box 7.3, p. 274 ). Com mand units, or redundant clusters of similar units, may turn out to be q uite general. The concept and the term come from the observation that certain single un its, upon sti mulation, are capab le of causing actions rese mbl ing major pieces of normal behavior. Some c<luse static postu re, others phasic sequences. In higher invertebrates, in which they Me prob<lbly all potentially identifiable, non­identical subsets of com mand cells may act together to determine the form of the behavior. Th ey are presumed to trigger or release the action, not to instruct the motor neurons in the te mporal pattern of their firing; that pattern is alre.1dy pre­formu lated by other neurons.

The best-known command fi bers in cf<lyfish are not of extraordinary size, are not motor neurons themselves, and in general do not even synapse d irectly on motor neurons. They are high er-order

•

interneurons that run through several ganglia or even the whole length of the neuraxis. They excite whole motor systems, small or large, controlling pos­ture or locomotion (Fig. 7.28). They are found repeatably in different anima ls. Each is uniquely characterized by posi­tion in the nerve cord, axon diameter, output function, and other properties . The pattern of output does not depend A much upon command fiber frequency or pattern, and the output may continue well after the stimulation of the com­mand fiber has ceased.

The command fibers are obviously labeled lines. The temporal pattern of activity in each is relatively unimportant. What is important is which command fibers are active, since they drive diverse behaviors. Each command fiber con­trolling posture of the crayfish abdomen seems to have unique but overlapping output fields. Perhaps the severa l fibers commanding swimmeret beat are also unique and produce somewhat different outputs (Fig. 7.29). Whether they do or not, swimmeret movements can be differentially controlled by combining the action of fibers driving the OScillatory mechanism with those affecting posture. Command fibers m.1Y turn motor system s on or off or bias them for pur­poses of steering or orientation.

Some giant fibers (see Box 7.3) are command units- namely, those that are not motor neurons but interneurons re­ceiving from many small cells. They are specialized for prompt, synchronous, and brief action s that do not continue after the giant fiber stops firing. local sites in the hypothalamus of mammals, where stimulation rather reliably triggers char­acteristic behavior, may represent a cluster of cells acting as a kind of com­mand center (pp. 316-319).

CM10

CJ 62 / 1 , 60 0" '75 , , '

-r 63 , ,; ; ,

L. __ ,. ~; 1 74

4 1 65 "\ - - ; 61 --"\ --7,. ... ---}- -­, 67 , , , 73 ,

66 ,, /69 , 71 '>' 72 --.J ~- ,

68 70

B

Figure 1.28 A. A command fibe r for posture. This fiber, called CMlO, travels in area 75 of the circumesophageai connective of the cray fi sh. When stimulated at a minimum of 20 shoch per sec, it releases the "defensive postu re" shown in H, involving stereotyped positions of all appendages. This is one of the static, postura l responses trigge red by anyone of a small number (3-5) of specific command neurons; othe rs are dynamic, rhythmic move­ments, such as swimmere! beat ing (e). [l'art A, Wiersma, 1958.J

279

280

Ch"pter 7 Integration at the Inte rmed iate Levels

Command neuron

Coordinat ing neuron

~~r-,--~(--___ ~

~

Fig ure 7 .29 Circu it for swimmer,,! beating in crdyflsh. T he segmentally repeated circuit is connected both by the multisegmental command neuron dnd by the intersegmental coordirMting neu­ron tha t controls the ,lelual t ime relations 0( the wave of beating. [Stein, 1971.]

What sets off a command cell? In general, we do not knolV, but an in­formed guess might be that it usually requires a number of backgrou nd condi­tions plus some triggering input. This means it is highly integrative, acts as a recognit ion un it (p. 236) to detect these criteria, a nd is therefore a decision unit (p. 238) in a significant sense. The input it requires may come from sense organs, but in some cases it seems likely that centrally a risi ng changes in mood or re.ldiness might do the same th ing (see vacuum activity, p. 315, central spon­ta neity, p. 325).

C. Hierarchical Structuring of Molor Systems

If rhythmic motor output can be driven by nonoscillatory impu lse trains in the command fibers, where does the oscilla-

tion arise? Cou ld it be due to interactions between the motor neurons? Several sorts of motor neuron interaction a re known . Mutual excitation and reciprocal inhi bition between motor neurons in the stomatogastric ganglion of crabs and lobsters are respons ible, at least in part, for the way in which the output pattern of that ganglion rein forces and ad justs a spon taneous rhythm or command fi ber response (p. 279). Some motor neurons exciting the same muscle in insects are electrotonically coupled and tend to fire together. The sa me is true for contra­laterally paired inhi bitory motor neurons in nervati ng the abdominal postural muscles in crayfi sh . Motor neu rons in­nervati ng th e same flight muscle in cer­tain flies inh ibit each other. Synergistic motor neurons in vertebrate spina l cords are a lso inhibitorily lin ked, through known short-.lxon inte rneurons, the Ren­shaw cells (p. 269).

If interaction between motor neurons were genera lly responsible for thei r rhyth mic activity, one might expect that ant idromic stim ulation of sets of motor neurons could reset or modify the output of the whole network. In genefil!, this is not true. At a more detailed level of analysis one wou ld expect to find th.lt intracellular stimu lati on of one motor neuron would give rise to synaptic po­tentials in functionally related ones, but aga in this is not generally true, or else the effects are quite weak. Current thought on the matt er is that motor neuron interactions are usually in­adequate to account for the patterns of output in motor systems.

This concl usion leaves us look ing for a process, or even a structure, that mediates between the com mands and the motor d ischarge itself. The search ison for interneurona l pacemaker cells or net­works th"t produce rhythmicity(Fig. 7.29}. Cons istent with the notion that there is a hierMchy in molor systems, with input co mmands driving osci ll ators that drive motor neurons, is the fact that the same set of motor neurons ca n be used in more tha n one behavioral pattern. Frogs swim or jump with synchronous ou tput to homologous cont ralateral muscles, bu t alterna te them d uring walking. Insects use some of the same muscles to move the wings and legs, but they do so according 10 different synergistic/ antagonist ic relationships, depend ing upon whether the com mand says "walk," "jump," "sing," or "fly." We are pushed into thinking th"t even in the lower ga nglia or spina l cord there is a mu ltiplicity of p"Uern gener.ltors or oscillators that can each be turned on or off or be modul.1ted in frequency or .1mp­lilude by command input, and thai these pattern generators each converge

upon identical or overlapping pools of motor neurons. Thus the motor neurons Me, to use Sherringlon's phr,1se, the "final com mon paths" transm itting to the muscles signa l piltt em s th"t are pro­duced at a higher level.

If we turn from a preoccupat ion with the genesis of rhythm to the question of how alternative motor patterns that in­volve higher level switching are selected and programmed, we have to dea l mai nly with conceptual models that seem compatible with principles of physiologica l organization.

The rather wide ly accepted notion today is that actions commanded by higher centers are not specified in detail by those centers, nor primarily deter­mined by peripheral stimuli, but trig­gered in a preprogrammed language ca lling up combinations of elementary acts in a hierarchy of levels. The highest command, as well as the lower-level instructions to still lower levels, may simply specify the sequence, strength, " nd duration of the nex t lower compo­nents, finally elicit ing moveme nts as though the adequate periphera l stimuli had occurred that can reflexly elicit them. Ana lysis of six ga its of horses shows that they could be produced simply by ca lling up components eq ual to certain loca l spinal and long spinal reflexes in formulated sequences and durations, given a few fixed rules (Easton, 1972). Whether the horse actually works this W,lY, we ca nnot tell, but the !'ules are simple, and the model might work. Of course, superimposed on wha tever "c" lIing up" the bra in initiates, proprioceptive input as well as visual and s01llesthelic ]'eafference (sensory input c"used by one's own actions) wi ll be important in shaping and correcti ng the centrally patterned prograJll, by

281

Section VI C~"'tr.l.lly Sco<('d Bth~Yior'

r~tt (' rnlng In Sp~c ... ~nd Timt

2.2

Command

A

!'DI---==­+.-----...----''-<G)I----==- l L""".

E)tci lalory Inhibitory

Feeding pacemaker

Ton ic

Depressors

Phasic (campanlfo rm

sensi lla)

Components 01 intrinsic program

ConUllcUon c~' .. ' ''~~ j

Positive leedback loop sustaining re tractor burst

An tagonist inhi bi tion

B

S tress ing 01 mechanoreceptor,

Retra ction 01 buccal mass ~j,,.;..,+ .,....

Mechanore<:eptor J activity

I -Destressing 01

mechan'oreceptors

~----~~----Delayed negative leedback loop ending mechanoreceptor burst

,

s imple adjustments of strengt h and ph.1Se of pdrticular components.

Motor patterning systems have been so incompletely s tudied that we kn ow vel·Y little of th eir actua l mechanisms. Parti,ll models with some supporting evidence ca n be made in many cases. (Examples are shown in Figure 7.30 and in the figure in Box 7.3, p. 275; see also p. 333.) Since it has been so diffi cul t to ma ke an ana lysis of even relatively sim ­ple motor syste ms in terms of individual neuron activi ties, perha ps we shou ld ex­pect 10 find eventua lly Ihat presen lly un known concepts of neural function are involved.

Figure 1.30 (f~,iQg ,.~gt )

D. Centrall y Scored Pattern by Sensory Tape

Another possible mecha nism of central programming, besides the motor score, has been suggested by Hoyle. He calls it a sensory la pe, or 10 use the phrase of the ethologists, a sensory templa te (sec p. 309). Sensory tapes or templates have not been demonstrated, though strongly inferred for the control of some bird song. The idea is worth more discuss ion. Suppose the CNS contained an instruc­tion tha t sa id, " Produce a motor output that results in a speci fi ed feedback from

Circuits for inso:><:t w,llk ing (AI and sn~i l feedi ng (8). A. Hypothetic .. 1 scheme for the ob5erved d is· charge pdUerns of levator (5 and 6) and depressor motor axons (D). A bursting interneuron (bi) is excited by the comm~nd neuron and in lurn excites the levators whi le inhibiting depressor motor neurons. The comm,md nber is believed to exc ite the depressor, so t h ~1 ~n incre,lse in command input decreases interburst inte rvdl while producing ~ less marked decruse in burst durdtion. Cerl.l in sensory input toniCollly ( .. cilitates the bi ~nd inhibi ts D; other input ph .. sic.Uy excites D dnd inhib its bi. [pe .. rson .. nd lies, 1913. 1 D. Diolgr .. m showing the funct iondl interconnec tions that give rise to Ihe tempor .. 1 re l .. t ions o( dCl ivity in the retr .. ctor .. nd protr .. ctor elements o f the 2S pairs of lIIust tes re­sponsible fo r feed ing in the sn .. il Ht/isortllf. The neurons of the p.Ktm"ker generdte dn .lutonomous rhythmic output th~t olppeMS with differen t "mounts o f ph .. se shift in the \'o1rious motor neurons. The rhythm is symbo li ~ed by the sawtooth Wolve ~nd the two represent,ltive populations of motor neurons at the lop of the di,lgr,lIn (upper brace, right -hand side of diagram). T he retrdcto r neurons drive the retractor muscles. Cont ract ion in these muscles exo: ites mech~noreceptors, ~nd their oulput is fed back positively 'lid exc itatory synapses on the retractor motor neurons. T his posi tive-feedba<:k loop (Io\ve r br .. ee, r igh t-hand side) sust" ins the re tr .. cto r burst. Contrdction of the retractor muscles p roduces, .. fter .. de l .. y (due 10 exci t .. tion-contract ion coupling .. nd the Vi5Coel .. ~ic .. oo inert i .. 1 prop­erties of the system), .. retrol("t ion of the bucc .. 1 mdSS. When it is retr.acted, cont r .. ction of Ihe muscles no longer stresses the mech,Uloro:><:eptors, .. nd these shut off; Ihis series of events constitu tes a nega­ti ve-fCi'dback loop wi th deldY, ,md it limits the retrdctor burst by opening the positive-feedbdck loop ,1(ler relraction is complele (low brdce). The protraclor motor neurons and muscles dre excited 011 Ihe opposite phase of the cycle from Ihdt o f retractors. They fire un til inh ibited by inpu t (rom mecl1dIlO. receptors, which Me stretched by the contracting re tractor muscles. This accomplishes a unidlrec· Uondl ,mtdgonist inhibition with de lay, which .. lIows the protract ion ph .. se to be sustained unt il re-tr .. ctor tension is developed. Ant.agonist inh ibition in the re\"('rse d irect ion (i.e., protr.actors inhibiting retr.lctors) is absent, ~nd this allows the O\"erlapping activity in the two groups of muscles .It the beginning of the retraction phdse. (Kater and Rowell, 1913.1

283

Se(tlon VI Centrally Scored Dehulor:

Patterning in Sp.ce and Time

264

Genera! driver

A

MOlor tapes

Specillc driver

Other

X"","U

2nd-order driver

p"., Extensor

Jo int

Proprioceptive afference

SenlOlY tapes

In ternal stale ,..'--''--', Tape "clues"

Compare

leleClo r

Compule ' rom di fference -

Specilic driver

•

proprioceptors or other sense organs." This ins tructi on would not lead inevit­ably to a sing le s tereotyped motor out­put, as from a molar score ge nerator, but it could guide motor out put to achieve a goal (Fig. 7.31). Ou tput might at first give in correct results, but feedb<1Ck cou ld modulate th e output on successive cycles of loop operation . Consistent with th is notion of central progra mming by com­parison of sensory feedback with a centrally slored goa l pattern is the fact that diverse motor oUlputs may be asso­ciated wit h apparently identical leg movements during walk ing in insects. In some cases the flexors and extensors alternate. In others, one muscle contracts tonically whil e the other oscil lates. The resulting movemcnt is thc same.

The only reasonably strong case of a sensory template is found in bird song control (see Fig. 8.18). In the invertebrate cases in wh ich a sensory principle may be operat ive, it seems to be supe r­imposed upon a motor score type of central generator. Insect flight is basically programmed by .1 motor score, but ex ­teroceptive as well as proprioceptive inputs can modify that score for pur­poses of s tability, s teeri ng, or com pensa­tion for inherent error or damage to bod y paris. Perhaps in these systems the no-

Figure 1.31 Ullting ,111<')

tion of a sensory template n~ally redu ces to reflex modulation of a motor score.

E. Coordinated Movement to Gross Stimulation of the Brain

A clear progression is ev ident if one compa res the responses to crude elec­trica l stimulation of s tructures at succes­sive neural levels. Ventral roots give a segmental, loca l contraction closely re­lated to the duration and strenglh of stimula tion. Lower motor centers in the cord or bra in stem give li tt le more, though the d istribution m.1Y be mOfe functional (e.g. flexion of certain joints). The responses relevant to this section are the quite normal actions involving sequences of 'movements, such as ca n be elicited from the hypothalamus of mammals (p. 471). These are so natural as 10 suggest a centra l pattern, si nce the stimuli are like lightning bol ts. A pocket mouse may s tuff invisible seeds inlo its cheek pouches at a high rate, a cal may arch its back, hiss, unshealh its claws, erect its hair. The value for our purposes is the same whether we assume that the stimuli trigger motor patterns or sensory " hall ucinations": the patterns are cenlral and need only an adequate trigger.

Two types of cont rol by centrally determined sequ('nces. A. System driven by sequellces Ih ~t deter­mine motor output directly. Driving comm~nds c~n be gener~ 1 or specific to dny degree of det.!il, ,md .It a lower level they Cd ll be modified by prop rioceptors. B. System driven by tdpes of sensory feedbdc k th .. t must be expected. A compd(~to r m .. kcs the dctl1~ 1 (OIllIllJnds on the bdsis of the differences between the proprioceptive dfferenee dnd the inst ruction from the tJpe. There c .. n dlso be propriocept ive (onlrol lower down, .. s before. Thi~ syslem is more "d"ptdble, but requires much more circui try in dddit ion to d COlllpdril tor ~nd .. COlllllldnd center, which th('m~I"es musl produce highly compleK .Id.l lJlil'e sequences of impulses. Most inverlebr.lte responses inveSlig" tcd usc .In " in­line'" syslem, ,15 in A. IHoyle, 19M. 1

285

Seella n VI Celltrd lly Score.! Behavior:

r dttern ing in Sp.lce ~ nd Time

286

Box 7. 4 Chronology and Background of Ideas on Ihe Ph ysiology of Ihe Nervous System

We present here i\ selection of highlights in the history of ide,IS, from the ('.Hlies! till1es up to 1929. The inter­pretation of the brain in terms of cells is highlighted in Box 3.1, p. 102. The roots of bra in chemistry, membrane biophysics, ph,umacology, sensory and psycho­physiology, .md behavioral analys is arc not attempted here.

The redder is urged to look furthe r, for more balance and adcquille representation. Some useful, more-or- Iess condensed accounts aTC by Nordcnskiold (1935), Dampier ( 1948), Singer (1959), Brazier (1961). Sirks and Zirkle (1964), Gardner (1965), Clarke and O'Millley (1968), and McHenry (1969). Specia l aspects are dealt with in FeiHing (1 970), Brazier (1961), and Swazey and Worden (1976).

1700 B.C. An Egyptian document, transl,l ted cen­turies later ,lnd published ,lS The Edwin Smith Surgicdl Papyrus, includes 13 Cdse descriptions of head injuries. Aphasia, paralySiS, and seizures were described, and suggested the fun ct ions of the brain. Nevertheless, diseolse cont inued to be generally oltiributed to suprd­naturoll influences olnd whims of the gods.

800 B.C. Homer's works and the flowering of Greek iHt and intellectua l life led slowly and incomple te ly to the idea thilt the world is knowilble.

500 B.C. Alcmaeon performed the first recorded d issection of ,l human body. He p,l id some ,ltiention to the brain and discovered the optic nerves. His teacher, Pythagoras, tilught th.t t the brain is conc::erned with reasoning. In the next century more dissections and similar speculations were made by others.

400 8.C. Hippocr,ltes of Cos countered the mystics and the entrenched supranaturalists to introduce ra­t iona l medicine. This requ ired the systematic accumu­lation of cl inic::al experience, and his desc::ription of epilepsy went unsurpassed until the work of Hughlings Jackson. However, Hippocratic teaching was dominil ted by the idea tha t fun ction derives from the combination of four humors; blood, phlegm, ,lnd black and yellow bile. The brilin is the org.tn of intelligence .md dre,lms,

but it a lso secretes phlegm OWd cools the blood. Even in the Golden Age the Greeks- did nol easily fo llow his exam ple. The lack of .lutopsies del'lyed progress.

340 B.C. Aristotle systemolt k.llly pursued compar,l­live an.t tomy and in his 19 books set a high wolter mark of natura l knowledge thilt l.lsted until the RenaisS.lnce, but he did litt le to change ideas on the bra in.

300 8.(;. Herophilus and the greil t school of Alexandria in Egypt dissected many cadavers Jnd really founded anatomy. He distinguished sensory and motor nerves and showed that they connect from spinal cord to periphery.

250 8.C. Erisistrat us postulated a mechanism of the brain function: blood and two kinds of air arc carried in the veins, arteries, and nef\'es; air is changed to vita l spi rits in the hear t and these to animal spirits in the brain ventricles, whence they go via the nerves to distend and shorten the muscles. -.

200 6 .C. Galen culminated the classic period, writing more than 400 works that were dennitive for a millenium. By now much of the naked-eye anatomy of the nervous system had been discovered, including most of the cranial nerves. Among the few advances in the understanding of function were the descriptions of symptoms fo llowing section OInd hemisection of the spi nal cord in lower mammol ls.

400- 800 The D.uk Ages lasted more than 12 generat ions. Greek knowledge WolS forgotten in Europe. Men did not ask to understolnd themselves or nature but to be told the supra natura l or religious meanings of things.

600-1200 Islam spread from Asia minor to Spain, carrying Greek knowl edge unknown in the Christian world. Jewish tfolders introduced Arabic transla tions to Europe. Long-lost Latin versions of Greek writings were rediscovered in monastery storerooms. Men did nol yet ask abou t nature but Jbout thei r heritage.

]200-1300 This interest in book learning and the new wave of ideas induced schol.lSticism, which in turn

•

110x 7.4 (n",'hilltd)

brought on humanism as a redetian. Univers ities sprang up widely during the last period of thc crusades.

HOO- 15OO The invention of movable type stimu­la ted printing. Voyages of explora tion expressed the new .ttti tudc IOWo1rd d iscovery.

1478 Mondino represents the height of dassicill, dulhoritar i.J[l (Galenic) anatomy. His manual, il lustra ted crudely and ascribing func tions such as f,l1l tasy to the anterior part of the la teral ventricle, was little influenced by actual dissection. Nevertheless. it WdS used for 200 Y C<lTS.

1500 l eonardo da Vinci manifested the new curios­ity by making his own dissections and drawing more accurately th,1n anyone before; but, fai li ng to publish, he had li tt le influence on the progress of .-millomy.

1543 Vesal ius broke wilh the tradition of Aris totelian J nd c..lenic .Julhority, inaugurating the modern ide.J of the authority of origina l observations. His landmark work, Dr H uma'ii Corporis Fllbrica, con­ta ins many plates of brain dissections showing numer· ous fe.Jture s for the fiTst ti me.

1608 Ha rvey WJS the first to reason that the blood circul.Jtes. He multiplied the c.Jpolcity of the heMt (with its one-w.JY va lves) by the heart rate, both long-known quantities. The method of inductive reasoning, lost since the Greeks, was re-established .

1662 Descartes, the lCiider of 17th-century physi­ological thought, crudely conceived the idea of reflex act ion powered by a Galenic mechanism. He broke new ground also in the mind-body problem, placing the seat of the soul in the pineal.

1664 Thomols W ill is published o ne of the first separate works on the brain, the most complete and accura te so far, introduci ng several of our current terms. He suggested tha t the cerebrum presides over voluntary motions and the cerebellum over involuntary move­ments; he manipu lated the cerebellum in a living m.Jmm.J I itnd noted tha t the hE'oIrt s topped.

1691 Robert BoylE' pointed to the existence of a motor cortex. A knight suffering a depressed skull

fr,Kture showed a sustained para lys is of a rm and leg; the pOlrOl lysis disappe.ued promptly after surgical re­moval of iI spicule of bone.

1.730 StE'phE'n Hales openE'd the door to reflex physiology by nol ing that the legs of a decapitated frog would withdraw upon pinching but that such "re­actions" disappeared when the spinal cord was de­stroyed. The terms "stimu lus," "response," "reflex," "afferent," and "efferent" came into use by the 1770's. Rohert Whytt (175 1) played an important role in the drama.

]740 Swedenborg considered the basal git l\glia the seat of primary sensibility of body and soul and the route of "all determinations of the wilL" He distin­guished upper and lower motor cen ters and correctly subdivided the motor cortex.

1791 Galvani starled electrophysiology by il\­advertenlly stimulating the muscles of d issected frog legs when they completed a circuit with two d issimilar metals. Volta used the discovery of a source of electric potentia l to develop the battery and voltaic pile. Ga lvani, mistakenly believing tha t the poten tial came from the tissue, went on to discover bioelectr icity by observing that a nerve is excited when it completes a circuit between an injured and an unin jured tissue. The use of a nerve-muscle preparation as a biological de tector, amplifier, and ind icator was an ingenious physiological techni(IUe that permitted the discovery of mi llivolt level bioelect ricity many years before the g<1lvanometer was invented. Elect ricity came just in time to fill the gap as improved anatomy excluded the hyd raulic model of nerves and the new physics and chemistry ra ised doubts about "a nima l spirits." T he s t<1ge was set for a r<1tiollal, mechanistic physiology.

·1809 Rolando removed the cerebellum in fish, rep­tiles, and mamma ls and saw disturbances in voluntary movements without influencing sensa ti on .

1822 Francois Magendie made fi rm an eMlie r claim by Charles Bell that dorsa l roots are sensory and a lso showed unequivoca lly that ventral roots are motor. like a ll such experiments in these pre-anesthesia days, his work was based on vivisection.

(ColllillJlfd QII II fXl l!IIg~.)

287

288

UOI( 7.4 (rOlllj'HIt<I)

1823 Pierre Flourcns showed that vision depends on th e cortex; ablation on one s ide in pigeons, T,lbbits, and dogs was found 10 cause contrali1ter,l l blindness.

1826 Johannes Mli ller, wide-rangingGcrmilll natura l philosopher, sensory phys iologist, and ("omp,-u .ltive anatomist, enunci,lied the " law of specific nerve energies," which s tates that each sensory nerve gives rise to its own charilderistic sensation, however it is s timul<l ted . For example, electrical, mechanicdl, or chemical stimulation of the optic nerve (.luses a sensa­t ion of light.

1833 Marshall Hall recogniled segmcntill, intcr­segment<l l, and suprasegmenlal reflexes. '1"he spinal cord is a chain of segments whose function.ll units iH C

sep.H.lte reflex arcs which interact with one another and with the higher centres of the nervous system to secure coordinated movement" He recognized the tempor.uy depression of reflexes below a spinal transection and c,tl led it spinal shock.

1848 Du Bois Reymond showed that activity in a nerve is invMiably accomp.mied by an e lectrica l change (" negative v<lria tion"). He described the properties of neurOlnuscul<lr transmissiOn and opposed the prevail­ing doctrine of vitalism.

1850 Helmholtz measured the veloci ty of conduc­tion in nerve tissue and began what would become a continuing effort to improve the instruments of electro­physiology.

1651 CI.lude BernMd developed the la ndm.uk con­cept that the body maintains a constant inte rnal en­vironment for the ce lls by means of the extracellular fluids . Among other things related to sympathetic func­tion, he described the vasomotor nerves, which pldy a key part ill this regulation, late r ca lled homeostas is. An en thus ias tic experimen talis t, he wrote influentially on the experilllentdi method, advocating rigor, con trols, and form uldtion of testable pred ictions.

1861- 1898 Hughlings Jackson, British neurologist, developed concepts on the underlying principles of brain function from cl inical observa ti ons. One was the concept of "release" to olccount for various signs of

in jury to higher p.uts of the brain, s llch itS spast icity, thai a rc more pos it ive th,ln negative; the resulting OVC I·act ivi ty of the surviving lower centers bespeaks .1

normal restrolint imposed by the higher. A seco nd concept lVds that although e \'olution has been a process of increased differentioltion and heterogcneity, wi th integration keeping pace, d isedse reverscs this, such that higher parts go first and the lower take control. A th ird concept, growing out of intense study of patients with speech disorders, those wi th sensory, motor, or psychic epilepsy ilnd hemiplegi.Js, was that the cortex has Illany locollized func tions.

1863 Sechenov stud ied " reflexes of the brol in," me.lning cerebrol l activity thai .lTises from sensory st imulat ion and med iates psychic experience and causes volun tMy action, subject to modulation by other bra in cente rs, incl uding inhibition by the midbrain. This he obtained by placing salt on the optic lobes. He recog­nized temporal summation of subth reshold stimuli; also muscle sense, later ca lled proprioception . He empha­s ized the physicochemiCdI anollysis of metolbolism and excitoltion. Trained with CI,lude Bernol rd in Paris dnd Ou Bois Rcymond in Berl in, he is reg.lTded as the father of Russian physiology.

1865 Pfluger systcmoltically investi g~ted inhibition, mol inly vid autonomic nerves. Searching fo r inhibitory nerves to skele ldl muscle, Paylov (1885) found them [n the fresh-water mussel, AIIQdQrlili. a biva lve mollusc; Biedermann (1887) found them in the crayfish. They ~re still unknown in vertebrJ.tes.

1870 Gudden's finding thitt specific thalamic nuclei degencrate when certilin areas of the cerebral cortex arc dcstroyed was a milestone in experimental anatomy, as well as col iling attention to retrograde degenera tion and opening the modern period of study of the thalamus.

t874 Bartholow, in the U.s.A., stimu),lted and mapped the motor cortex in man. He found, incident­a lly, ihat the brain itself is insensitive to manipulolt ing and cutting.

1875 Richard Caton, in Engl.lIld, observed clec tr ic~l

waves from the exposed brains of rolbbits dnd monkeys;

,

80)( 7.4 (Wllf;,mrd)

his finding was overlooked, but the waves were re­discovered later in Russi,1 (1877), Poland (1890), .lnd Austria (1890). Caton IV.lS looking for action potcnti,ll s ill the ur,lin, inspired by Du Bois Reymond's in nerve, hoping they would provide a Inelhod for loc.tl izing sensory arCilS. [n this he succeeded, d iscovering evoked potentiolls and, inddenlally, DC shifts with activity, as well as the ongoing EEG

1898 Lmgley introduced the term "autonomic" ilnd 7 yc,lrS lale r, "sympathetic" and " p.Holsympathelic."

1902 Pavlov, investigating the phys iology of d iges­tion, saw clearly the rOild he would follow for 34 years, analyzing the psychologica] propert ies of conditioned reflexes. His influence on neurophysiology was "" Imost nil" (Fu lton, 1949) until recen t yeMs.

1903 Brodm,lnn, Vogt, and Ca mpbell e<1ch made their first communiutions on the archi tectonics of the cerebr<1l cortex, m<1pping the dist ribution of different types of cortiC<11 struclure. It lYas some ye.us, however, before the 6-layered str<1tificoiltion lYas fully exploited; then its embryologic<11 origin lYoiIS emphasized. By 1929 Ariens K<1ppers and others had ildded an evolutionary origin from a primitive 3.I<1yered mantle. Enthusiasm for the 6·Jayered structure long delayed a concern for the neurona l organ iz.ltion and connecti ons.

1906 Sherrington, in England, published his land· m.uk treollise, Thf IlIlfgralivt ATlioll of illt Navolls 51/SItIII, in which he systematic.llly .malyzed holY the nervous system works, by close eXolm inoltion of si mple and compound reflexes. Primarily he used mechanical re· cording of contraction of individual muscles. Most of thc ideas on pp. 271-27Z M e his.

1909 Karplus .Ind Kreidl beg.ln the first experi. ment.ll study of the hypothalamus, their results .lppe.u· ing in iI long series of papers. Many othe rs joined in, including Ihe ce lebr.lIed surgeon Harvey Cush ing, who in 1912 discovered that removal of the pituit.uy or merely tr<1nsecting its st.llk c.luses an adiposogenit<11 dystrophy. Nevertheless. the modern period of re· seMch, in IYhich the hypotha lamus is related to "utonomic, emotional, and instinctive functions and the

control of the pituitary, did not really begin until the 1930's.

19"1"1 Henry Head and Gordon Holmes, using psychologic" l concepts ilnd testing, studied sensory deficits after clinical lesions. They were more concerned lYi th thc [ldture rather than the locus of cortical sensory processes in man. They sholYed th.l t the cortex is especi.,11y involved in discri minat ive "nd higher aspects of perception. Head is remembered a lso for severing a nerve in his own ,J. rm to study the loss "nd return of sensation with rcgener,J. tion.

1917 Keith Lucas firmly established the all·or· none law and quantitative relations in exci tation, such as the minimal s lope of a slowly riSing current necessary for excitation.

1924 Kato, in J"P<1I1, sett led a con troversy by show· ing nondecremental conduction in nerve.

1924 G,J.sser and Erlanger, in the U.s.A., used the cathode fay osci lloscope to describe the components of the compound action potential of the whole nerve.

1924 Rudolph Magnus published his la ndmark monogrdph on posture. Starting his investigations 16 years before with Sherrington, he edrly realized the importa nce of twisting the head on the neck, IYhich reduces the tonus of posturalmusde and thei r reAexes on the side of the forward car in the human or the lower ear in the dog.

1926 Adridn and Zo\terman recorded from si ngle scnsory nerve fibe rs and found Iha t, as in motor fibers, the repet ition rate of impulses is graded. Stronger stimuli result in a greater frequ ency of spikes.

1929 Denny. Brown recorded from a single motor neuron ,lctivdled by a n0T1n~1 reflex stimu lus; he used the stre tch reflex in a decerebrate Cdt.

[929 Berger reported that brJin W.lVes could be recorded through the skull in humans. After" general scepticism. this was confirmed by Adri.ln in England and t,lken up by Davis and others in the U.s.A.

289

290

Chapler 1 Inlegra tion al the Intermediate Levels

SUGGESTED READI NGS

Autrum, H., R. Jung, W. R. Loewenstein, D. M. MacKay, il,o.d H. L Teuber. 1911. Hllndboak of 5'"50/,y Ph ysiology. Springer-Verlag. New York. fProjectea to make up e ight volumes, some in paris, th is will be a comprehE'ns ive treatise.]

Bach-y-Rita, P., C. C. Collins, and j. E. Hyde. 197 1. Tht (,mlral of Eyt MClllfllltIJ/S. Academic Press, New York. [A symposium on a system especial ly favorable for revea ling principles. Other more recent symposia and reviews can be fou nd on this system. ]

Gr,mit, R. 1970. TIl t BlI5is of M olor Conlro l. Academi c Press, New York. [A systematic ana lysis from sensory receptors to higher cerebr'!l level s controlling motor neurons.]

Grodins. F. S. 1963. CD lli rol Tlltory ,wd BitJlogiltl/ SyS/lm $. Columbi(l Univ. Press, New York. [A compact textbook, useful both as an introduction to this (lpprO<lch and to some ins ightfu l examples, which .ue partly neur.l l.!

Horridge, C. A. 1968. 'll/mlll.rou s. W. H. Freeman and Company, S.lIl Francisco. [A collection of diverse examples of neuf;11 mechanisms that involve interneurons a t several levels of analysis. ]

Reiss, R. F. 't964. NlUrtll TlltDI'y ami M odtling (Proc. t962 Ojai Synlpos ium ). Stanford Univ. Press, Stanford. [A symposium combining experi mental and theore tical trea tments of examples of sensory and motor beh.lvior. [ ""'-

Schmi tt, F. 0 ., and F. G. Worden. 1974. Tlt t N furDsrinl(fs: TI,jrd Study PrDgrll lll . (See Suggested Read ings in Chapter 8 fo r comments on this and two preceding volumes in the series.]

Stein, It 8., K. G. Pe.lTSOn, R. S. Smith, ,m d J. B. Redford. 1973. CO Il/rol of pos/u,-e alld LOCDrIl O/iDli . Plenum Press, New York. ]A symposium includ ing invertebrate and vertebrate sensory, ccntr,ll, reflex, and rhythmic control.!

St.uk, L 1968. NfrfrDIDgirll l CDII/ ru/ Sys/rllls: Slrulits ill BiolHgillrniag. Plenum Press, New York. [Exemplary ,111,l lyses from an engineering ilpproach of some subsystems controll ing seei ng and manipulating.]

Wiener, N., and J. P. 5chad~. 1965. Cybmltiirs of Ol t Nrrvolls 51/51t lll (Progress in the Br.l in Resea rch, vol. 17). Elsevier Publishi ng Company, Amsterdam. IAlt hough many cha pters (Ire dated or mainly of historical interest, the collection exemplifies how diverse are the problems and stages of progress under this rubric. Many modern symposia and articles are readily fo und, in addition to the journ.l ls whose ti lles Cil rry Wiener's term.)

Wiers nlol , C. A. G. 1967. IIII'tr/tbral, Nuvcllts SysltJIl5; Tluir SigllifiCIIll ft fDr MllmmR/ia li Nfur". Vll 1/sivlQgy. Un iv. Chicago Press, ChicJgo. [A symposium deal ing a t many levels wi th results relevant to Ill<lmma li,11l neurophysiology.]