Visual inter-hemispheric processing: Constraints and potentialities set by axonal morphology

J. Physiol. (Paris) 93 (1999) 271-284© 1999 Editions scientifiques et médicales Elsevier SAS. All rights reserved

Visual inter-hemispheric processing: Constraints and potentialitiesset by axonal morphology

Jean-Christophe Houzela*, Chantal Milleretb

aMax Planck Institut fiir Hirnforschung, Deutschordenstr. 49, 60528 Frankfurt/Main, GermanybLaboratoire de Physiologic de la Perception et de l'Action, CNRS UMR9950,

Collège de France, II, Place Marcelin-Berthelot, 75005 Paris, France

(Received 5 January 1999; accepted 3 February 1999)

Abstract - The largest bundle of axonal fibers in the entire mammalian brain, namely the corpus callosum, is the pathway throughwhich almost half a billion neurons scattered over all neocortical areas can exert an influence on their contralateral targets. These fibersare thus crucial participants in the numerous cortical functions requiring collaborative processing of information across the hemispheres.One of such operations is to combine the two partial cortical maps of the visual field into a single, coherent representation. Thispaper reviews recent anatomical, computational and electrophysiological studies on callosal connectivity in the cat visual system. Weanalyzed the morphology of individual callosal axons linking primary visual cortices using three-dimensional light-microscopictechniques. While only a minority of callosal axons seem to perform a strict ‘point-to-point’ mapping between retinotopicallycorresponding sites in both hemispheres, many others have widespread arbors and terminate into a handful of distant, radially orientedtufts. Therefore, the firing of a single callosal neuron might influence several cortical columns within the opposite hemisphere. Computersimulation was then applied to investigate how the intricate geometry of these axons might shape the spatio-temporal distribution oftrans.callosal inputs. Based on the linear relation between diameter and conduction velocity of myelinated fibers, the theoretical delaysrequired for a single action potential to reach all presynaptic boutons of a given arbor were derived from the caliber, g-ratio and lengthof successive axonal segments. This analysis suggests that the architecture of callosal axons is, in principle, suitable to promotethe synchronous activation of multiple targets located across distant columns in the opposite hemisphere. Finally, electrophysiologicalrecordings performed in several laboratories have shown the existence of stimulus-dependent synchronization of visual responsesacross the two hemispheres. Possible implications of these findings are discussed in the context of temporal tagging of neuronalassemblies. © 1999 Éditions scientifiques et médicales Elsevier SAS

axonal tree / corpus callosum / spike propagation / temporal coding / visual cortex

1. Introduction

1.1. Two brains for one body?

Before investigating possible correlations betweenstructure and function at the level of single axonaltrees, let us briefly consider more generally how themacroscopic structure of organisms relates to theirbehavior and, in turn, to the structure of their brains.Despite the fascinating diversity of forms encounteredin the animal kingdom, both sensory and motor organsof almost all species, at least all arthropods andvertebrates, come pairwise and are located on eachside of their longitudinal axis. Only those organsinvolved in internal regulation fail to comply with thisrule, suggesting that the symmetric layout applies

* Correspondence and reprints, present address: Lab. Neuro-plasticidade, Dept. Anatomia, Centro de Ciências da Saúde,UFRJ, 21941-590 Rio de Janeiro, Brasil.E-mail: [email protected]: CVM, central vertical meridian of the visualfield; 17/18, cortical region forming the transition betweenthe cytoarchitectonically defined visual areas Al7 and Al8.

manifestly to the apparatuses enabling organisms torelate with their environment. Indeed, our sensesbasically proceed by a balance between pairs ofsensors, as our acts result from a dynamic equilibriumbetween pairs of effecters, and our decisions oftenfollow judgments from contrasting points of view. Ifone considers, as modem cognitive neuroscienceteaches us to do, that the organization of mentalrepresentations somehow reflects the structure of thebrain, it is not surprising that our encephalon alsocomes in two halves. Yet, the essence of this biparti-tion remains far from being clear.

From early on, neurology and neuropsychologyindicated a marked dichotomy between a ‘dominant’hemisphere, usually the left one in right-handers,committed to analytical, sequential, verbal, local, ra-tional or objective processing, and a ‘minor’ hemi-sphere concerned rather with operations requiringsynthetic, global, spatial, more intuitive or affectiveabilities. Such a trend to conceive each hemisphere asan entity able to achieve a variety of perceptual ormnemonic tasks on its own was encouraged by famousobservations in commissurotomized patients [6, 17,

272 Houzel and Milleret

66]. Nowadays, techniques allowing the monitoring ofactivity in the living brain demonstrate that well-defined tasks are accompanied by localized changes inactivation levels. A marked asymmetry in the activa-tion patterns is observed for most tasks, confirming theundeniable hemispheric functional specialization.However, even simple tasks do not involve exclusivelyone hemisphere; since in real life we keep swappingbetween left-dominated and right-dominated opera-tions, the harmony of cerebral functions relies on adynamical cooperation between both sides of thebrain, and calls for a considerable traffic of informa-tion across the two hemispheres.

In addition to sub-cortical and anterior commis-sures, the major path for interhemispheric communi-cation is the corpus callosum. Contrasting with theamount of research devoted to this pathway and theamazing popularity of the concept of interhemisphericcooperation, it is striking to realize how little we knowabout the neuronal mechanisms of interhemisphericintegration (see [31]). This situation is probably due tothe abundance of callosal fibers and to their manifoldfunctions: callosal connections exist for sensory, mo-tor, associative, frontal or limbic cortices, and they linkheterologous as well as homologous areas. In order tostudy the role of the corpus callosum in interhemi-spheric integration, it seems wise to choose a systemwhere straightforward questions can be addressed. Inthis respect, the primary visual cortex offers theadvantage that, as recognized a century ago by Ramóny Cajal [57], at least one function of callosal connec-tions might be inferred from the topographic organi-zation of the visual system.

1.2. Need for interhemispheric cooperation in vision

In all mammals, each hemisphere handles visualinformation seen in the opposite hemifield. In inferiormammals, such as rodents or lagomorphs, which havelateral eyes and limited binocular vision, axons fromalmost all retinal ganglion cells cross the midline at theoptic chiasm and terminate within the contralateralgeniculate nucleus. In higher mammals with frontaleyes and a large binocular visual field, such as cats orprimates, axons arising from ganglion cells located inthe nasal half of the retina cross the midline andproject to the contralateral thalamus, whereas axonsfrom the temporal retinae remain on the ipsilateralside. Since each hemisphere only gets primary sensoryinputs from one half of the visual field (figure 1), itfollows that the visual world is represented in twophysically discontinuous cortical maps, split across thetwo hemispheres along the central vertical meridian ofthe visual field (CVM).

Despite this conspicuous gap in the cortical repre-sentation of the visual field, we do not perceive any

fracture along the vertical midline. On the contrary, themain function of the elaborate circuitry responsible forgaze control is to capture relevant objects within thevery center of the visual field where detailed analysiscan be best achieved. As Ramón y Cajal stated in hischapter on ‘the necessity for the corpus callosum’ [58],this “suggests the postulate that correct mental per-ception of the visual space cannot be achieved withoutthe existence of a bilateral perceptual brain centerwhose halves act in a concerted way with one another,in a way that unifies and places in the same orientationthe two images that left and right retinas project”. Inother words, the perceptual representation of the visualfield has to be integrated across the hemispheres. In thefollowing sections, we will briefly review classicalanatomical and physiological data on callosal connec-tivity between primary visual cortical areas and sum-marize more recent data about single axon morphologyand multiple recordings.

2. Basic layout of callosal connections betweenprimary visual areas

In each hemisphere, one half of the visual field isrepresented in several different maps, each of themcorresponding to a cytoarchitectonic area. Amongthose numerous maps, only the ‘primary’ ones (Al7 orVl, and Al8 or V2) cover the entire hemifield andrespect a strict retinotopic organization. Furthermore,the projections in Al7 and Al8 are contiguous andarranged in a mirror-like fashion, both maps sharing acommon representation of the CVM (cats: [70]; hu-mans: [63]). As a consequence, the 17/18 border zonematches the CVM projection line, which is also theedge along which the cortical representation shiftsfrom one hemisphere to the other (figure I).

Remarkably, the 17/18 border also coincides withthe narrow portion of primary visual areas wherecallosal connections are present. Indeed, retrogradeand anterograde tracing techniques demonstrated thatwhereas most parts of areas 17 and 18 are free ofinterhemispheric projections, cell bodies and axonalterminals of callosal neurons are densely packed alongtheir common border (reviewed in [28]). Seen from thecortical surface, the callosal zone forms a ribbonrunning along the CVM representation [40, 53]. Thecallosal ribbon is slightly wider in the representation ofthe area centralis, where the magnification factor ismaximal, but as a whole, taking into consideration thesmaller size of central receptive fields, the portion ofvisual field represented within the callosally connectedzone has the shape of an hour-glass centered on theCVM and spanning several degrees apart from it onboth sides [55]. In the cat, additional patches ofcallosal connections are found in more lateral parts ofA18, but those regions were shown to be islands of

Callosal axons and visual interhemispheric processing 273

Mammals with frontal, binocular vision(Man, monkey, cat,..)

representation

corpus callosum

A18 A l 7 A l 7 Al8

optic radiations

thalamus

optic chiasmoptic nerve

retina

Mammals with lateral, monocular vision(rabbit, rat,..)

corpus callosum

A18 A17 A17 A18

LEFT RIGHT LEFT R I G H T

visual field

Figure 1. Visuotopic organization of the retino-geniculo-cortical pathway. Bottom. The central 150º of the visual field arerepresented as the arc of a circle, with its left and right halves in shades of green and purple respectively, the intensity of thecolor being proportional to eccentricity. The blue and orange arcs indicate the portions of the visual field seen through the leftand right eye, respectively. The large black arrow represents a visual scene crossing the vertical meridian (dotted line). Center.Schematic of the visual pathways. Colors indicate the respective projections from the temporal retina of the left eye (light blue),the nasal retina of the left eye (dark blue), the nasal retina of the right eye (red) and the temporal retina of the right eye (yellow).Thin lines show the resting position of the optic axes. The head and the tail of the arrow represent the fragments of the visualscene projected at successive processing stages (retina, thalamus, cortical areas 17 and 18). Whatever the species, eachhemisphere only receives direct information about the opposite half of the visual field. The cortical projection of the centralvertical meridian (thick dotted line) coincides with the border between areas 17 and 18. Top. Theoretical perceptualrepresentation of the visual field. Dense, reciprocal interhemispheric connections between the left and right 17/18 borders aresupposed to link to the two partial maps and thereby allow coherent processing of visual scenes.

displaced cells representing fragments of theCVM [45, 60]. These fundamental observations cor-roborate the hypothesis that the primary function of

callosal connections between Al7 and A18 is to linkthe midline representations across the two hemi--spheres.


Accordingly, electrophysiological recordingsshowed that the receptive fields of trans-callosallyactivated neurons [10], as well as of single callosalfibers [3, 25] or of antidromically identified callosalefferent neurons [21, 27] are located on the CVM or inits immediate vicinity. Furthermore, the use of thesplit-chiasm preparation made it possible to demon-strate that trans-callosal and ipsilateral, geniculo-cortical inputs converging onto a given target neuronare precisely matched in their selectivity for stimulusorientation, direction and velocity, and that receptivefields plotted through both pathways are virtuallysuperimposed [4, 47] .

Taken as a whole, classic anatomical and physio-logical data seem to indicate that interhemisphericconnections are extremely precise and meet the re-quirements for achieving a point-to-point mappingbetween corresponding loci of the two hemispheres.However, detailed investigations using modern mor-phological techniques revealed that the organization ofcallosal connections is more complex than previouslythought.

3. Diversity of cellular componentswithin the callosal pathway

3.1. Callosal connections stem from neuronswith different morpho-chemical phenotypes

During the last decade, the technique consisting ofin vivo retrograde labeling followed by in vitro intra-cellular filling with lucifer yellow favored a morethorough analysis of cellular morphology than possiblewith horseradish-peroxidase techniques. Studies usingthis new method indicated that, as expected from theirelectrophysiologically and histochemically recognizedexcitatory function, the vast majority of callosal neu-rons are large pyramidal cells; however, as much asone-third of callosal cells have a different pheno-type [8, 71]. Among those are spiny stellate, but alsosmooth stellate and fusiform cells, which suggests thatsome callosal neurons could use inhibitory transmit-ters. Despite the failure to identify GABA-containingcallosal neurons, using either immunocytochemistry orselective uptake of radio-labeled transmitter ([13]; butsee [11]), the hypothesis of an inhibitory componentremains compatible with the occasional observation ofsymmetric callosal synapses as well as with theelectrophysiological disclosure of short-latency tran-scallosal inhibition after pharmacological or cryogenicmanipulations [56, 67]. Therefore, given their variousmorphological and chemical phenotypes, callosal neu-rons do not constitute a homogenous population.

3.2. Visual callosal axons display a significantvariability in their caliber

Electron microscopic studies in cats have shownthat about two-thirds of visual callosal axons areunmyelinated fibers between 0.08 and 0.40 µrn indiameter, whereas the remaining third consists ofmyelinated axons ranging from 0.25 to 4.0 µrn [2].This indicates that callosal axons display a consider-able variability in caliber. However, these values werederived from samples of the posterior part of thecorpus callosum, where axons from all visual areas areknown to travel [36]. Since this caliber heterogeneitymight have important functional implications (seebelow), it was interesting to investigate whether it alsoheld for axons originating from well-defined portionsof primary visual cortex.

The recent development of sensitive anterogradetracers allowed the inspection of individual callosalaxons (see [24] for technical details and extensivedescription of the results). Briefly, small amounts ofbiocytin were injected extracellularly close to the areacentralis projection along the 17118 border in adultcats; after adequate survival time, animals were killedand perfused with buffered aldehydes; the tracer wasthen revealed in serial sections, using a standardavidin-biotin-peroxidase histochemical procedure. Theresulting material was highly contrasted and allowedthe unambiguous distinction of stained axonal pro-cesses. Typically, a tracer injection yielding an uptakezone about 500 µrn wide and spanning throughout theentire cortical thickness labeled callosal axons whosediameters were found to range between 0.25 and2.25 µrn, when measured at the midline (figure 2).Although this method obviously overlooks axonalsegments falling below the resolution of the light-microscope (ca. 0.25 µrn), it demonstrates that evenwithin a restricted volume, corresponding roughly tothe size of a functional cortical module, callosalneurons send axons with calibers differing by at leastone order of magnitude.

Considering this wide range in axonal caliber andthe diversity of cellular phenotypes, it is likely that thecallosal pathway actually comprises several channelswith diverse morphological and physiological features,playing distinct roles in the interhemispheric process-ing of visual information. Therefore, it seemed crucialto investigate the relations established by single cal-losal neurons. Since biocytin yields a complete stain-ing of neurons, it allows one to follow all the branchesstemming from a given neuron and, in principle, toreconstruct its entire axonal tree throughout a stack ofconsecutive sections. From a practical point of view,such an approach is surely exacting; providentially,techniques for computer-assisted microscopy under-went considerable progress during the last decade and


diameter (µm)

Figure 2. Variability of axonal caliber within the corpuscallosum. A volume of 0.8 µl biocytin (5%) was injected atthree different depths along a vertical penetration within theapex of the lateral gyrus, close to the 17/18 border (triangle).The resulting uptake zone, characterized by sharp edges, isillustrated by a frontal section through its center (A) and bya schematic 3D reconstruction (B) where its tangential andlaminar extents can be assessed. This localized injectionlabeled 48 axons, some of which can be seen in a montage ofhigh power micrographs from the midline of the callosum(C). The axons ranged between 0.25 and 2.25 µm in diame-ter (hatched bars in D). For comparison, the distribution ofcalibers in our sample of 20 completely reconstructed axons,stemming from 12 injections in seven normally raised adultcats, is indicated by the solid bars. n, number of axons. NB:axons falling below the resolution of conventional lightmicroscopy are not taken into account.

now enable one to examine the precise three-dimensional morphology of long-range projecting ax-ons. Using the Neurolucida™ equipment [18], weperformed such reconstructions for a sample of 20callosal axons from cat visual cortex [24](Houzel andMilleret, unpublished results). Briefly, axons labeledby extracellular injections of the tracer were randomlyselected at the midline and traced up to their distal-most endings in the target hemisphere, with a finalmagnification of 2000x, yielding a resolution of0.23 µm. The acquisition program registered the spa-tial coordinates and geometrical properties of axonalsegments, which could then be followed from one75 µm-thick section to the next. Using a graded ocularunder high magnification the diameter of each axonsegment was recorded. Particular care was taken toregister changes in diameter which occurred frequentlyat branching points, and to plot all differentiationscharacteristic of presynaptic structures, such as short-stalked boutons or en passant swellings. Tissue defor-mations resulting from the histological procedure werealso estimated and corrected accordingly. Analysis of

the spatial geometry of the reconstructed trees wasdone using ad hoc software (‘Maxsim’, see below).

3.3. Callosal axons difer in the size of their terminalterritories

A puzzling result of our study was the diversity inthe size of the terminal arbor of individual axons. Onlya minority of them (three out of 20) terminated into asparsely ramified tuft and distributed their boutonswithin a single conical volume (see arbor A in fig-ure 3). The estimated cortical surface covered by suchaxons was found to be restricted to a few hundred µm2.Interestingly, these narrow arbors were also restrictedin their laminar distribution and occurred in anteriorportions of the 17/18 border, i.e., in regions withrelatively lower magnification factor than within thearea centralis projection zone.

All other axons started to divide in the white matter,where they gave rise to a few branches which ramifiedextensively within the cortex. After all boutons of agiven axon were projected radially onto the corticalsurface, the smallest possible convex polygon encom-passing all of them was usually larger than 0.25 mm2

(12/17 axons), and could reach as much as 9 mm2.

Thus, callosal axons linking primary visual areasdisplay a high variability in the extent of their terminalterritory. Given the local magnification factors ofvisual maps in the origin and target zones, this impliesthat most callosal neurons do not establish a strictpoint-to-point mapping between topographically cor-responding loci of the two hemispheres. As a rule, theheaviest site of termination is aimed at the homotopiccontralateral site, but many axons form diverging,heterotopic connections between one cortical locus atthe 17/l 8 border and several loci spread throughout thecontralateral A 17, A 18 and 17/l8 border (see examplesC and D in figure 3). Conspicuous branches bearingnumerous preterminal boutons (see micrograph infigure 3D) are thus aimed at cortical regions represent-ing quite peripheral sectors of the visual field, andwhere physiological techniques failed to reveal anysupraliminar influence from the contralateral visualcortex. This observation suggests that trans-callosalinputs do not merely contribute to the construction ofthe receptive fields of target neurons, but are likely toserve alternative or additional functions.

3.4. Individual callosal neurons terminatewithin distinct columnar and laminar compartments

Despite the diversity in the overall tangential extentof the terminal territory of these profuse arbors, theirboutons were invariably clustered into several corticaldomains which, having the shape of radially-orientedcylinders, will be referred to as ‘terminal columns’although there is as yet no downright relationship


Figure 3. Architecture and topography of visual callosal axons. Four axons (A-D) of increasing projection territory are shownwith different magnifications and viewing angles. Shaded band indicates layer IV, triangles point to the center of the 17/18transition zone; ruler marks are placed every 500 µm. Left insets. Location of the respective tracer injection within the oppositehemisphere, labeled zone in black. Right columns. Schematic of the injection sites (inset) and of the modular clustering ofcallosal terminals. Boutons were projected onto the pial surface of a smoothed cortex for two-dimensional cluster analysis, andgrouped according to their laminar position. The shading of each terminal sector is proportional to its contribution to the entireterminal territory of the axon (percentile of the total number of presynaptic boutons). A. Simple axon with restricted, homotopictermination. B. For this axon, two terminal columns, aligned along the rostro-caudal axis, appear upon changing the viewingangle. Rotation is indicated by the dorsal direction (arrowhead) and the standard Horsley-Clarke frontal plane (gray triangle)which was used for sectioning the brain. On the enlarged 3D views, arbor and boutons are plotted separately to ease identificationof radial clusters. C. Axon with two terminal columns aligned along a medio-lateral direction; enlarged details from the frontview show that the proximal and narrow column (in A18) is fed by a thinner branch than the more distal and larger bouquet (atthe 17/18 border). D. This axon was labeled by the same injection than C; it terminates into five widely spaced columns. Detailsfrom the lateral-most column show branchinns patterns and preterminal structures (arrow in the microphotograph point to enpassant boutons).


between them and the radially oriented functionalmodules characteristic of visual areas, such asorientation- or ocular dominance columns. Actually,callosal terminal columns are very irregular since, inour limited sample, their diameter varied from 100 to800 µrn, the spacing between two neighboring col-umns of an axon was found to range between 100 and2000 µm and the number of terminal columns per axonranged from 2 to 8. Furthermore, the boutons were notdistributed homogeneously throughout the depth ofterminal columns, but rather concentrated in certainlayers. The majority of all columns (35 out of 50 for allthe 20 axons) displayed a high density of boutons insupragranular layers, often associated with a second-ary but substantial infragranular component (16/35).Other columns contained boutons spread throughoutall layers (9/50), while fewer were restricted to layersIV (3/50) or V-VI (3/50). Axons usually terminatedinto several columns of different laminar distribution.The only clear trend in this intricate pattern was that allsimple axons, terminating into a single column, wereinvariably restricted to the thalamo-recipient layer IV.To clarify possible hierarchical principles for thispathway, it will be necessary to correlate the spatialdistribution of contralateral terminals of callosal neu-rons with the laminar location and the morphologicalphenotype of their cell bodies in the source hemisphere(see Conclusion).

One important issue is the functional meaning ofthe discrete, columnar layout of callosal projections.Thirty years ago, applying the newly developed Nautatechnique to visualize fibers degenerating after sectionof the entire corpus callosum, Heimer et al. [22]already noted that the overall distribution of callosalterminals was uneven over the cortical surface. Closerinspection with progressively more sensitive methodssuggested a tendency of callosal neurons and terminalsto form patches along the Vl/V2 border [7,71], but thesignificance of this pattern remained elusive. Recently,Olavarria and Abel [52] showed that callosal connec-tions link preferentially cortical domains characterizedby strong cytochrome-oxidase activity which, at leastin primates, are undoubtedly related to segregatedprocessing streams [9]. Although this correlation pro-vides a first hint that the patchiness of interhemisphericlinkage might relate to the functional compartmentali-zation of visual areas, it is not known how far thismight hold for the relations established by singleaxons. This aspect could be investigated further bycombining anatomical experiments with the functionaldetermination of cortical domains (see Conclusion).For the time being, the careful observation of purelymorphological data might provide additional insightsinto the consequence of this columnar pattern forinformation processing.

3.5. Callosal axons differ in their architecture

Axonal bifurcations were usually accompanied bythe shrinkage of daughter branches, whose respectivediameters could be reduced by the same factor or not.Detailed inspection of such branching points wascarried out for entire trees, and yielded a comparisonof the relative caliber of ramifications supplying theirdistinct terminal columns. Since many arbors wereexhaustively described in a previous paper [24], wewill simply use the three representative examplesillustrated in figure 3 to underline the main conclu-sions.

The axon shown in figure 3B forms two radialclusters of boutons which, as can be easily perceivedfrom the top and side views, are supplied by twoindependent branches that parted from the main trunkwithin the white matter. Since those branches had verysimilar calibers, and ran almost side by side for aconsiderable distance (ca. 2 mm), the architecture ofsuch axons could be qualified as ‘parallel’.

In contrast, axon C, which also terminated into twocolumns, was characterized by a tangentially runningtrunk from which arose a set of radially ascendingbranches of different calibers. As can be seen in theenlargement of the front view (‘main branchings’;figure 3C, the proximal branch supplying a narrowcolumn within A18 was much thinner (0.26 µm) thanthe trunk, which maintained its original caliber(1.12 µm) for several millimeters, until it ramified intoa second, more distal tuft at the 17/18 border. Inopposition with the previous type, the architecture ofsuch axons was qualified as ‘serial’.

Many axons that terminated into more than twocolumns displayed more complex branching patterns.One such tree is shown in figure 3D. It terminated intofive columns spread throughout areas 17 and 18. Thedetailed front view reveals that the most proximalbouquet (in A18) was supplied by 2-3 offshootsforming a typical ‘serial’ column. On the other hand,the top view indicates that the remaining part of thistree was characteristic of what was defined as a‘parallel’ architecture, since the two broad columns(on the 17/18 border) are fed by branches withcomparable calibers, traveling in a similar directionover a considerable distance. Finally, one can distin-guish an additional branch running between these twocolumns, with a rostro-caudal direction, i.e., exactlyalong the 17/18 border. This suggests some degree ofreconvergence and the existence of cross-talks be-tween spatially distant terminal columns (see alsofigure 4D).

Therefore, callosal axons display one more variablefeature, namely their architecture; in other words, theydiffer in the geometrical properties of the branchessupplying their different columns. Since conduction


Vc physio(m/s)

t-

1.427.013.0 A

- kfront view

C

DFigure 4. Simulation of the spatio-temporal activation profiles for individual callosal axons. A. Quantitative values returned byour simulation or measured physiologically by different authors (shaded columns). From left to right: diameter of the axonaltrunk of the reconstructed arbors (as measured at the midline); raw conduction velocity, derived from the linear equation ofWaxmann and Bennet [73]; conduction velocity after correction for tissue shrinkage; conduction velocity measured in vivo byHarvey [21]; latency required for a simulated action potential to travel from the midline to the first bouton of the arbor; latencyestimated after antidromic activation by McCourt et al. [43]. B. Activation profile of a ‘parallel’ tree (see figure 3B for itsmorphology). In the front view as well as in the progressively enlarged side views, axonal segments and boutons are coloredaccording to their time of activation (µs), considering that the spike crossed the midline (arrow) at time zero. Histograms (100µs binwidth) indicate the distribution of activation times for the entire arbor (top) and for mdlvidual columns (a and b). C.Activation profile of a ‘serial’ axon (morphology in figure 3C). Different side views corresponding to successive time points(tl-t7) are shown. The portion of the arbor (t1-t3) or the fraction of boutons (t4-t7) activated during a temporal window of 100µs is shown in red, while the rest of the tree is in gray. D. Activation of a complex axon (morphology in figure 3D). Front viewand lower top view: segments are color-coded according to their activation time, as in B. In the upper top view, colors indicateethe primary level of bifurcation to ease the recognition of the converging branch (see text).


velocity is tightly related to axonal caliber, thesegeometrical properties have direct consequences onthe timing of trans-callosal impacts.

4. Possible implications of axonal architecturesfor temporal processing

4.1. Investigation of the computational propertiesof axonal trees

As already mentioned by Segev and Schneidman[61], a decisive step in modeling the electrical prop-erties of axons was the work of Goldstein andRall [19], indicating how branching points or abruptchanges in diameter modify the impedance of conduct-ing segments and thereupon affect the shape as well asthe velocity of traveling action potentials. Withcomputer-simulation of compartmental Hodgkin-Huxley models, the theoretical consequences of geo-metrical irregularities on spike propagation and boutonactivation were re-investigated in great detail [38, 39].Particularly elegant, dynamic implementation of suchmodels recently allowed the examination of morecomplex, and hence more realistic axonal trees [41].These approaches suggest that minor irregularities,such as varicosities or branching points, are likely todelay or accelerate the propagation of a spike. How-ever, they also indicate that most of the time requiredto activate an axonal tree results from the conductionalong its long, homogenous branches: “in other words,the pure delays from the axonal cables (placing thebranches end to end and neglecting the interveningbranch points and varicosities) account for most (67-78%) of the delay” [42]. Therefore, despite the lack ofinformation about many parameters important foraccurate modeling, such as the local density of ionicchannels at branching points and at nodes of Ran-vier [.5, 741, it seemed reasonable to use currentlyavailable data to derive average conduction delays.Furthermore, all callosal axons we reconstructed had atrunk diameter greater than 0.25 µm, and were thuspresumably myelinated. Since the conduction velocityof myelinated axons increases linearly with fiberdiameter, our task was greatly simplified.

Rather than simulating the successive bioelectricalevents responsible for spike propagation, we chose asimple approach consisting of the following steps:First, electron microscopic examination of somesamples (Aggoun-Zouaoui et al., unpublished obser-vations) indicated that biocytin specifically stained theaxoplasm; that all labeled callosal axons above0.25 µm diameter were indeed myelinated; and that theratio of axoplasm diameter to fiber diameter (includingthe myelin sheath) was systematically close to 0.7,agreeing with previous measurements in rabbits andcats [2, 75]. Second, based on the linear relation

derived from visual callosal axons by Waxmann andBennet [73] (Vc = 5.5 x Df, where Df is the fiberdiameter, i.e., axoplasm diameter/0.7), we calculatedthe conduction velocity (Vc) of each axonal segment.Third, we summed these conduction delays, segmentper segment, and computed the resulting activationtime for every single bouton of the reconstructedarbors, considering that an action potential crossed themidline at time zero. Finally, activation times wererecorded and color-coded in the 3D graphical repre-sentation of each axonal tree, which could be enlargedand rotated as required. These steps were performedusing a program specifically developed for this pur-pose, whose implementation was described elsewhere(Maxsim [68]).

4.2. Spatio-temporal profiles suggest that callosalaxons promote the synchronous activationof their terminal columns

This simulation paradigm allowed the investigationof theoretical spatio-temporal activation profile ofseveral reconstructed axons. Since an exhaustive re-port of the results was published elsewhere [30], wewill recapitulate only the main findings, using the fourrepresentative axons whose morphology was describedabove.

For the simple axon of mid-range caliber (1.4 µm)described in figure 3A, the conduction velocity was11.2 m/s. Its proximal-most bouton was located13 mm away from the midline and was activated 1840µs after the spike crossed this point; the distal-mostbouton, only a few hundred pm apart, was reached 316µs later, indicating that the narrow terminal territory ofthis axon was entirely activated within a very shorttime interval. In contrast, the activation of the boutonsof axons with more widespread arbors spanned asmuch as 2700 µs. The table in figure 4A gives theranges and mean values returned by the simulation forall axons; they are compatible with experimentallymeasured conduction velocities and interhemisphericlatencies.

The activation profile of a typical ‘parallel’ axon isillustrated in figure 4B, by enlarged side views of thearbor and the boutons. Not surprisingly, this architec-ture leads to quasi-simultaneous invasion of bothterminal columns, all boutons being activated within400 µs. A quantitative estimation of the synchronicityis given by the distribution of activation times, indi-cating a substantial overlap in the activation of thecolumns. The paired coactivation index, expressed bythe fraction of boutons of the axon which were activewhile boutons of the other column were also active,was 60%.

The next panel of figure 4 shows the activationprofile for an axon with ‘serial’ architecture (whose


morphology is illustrated in figure 3C). The distribu-tion of activation times indicates a fair degree oftemporal overlap between the two terminals columns,despite their significant spatial separation. A sequenceof the successive simulation frames shows that thecoactivation is due to the marked reduction in diameterof the branch supplying the proximal column, whichintroduces a supplementary conduction delay andleads to the simultaneous invasion of both columns.Moreover, since the distal column contained aboutfour times as many boutons than the proximal one, thecoactivation index was very high (85%).

As mentioned earlier, the architecture of manycomplex axons was characterized by both serial andparallel components. In addition, we noted the occur-rence of cross-talks between several columns. Figure4D emphasizes one such case, described previously infigure 3D: in the top panel, branches are color-codedaccording to the level of the bifurcation they stemfrom. One can easily recognize a branch of highcaliber emerging from the posterior column (d) andrunning along the rostro-caudal axis for about 1.5 mmbefore converging onto the anterior column (c) whereit finally ramified into a terminal tuft. However, if theaxonal segments are color-coded according to theirrespective activation time (lower panel), one cannotdistinguish anymore the boutons supplied by thisconverging branch from those belonging originally tothe anterior column, because they are all activatedsimultaneously. In addition, these two columns show aconsiderable degree of coactivation with the moreproximal column (a), which was supplied by typicalserial branches. As a result, the coactivation index forthis axon was 95%.

With the limited sample of myelinated axons exam-ined arising from areas 17 and 18, our study is hardlyrepresentative of the diversity of the callosal pathway.Given this restriction, one may nevertheless concludethat many visual callosal axons are likely to conveyinformation with delays comparable to those observedfor intrahemispheric connections [50]. However largethe terminal territory of a callosal axon, its architectureis such that action potentials can invade it within nomore than a couple a milliseconds. Moreover, thisarchitecture appears well suited to lead to the simul-taneous activation of targets distributed across segre-gated terminal columns. Synchronization might beachieved through equalization of conduction times inparallel arbors, or through appropriate modulation ofcalibers in serial axons; moreover, the degree ofcoactivation might be enhanced, or adjusted, by con-verging branches with fitting geometrical properties.Given the theoretical and practical limitations of ourapproach, it is impossible to provide an accurateestimation of the actual time scale for this coactiva-tion. Time bins of 10 and 100 µs were chosen to

calculate coactivation indices and to represent thedistribution of activation times, respectively. There-fore, the synchronization is likely to operate at themillisecond range, and probably below. This issue isnevertheless complicated by the absence of a consen-sus regarding the temporal precision relevant forneuronal processing (for example, see [69]).

Another critical issue is whether one might be ableto validate experimentally the results of such simula-tions. It is probable that with the further developmentof voltage-sensitive dyes and fast optical systems (see[44]), these methods will become applicable to thinaxons and could thus be used for in vitro pre-parations [37]. This might be of great help to investi-gate structure-function relationships in simple axons.However, it is unlikely that, in a foreseeable future,any tool will render possible the direct visualization ofspatio-temporal patterns of activation in complex,long-range projecting axons in situ, i.e., within theirphysiological environment including the network ofafferent and efferent connections.

4.3. Functional significance of the propertiesof callosal axons

In the eventuality of a genuine coactivation ofcallosal terminals within the millisecond range aspredicted in our study, how could it participate in theinterhemispheric integration of visual scenes? Tenyears ago, while recording simultaneously the activityfrom several columns of the cat visual cortex, Gray etal. [20] observed that neuronal responses to lightstimuli were often oscillatory, and could be preciselysynchronized, on a millisecond time scale, providedthat the neurons were driven by a coherent stimulus.Stimulus-dependent synchronization was later ob-served for numerous areas and various species and isthought to provide a general code for the dynamicgrouping of neurons into functionally coherent assem-blies required to establish relations among the variouscomponents of sensory patterns. The synchronousdischarges of a group of neurons would be effectivelysummed up by target cells, thereby increasing thesaliency of the responses of the assembly for subse-quent joint processing. Such a flexible temporal codecould operate at a fast time scale and provide asolution to the ‘binding’ problem (reviewed in [65]).

Since callosal axons are capable of coactivatingtarget neurons located in several columns of therecipient hemisphere, they might well contribute to theprecise synchronization of distributed neuronal re-sponses. Moreover, callosal connections fulfill allthree postulated conditions for serving stimulus-dependent synchronization [64]: i) in contrast to thefeed-forward connections responsible for the genera-tion of neurons with feature-selective receptive-fields,


interhemispheric projections are basically reciprocal(see Introduction); ii) since they develop ratherlate [26, 47], callosal connections are highly suscep-tible to experience-dependent modification. Not onlytheir global layout [29], but also their functional char-acteristics [46] and their fine architecture [23] can beprofoundly altered in response to early manipulation ofvisual inputs. Hence, they are capable of reflectingfrequently occurring feature constellations, as requiredfor cortico-cortical networks involved in perceptualgrouping; iii) finally, in order to account for newfeature configurations, assembly-forming connectionsshould retain some degree of flexibility throughoutadulthood. Indeed, interhemispheric connections wereshown to be susceptible not only to deafferentation,but also to sensory experience in adult cats [45, 48].

4.4. Callosal axons and interhemisphericsynchronization

Precise temporal coupling of oscillatory visual re-sponses was found to occur not only between distantcolumns of the same hemisphere but also acrosshemispheres, between neurons located close the 17/18border zones. In both cases, the degree of synchroni-zation reflected global stimulus properties such ascontinuity, colinearity or common fate [14]. A recent,careful examination of the temporal patterning ofvisual responses revealed that interhemispheric syn-chronization might occur with different timescales [51]. In addition to the narrow peaks typical ofmillisecond-range coupling, cross-correlations com-puted over large peristimulus periods revealed theexistence of broader peaks, ranging between 30-100or 100-1000 ms. After section of the posterior corpuscallosum, precise synchronization was abolished,intermediate-range peaks were markedly reduced, butsome looser coupling was still present. In contrast,extensive lesions of extrastriate cortex did not affectthe sharp synchronization, whereas they strongly re-duced the occurrence of intermediate and loose cou-pling [49]. This indicates that the loose coupling ofvisual responses involves polysynaptic loops and feed-back projections from heterotopic areas, whereas theprecise interhemispheric coupling, presumed to servethe dynamic formation of assemblies, is mediated byreciprocal callosal connections between primary visualareas.

The properties of callosal axons described in thispaper are, alone, not sufficient to account for theoccurrence of precise, zero-phase lag synchronizationbetween the two hemispheres. They indicate only thatcallosal neurons located within one hemisphere arecapable of providing synchronized inputs to severalcolumns of the opposite hemisphere. As noted above,these inputs are not limited to regions where supra-

threshold contralateral influences can be recorded.Interestingly, recent in vitro work suggests that pre-cisely timed, subthreshold oscillations of membranepotential affect the spatio-temporal integration of ad-ditional converging inputs, and thereby facilitate thesubsequent synchronization of neuronal d i s -charges [72]. One can hypothesize that heterotopiccallosal connections establish temporal relations es-sential to the integrate representation of sensory fea-tures, while homotopic connections allow accuratetopographic correspondence between left and rightmaps and thereby contribute to the generation offeature-selective receptive fields, such as in disparity-sensitive neurons. We have shown that neurons withnarrow contralateral arbors link topographically corre-sponding loci, and that contralateral and ipsilateralinputs converging onto callosal-recipient neurons areprecisely matched in their functional characteristics,including retinotopic location. It remains to be estab-lished whether neurons with larger terminal territories,often consisting of both homotopic and heterotopiccomponents, contribute to one or both tasks. However,the diversity of the cellular components participatingin interhemispheric networks favors the hypothesis ofseveral trans-callosal channels.

There is no unique strategy to produce zero-phaselag synchronization between remote groups of neu-rons. Such precise synchronization might be achievedthrough common inputs, through reciprocal connec-tions and/or through local excitatory and inhibitoryinterconnections within more intricate networks. Com-puter simulations suggest that any of these threenon-mutually exclusive mechanisms could result in thetemporal patterns observed experimentally [32, 33] .Whatever the solution(s) actually used by interhemi-spheric connections, it is essential to keep in mind that‘callosal projecting’ neurons do not only project to thecontralateral hemisphere, but also have local axonalcollaterals providing inputs to remote targets withinthe hemisphere of origin (see below). If, as one mightinfer from data on cortico-cortical connections, theconduction delays are equalized for both terminationsites, such bi-hemispheric projecting neurons could beresponsible for inter-hemispheric synchronization.They would be crucial participants in the integration ofsensory features across the two hemispheres, andthereby in forming a unified cortical representation ofthe visual world, across the central vertical meridian.

5. Conclusion: directions for future research

As suggested in the previous section, informationabout the relations between the contralateral and theipsilateral axonal arbors of callosal neurons are re-quired for understanding the neuronal mechanisms ofinterhemispheric cooperation. Evidence for ipsilateral


axonal collateralization come from double-labelingexperiments [1] or antidromic activation studies [15]revealing that some callosal neurons in the rat frontalcortex also project to ipsilateral cortical or subcorticalstructures. These findings are compatible with therecent observation that callosal neurons survive thetransection of their callosal axon [54]. For visual areas,it was shown that a few extrastriate callosal neurons ofcats projecting to the contralateral 17/l 8 border alsoproject to the homologous ipsilateral region [62]. Theexistence of ipsilateral terminals of callosal neurons inprimary visual cortex is supported by the observationof the initial segments of axonal collaterals directed tothe ipsilateral hemisphere in cats [27] and of localsynapses labeled within the cortex contralateral toperoxidase injections in mice [12]. However, due totechnical limitations, our current knowledge of thetopology of the ipsilateral axonal arbors of callosalneurons is extremely limited. It could now be comple-mented by single neuron tracing work.

It will also be mandatory to investigate to whatextent the organizing principles observed in cats mighthold for other species as well, particularly in primates,whose visual cortex laminar and modular compart-mentalization is much better defined, and in otherwell-studied inferior mammals such as the rat. Re-markably, callosal linkages between primary visualareas of rodents are not strictly limited to the 17/18border; numerous layer V neurons located within theperipheral visual field representation of Al7 have acontralateral projection. Based on targeted retrogradetracer injections, it was recently proposed that theseconnections do not link retinotopically correspondingloci but rather mirror-symmetric points from bothhemispheres [34]. In this respect, such projectionswould resemble the interhemispheric connections be-tween cat or primate extrastriate areas and might playa role in the computation of optic flow and/or insymmetry detection. It is likely that the axons respon-sible for these projections have distinctive architec-tures.

Further, the possible relations between callosalterminals columns and functional domains of visualareas should be clarified. Straightforward correlationscould be obtained by combining single neuron tracingwith optical recording of intrinsic cortical signals. Thistechnique could be applied simultaneously for bothhemispheres, and would provide answers to bothprevious questions.

Ideally, further experiments should be conducted inspecies with less convoluted brains. This is essential inorder to obtain direct correlations in each individualanimal without having to rely on complex methods fortissue flattening or deconvolution algorithms, whichinevitably introduce deformations. Small-sized brainswould offer the additional advantage of reducing the

time needed for reconstruction. In addition to the rat,the marmoset monkey seems to be an ideal preparationin this respect. Moreover, studies would benefit ofrecent electrophysiological work which provided themost complete and accurate topographic maps ofvisual areas ever published for a primate [16, 35, 59].

Acknowledgments

We are indebted to Laurent Tettoni, Patricia Lehmann andGeorgio Innocenti for contributing to most of the workpresented; Jack Glaser for unfailing support with the Neu-rolucida equipment; Lam Hofer for providing access to theIndigo Station at MPIH; Renate Ruhl for invaluable helpwith the illustrations; Suzana Herculano-Houzel for insight-ful comments and thorough English revision; and Profs.Pierre Buser, Alain Berthoz and Wolf Singer for continuousencouragement. Work was supported by the Centre Nationalde la Recherche Scientifique, Collkge de France, EuropeanTraining Program, Fond National Suisse pour la Recherche,Fondation de France and the Max Planck Gesellschaft.Please contact J.C.H. for distribution of the Maxsim soft-ware.

References

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

Audinat E., Condé F., Crépel F., Cortico-cortical connectionsof the limbic cortex of the rat, Exp. Brain Res. 69 (1988)439-443.Berbel P., Innocenti GM., The development of the corpuscallosum: a light and electron microscopic study, J. Comp.Neural. 276 (1988) 132-156.Berlucchi G., Gazzaniga M.S., Rizzolatti G., Microelectrodeanalysis of transfer of visual information by the corpuscallosum, Arch. Ital. Biol. 105 (1967) 583-596.Berlucchi G., Rizzolatti G., Binocularly driven neurons in thevisual cortex of split-chiasm cats, Science 159 (1968)308-310.Black J.A., Kocsis J.D., Waxmann S.G., Ion channels organi-zation of the myelinated fiber, Trends Neurosci. 13 (1990)48-54.Bogen J.E., Fisher E.D., Vogel P.J., Cerebral commissuro-tomy: A second case report, J. Am. Med. Assoc. 194 (1965)1328-1329.Boyd J.D., Matsubara J.A., Tangential organization of callosalconnectivity in the cat’s visual cortex, J. Comp. Neurol. 347(1994) 197-210.Buhl E.H., Singer W., The callosal projection in cat visualcortex as revealed by a combination of retrograde tracing andintracellular injection, Exp. Brain Res. 75 (1989) 470476.Bullier J., Nowak L.G., Parallel versus serial processing: newvistas on the distributed organization of the visual system,Curr. Opin. Neurobiol. 5 (1995) 497-503.Choudhury B.P., Whitteridge D., Wilson M.E., The functionof the callosal connections of the visual cortex, Quart. J. Exp.Physiol. 50 (1965) 214-219.Cobas A., Fairen A., Alvarez-Bolado G., Sanchez M.P.,Prenatal development of the intrinsic neurons of the ratneocortex -A comparative study of the distribution of GABA-immunoreactive cells and the GABA-A receptor, Neuro-science 40 (1991) 375-397.


[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

Czeiger D., White E.L., Synapses of extrinsic and intrinsicorigin made by callosal projection neurons in mouse visualcortex, .I. Comp. Neurol. 330 (1993) 502-513.Elberger A.J., Selective labeling of visual corpus callosumconnections with aspartate in cat and rat, Visual Neurosci. 2(1989) 81-85.Engel A.K., König P., Kreiter A.K., Singer W., Interhemi-spheric synchronization of oscillatory neuronal responses incat visual cortex, Science 252 (1991) 1177-1179.Ferino F., Thierry A.M., Saffroy M., Glowinski J., Interhemi-spheric and subcortical collaterals of medial prefrontal corti-cal neurons in the rat, Brain Res. 417 (1987) 257-266.Fritsches K.A., Rosa M.G.P., Visuotopic organization ofstriate cortex in the marmoset monkey (Callithrix jacchus), J.Comp. Neurol. 372 (1996) 264-282.Gazzaniga M.S., Bogen J.E., Sperry R.W., Laterality effects insomesthesis following cerebral commissurotomy in man,Neuropsychology 1 (1963) 209-215.Glaser J.R., Glaser E.M., Neuron imaging with neurolucida: APC-based system for image combining microscopy, Comput.Med. Imag. Graphics 14 (1990) 307-317.Goldstein S.S., Rall W., Changes of action potential shape andvelocity for changing core conductor geometry, Biophys. J. 14(1974) 731-757.Gray CM., König P., Engel A.K., Singer W., Oscillatoryresponses in cat visual cortex exhibit inter-columnar synchro-nization which reflects global stimulus properties, Nature 338(1989) 334.Harvey A.R., A physiological analysis of subcortical andcommissural projections of areas 17 and 18 of the cat, J.Physiol. (Lond.) 302 (1980) 507-534.Heimer L., Ebner F.F., Nauta W.J.H., A note on the termina-tion of commissural fibers in the neocortex, Brain Res. 5(1967) 171-177.Houzel J.C., Organisation anatomo-fonctionnelle, developpe-ment et plasticité des connexions inter-hemispheriques entreles aires visuelles primaires chez le chat, Thèse de doctorat,Université Pierre & Marie Curie, Paris, 1996.Houzel J.C., Milleret C., Innocenti G.M., Morphology ofcallosal axons interconnecting areas 17 and 18 of the cat, Eur.J. Neurosci. 6 (1994) 898-917.Hubel D.H., Wiesel T.N., Cortical and callosal connectionsconcerned with the vertical meridian of visual field in the cat,J. Neurophysiol. 30 (1967) 1561-1573.Innocenti G.M., Postnatal development of interhemisphericconnections of the cat’s visual cortex, Arch. Ital. Biol. 116(1978) 463470.Innocenti G.M., The primary visual pathway through thecorpus callosum: morphological and functional aspects in thecat, Arch. Ital. Biol. 118 (1980) 124-188.Innocenti G.M., General organization of callosal connectionsin the cerebral cortex, in: Jones E.G., Peters A. (Eds.),Cerebral Cortex, Vol. 5, Sensory-motor areas and aspects ofcortical connectivity, Plenum Press, New York, 1986, pp.291-353.Innocenti G.M., Frost D.O., Effects of visual experience onthe maturation of the efferent system of the corpus callosum,Nature 280 (1979) 23 l-234.Innocenti G.M., Lehmann P., Houzel J.C., Computationalstructure of visual callosal axons, Eur. J. Neurosci. 6 (1994)918-935.Kennedy H., Meissirel C., Dehay C., Callosal pathways andtheir compliancy to general rules governing the organizationof corticocortical connectivity, in: Dreher B., Robinson S.R.

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[401

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

(Eds.), Vision and visual dysfunction, Vol. 3, Neuroanatomyof the visual pathways and their development, McMillanPress, London, 1991, pp. 324-359.König P., Engel A.K., Correlated firing in sensory-motorsystems, Curr. Opin. Neurobiol. 5 (1995) 511-519.König P., Schillen T.B., Stimulus-dependent assembly forma-tion of oscillatory responses: I: Synchronization, Neural.Comput. 3 (1991) 155-166.Lewis J.W., Olavarria J.F., Two rules for callosal connectivityin striate cortex of the rat, J. Comp. Neurol. 361 (1995)119-137.Liu G.B., Spitzer M.W., Pettigrew J.D., Rosa M.G.P., Opticalmapping of extra-striate visual areas in the marmoset, in:Third Australian meeting of Brain Mapping, Canberra, 1998.Lomber S.G., Payne B.R., Rosenquist A.C., The spatialrelationship between the cerebral cortex and fiber trajectorythrough the corpus callosum of the cat, Behav. Brain Res. 64(1994) 25-35.Lüscher C., Lipp P., Lüscher H.R., Niggli E., Control of actionpotential propagation by intracellular Ca2+ in cultured ratdorsal root ganglion cells, J. Physiol. (Lond.) 490 (1996)319-324.Lüscher H.R., Shiner J.S., Computation of action potentialpropagation and presynaptic bouton activation in terminalarborizations of different geometries, Biophys. J. 58 (1990)1377-1388.Lüscher H.R., Shiner J.S., Simulation of action potentialpropagation in complex terminal arborizations, Biophys. J. 58(1990) 1389-1399.Malach R., Patterns of connections in rat visual cortex, J.Neurosci. 9 (1989) 3741-3752.Manor Y., Gonczarowski J., Segev I., Propagation of actionpotentials along complex axonal trees: model and implemen-tation, Biophys. J. 60 (1991) 1411-1423.Manor Y., Koch C., Segev I., Effects of geometrical irregu-larities on propagation delay in axonal trees, Biophys. 3. 60(1991) 1424-1437.McCourt M.E., Thalluri G., Henry G.H., Properties of area17/18 border neurons contributing to the visual transcallosalpathway in the cat, Vis. Neurosci. 5 (1990) 83-98.Meyer E., Müller C.O., Fromherz P., Cable properties ofdendrites in hippocampal neurons of the rat mapped by avoltage-sensitive dye, Eur. J. Neurosci. 9 (1997) 778-785.Milleret C., Buser P., Reorganization processes in the visualcortex also depend on visual experience in the adult cat, in:Hicks T.P., Molotchnikoff S., Ono T. (Eds.), Progress in BrainResearch, Vol. 95, The visually responsive neuron, Elsevier,Amsterdam, 1993, pp. 257-269.Milleret C., Houzel J.C., Rules for experience-dependentshaping of interhemispheric connections to visual corticalareas 17 and 18 in the cat during development, in: ColloqueMedecine et Recherche, ‘normal and abnormal developmentof the neocortex’, IPSEN, Paris, 1996.Milleret C., Houzel J.C., Buser P., Pattern of development ofthe callosal transfer of visual information to cortical areas 17and 18 in the cat, Eur. J. Neurosci. 6 (1994) 193-202.Milleret C., Watroba L., Buser P., What is the actual width ofthe ‘primary’ callosal visual field in cat?, Soc. Neurosci. Abstr.28 (1998) 254.12.Munk M.H.J., Nowak L.G., Nelson J.I., Bullier J., Structuralbasis of cortical synchronization. 2. Effects of cortical lesions,J. Neurophysiol. 74 (1995) 2401-2414.Nowak L.G., Bullier J., The timing of information transfer inthe visual system, Cereb. Cortex 12 (1997) 205-241.


[51] Nowak L.G.. Munk M.H.J., Nelson J.I., James A.C., Bullier J.,. _Structural basis of cortical synchronization. 1. Three types ofinterhemispheric coupling, J. Neurophysiol. 74 (1995)2379-2400.

[52] Olavarria J.E. , Abel PL. , The distribution of callosal connec-. 1

tions correlates with the pattern of cytochrome oxidase stripesin visual area V2 of Macaque monkeys, Cereb. Cortex 6(1996) 631439.

[53] Olavarria J.F., VanSluyters R.C., Overall pattern of callosalconnections in visual cortex of normal and enucleated cats, J.Comp. Neurol. 363 (1995) 161-176.

[54] Orihara Y.I., Kishikawa M., Ono K., The fates of the callosalneurons in neocortex after bisection of the corpus callosum,using the technique of retrograde neuronal labeling with twofluorescent dyes, Brain Res. 778 (1997) 393-396.

[55] Payne B.R., Neuronal interactions in cat visual cortex medi-ated by the corpus callosum, Behav. Brain Res. 64 (1994)55-64.

[56] Payne B.R., Siwek D.F., Lomber S.G., Complex transcallosalinteractions in visual cortex, Visual Neurosci. 6 (1991)283-289.

[57] Ramón y Cajal S., Estructura de1 quiasma óptico y teoriageneral de los entrecruzamientos nerviosos, Rev. Trim. Mi-crografica 3 (1898) 2-18.

[58] Ramón y Cajal S., Histologic du sytème nerveux de 1’Hommeet des vertebrés, Maloine, Paris, 1911.

[59] Rosa M.G., Fritsches K.A., Elston G.N., The second visualarea in the marmoset monkey: visuotopic organization, mag-nification factors, architectonical boundaries, and modularity,J. Comp. Neurol. 387 (1997) 547-567.

[60] Sanides D., Commissural connections of the visual cortex ofthe cat, in: Steele-Russel I., Van Hof M.W., Berlucchi G.(Eds.), Structure and function of the cerebral commisures,McMillan, London. 1979, pp. 236-243.

[61] Segev I., Schneidman E., The axon as a computationaldevice: basic insights gained from models and questions forthe near future, J. Physiol. (Paris) 93 (1999) 263-270.

[62] Segraves M.A., Innocenti G.M., Comparison of the distribu-tion of ipsilaterally and contralaterally projecting corticocor-tical neurons in cat visual cortex using two fluorescent tracers,J. Neurosci 5 (1985) 2107-2118.

[63]

[64]

[65]

[66]

[67]

[68]

[69]

[70]

[71]

[72]

[73]

[74]

[75]

Sereno M.I., Dale A.M., Reppas J.B., Kwong K.K., BelliveauJ.W., Brady T.J., Rosen B.R., Tootell R.B., Borders ofmultiple visual areas in humans revealed by functional mag-netic resonance imaging [see comments], Science 268 (1995)889-893.Singer W., Development and plasticity of cortical processingarchitectures, Science 270 (1995) 758-764.Singer W., Gray C.M., Visual feature integration and thetemporal correlation hypothesis, Annu. Rev. Neurosci. 18(1995) 555-586.Sperry R.W., Mental unity following disconnection of thecerebral hemispheres, Harvey Lecture 62 (1968) 293-323.Sun J.S.. Li B., Ma M.H.. Diao Y.C., Transcallosal circuitryrevealed by blocking and disinhibiting callosal input in thecat, Vis. Neurosci. 11 (1994) 189-197.Tettoni L., Lehmann P., Houzel J.C., Innocenti GM., Maxsim,software for the analysis of multiple axonal arbors and theirsimulated activation, J. Neurosci. Methods 67 (1996) 1-9.Tsodvks M.V.. Markram H.. Plasticitv of neocortical svnansesenables transitions between rate and temporal coding,-in: Vonder Malsburg C., Von Seelen W., Vorbrtiggen J.C., Sendhoff B.(Eds.), Artifial networks-ICANN, Springer Verlag, Bochum,1996, pp. 445-450.Tusa R.J., Palmer L.A., Rosenquist A.C., Multiple corticalvisual areas: Visual field topography in the cat, in: WoolseyC.N. (Ed.), Cortical sensory organization, Humana Press,Clifton, NJ, 1981, pp. 1-31.Voigt T., Levay S., Stamnes M.A., Morphological and immu-nocytochemical observations on the visual callosal projectionsin the cat, J. Comp. Neurol. 272 (1988) 450-460.Volgushev M., Chistiakova M., Singer W., Modification ofdischarge patterns of neocortical neurons by induced oscilla-tions of the membrane potential, Neuroscience 83 (1998)15-25.Waxman S.G., Bennett M.V.L., Relative conduction velocitiesof small myelinated and non-myelinated fibres in the centralnervous system, Nature 238 (1972) 217-219.Waxman S.G., Ritchie J.M., Molecular dissection of themyelinated axon, Ann. Neurol. 33 (1993) 121-136.Waxman S.G., Swadlow H.A., Ultrastructure of visual callosalaxons in the rabbit, Exp. Neurol. 53 (1976) 115-127.

Visual inter-hemispheric processing: Constraints and potentialities set by axonal morphology

Documents

Transcript of Visual inter-hemispheric processing: Constraints and potentialities set by axonal morphology