Resting discharge patterns of macular primary afferents in otoconia-deficient mice.

Vestibular primary afferents in the normal mammal are spontaneously active. The consensus hypothesis states that such discharge patterns are independent of stimulation and depend instead on excitation by vestibular hair cells due to background release of synaptic neurotransmitter. In the case of otoconial sensory receptors, it is difficult to test the independence of resting discharge from natural tonic stimulation by gravity. We examined this question by studying discharge patterns of single vestibular primary afferent neurons in the absence of gravity stimulation using two mutant strains of mice that lack otoconia (OTO-; head tilt, het-Nox3, and tilted, tlt-Otop1). Our findings demonstrated that macular primary afferent neurons exhibit robust resting discharge activity in OTO- mice. Spike interval coefficient of variation (CV = SD/mean spike interval) values reflected both regular and irregular discharge patterns in OTO- mice, and the range of values for rate-normalized CV was similar to mice and other mammals with intact otoconia although there were proportionately fewer irregular fibers. Mean discharge rates were slightly higher in otoconia-deficient strains even after accounting for proportionately fewer irregular fibers [OTO- = 75.4 +/- 31.1(113) vs OTO+ = 68.1 +/- 28.5(143) in sp/s]. These results confirm the hypothesis that resting activity in macular primary afferents occurs in the absence of ambient stimulation. The robust discharge rates are interesting in that they may reflect the presence of a functionally 'up-regulated' tonic excitatory process in the absence of natural sensory stimulation.

Age-related differences in parameters of vestibular evoked myogenic potentials.

CONCLUSIONS: The statistically significant correlations between vestibular evoked myogenic potential (VEMP) parameters and age may be due to hair cell loss of the otolith organ and/or to degenerative changes of the vestibular neural pathway. These findings indicate that age should be taken into account when interpreting VEMP results. It is also important to determine a standard method for performing VEMP and a universal index for comparison among laboratories. OBJECTIVES: VEMP, which measures the surface electric potential from the cervical muscle evoked by sufficiently loud sounds, is a useful tool to evaluate vestibule-colic reflex function. We have assayed the effect of age on VEMP results. SUBJECTS AND METHODS: After excluding subjects with a previous history of dizziness, middle ear pathology, or other inner ear symptoms, a total of 97 healthy volunteers (194 ears) were included. All VEMP parameters were analyzed to find differences related to side and gender, as well as the relationship between age and each VEMP parameter. RESULTS: Age was correlated with all VEMP parameters. Latency of p13, n23 showed a negative correlation and amplitude of p13-n23 showed a positive correlation with age. Differences between the right and left sides were not significant.

A place principle for vertigo.

OBJECTIVE: To provide a road map of the vestibular labyrinth and its innervation leading to a place principle for different forms of vertigo. METHOD: The literature describing the anatomy and physiology of the vestibular system was reviewed. RESULTS: Different forms of vertigo may be determined by the type of sense organ, type of ganglion cell and location in the vestibular nerve. CONCLUSION: Partial lesions (viral) of the vestibular ganglion are manifested as various forms of vertigo.

Regeneration of vestibular otolith afferents after ototoxic damage.

Regeneration of receptor cells and subsequent functional recovery after damage in the auditory and vestibular systems of many vertebrates is well known. Spontaneous regeneration of mammalian hair cells does not occur. However, recent approaches provide hope for similar restoration of hearing and balance in humans after loss. Newly regenerated hair cells receive afferent terminal contacts, yet nothing is known about how reinnervation progresses or whether regenerated afferents finally develop normal termination fields. We hypothesized that neural regeneration in the vestibular otolith system would recapitulate the topographic phenotype of afferent innervation so characteristic of normal development. We used an ototoxic agent to produce complete vestibular receptor cell loss and epithelial denervation, and then quantitatively examined afferent regeneration at discrete periods up to 1 year in otolith maculas. Here, we report that bouton, dimorph, and calyx afferents all regenerate slowly at different time epochs, through a progressive temporal sequence. Furthermore, our data suggest that both the hair cells and their innervating afferents transdifferentiate from an early form into more advanced forms during regeneration. Finally, we show that regeneration remarkably recapitulates the topographic organization of afferent macular innervation, comparable with that developed through normative morphogenesis. However, we also show that regenerated terminal morphologies were significantly less complex than normal fibers. Whether these structural fiber changes lead to alterations in afferent responsiveness is unknown. If true, adaptive plasticity in the central neural processing of motion information would be necessitated, because it is known that many vestibular-related behaviors fully recover during regeneration.

Organization of the efferent vestibular nuclei and nerves of the toadfish, Opsanus tau.

The efferent vestibular nuclei and nerves were studied in the toadfish, Opsanus tau, with morphological and electrophysiological techniques. The origin and course of the efferent vestibular nerves was extensively documented. One major morphological observation was that the efferent nerves comprise a peripheral network that is anatomically distinct, and separable by dissection from the primary afferents innervated by each end organ. These anatomically distinct nerves are likely to be a major asset in physiological studies of efferent vestibular function. The retrograde transport of horseradish peroxidase (HRP) from each of the nerves innervating the vestibular and lateral line organs was used to delineate the subgroups of efferent neurons projecting to these end organs. The efferent vestibular nuclei are located in the posterior medulla in and around the median longitudinal fasciculi (MLF). We divided the nuclei cytoarchitecturally into lateral, medial, and dorsal subdivisions. The lateral cells had bilateral dendritic trees while the dorsal cells had ipsilateral, unilateral dendritic trees. There was a higher proportion of lateral cells that innervated the canal organs and the utricle while the dorsal cells tended to innervate the other organs. The total number of cells obtained by summing those from separate nerve label was twice the total cell count present in the nuclei. Indirectly, this indicates that some cells project to more than one end organ. Efferent neurons were penetrated with glass microelectrodes, and their end organs and patterns of connectivity with other end organs were investigated by stimulating various vestibular nerves. Posterior semicircular canal efferent cells are electrically coupled to each other and could be activated electrically or chemically by stimulating other ipsilateral or contralateral vestibular nerves. It is suggested that electrical coupling might be responsible for the uniform behavior of these cells under certain conditions. Morphological and physiological experiments suggested that the semicircular canals are innervated by their own, exclusive populations of efferent neurons while other end organs may share efferent innervation. Single cells were injected intracellularly with HRP and their morphology was studied and characterized by light microscopy. Intracellular label confirmed the morphological features demonstrated by retrograde transport of HRP and also revealed that some cells had central axon collaterals that terminated within the MLF. These morphological and physiological results provide a basis for understanding the behavior of efferent vestibular neurons in the alert animal.

Comparative scanning electron microscopic investigations of the sensory epithelia in the teleost sacculus and lagena.

Scanning electron microscopic studies were conducted on the sensory epithelia of the auditory portions of the ears in teleost species representing wide taxonomic diversity. A number of the features of the ears investigated resembled features found in other teleost species, although some major exceptions to earlier patterns were found, particularly in the saccular sensory epithelium. The saccular maculae of all but one species contained basically similar ciliary bundles on the sensory hair cells while there was some significant variation on the lagenar maculae. Hair cell orientation patterns on the sacculus contained four orientation groups in all of the species, other than the mormyrid, Gnathonemus, which only had two groups. Lagenar maculae had two orientation groups, and the orientation patterns were similar to one another. The most divergent form of lagenar macula was found in gnathonemus. These data, combined with data from earlier investigations, provide a broad overview of the surface features of the ear in teleost fishes. Most significantly, it now appears that there are at least five different saccular hair cell orientation patterns among teleost fishes, and all of these patterns are found spread through many major teleost taxa. While there is some similarity in ear structures among some groups of closely related species, such as the Elopomorpha and the Gadiformes, it is becoming more apparent that there is extensive convergence in a number of features of he teleost ear that most likely reflect similar selective pressures during the evolution of the ear. The nature of these selective pressures, however, are not well understood.

Analysis of temporal and spatial patterns of rat vestibular hair cell differentiation by tritiated thymidine radioautography.

The location in time and space of the terminal mitoses of type I and type II sensory hair cells (HCI and HCII) of the developing crista ampullaris in rat lateral semicircular canal and macula utriculi was determined by radioautographs of specimens exposed to tritiated thymidine from the 13th to the 20th day of gestation. Qualitative analysis and statistical treatment of the percentages of labeled HCI and HCII show that the terminal mitoses occur first in the macula utriculi with a maximum percentage of the 14th day of gestation, for the HCI, and on the 15th day of gestation, for the HCII. In the lateral crista, the maximum percentage of labeled HCI occurs on the 17th and 18th day of gestation and on the 19th day of gestation for the HCII. A spatial distribution of this labeling activity is also described: the older cells are located at the top of the crista and at the level of the striola of the macula utriculi while the younger cells are found at the bottom of the crista and on the sides of the utricle. A study of the vestibular receptors in the fetuses shows that synaptic contacts already exist on the 18th day of gestation in the macula utriculi at the level of the striola and on the 19th day at the top of the crista; the cells situated on the periphery are still immature. The first hair cells to undergo their terminal mitoses are, therefore, connected first. These results also suggest that the two types of cells are genetically programmed and that the HCI start functioning first during the development of the labyrinth.

Central projections of the utricular nerve in the gerbil.

The central projections of primary afferent fibers in the utricular nerve, which convey linear head acceleration signals to neurons in the brainstem and cerebellum, are not completely defined. The purpose of this investigation was twofold: 1) to define the central projections of the gerbil utricular afferents by injecting horseradish peroxidase (HRP) and biotinylated dextran amine (BDA) into the utricular macula; and 2) to investigate the projections of individual utricular afferents by injecting HRP intracellularly into functionally identified utricular neurons. We found that utricular afferents in the gerbil projected to all divisions of the vestibular nuclear complex, except the dorsal lateral vestibular nucleus. In addition, terminals were observed in the interstitial nucleus of the eighth nerve, nucleus Y, external cuneate nucleus, and lobules I, IV, V, IX, and X of the cerebellar vermis. No projections appeared in the flocculus or paraflocculus. Fibers traversed the medial and intermediate cerebellar nuclei, but terminals appeared only occasionally. Individual utricular afferents collateralize extensively, projecting to much of the brainstem area innervated by the whole of the utricular nerve. This study did not produce complete filling of individual afferent collateral projections into the cerebellar cortex.

Computer-assisted three-dimensional reconstruction and simulations of vestibular macular neural connectivities.

The macular neuroepithelium is morphologically organized as a weighted neural network for parallel distributed processing of information. The network is continuous across the striola, where some type II hair cells synapse with calyces containing type I cells with tufts of opposite directional polarities. Whether other hair cell to calyx appositions that lack synapses interact because of intercellular potassium accumulation remains an open question. A functionally important inference of macular organization is that just as arrays of hair cells communicate an entire piece of information to a nerve fiber, so do macular subarrays of nerve fibers (not single units) carry the whole coded message to the brain stem. Moreover, the size of the network subarray can expand or become more limited depending upon the strength and/or duration of the input. It is the functioning of the network and its subarrays that must be understood if we are to learn how maculas carry out their work and adapt to new environments. Simulations of functioning maculas, or subparts, based on precise morphology and known physiology are useful tools to gain insights into macular information processing. The current simulations of afferent collateral electrical activity are a prelude to development of a 3-D model. The simulations demonstrate a relationship between geometry and function, with the diameter of the stem apparently being a major determinant of electrical activity transmitted to the base in the case of collaterals with short stems. Thus, while changes in synaptic number and/or size may be an important adaptive mechanism in an altered g environment, changes in diameter of the stem is another means of altering outflow. Research on the effects of microgravity should be extremely useful in examining the validity of this and other concepts of neural adaptation, since maculas are biological linear accelerometers ideally suited to the task. Maculas are also extremely interesting to study in detail because of the richness of connectivities and submicroscopic organization they present. Many of their features are common with more complex parts of the brain. It seems possible that knowledge of the three-dimensional geometric relationships operative in a functioning macula will contribute much to the understanding of the dynamics underlying more complex behavior. Computerized approaches greatly facilitate this task and provide an objective method of analysis. It is likely that, in the end, simple rules will be found to govern optimal neural architectural organization, even at higher cognitive levels. The architecture only appears complex because we do not yet grasp its meaning.(ABSTRACT TRUNCATED AT 400 WORDS)

Evidence of a sensory processing unit in the mammalian macula.

We cut serial sections through the medial part of the rat vestibular macula for transmission electron microscopic (TEM) examination, computer-assisted 3-D reconstruction, and compartmental modeling. The ultrastructural research showed that many primary vestibular neurons have an unmyelinated segment, often branched, that extends between the heminode (putative site of the spike initiation zone) and the expanded terminal(s) (calyx, calyces). These segments, termed the neuron branches, and the calyces frequently have spine-like processes of various dimensions with bouton endings that morphologically are afferent, efferent, or reciprocal to other macular neural elements. The major questions posed by this study were whether small details of morphology, such as the size and location of neuronal processes or synapses, could influence the output of a vestibular afferent, and whether a knowledge of morphological details could guide the selection of values for simulation parameters. The conclusions from our simulations are (1) values of 5.0 k omega cm2 for membrane resistivity and 1.0 nS for synaptic conductance yield simulations that best match published physiological results; (2) process morphology has little effect on orthodromic spread of depolarization from the head (bouton) to the spike initiation zone (SIZ); (3) process morphology has no effect on antidromic spread of depolarization to the process head; (4) synapses do not sum linearly; (5) synapses are electrically close to the SIZ; and (6) all whole-cell simulations should be run with an active SIZ.

Responses of vestibulospinal neurons to neck and macular vestibular inputs in the presence or absence of the paleocerebellum.

1. The role of the paleocerebellum in determining the responses of lateral vestibular nucleus (LVN) neurons either to independent or combined stimulation of macular vestibular and neck receptors has been investigated in decerebrate cats. Sinusoidal rotation around the longitudinal axis at 0.026 Hz, 5-10 degrees, represented the constant input parameters. Among the tested neurons, 100 and 131 units were recorded in animals with intact cerebellum and following partial cerebellectomy, respectively. The units were classified according to their anatomical location in either the rostroventral (cLVN) or dorsocaudal (ILVN) part of the LVN; units also were activated antidromically from the spinal cord. 2. The majority of units responded to stimulation of macular receptors both in preparations with intact cerebellum (75.0%) or with partial cerebellectomy (71.8%), the response being primarily in phase with the direction of animal orientation. The proportion of responding units and their response sensitivity were greater in the cLVN than ILVN in each preparation; no significant differences in mean firing rate and response sensitivity were observed between the two preparations for each subdivision of the LVN. In animals with cerebellum intact, the majority of cLVN and ILVN units were excited during side-down tilt; following partial cerebellectomy, this predominant response pattern still was present in cLVN but was reversed in ILVN. 3. About one-half of the units responded to sinusoidal stimulation of neck receptors in both preparations, the response being mainly in phase with the direction of neck orientation. In the intact cerebellum preparations, the proportion of cLVN units responding to neck rotation was greater than that of ILVN units, but no difference in response sensitivity was observed between these units. Following partial cerebellectomy, the proportion of cLVN units responsive to the neck input was reduced but that of ILVN units was not; however, the average response sensitivity was halved for both cLVN and ILVN units. In preparations with cerebellum intact, most of the cLVN units were excited during side-down neck rotation, whereas ILVN units were excited mainly by rotation in the opposite sense; following partial cerebellectomy, the majority of units were excited during side-up neck rotation, not only in ILVN but also in cLVN. 4. Units receiving a convergent input from both receptors were more numerous in cLVN (72.7%) than ILVN (41.8%) in preparations with intact cerebellum; following partial cerebellectomy, this disproportion of responsive units in the two divisions (45.3% and 44.9%, respectively) disappeared, due to a reduced number of cLVN units responding to the neck input. In both preparations, the macular input had a relatively greater influence on the cLVN whereas the neck input was more effective on the ILVN. 5...

Correspondences between afferent innervation patterns and response dynamics in the bullfrog utricle and lagena.

Otoconial afferents in the bullfrog were characterized as gravity or vibratory sensitive by their resting activity and their responses to head tilt and vibration. The responses of gravity afferents to head tilt were tonic, phasic-tonic, or phasic. A few afferents, termed vibratory/gravity afferents, had gravity as well as vibratory sensitivity. Functionally identified otoconial afferents were injected with Lucifer Yellow and subsequently traced to their peripheral arborizations. Morphological maps, previously constructed with the scanning electron microscope, were used to identify microstructural features of the sensory maculae associated with the peripheral arborizations of dye-filled afferents. The utricular and lagenar macula each is composed of a specialized central band surrounded by a peripheral field. The central bands are composed of densely packed medial rows and more sparsely packed lateral rows of hair cells. Hair cells exhibit a variety of surface topographies which correspond with their macular location. The response dynamics of afferents in the utricle and lagena correspond with the macular locations of their peripheral arborizations. Tonic afferents were traced to hair cells in the peripheral field. Phasic-tonic and phasic afferents innervated hair cells in the lateral rows of the central band, the former innervating hair cells at the edges of the central band and the latter innervating hair cells located more medially. Afferents with vibratory sensitivity were traced to hair cells in the medial rows of the lagenar central band. The response dynamics of afferents corresponded with the surface topography of their innervated hair cells. Tonic and phasic-tonic gravity afferents innervated hair cells with stereociliary arrays markedly shorter than their kinocilium (Lewis and Li types B and C) while phasic gravity and vibratory afferents innervated hair cells with stereociliary arrays nearly equal to their kinocilium (Lewis and Li types E and F). Vibratory sensitivity was uniquely associated with hair cells possessing bulbed kinocilium (Lewis and Li type E) while afferents sensitive to both gravity and vibration innervated hair cells from both of the above groups. We argue that afferent response dynamics are determined, at least in part, at the level of the sensory hair bundle and that morphological variations of the kinocilium and the otoconial membrane are dictated by specialization of sensitivity. We propose that morphological variations of the kinocilium reflect variations in its viscoelastic properties and that these properties determine the nature of the mechanical couple between the stereociliary array and the otoconial membrane.

Central projections of vestibular afferents innervating the macula of the saccule in gerbil.

The central distribution of vestibular afferents that innervate the saccule has been investigated in the gerbil using transganglionic transport techniques. Following horseradish peroxidase injection into the saccular neuroepithelium, labeled ganglion cells were clustered at the junction of the superior and inferior ganglion. Labeled fibers entered the vestibular nuclear complex and divided into rostral and caudal branches. Terminal fields were observed in the interstitial nucleus of the vestibular nerve among the entering fibers and in the lateral vestibular nucleus. Rostrally, fibers terminated in cell group y and the nodulus; caudally, fibers ended in the descending and medial vestibular nuclei.

The distinctive central utricular projections in the herring.

The utriculus of the inner ear of clupeid fishes comprises three maculae, each separately innervated. The three nerves were labelled with horseradish peroxidase in the herring and were found to project ipsilaterally to the cerebellum and octavus area and bilaterally to the lateralis zone of the brainstem. The anterior macular nerve terminates dorsally in the octavus area and exclusively in a rostral portion of the anterior octavus nucleus, whereas the middle and posterior macular nerves end ventrally in the octavus area and in the reticular formation. Large middle macular fibers synapse on the Mauthner-cell lateral dendrite.

Innervation patterns and spontaneous activity of afferent fibres to the lagenar macula and apical basilar papilla of the chick's cochlea.

To investigate the origin of non-auditory fibres in the apical area of the avian cochlear ganglion, we recorded from nerve fibres in the young chick (87% of animals were aged between 5 and 10 days post-hatching). After characterization of their spontaneous activity patterns and, if present, their responses to sound, some fibres were stained with cobalt-ion injections and traced to their peripheral terminals. All stained fibres which were traced to the lagenar macula (N = 13) were non-auditory. They did not increase firing rate or phase-couple to sound stimuli. Their spontaneous activity was either regular (12 cases) or irregular (1 case). Regularly-firing cells all innervated several to very many hair cells, whereby there was no great difference in the pattern of spontaneous activity between those making calyx endings on relatively few hair cells in the striola region and those making small bouton endings on up to 80 hair cells outside the striola. All fibres that responded in any way to sound were irregularly spontaneously active. Three fibres, two of which only responded to sound with phase-coupling, innervated several hair cells in the apical, abneural region of the basilar papilla. Two other fibres traced to the basilar papilla are of previously undescribed types.

Ultrastructural localization of ChAT-like immunoreactivity in the human vestibular periphery.

Acetylcholine (ACh) has long been considered a neurotransmitter candidate in the efferent vestibular system of mammals. Recently, choline acetyltransferase (ChAT), the synthesizing enzyme for ACh, was immunocytochemically localized in all five end-organs of the rat vestibule (Kong et al. (1994) Hear. Res. 75, 192-200). However, there is little information in the literature concerning the cholinergic innervation in the vestibular periphery of man. In the present study the ultrastructural localization of the ChAT-like immunoreactivity in the human vestibular periphery was investigated in order to reveal the cholinergic innervation in the human vestibular end-organs. A modified method of pre-embedding immunoelectron microscopy was applied. It was found that the ChAT-like immunoreactivity was located in the bouton-type vesiculated nerve terminals in the vestibular neurosensory epithelia of man. These ChAT-like immunostained nerve terminals make synaptic contacts either with afferent chalices surrounding type I vestibular sensory hair cells, or with type II vestibular sensory hair cells. These results show that the ChAT-like immunoreactivity in the human vestibular periphery is confined to the efferent vestibular system. The ChAT-containing efferents innervate both type I hair cells and type II hair cells, making postsynaptic and presynaptic contacts, respectively. This study presents evidence that ACh is a neurotransmitter candidate in the efferent vestibular system of man.

Ultrastructural localization of GABA-like immunoreactivity in the human utricular macula.

In the vertebrate vestibular periphery, gamma-aminobutyric acid (GABA) has long been presumed to be a neurotransmitter candidate. However, experimental reports about the localization and function of GABA in the vestibular systems of vertebrates are contradictory. In addition, there is no information in the literature concerning the localization of GABA in the human vestibular periphery. The present study investigates the ultrastructural localization of GABA-like immunoreactivity in the human utricular macula. A modified pre-embedding immunostaining electron microscopy technique was applied using two different commercially available polyclonal antibodies to GABA. GABA-like immunoreactivity is confined to the vesiculated nerve fibers and terminals of the human vestibular neurosensory epithelia. The GABA-containing nerve terminals make asymmetrical axo-dendritic synapses with the afferent chalices surrounding the type I sensory hair cells. Type I and type II hair cells as well as afferent chalices are devoid of GABA-like immunoreactive staining. The present study demonstrates that GABA exists in the human vestibular periphery, and that GABA is a neurotransmitter candidate of the human efferent vestibular system.

Neuroanatomical correlates of vestibular function.

Neuroanatomical data revealed by new neuroanatomical techniques over the past decade, have clarified the central connections of the peripheral vestibular sense organs. A knowledge of this information is important in the clinical evaluation of vestibular disorders. The reflex connections of the semicircular canals serve vestibulocerebellar, vestibulo-ocular, commisural, and vestibulospinal pathways. The central connections of the utricular macula are primarily in a descending direction to the motor neurons of the entire length of the spinal cord. These form the basif for vestibulospinal reflexes which are characteristically associated with utricular function. Additional central connections provide the basis for vestibulo-ocular responses to utricular activity. Finally, the saccule has reflex connections both to the upper spinal cord to a minor degree and to extraocular nuclei that serve vertical and oblique eye movements. The minor vestibular group Y nucleus provides a commissural reflex connection for the saccule.

Ultrastructural and cytochemical evidence for single impulse initiation zones in vestibular macular nerve fibers of rat.

Cupric ion-ferricyanide labeling methods and related ferrocyanide-stained tissues were used to locate and characterize, at the ultrastructural level, presumptive impulse initiation zones in the three types of vestibular macular nerve fibers. Large-diameter, M-type vestibular nerve fibers terminate in a calyx at the heminode, and labeling is coextensive with the base of the calyx. Intermediate, M/U-type nerve fibers have short, unmyelinated preterminal segments that sometimes bifurcate intramacularly, and small-diameter, U-type nerve fibers have long, unmyelinated preterminal axons and up to three branches. Preterminals of these nerve fibers display ultrastructural heterogeneity that is correlated with labeling patterns for sodium channels and/or associated polyanionic sites. They have a nodelike ultrastructure and label heavily from near the heminode to the base of the macula. Their intramacular branches, less organized ultrastructurally, label only slightly. Results indicate that vestibular nerve fibers have one impulse initiation zone, located near the heminode, that varies in length according to nerve fiber type. Structural heterogeneity may favor impulse conduction in the central direction, and length of the impulse initiation zone could influence nerve discharge patterns.

3-D components of a biological neural network visualized in computer generated imagery. II. Macular neural network organization.

Computer-assisted reconstructions of small parts of the macular neural network show how the nerve terminals and receptive fields are organized in 3-dimensional space. This biological neural network is anatomically organized for parallel distributed processing of information. Processing appears to be more complex than in computer-based neural networks, because spatiotemporal factors figure into synaptic weighting. Serial reconstruction data show anatomical arrangements which suggest that 1) assemblies of cells analyse and distribute information with inbuilt redundancy, to improve reliability; 2) feedforward/feedback loops provide the capacity for presynaptic modulation of output during processing; 3) constrained randomness in connectivities contributes to adaptability; and 4) local variations in network complexity permit differing analyses of incoming signals to take place simultaneously. The last inference suggests that there may be segregation of information flow to central stations subserving particular functions.