John Zachary Young

The retina of cephalopods and its degeneration after optic nerve section

Each retinal cell of Octopus carries a rhabdomere on two opposite faces. Rhabdomeres from four cells combine to make a square rhabdome. The cells are mainly arranged with their axes in approximately either the vertical or horizontal plane as the eye is usually held in the head. Counts show that there are about twice as many retinal cell nuclei as there are rhabdomes. There are altogether about 2 x 107 retinal cells in each eye, with a density of about 50 000/mm2. The retinal cells at the centre of the retina are longer and thinner than those at the periphery. There is a strip of longer, thinner cells running horizontally along the equator. These often have less pigment in their distal ends than do the cells dorsally and ventrally, but other distributions of the pigment are seen, depending on the previous illumination. There are several types and sizes of retinal cell and not all are associated in fours to make rhabdomes. The proximal segments carry fine collateral twigs, these interdigitate and may allow mutual interaction between neighbours. The main meshes of the retinal plexus are not formed by fibres of the retinal cells but by the axons of cells in the optic lobes, presumably efferents. After severing the optic nerves to any region of the retina all the retinal cells undergo retrograde degeneration, leaving only the supporting cells intact. The retinal nerve plexus disappears almost completely, but a few fibres remain. At the boundary between a region with severed and intact nerves the plexus continues for some distance into the denervated region. After removal of all the optic lobe except a portion of its outermost (plexiform) zone the retinal receptors do not degenerate completely but are reduced in length. Their axons have not been interrupted by the operation and this is therefore a partial transneuronal retrograde degeneration.

The Statocysts of Octopus vulgaris

The statocyst of octopods is a sack containing endolymph, lying in a larger cavity, containing perilymph and crossed by strands containing blood vessels. The arrangement shows remarkable parallels with vertebrates. The crista runs a course in three planes, each section being innervated by a separate nerve. It thus presumably serves as a receptor resolving angular accelerations. Each of the parts is further divided into three subsections, with its own flap of massed, loaded hairs. In front of the vertical crista is a rigid plate of cartilage, the anticrista, perhaps serving to prevent stimulation of that part of the ridge when the octopus suddenly accelerates during an attack. The crista contains large hair cells at its centre, there being alternately sections with one row and two rows of these. Flanking the large hair cells are inner and outer rows of smaller hair cells. These show characteristic differences and are asymmetrical above and below. All the hair cells are held to be neurosensory cells, carrying axons. In addition there are present multipolar cells, which besides carrying hairs also have many dendrites, making contact with other hair cells. Some of these multipolar cells probably lack hairs and thus form true peripheral neurons, stimulated by the overlying hair cells. The latter still, however, continue to be primary receptor cells with their own axons. If this is so there are two channels from these hair cells to the C. N. S. Each fires through its own axon and several together fire through each multipolar cell. A third set of nerve fibres is present in the crista, having cell bodies within the C. N. S. and endings around the apical parts of the hair cells and around the nerve cells. These are probably efferent fibres. The macula is placed vertically on the antero-medial wall, with a statolith hanging on it. It contains hair cells with short hairs embedded in the statolith. Multipolar cells lie below the hair cells and have axons but probably no hairs. There is a plexus of presumably efferent fibres. The membranous wall of the statocyst contains muscle fibres, a plexus of nerve fibres, and receptor cells with hairs and (sometimes) dendritic processes. These may be pressure receptors recording changes of pressure within the mantle and body cavity. The outer wall of the statocyst, though mainly cartilaginous, has a membranous region. This plexus and its receptors are especially well developed in a protruding posterior sack. Kölliker’s canal lies in the wall of the statocyst and opens into its cavity, the other end being a closed tube. Its cells are ciliated. It may serve to regulate the endolymphatic pressure by secretion or absorption, assisted by the action of the muscle fibres and perhaps by the receptors of the wall of the sack.

A memory system in Octopus vulgaris Lamarck.

An octopus that has attacked a crab shown with a square and received a shock rapidly learns not to attack when this situation appears again, while continuing to attack crabs shown alone. The memory preventing attack on crabs shown with a white square may last for 2 or 3 days if the crab and square are not shown during that period. If the situation is shown three times a day the memory may last for 6 days or longer. The memory is not erased by anaesthesia nor by electrical stimulation of the supra-oesophageal lobes. After complete removal of the vertical lobe, or of the medial superior frontal lobe, or section of the tract between the two, the memory preventing attack is lost and cannot again be acquired. Animals operated in this way attack a crab and square if shown at 2-hourly intervals in spite of the numerous shocks they receive. A transitory memory lasting a few minutes can still be set up if the frequency of presentation is increased to about once every 5 min. Partial removal of the vertical lobe system does not interrupt the memory. A memory set up by the use of one eye is not abolished if the optic lobe of that side is later removed. The memory is not interrupted by slashes in both optic lobes. After lesions to the lateral parts of the superior frontal lobes an octopus makes few or no further attacks on crabs, unless these are placed close to the animal. The effect of such an operation is to upset the balance of central neural activities in such a way that a region responsible for inhibiting attacks on distant objects assumes control. This inhibitory region may be the first subvertical lobe, whose action is normally balanced by the lateral superior frontal lobes and the vertical lobe. The tangle of fibre bundles within the optic lobes allows for a wide degree of interaction between impulses arriving from different parts of the retinal surface. In addition, these lobes receive afferent fibres from the arms. They thus provide a system within which associations between given sets of inputs can be set up in such a way as to ensure that there is no attack when a similar set of inputs occurs again. Further plexiform arrangements are found in the pathway from the optic to the superior frontal lobes and from the latter to the vertical lobe. These plexuses make possible the interaction in each succeeding lobe of impulses arriving from distant parts of the preceding lobe. Each lobe can thus serve to record the pattern of associations present in the previous one. Since the arrangement is circular the pattern originating in the optic lobe is then re-presented back to it. It is suggested that the vertical lobe system serves to prolong memories set up in the optic lobes by re-presenting them from within, and thus allowing them to persist for long enough to produce some change of a more permanent nature.

The Optic Lobes of Octopus vulgaris

The optic lobes provide a system for coding the visual input, for storing a record of it and for decoding to produce particular motor responses. There are at least three types of optic nerve fibre, ending at different depths in the layered dendritic systems of the plexiform zone. Here the optic nerve fibres meet the branches of at least four types of cell. (1) Centripetal cells passing excitation inwards. The dendrites of these are very long, with fields orientated more often in horizontal and vertical than in other directions. (2) Numerous amacrine cells, with cone-shaped dendritic fields but no determinable axon. (3) Centrifugal cells conducting back to the retina. (4) Commissural fibres from the opposite optic lobe, and other afferents. After section of the optic nerves the plexiform layer of the corresponding part of the optic lobe becomes reduced, but the tangential layers of dendrites remain. There is a reduction in the thickness of the layers of amacrine and other cells and a shrinkage of the whole lobe. Conversely the tangential layers can be degenerated, leaving the optic nerve fibres, by severing the arteries to the optic lobe. The centre of the optic lobe contains cells with spreading dendritic trees of many forms. Some run mainly tangentially, others are radial cones. Those towards the centre send axons to the optic tract. Small multipolar cells accompany the large neurons of the cell islands. About 2 x 107 optic nerve fibres visible with the light microscope enter the lobes but only 0-5 x 106, or less, leave in the optic tract, these being distributed to some ten centres in the supraoesophageal lobes. It is suggested that the variety of shapes of the dendritic trees within the optic lobes provides the elements of the coding system by which visual input is classified.

The Central Nervous System of Nautilus

The central nervous system of Nautilus shows greater similarity to that of coleoid cephalopods than appears at first sight. In the area where the three main cords of the nervous system meet there is a region comparable in position to the magnocellular lobe of coleoids, and it contains large cells. It receives some static nerve fibres and is the origin of the nerves of the ocular tentacles. The anterior suboesophageal cord is not a single entity. The brachial nerves and nerves of the hood arise from its anterior part, which is directly continuous with the cerebral cord. The funnel nerves arise from a distinct part, continuous with the magnocellular and palliovisceral regions. If the tentacles are innervated from a region derived from the cerebral cord then they cannot be closely compared with the foot of other molluscs. The cerebral cord shows no clear internal division into lobes, but it is nevertheless organized on a plan recognizably like that of coleoids. Its anterior portion contains large cells and gives rise to the connectives that control the buccal mass. It receives the labial nerves and probably gustatory fibres. In the hinder part of the cerebral cord four regions are recognized. An outer dorsal plexiform zone receives afferents from many sources and perhaps serves to allow responses to combinations of inputs. It is especially developed as lateral cerebral lobes at the entry of the brachial nerve fibres. This zone may be compared with the inferior and superior frontal lobes of Octopus. Fibres pass from the plexiform zone through a layer of small cells to a laminated zone of specialized neuropil. This region corresponds approximately to the vertical lobe of coleoids, but the similarity is not very great. The centre of the cerebral cord contains larger cells, probably providing the output channels to other centres. The ventral portion contains commissural bundles. The olfactory lobes are relatively larger and the optic lobes smaller than in coleoids. Both are lateral continuations of the cerebral cord and have the same basic structure as the latter. The optic nerve fibres do not form a chiasma between the retina and the optic lobe. The optic lobe shows a general similarity to that of coleoids but there is no external granular layer and no peduncle lobe. There is no distinct optic gland but cells that perhaps represent optic gland tissue occur between the optic and cerebral lobes. The statocyst is a simple sack with no signs of macula or crista. Its duct remains open in the adult. The static nerve fibres run partly to the magnocellular lobe, partly to the cerebral cord. The plan of the cerebral cord of Nautilus thus appears as a general sketch of the system that exists in coleoids. The ‘higher’ centres for producing responses from combinations of inputs and perhaps for memory storage are only beginning to emerge from an undivided centre for the reflex control of the operations of feeding. The fact that Nautilus has remained macrosmatic and has poor vision may be connected with the relative simplicity of its higher centres. Nevertheless, its nervous system contains vastly more channels and complex parts than are found in any non-cephalopod mollusc.

The Buccal Nervous System of Octopus

The operations of killing and eating food by an octopus are under the control of a series of nervous centres. The poison centre lies most posteriorly and is probably activated first, since it lies close to endings of fibres from the arms. The fibres of the nerves to the posterior salivary gland run without synapse from the superior buccal lobe to the glands, passing first far forward and then back along the duct. There is thus no peripheral synapse on this path, perhaps because no continuing rhythmic operations are involved in the secretion, and no reflex guidance is needed. The actual injection of the poison by the salivary papilla is controlled through the subradular ganglia. The cerebro-subradular connectives arise from the front of the superior buccal ganglia, near the entrance of the labial nerves, and run direct to the subradular ganglia, bypassing the inferior buccal ganglion. The interbuccal connectives also arise from the front of the superior buccal lobe and run to the inferior buccal ganglion. The inferior buccal ganglion sends nerves to the muscles of the jaws and radula and to the anterior salivary glands, buccal palps and oesophagus. Through the sympathetic nerve it communicates with the gastric ganglion. The inferior buccal ganglion has a complicated internal structure. From its outer surface arise numerous strands of the juxtaganglionic tissue, which end at the surfaces of the buccal sinus. The proportion of large cells decreases in the sequence posterior buccal, superior buccal, inferior buccal, subradular and gastric ganglia.

The Croonian Lecture, 1965: The Organization of a Memory System

The study of memory is unfortunately a difficult and confused subject. Its importance is beyond question, if only because memory lies near the centre of human abilities. Yet there is little agreement even as to the practical results, if any, that may be expected from its study. Education and psychology proceed empirically for want of a general theory of memory, but it is not clear whether such a theory may be expected to come from psychology as studied in man, from experimental study of animals, or from some discoveries of logic, mathematics or engineering. Yet the subject has certainly not been neglected. Immense efforts have been made by psychologists, clinicians, physiologists and workers in other disciplines. In recent years occurrences of various sorts have been reported in the nervous system during learning, from electrical changes, to changes in the base composition of the ribosenucleotides of single nerve cells. In fact there is such a large literature that one cannot avoid feeling that anything that an anatomist may say on the subject will be irrelevant, superficial, naive or worst of all, confusing. I am encouraged to take the risk by the fact that study of the connexion pattern of the nervous system of the octopus has given me a feeling of beginning to understand a little about the subject of memory. To study the material organization behind any subject or problem is surely the basis of a scientific approach. I shall go so far as to suggest evidence for the existence of a unit of memory or mnemon. The suggestion is made with great hesitation and in full awareness of its dangers. The technique of pushing the analysis of the system as far as possible on the basis of the connexion pattern seems to have brought increasing clarification. Recognition of the units is the next further step. Many of those who have so ably assisted in this work will probably not approve of the identification of a unit, still less of inventing a name for it. Many will regard it as a guess, especially since it is not supported by evidence from microelectrodes. Yet perhaps we need to try to identify units, not only as a basis for discussion but in order to find out where to look in our search for the electrical, chemical and other correlates of memory.

RESPONSES OF MUSCLES OF THE SQUID TO REPETITIVE STIMULATION OF THE GIANT NERVE FIBERS

With increasing frequency of stimulation of a giant nerve fiber in the squid, Loligo pealii, the only increase in the tension developed by the circular muscle fibers of the mantle is a small amount (5 to 10 per cent) over the range of incomplete relaxation. The absence of any increased response at higher frequencies shows that in the fresh muscle a single nerve impulse is capable of activating every muscle fiber which it reaches. However, the isolated muscle very readily becomes fatigued when stimulated at high frequency and thereafter greater tension is produced at the higher rates. In the normal animal there would be no use for peripheral facilitation and each contraction of the mantle is produced as an all-or-nothing response.

The Centres for Touch Discrimination in Octopus

The inferior frontal system, concerned with learning chemotactile discriminations, shows four distinct regions. The posterior buccal lobes contain both large and small cells and are the centre of the system. They receive fibres from the arms (without interweaving), from the lips, and from the buccal mass. They send fibres downwards to the arm centres and backwards to the optic and superior frontal/vertical systems. This is therefore probably both a reflex centre for response to some simple chemotactile stimuli and also the main output pathway for the whole system. In the lateral inferior frontal lobes the fibres from the arms interweave and mix with those from other sources. Their efferent fibres pass to the posterior buccal lobes and to the same destinations as the efferent fibres of the posterior buccal lobes. The organization of the lateral inferior frontal thus allows responses to combinations of chemotactile inputs. The median inferior frontal lobe receives the same input as the lateral inferior frontals, and its interweaving bundles allow for further spreading and combination between afferents. Its efferent axons pass only to the subfrontal lobes. The subfrontal lobes, besides the input from the median inferior frontal lobe, receive fibres from below. Their cells are mostly very small, with axons ending within the lobe A few larger cells with axons running to the posterior buccal lobes carry the output. The tactile system is thus essentially similar to the visual one, with a pair of lower centres (posterior buccal and lateral inferior frontal) and a pair of upper ones (median inferior frontal and subfrontal) . Embryologically these all differentiate from a single lobe, and the small cells of the upper lobes form a continuous layer with the relatively fewer small cells of the lower lobes. The main difference between the visual and tactile systems is the absence from the latter of a differentiated region corresponding to the optic lobe. From the evidence of Wells the change that constitutes a memory record occurs in the region of the posterior buccal lobe that contains both large and small cells. This region is under the influence of the circuit through lateral and median inferior frontal and subfrontal lobes.

The diameters of the fibres of the peripheral nerves of Octopus

The diameters and numbers of fibres have been measured throughout the peripheral nervous system. The nerves of the muscles that act upon the outside world contain few fibres, having very large and medium-sized fibres but no very small ones. Thus the muscles of the head receive 6000 fibres, the largest of 30 μm diameter. The eye muscles receive 3300 fibres, reaching 24 μm. The stellar nerve fibres are more numerous (150 000), but smaller (< 20μm). The preganglionic fibres joining the c. n. s. to the stellate ganglia are fewer than the postganglionic ones (4000, < 16 μm). In some of the somatic motor nerves the longest bundles contain the largest fibres. However these are accompanied by a distinct group of smaller fibres, whose significance is uncertain. It is not clear that there is a distinct proprioceptor group. The fibres to the chromatophores are numerous (30000) and of medium diameter (< 12 μm). The visceral motor and vasomotor nerves, however, contain hundreds of thousands of minute fibres (< 3 μm). The significance of these numerous small fibres can hardly be to obtain great resolution of movement in such acts as secretion of saliva. Presumably the great number has a special significance. The fibres to the muscles of the buccal mass are more numerous and smaller than the somatic motor fibres, but fewer and larger than those for the viscera. The muscles of the arms and suckers have 3.0 x 106 fibres, all < 6 μm, originating from neurons within the arms. They are controlled from the brain by relatively few but large fibres (32000, < 26 μm). There is also a reduction of 100 times on the afferent pathway of the arms, from some 18 x 106 receptors at the periphery of the suckers to 140000 fibres entering the brain. The brain thus serves for major decisions, whose detailed execution is left to peripheral reflex centres in the arms. The optic nerve fibres are very numerous, and all small (20 x 106, < 1.3 μm) presumably for economy of space and material. By contrast the static nerves, although they are short, contain a small number of large fibres (1400 between 6 and 22 μm) as well as several thousand smaller ones. The presence of sets of fibres with their distinctive diameters, conduction velocities and other properties is evidently a fundamental feature of the design of the nervous system of cephalopods as it is of vertebrates, although the significance of many features remains to be explored.