Russell and Semen (Russell, 1921) originally proposed that within the nervous system 'mnemic phenomena' occur. As the system responds to stimuli it undergoes dynamic transitions between 'indifference-states', which can themselves be seen as 'excitement- complexes'. The difference between some equilibrium state and the excitement- complex generated by a stimulus is the 'engram' due to the stimulus. Semon had little idea what the nature of an engram was, his idea was not a result of direct observations of nervous system function. However Russell proposed that physiology would provide grounds for the support of Semon's hypothesis. Early experiments in the nervous system focused on the networks of connections between brain cells, the material substrate for mnemic phenomena. Later physiological experiments provided insights into how cells behave as a result of the connectivity characteristics of the network, and thus provided some insights into what might be the nature of an excitement-complex, at the level of the single neurone. While both these types of investigation continue to illuminate the subtlety of function in the nervous system, in the last decade many investigators have attempted to more fully describe an excitement-complex and determine the nature of an engram, the code used by the nervous system to generate an excitement-complex. It was the aim of the physiological experiments on which this thesis is based, to investigate how a visual stimulus is represented in the activity of populations of cells in the primary visual cortex, so the ideas of Russell and Semon concerning the engraphic operation of the nervous system provide a useful starting point.
The broad aim of this investigation was to examine what strategies individual cells and the network in which they exist use to encode the spatio-temporal characteristics of complex contextual stimuli. Cells were stimulated in a variety of ways with sinusoidal grating stimuli presented in a bipartite fashion centred over a cell's excitatory field, the dimensions of which were determined by presenting circular patches of grating with different diameters. The effect of spatial context was investigated by presenting each cell with up to 200 differently configured stimuli. Centre diameter, orientation and surround orientation were varied to make each configuration of a unique bipartite grating. Single unit visual responses were analysed
to determine the effect changing these stimuli had on the number of spikes fired, the temporal structure of discharges, and the way one response synchronised with other simultaneously recorded ones.
4. Methods.
In this study two animal systems were investigated. The data analysis reported below was obtained during recording sessions carried out on seven female monkeys (Macaca Fascicularis) weighing between 6 and 10 Kg, and ten female domestic cats weighing between 2 and 3 Kg. In some cases these preparations were shared with other projects running in the laboratory (e.g. Cudeiro and Sillito, 1996).
Preparation of mammalian models.
Cats.
Experiments were carried out using acute feline preparations, during an experiment these were paralysed and anaesthetised using a gaseous mixture of 70% nitrous oxide, 30% oxygen and halothane. Anaesthesia was induced with a N20\02 and 5% halothane mixture, initially this was delivered into an enclosed cat box, and then through a face mask when the animal had been transferred to the surgery table. During surgery halothane was delivered at a concentration of 2-3%. Before each of the following surgical procedures a local anaesthetic (lignocaine hydrogenchloride and adrenaline acid tartrate) was also administered at each site where an incision was to be made. After each procedure all wound margins were treated by topical application of antibiotic powder (chlortetracycline hydrogenchloride), and then they were sutured. Surgery began with a tracheotomy that allowed a Y-shaped canula to be inserted, this was secured and for the duration of the experiment the animal was artificially respired with a small animal ventilator (SAR-830 from CWE Inc.). Initial indicators of the preparations physiological condition and anaesthetic depth were respiration rate and reflex activity these were closely monitored, subsequently end-tidal carbon dioxide concentration and respiration rate were monitored electronically using equipment manufactured by Hewlett Packard. Carbon dioxide concentration was maintained between 3.8 and 4.2% and respiration rate at 20-30 m in '\ deviations from these limits were signalled with alarms, the concentration of halothane was adjusted between 2- 3% during surgery to ensure that the preparation was maintained within these limits.
A bilateral cervical sympathectomy was conducted to remove innervation to eye muscles that might cause involuntary eye movements during the experiment. Following this a canula was inserted in the femoral vein, this was subsequently used to deliver drugs to the preparation. Electrodes were then attached to four points on the body, in order that an BCG signal could be obtained and the cardiac pulse rate could be monitored, this was done using equipment manufactured by Hewlett Packard. Body temperature was monitored and controlled to maintain it at 38°C using a thermostatically controlled electric blanket.
After these initial stages of surgery animals were stabilised in a stereotaxic frame with ear, eye and mouth bars. The head was positioned precisely in the centre of the frame by manipulating the positions of the ear bars. Eye and mouth bars were used to prevent rotation around the ear bar axis. This was done so that particular areas of the central nervous system could be localised, using predetermined co-ordinates. When the preparation was stable within the stereotaxic frame, paralysis was initiated, this was achieved using a neuromuscular blocker, supplied through the venous canula, the agent used was gallamine triethiodide (flaxedil). It was delivered as an initial dose of 40 mg K g '\ and subsequently for maintenance, at a rate of 10 mg Kg'^h'^ in 4.3% dextrose saline.
An incision was made in the scalp above an appropriate site depending on whether the aim of the experiment was to recorded from the cortical area VI or the lateral geniculate nucleus. Subcutaneous tissue was reflected in order to expose the skull. Craniotomy sites were determined using stereotaxic co-ordinates, relative to a zero point defined by the position of two ear bars, if the target was V 1 the craniotomy was made on the posterior surface of the skull, near the mid-line, the craniotomy site made when LGN recording was made on the dorsal surface of the skull. Sub-cutaneous tissue was than removed to reveal the skull. A craniotomy was performed using a dental drill, a small aperture was created to reveal the dura, a membrane that covers the cortex, the dura was dissected, so that electrodes could be directly inserted into the tissue. Around this exposed area a cylindrical aluminium chamber was secured using dental acrylic, after electrode insertion this chamber was filled with a agar/cerebro
placed on top of this, to prevent fluid leakage and tissue movement. After the chamber had been sited a silver plated electrode was attached to the skull, the scalp incision was then treated with antibiotic powder and sutured around the chamber. The electrode was used to record the electro-encephalogram (EEG) of the animal, the pattern of activity indicated the anaesthetic depth, recordings were made when spindle activity was apparent.
The positions of the area centralis and blind spot were determined by reflecting the tapetum onto a tangential projection screen, using a Keeler pantoscope. It was possible to see these landmarks after dark adapting, they were marked on the screen and used as spatial reference points in later analysis. After this the eyes were washed with 0.9% saline, then they were treated with phenylephrine hydrogenchloride to retract the nictating membranes, pupils were dilated using atropine methonitrate, the corneas were then protected with contact lens. The eyes were focused using 2 mm apertures placed concentrically around the area centralis and a further set of lens, on the stimulation screen 0.57 m away. At regular intervals during the experiment this process was repeated to check that the corneas were clear and healthy.
During sessions of visual stimulation and recording from neurones the preparation was lightly anaesthetised, according to the EEG criterion, the concentration of halothane was reduced to between 0.1 and 0.5%.
Primates.
Experiments were also carried out on acutely prepared primates. Before surgery each experimental animal was given a pre-med injection, of 0.05 mg Kg'^ of acerpromazine maleate (manufactured by C-vet). Subsequently atropine sulphate (manufactured by C-vet) administered intra-muscularly at a concentration of 40 mg K g '\ this was done to ensure a clear respiratory tract. Anaesthetesia was then induced using ketamine (Ketaset), supplied intra-muscularly at a concentration of 10-15 mg K g '\ Before each surgical incision was made a local anaesthetic was used, this was marcain (bupivacaine hydrogenchloride BP), subsequently wound margins were treated with a
topical application of cicatrin anti-biotic powder (neomysin sulphate BP). The first procedure performed was a tracheotomy, this allowed a Y-shaped canula to be inserted in order to facilitate artificial respiration and the provision of anaesthetic gases in a mixture containing 70% nitrous oxide and 30% oxygen. A small animal ventilator was used to respire the preparation during each experiment (SAR-830 manufactured by CWE Inc.). Next canuli were inserted in both femoral veins, drugs would subsequently be delivered through these.
After these initial procedures the preparation was transferred to a stereotaxic frame for the remainder of the experiment. At this point the preparation was transferred onto a maintenance anaesthetic. The transfer occurred when muscle tone returned as the preparation came round from the ketamine that had been delivered. The maintenance anaesthetic, sufenta forte (sufentanil citrate) (Levitt et al, 1994) was delivered as an initial bolus over a couple of minutes through a venous canula, until cessation of respiration, at this point the ventilation began using the pump, sufentanil was then delivered at a rate of 4 pgKg'^h'^ continuously throughout the experiment. A neuromuscular blocker was used to prevent movement, the agent used for this purpose was norcuron (vercuronium bromide) (Levitt et al, 1994), again this was supplied through a venous canula, and at a rate of 0.1 mg Kg'^h'\ The fluid and nutrition requirements of the animal were managed using the solutions containing the maintenance anaesthetic and paralytic agent. Supplements of lactated ringers (Hartmanns solution) and glucose were supplied in these solutions. To reduce swelling due to cerebral oedema, 0.5 mg Kg'^ of a steroid, dexamethasone was administered, antibiotics were also given daily to prevent the development of infection.
Anaesthetic level was adjusted throughout the experiment in accordance with data obtained from various monitors of the preparations condition. ECG electrodes were attached to four points on the preparations torso and limbs. The intersystolic heart rate was monitored, deviations out of the range 120-180 bpm were alarmed, using an electronic monitor manufactured by Hewlett Packard. The state of the preparation was also monitored, using end-tidal CO2 concentration as a parameter, changes in this
preparation was stable. EEG was recorded using electrodes that were attached to the scalp, and core body temperature was monitored using a rectal probe attached to a thermostatically controlled electric blanket. Body temperature was maintained between 37.5-38.5°C.
The head was initially secured in the centre of a stereotaxic frame using ear, eye and mouth bars, a head bar was then attached, and the eye bars were removed. This precise head positioning allowed the positions of recording sites to be determined with stereotaxic co-ordinates. This meant that in each experiment cells with receptive fields in similar zones of visual space were recorded. A recording site was prepared by making a craniotomy, 2-3 mm in diameter above the appropriate portion of cortical area V I, over a zone concerned with perception of parafoveal visual space approximately 5° from the fovea, the dura was exposed and a very small aperture was dissected for the insertion of electrodes. A chamber was then secured around the craniotomy with dental acrylic (simplex rapide), after electrode insertion this was filled with agar/CSF and then wax to plug the aperture and maintain a constant cranial pressure and thus minimise brain movement.
Features of the retinal surface were visualised using a reverse ophthalmoscope, this process involved visualising the fovea and the head of the optic nerve separately, by placing a cross-hair over each in turn with the illuminating scope, and then placing a mirror in front of the illuminating beam so that the cross-hair positioned over the landmark was projected back onto the projection screen. Eyes were treated with atropine methonitrate and protected with contact lenses, they were then focused with 2 mm apertures and lenses placed in front of the apertures, on a stimulation screen placed at either 57 or 114 cm.
Electrodes.
Extracellular recordings were made using tungsten-in-glass microelectrodes (Merrill and Ainsworth, 1972), these were manufactured within the laboratory. Tungsten wire (figure 5a) from Clark Electromedical Instruments, of diameter 125pm was
F ig u re 5.
a.
c.
b.
d.
Figure 5.
A diagram of the various of stages in the manufacture of tungsten-in-glass microelectrodes, a. An unetched tungsten wire, b. a typical etched electrode, and the capillary into which it is inserted, c. The process of passing an etched wire inside a capillary through a heated coil to melt an insulating glass coating on to the tungsten surface at the tip. d. An insulated tungsten wire which has been exposed at the tip. It is this tip that records from neurones.
electrolytically etched, by repeated immersion in a solution containing NaN0 2 (71g\100mls) and KOH (34g\100mls). The resulting sharp point (figure 5b) was then coated in glass by passing it through a capillary (Clark Electromedical Instruments) and slowly through a heated coil (figure 5c). The last 8-20pm of the tungsten tip was then exposed by dipping the glass covered tip in molten boric acid. Removing the current from the boric acid caused it to cool, contract and pull the glass coating away from the tungsten tip. This tip could then be etched to the right length and shape electrolytically using a dilute KNO2 solution, these last two stages, being very delicate, were performed under a microscope (figure 5d). The next stage of the process involved plating the tip, first with gold, by dipping it in a gold solution, and then applying a current of 0.05pA across the electrode and solution for 5 seconds. A second surface layer of platinum plating is then applied electrolytically by dipping the electrode in a solution containing gelatine and Kolrausch solution (which contains platinum) and applying a current of 0.15-0.25pA for about 10 seconds. This deposited a layer of platinum that is l-2pm thick on the tungsten tip (figure 6a), which reduces its impedance. Before plating electrodes can have an impedance of 2-6MQ at IkHz, however plating with gold and platinum reduces this to 300-800KQ. Such an impedance decrease, increases an electrode's ability to detect the spikes fired by individual cells. In these experiments, electrodes with platinum plated tips, between 8 and 12 pm long were generally used, these were inserted in pairs or in triplets. Pairs were made by positioning individual electrodes in a precisely machined perspex holder so that their tips were about 1 mm apart (figure 6b). To make a triad, three individual electrodes were glued to a central rigid core, a spinal needle was generally used, each electrode tip protruded 4mm from the spinal needle tip, so that this, while providing mechanical support, did not disturb the recording site. The spacing between each electrode tip precisely controlled during triad manufacture (figure 6c) by manipulating them under a microscope, using two Leitz manipulators and a Prior manipulator, before being glued in place using quick hardening adhesive. Connections were made with the signal processing system described below using gold plugs crimped onto the other end on the tungsten wire.
Figure 6.
b.
1 m mC.
□
n
8Figure 6.
A diagram showing a close-up of the recording tip of a tungsten-in glass microelectrode (a), the uninsulated tungsten tip is plated in gold and platinum to reduce its impedance, a facilitate the recording of cells. In b and c two configurations of electrodes are featured, a pair where the tips are separated by 1mm and a triad where the separations are smaller. The electrodes were spatially distributed in these ways to facilitate recording simultaneously from nearby and neighbouring orientation columns, to detect connectivity over distances which are associated axon projection in the anatomical literature.
Insertion of the electrodes, with aim of recording in both the lateral geniculate nucleus and the primary visual cortex in both cats and primates, took place with reference to a three dimensional stereotaxic co-ordinate scheme. The zero point for this co-ordinate space was defined as the point at which the axis containing the ear bars met the cranial mid-line, the mid-line was apparent as bony sagital suture on the dorsal surface of the skull. This position as well as all others were recorded using reading taken from vernier scales, in the longitudinal anterior/posterior (AP) axis and medial/lateral (ML) axis, these scales were an integral part of the manipulator used to position the electrode. When attempting to record from the cat LGN, the trajectory of the electrode penetration was tilted in the rostro-caudal direction in order to prevent damage to visual areas in the parietal and occipital lobes. To determine the entry point that was necessary using this trajectory, in order to record from the LGN, a perspex model was used, electrode tips were positioned at a predetermined point within this model that represented the representation of central visual space within the LGN, within 6° of the area centralis. This achieved, the AP and ML values were noted, these values were then transposed onto the cranial surface, it was at this point that a craniotomy was performed and the dura dissected. Typical co-ordinates required were AP=+6.5 mm and ML=+8.5 mm, the electrode was required to proceed below the cortical surface for approximate 5.5 mm before it entered the LGN. Cortical recordings were made by making penetrations that were approximately perpendicular to the surface, and these proceeded up to a depth of typically 2 mm when they entered the white matter. The co-ordinates used to record from a zone containing cells with receptive fields with 6° of the area centralis in the cat were AP=+2 to -6mm and ML=+1.5 to +2.1 mm (Grieve and Sillito, 1991).
During an experiment an electrode or a composite arrangement was positioned in a rigid perspex holder, this was then attached to the electronic stepping microdrive (Digitimer, Scat). This arrangement was then placed directly above the insertion site using a microscope, the microdrive was then advanced to the point where the electrode tips could be seen to enter the cortex. At this point the microdrive counter was set to zero. The microdrive enabled the electrode to be advanced in a highly