Electrophysiological recordings were performed in six pigmented and three albino adult female ferrets (Mustela putorius furo) 1–2 years of age. All experiments were approved by the local authorities (Regierungspräsidium Arnsberg) and were carried out in accordance with the Deutsche Tierschutzgesetz of 12 April 2001, the European Communities Council Directive of 24 November 1986 (S6 609 EEC), and the National Institutes of Health guidelines for care and use of animals for experimental procedures. The animals were bred and raised in the animal facility of the Department of General Zoology and Neurobiology, Ruhr Universitaet Bochum, and were group-housed in an enriched environment with access to an outdoor enclosure.
Animals were initially treated with 0.05 mg atropine sulphate (Braun) i.m. and anesthetized with 20 mg/kg ketamin (Ketavet®) and 2 mg/kg xylacine (Rompun®) i.m. They were intubated through the mouth, and a catheter was introduced into the cephalic vein. After additional local anesthesia with bupivacain hydrochloride, animals were placed into the stereotactic frame and artificially ventilated with air containing 0.2–0.6% halothan as needed throughout the entire experiment. Heart rate, endtidal CO2, and body temperature were monitored and maintained at physiological levels. The depth of anesthesia was controlled based on the heart rate and cardiac reaction to tactile stimuli. Pupils were dilated with tropicamide (Mydriaticum® Stulln). Corneae were protected with contact lenses chosen with a refractometer (Rodenstock) to correct the optics of the eyes to the distance of the stimulation monitors (see “Visual stimulation”). The skin overlying the skull was cut, the temporalis muscle was deflected, and a craniotomy was performed allowing access to areas 17 and 18 of visual cortex. Additionally, a head post was implanted, by which the animals were held during the recording. During the recordings, the animals were paralyzed with alcuronium chloride (Alloferin®) by an initial i.v. injection of 0.1 mg/kg and an i.v. infusion of 0.1 mg/kg per hour. After completion of the recording sessions, the wound was closed in appropriate layers and covered with antibiotic ointment (Nebacetin®). After full recovery, the animals were returned to their home enclosure and treated with analgetics (Carprofen, Rimadyl®) for 2 days and broadband antibiotics (Enrofloxacin, Baytril®) for 1 week after surgery. Electrophysiological recordings were repeated three to five times with 3 weeks recovery in-between. Because the quality of recordings during longterm acute experiments turned out to decline with time, we instead chose repeated shorter recording sessions (12–13 h duration). With this method, the quality and the amount of data recorded from individual animals dramatically increased. The procedure was well tolerated by the animals as shown by their normal feeding and play behavior.
Electrophysiological recordings in visual areas 17 and 18 were performed with tungsten in glass microelectrodes. To achieve penetrations from the mediodorsal to the lateroventral parts of the areas, the electrode was angled 45° from medial to lateral in the coronal plane. Histological reconstructions of the recording sites indicate that our recordings sampled all cortical layers. Neuronal activity was conventionally amplified, passed through a window discriminator and stored for offline analysis.
To present stimuli in different horizontal disparities, we used an adapted mirror stereoscope (Ohzawa and Freeman 1986a, b). At each side of the animal, a screen (Eizo®, FlexScan F57, 60 Hz) was positioned at a viewing distance of 28 cm. Thus, each screen covered about 60° by 40° of the visual field. “Cortex” software (Laboratory of Neuropsychology, NIMH, Bethesda, USA) was used to control visual stimulation and data acquisition. Position and extent of monocular receptive fields of recorded neurons were determined by bright bars of various sizes moved and rotated on dark background manually by means of the computer mouse. Cells were classified based on their responses as belonging to areas 17 or 18. This electrophysiological classification of recorded neurons was applied to the further data analysis.
Two different stimuli were used. The first stimulus was a moving bright bar on dark background (size 1° by 20° of visual angle; moving in eight directions in 45° steps; speed 5, 10, 20, 50, and 100°/s; Michelson contrasts of 0.032, 0.063, 0.091, 0.189, 0.375, 0.559, 0.756, and 0.976; background luminance 0.8 Cd/m2). The second stimulus consisted of a cyclic luminance modulated grayscale sinusoidal grating (diameter 40° of visual angle; moving in eight directions in 45° steps; spatial frequencies of 0.05, 0.07, 0.1, 0.2, 1, and 2 cpd; temporal frequency of 1 Hz; Michelson contrasts of 0.016, 0.040, 0.063, 0.133, 0.298, 0.465, 0.778, and 0.976; mean luminance 19.1 Cd/m2).
Each stimulus presentation was preceded and followed by a 500 ms presentation of the dark or gray background, respectively. Neuronal responses for different stimulus parameters were recorded separately using a pseudorandomized block design with ten repetitions per condition. Great care was taken to always cover the whole receptive field with our stimuli. These short tuning programs allowed us to determine the stimulus parameters that led to a clear activation of each recorded cell.
Using the stimulus with optimal parameters (like orientation, size, contrast, and speed), the horizontal disparity was then altered by computer-controlled displacement of the left screen in horizontal direction. Only neurons with vertical or oblique preferred orientations (up to 45° from vertical) were included in this analysis. Thus, irrespective of the preferred orientation (vertical or oblique) horizontal relative disparities tested were the same. Each run consisted of increasing and decreasing values of horizontal disparity, i.e. a sequence of increasing disparities was followed by a series of decreasing disparities to exclude long-term drifts (bars: nine conditions with steps of 2°, or, when neurons had smaller receptive fields, with steps of 1°, 0.5°, or 0.25° horizontal disparity; gratings: 12 conditions with steps of 30° over a full range of 360° between the two dichoptically presented gratings. Thus, absolute steps of horizontal disparity depended on the spatial frequency of the grating.) Each condition was repeated ten times.
Data were analyzed with MatLab® (Version 7.0.1, The MathWorks Inc.) using the Curve Fitting Toolbox (Version 1.1.2). Spike trains were convolved with a normalized, symmetric triangle kernel function 30 ms in width (Nawrot et al. 1999).
For bar stimuli, there was a clear receptive field-dependent increase and decrease of neuronal activity per sweep. We fitted the density function with a Gaussian distribution to define an analysis window of ±one standard deviation of the Gaussian. Within that interval, we calculated the mean neuronal activity in spikes/s for each condition and not the total number of spikes in the response peaks to eliminate any bias due to different widths of the peaks with changing relative disparities.
For grating stimuli, we classified simple and complex cells via a Fourier analysis of the neuronal activity, calculating the F
0 ratio between the amplitude of the first harmonic of the response and the mean activity (Movshon et al. 1978; Dean and Tolhurst 1983; Ohzawa and Freeman 1986a, b; Skottun et al. 1991; but see Mechler and Ringach 2002). Neurons with highly modulated activity (relative modulation > 1) were classified as simple, others as complex. For the subsequent analysis of simple cells, the amplitude of the first harmonic was used; otherwise, the mean activity was used.
Because of no clear landmarks on the retina, we could not define the eye’s center of gaze. Following the assumption that differences of monocular receptive field locations in early visual areas converge to zero for aligned eyes (e.g. Bishop 1973), or converge to a certain value for unaligned eyes that represents the deviation of gaze in a given experiment, we defined the mean difference of monocular receptive field locations on both screens for each experiment (on average data from 14 neurons per experiment) as zero disparity and related the positions of the stimulus presentation to this defined zero disparity. We thus treat our disparity settings as relative disparity centered on zero as defined above. The spatial stability of the response through the right eye for which the monitor was never shifted served as a control for the lack of drifts of this eye during the recording of one neuron. In fact, we never observed such shifts throughout the experiment under the immobilization conditions we imposed. Only 2 h (minimum) after discontinuing the alloferin infusion and the recording did visible spontaneous movements of the eyes start.
To analyze a neuron’s sensitivity for horizontal relative disparity, we screened our data for runs with at least one dataset of a certain horizontal disparity in which at least one measured value exceeded the range of the 2.5-fold standard deviation of that group. These particular runs were excluded from further analysis because of possible unstable recording. We applied a nonparametric one-way ANOVA (Kruskal–Wallis, P < 0.05) to screen for significant differences of mean neuronal activities at different horizontal disparities. Furthermore, multiple comparison was used to find statistically different activations in maxima or minima of relative disparity tuning curves, respectively. The difference between two disparities leading to statistically different activations was defined as the minimal difference affecting activity (MDAA). This approach leads to a statistical determination of tuning width. For better comparison with other studies, we additionally fitted disparity tuning curves with a Gaussian.
After completion of the electrophysiological experiments, the animals were euthanized with an overdose of pentobarbital and perfused transcardially with 0.9% saline and 4% paraformaldehyde–lysine-periodate. After appropriate cryoprotection, frozen frontal sections were cut at 40 or 50 μm, respectively, and stained for Nissl, myeloarchitecture (Gallyas 1979, as modified by Hess and Merker 1983), and cytochrome C histochemistry (Wong-Riley 1979). Recording sites were reconstructed based on microlesions, the penetration scheme and the depth reading on the electrode microdrive.