European Journal of Neuroscience, Vol. 12, pp.

4757, 2000

European Neuroscience Association

Spatiotemporal prole of synaptic activation produced by

the electrical and visual stimulation of retinal inputs to the
optic tectum: a current source density analysis in the
pigeon (Columba livia)
Juan Carlos Letelier, Jorge Mpodozis, Gonzalo Marin, Daniver Morales, Carlos Rozas, Carlos Madrid and
Marcelo Velasco
Departamento de Biologa, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile
Keywords: CSD analysis, ganglion cells, optic tectum, pigeon, visual system

Abstract
The optic tectum of the pigeon is a highly organized, multilayered structure that receives a massive polystratied afference of at least
ve different populations of retinal ganglion cells and gives rise to various anatomically segregated efferent systems. The synaptic
organization of retino-tectal circuitry is, at present, mostly unknown. To investigate the spatiotemporal prole of synaptic activation
produced by differential (electrical and visual) stimulation of the retinal inputs, we performed a high-spatial-resolution current source
density analysis in the optic tectum of the anaesthetized pigeon. Electrical stimuli consisted of brief pulses of different durations
applied to the optic nerve head, while visual stimuli consisted of light ashes of different intensities. Electrical stimulation generated
sinks conned to retinorecipient layers. The temporal structure, spatial location and thresholds of these sinks indicated that they are
all due to primary tectal synapses of retinal bers with different conduction velocities. Sinks evoked by the fastest retinal axons were
more supercially located than sinks produced by slower retinal bers. Visual stimulation, on the other hand, resulted in a more
complex pattern of current sinks, with various sinks located in the retinorecipient layers and also well below. Visual stimulation
induced action potentials at supercial as well as deep tectal levels. We conclude that electrical stimulation activates most of the
populations of ganglion cells as well as their primary tectal synapses, but is unable to elicit a signicant activation of secondary tectal
synapses. Visual stimulation, on the contrary, activates just some of the incoming retinal populations, but in a way that produces
noticeable secondary activation of intratectal circuits. Laminar segregation of retinally evoked tectal activity, as reported here, has
also been found in other vertebrates. Similarities and differences with previous studies are discussed.

Introduction
The rst central stage of the retino-tecto-fugal pathway in all
vertebrates is the optic tectum (called in mammals `superior
colliculus'). The avian optic tectum is a large, topographically
organized, mesencephalic structure consisting of 15 clearly dened
plexiform and cellular layers (Fig. 1B) (Ramon y Cajal, 1911; Cowan
et al., 1961). In pigeons it receives at least 90% of all the ganglion
cell axons (total 2.4 million) that originate in the contralateral retina
(Binggeli & Paule, 1969; Mpodozis et al., 1995). These axons
terminate in unequal numbers in layers 2, 3, 4, 5 and 7. Layer 5
receives the major portion of the retinal input, as demonstrated by the
total volume and density of retinal arborizations it contains (Fig. 1A)
(Hunt & Webster, 1975; Karten et al., 1997). The retinal input to the
tectum is not homogeneous; it is made of at least ve populations of
bers with different conduction velocities, ranging from 20 m/s to 1
2 m/s (Mpodozis et al., 1995; Dye & Karten, 1996). These
populations may have different visual properties, as a variety of
responses have been described to exist among the optic nerve bers
of birds (movement-sensitive, direction-sensitive, pure `off', pure

Correspondence: Dr Juan Carlos Letelier, as above.

E-mail: letelier@uchile.cl
Received 7 December 1998, revised 31 August 1999, accepted 2 September
1999

`on' and `on/off') (Miles, 1972; Hughes & Pearlman, 1974; Holden,
1977). Furthermore, retinal endings present characteristically distinct
morphologies at each tectal layer (Ramon y Cajal, 1911; Hunt &
Webster, 1975; Karten et al., 1997). This evidence suggests that the
retinal input is spatially segregated and that each retinorecipient layer
may contain terminals of a single specic kind of ganglion cell
(Ramon y Cajal, 1911; Reperant & Angaut, 1977; Karten et al.,
1997).
In birds, the optic tectum gives rise to various anatomically
independent visually driven efferent pathways, which terminate in
different structures such as the dorso-medial isthmus, the lateral pons,
the dorsal and ventral pretectal complex, the nucleus rotundus and
other thalamic nuclei (Benowitz & Karten, 1976; Reiner et al., 1979;
Reiner & Karten, 1982; Hunt & Brecha, 1984; Uchiyama et al.,
1996). Each of these pathways arises from specic cell groups located
mainly in layers 914. In between the supercial retinorecipient
layers and these output layers, a complex network of radial and
horizontal local interactions exists (Ramon y Cajal, 1911; Stone &
Freeman, 1971; Hunt & Webster, 1975; Hardy et al., 1985).
Despite both the description of the tectal eld potentials (TFP)
evoked by electrical stimulation of the retinal input and the
characterization of tectal receptive elds begun 30 years ago
(Holden, 1968a; Jassik-Gerschendfeld & Guichard, 1972; Mori,
1973; Jassik-Gerschenfeld et al., 1975; Frost & DiFranco, 1976; Frost

48 J. C. Letelier et al.

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

FIG. 1. Tectal organization and TFP proles. (A) Labelling of the uppermost seven layers of the tectum after the injection of cholera toxin in the contralateral eye. The label is conned to the terminal arborizations
of retinal axons and the bands mark the retinorecipient layers; layer 5 receives a particularly dense retinal input. (B) Correlation between the tectal layers, revealed by Nissl stain (left panel) and the electrically
evoked TFPs obtained at representative depths (right panel). Compare the TFP prole at supercial layers (23) with the TFP recorded at layers 1213, the inversion point is located approximately at layer 6. Scale
bar in A, 200 mm.

CSD analysis in pigeon optic tectum

& Nakayama, 1983; Dye & Karten, 1996), the spatiotemporal
characteristics of the tectal processing of retinal inputs remains an
open eld of exploration. Intracellular recordings of tectal cells have
hinted at a complex circuitry in which optic nerve stimulation elicits
monosynaptic excitatory responses as well as polysynaptic excitatory
and inhibitory responses (Hardy et al., 1984; Leresche et al., 1986).
In this context, it is important to investigate whether the laminar
segregation of tectal afferences and efferences is reected as spatially
and temporally segregated current generators when the retinotectal
synapses are differentially activated. To this end we performed a
high-spatial-resolution current source density (CSD) analysis of the
local eld potentials in order to: (i) describe the spatiotemporal
structure of the current generators activated by electrical and visual
stimulation of retinal bers; (ii) compare synaptic events evoked by
both types of stimulation; (iii) investigate if different classes of
conduction velocity groups of retinal axons evoke laminarly
segregated current generators.

Methods
Animal preparation
Adult feral pigeons (Columba livia) of either sex, with a body weight
of 300350 g, purchased from a local provider and kept in an
institutional animal facility, were used. For surgical procedures, each
pigeon was deeply anaesthetized with equithesin injected intramuscularly (0.3 mL/100 g body weight). A canula inserted in the
pectoral muscle allowed supplementary doses. The pigeon was then
placed in a head-holding device at the orientation dened by the
stereotaxic atlas (Karten & Hodos, 1967). After opening the skin
posterior to the orbit, the temporal muscle was retracted and a
craniotomy was performed, exposing the centro-dorsal tectal region
for electrical recordings. During the experiment, the ECG was
continuously monitored and body temperature was kept at 4041 C
by means of a DC powered electric blanket. Anaesthesia was
maintained by injecting 0.3 mL of equithesin every 2 h. All
procedures were approved by the Ethical Committee of the
Facultad de Ciencias of the Universidad de Chile.
Stimulation and Recording
Tectal eld potentials were recorded using tungsten microelectrodes
(14 MW @1 kHz, Frederick Haer, ME, USA) along radially
orientated tracks that penetrated the optic tectum. The TFPs were
elicited by direct electrical stimulation of the optic nerve or by light
ashes. For electrical stimulation, the cornea was cut and the lens
carefully removed. Bipolar electrodes, consisting of two parallel
insulated tungsten wires separated by 1 mm, were placed near the
optic nerve head. Electrical stimuli consisted of rectangular current
pulses of 0.11 mA in amplitude, lasting 30, 100 or 300 ms. The
pulses were produced by a constant current stimulator (CCUI-8,
Frederick Haer), controlled by a pulse generator and slaved to the
main data acquisition program written in LabView 3.1 (National
Instruments, TX, USA). Our standard procedure was to place the
recording electrode just below the tectal surface and select the lowest
current of a 100-ms stimulation pulse that produced stable TFPs. We
then used 30, 100 and 300 ms pulses to record TFPs at every depth. In
experiments where light ashes served as stimuli, the eyelid was
maintained open by a suture, and two hemispherical diffusing screens
were inserted in the light path to ensure homogeneous illumination.
The light ashes were delivered by a conventional photographic ash
located at 2 m from the eye. The intensity of the ash was graduated
using calibrated neutral density lters in steps of 0.12 log units. Flash
intensities covering a total of 4 log units were used.

49

Tectal eld potentials were recorded each 30 mm (50 mm in the case

of visual stimulation) starting from the surface until a depth of
1400 mm, which corresponds to tectal layers 1415. At each depth,
ve stimuli were delivered at 0.033 Hz. The resulting TFPs were
collected and averaged. The TFPs were collected at 10 Ksamples/s
during 40 ms (200 ms for visual stimulation), starting 10 ms before
the onset of the stimuli. At the end and middle of each penetration,
small electrolytic lesions (5 s 3 5 mA) were made for further
histological analysis. To assess the stability of the preparation,
TFPs were collected during the electrode withdrawal and compared
with the initial ones. Only data from stable preparations were used for
the subsequent CSD analysis.
CSD Analysis
Once the TFP data were collected, an off-line numerical procedure,
performed with IGOR PRO (Wavemetrics, OR, USA) on a Macintosh
computer, was used to construct the CSD map. The following
procedures were applied: (i) following Golarai & Sutula (1996) the
set of average TFPs was expressed as a two-dimensional matrix V in
terms of time (t) and depth (d); (ii) in order to diminish the
unavoidable artifact that affects the most supercial TFPs, some new
TFPs, conceptually located outside the tectum, were dened by
extrapolation (Vaknin et al., 1988); (iii) each time transept (V as a
function of d, with t constant) was smoothed by convolving it with a
13-point Gaussian low-pass lter that had a 3-dB cut-off point at
3 cycles/mm to give a function Gt(d); (iv) the second spatial
derivative, Gt(d), of every transept was calculated by using the
ve-point formula (see Nicholson & Freeman, 1975; Mitzdorf, 1985),
sinks corresponding to depths where Gt(d) > 0 and sources where
Gt(d) < 0; (v) nally, we calculated, for each time t, the spatial
integral of sink and sources to check our procedures. Step (iii)
reduced the high spatial frequencies found in Gt(d) and diminished
the unwanted noise that numerical differentiation introduces (Fig. 2).
The CSD proles were computed in relative terms, as the
conductivity s of the avian tectum has not been determined; CSD
proles are thus expressed in V/mm2. We also assumed that the optic
tectum was a homogeneous structure with respect to the conductivity
tensor in the two tangential dimensions. The CSD distribution was
visualized by constructing colour-coded maps that graphically
displayed the sink and source proles in space and time (see below
Figs 4 and 5).
At the end of the experiment, the pigeon was injected with an
overdose of anaesthetic and perfused with 600 mL of avian saline,
followed by 600 mL of 10% formalin. The brain was removed and
processed according to Nissl or Giemsa protocols (Iniguez et al.,
1985). The electrode tracks were reconstructed and the absolute tectal
locations of the TFPs were estimated using the landmarks of the
electrolytic lesions. To visualize the retinal terminals in the optic
tectum, in some pigeons 50 mL of cholera toxin was injected in one
eye. After 5 days of survival, the animals were perfused and the
brains processed according to the standard ABC immuno-histochemical protocol (Karten et al., 1997).

Results
Description of the temporal course of electrically and visually
evoked TFPs
To differentially stimulate the various groups of retinal axons, single
electrical pulses of different durations were used. At short stimulus
duration (30 ms) the supercial TFP exhibited a single negative peak
that developed with a 4.3-ms onset latency (6 1.8, n = 18), had a peak
latency of 5.1 ms (6 1.2, n = 18), and lasted 8.7 ms (6 2.1, n = 18)

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

50 J. C. Letelier et al.

FIG. 2. Description of CSD calculation. (A) Set

of averaged TFPs recorded at 30-mm intervals
along a 1400-mm radially orientated track
penetration in the tectum (each trace is the
average of ve responses) evoked by a 30-ms
electrical pulse given at t = 0. (B) Voltage
transept collected at t = 5 ms (arrow at t = 5 ms
in A) for both traces. The thick line with lled
circles shows raw data, and the thin line shows
ltered data produced by a 13-point Gaussian
low-pass lter. (C) CSD prole corresponding
to the transept shown in B. This trace is the
second spatial derivative, computed with a
ve-point formula of the smoothed trace
shown in B. The almost periodic oscillation
visible after 800 mm is an unavoidable artifact
due to the numerical procedure used to
compute the second derivative. Graphs B and
C share the same depth scale, where zero
depth represents the tectal surface.

(Fig. 3A, left panel). At increasing depths, 300 mm, the TFP prole
changed. Its initial phase became positive and its maximum was
reached at 45 ms. After correlating the readings of the microelectrode advancer with the histological analysis of electrolytic
lesions, the inversion zone was located at the bottom of the
retinorecipient layers (Fig. 1B). In the inversion zone, TFPs showed
complex, low-amplitude (2050 mV) features superimposed on the
main positive component, and only visible for 60100 mm. Below the
retinorecipient area the inversion was complete and the small features
disappeared. Tectal eld potentials remained positive until the bottom
of layer 14 and their amplitude decreased gradually with depth.
At intermediate stimulus duration (100 ms) (Fig. 3A, centre panel),
the amplitude of the initial negative phase increased between threeand ve-fold. Furthermore, depending on the experiment, one or
more secondary negative peaks appeared or were hinted at by an
inexion point in the trailing edge of the rst negative peak, with
peak latencies ranging between 7 and 15 ms. As was the case with
short stimuli, at intermediate stimulus duration the rst negative peak
reversed its polarity at 300 mm. The secondary negative peaks also
changed their polarity in the same region and small features appeared
around the inversion zone. Below the retinorecipient layers, only a
single positive peak was visible.
With maximal stimulus duration (300 ms), the peaks hinted at with
intermediate stimulus durations became clearly visible (Fig. 3A, right
panel) as three to ve negative peaks could be elicited at the tectal
surface. The mean onset latency of the rst peak was 4.3 ms (6 1.5 ,
n = 18), while the secondary peaks had onset latencies of 8 ms (6 2.1,
n = 18), 11 ms (6 3.5, n = 18), 15 ms (6 4.0, n = 14) and 17 ms (6 4.0,
n = 11), respectively. The rst peak had the largest amplitude, 1000
1500 mV, similar to the amplitude reached with intermediate
stimulation, whilst the amplitude of the longer-latency peaks was
much smaller, in the range 100200 mV. This complex prole
changed polarity with depth. Albeit not every peak reversed at
exactly the same depth, at 600 mm the TFP polarity was completely
reversed. Below 1000 mm a large positive peak was visible and,
sometimes, a smaller positive peak could be detected at 20 ms
latency.
The invariant features of electrically evoked TFPs could be
summarized as: (i) an initial negative phase with a 45 ms latency that

reversed polarity at 300 mm and remained positive until the end of

tectal layers; (ii) three to ve peaks with longer latencies elicited with
longer stimulus durations, which reversed at the same region as the
rst peak; (iii) small features superimposed on the main phase around
the reversal zone; and (iv) below 1000 mm a single large positive peak
was the dominant characteristic of TFPs.
Compared with the responses produced by electrical stimulation,
light ashes produced TFP proles with a much longer latency,
stretched over a longer period of time (Fig. 3B). At the tectal surface,
two main negative components, an early one at 25.6 ms (6 4.0, n = 6)
and a late one at 53.6 ms (6 5.0, n = 6) were always visible, even at
threshold intensities. These two components lasted on average for 9.3
and 12.9 ms, respectively. At all stimulus intensities, but especially at
high intensity, many secondary components were superimposed on
the two main phases, thus producing TFP proles of complex
appearance. Visually evoked TFP series taken from different animals
presented high variability in the number, amplitude and latency of
these secondary features, making their characterization and analysis
difcult. The main negative components of the visually evoked TFPs
reverse their polarity at the bottom of retinorecipient layers. Below
the reversal zone, usually new negative components appeared so, at
increasing depth, the evoked TFP presented a mixture of positive and
negative waves. The number, position, amplitude and latency of the
negative components located below the retinorecipient layers were
also variable and dependent on the stimulus intensity.
Description of the CSD proles evoked by electrical and visual
stimulation
To reveal the spatiotemporal features of the current ows that
generate the TFPs, a current source density analysis was
performed on a series of TFPs. The colour maps exemplied in
Fig. 4 display the current sink and source patterns for electrical
stimuli of 30, 100 and 300 ms pulse durations. These three
sequences show that the main current sinks are restricted to the
supercial layers located above 500 mm in the retinorecipient
zone. Stimulus duration also affected the number of evoked sinks.
Short stimulus pulses elicited only one or two sinks, while longer
pulses elicited up to ve. At short stimulus duration, a single sink
(`S' in Fig. 4A) was detected at 230 mm in the retinorecipient

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

CSD analysis in pigeon optic tectum

51

FIG. 3. Examples of tectal eld potentials evoked by electrical and visual stimulation. (A) Proles of electrically evoked TFPs at different depths and stimulus
durations. With a 30-ms pulse (left panel) the TFP shows, near the surface, an initial negative phase with a peak latency of 5 ms. Small, low-amplitude features are
visible near the inversion zone (). At 390 mm the initial phase becomes and remains positive until layer 13. With 100-ms pulses (centre panel) the TFP shows the
same qualitative prole but with a larger amplitude. The existence of secondary negative peaks is hinted at by the inexion points (*). With 300-ms pulses (right
panel) the supercial TFP prole shows four negative peaks with different amplitudes (r). The rst peak becomes positive at 390 mm. At the bottom of layer 13
(1400 mm) the waveform shows only two positive peaks, with different amplitudes. (B) The left and right panels show visually evoked TFPs triggered by light
ashes with a 1 : 2 intensity ratio. Visually evoked TFPs are more complex than electrically evoked ones. At low ash intensity (B, left panel) two clear negative
phases, separated by 30 ms, are visible near the surface (*). At 390 mm a positive peak can be detected but, as its latency does not correspond to the latency of the
negative peak recorded at the tectal surface, this positive peak is not the simple inversion of the negative peak found at the surface. Below the retinorecipient layers
the TFP prole is not completely positive. Negative components with 5070 ms latencies remain and do not change polarity as the electrode goes deeper into the
tectum. At higher ash intensity, a more complex TFP prole appears (B, right panel). Near the surface the number of negative components increases and the
latency of the rst negative component decreases. The complete visually evoked TFP lasts for 120 ms and contains many peaks. At 390 mm (corresponding to layer
5) the initial negative peak remains, but the rst positive peak is much larger. Below the retinorecipient layer (at 600 mm) the rst peak is totally inverted but the
secondary negative peaks are not.

area, extended for 100 mm, had an onset latency of 3.0 ms and
lasted for 6 ms. The histological analysis showed that this depth
corresponded to tectal layer 5 with the lower source located
between layer 7 and layer 9. No additional sinks were detected
below the retinorecipient layers. In two cases a very small sink
could be detected at the very surface with an onset latency of
3.0 ms (Fig. 4A, sink p).
At intermediate stimulus duration (100 ms), the CSD pattern
usually exhibited more sinks with longer latencies. In some
preparations two sinks were visible, while in others up to four sinks
appeared. The rst sink, S, resolved itself in two components (S1 &
S2 in Fig. 4B) with peak latencies of 4.2 and 6 ms. Two additional
sinks, with peak latencies of 7 and 12 ms, could be distinguished (S3
& S4). These subsequent sinks tended to appear 50100 mm below the
rst sink (S1) (Fig. 4B).
With longer pulses (300 ms) a total of ve sinks was normally
obtained. A new sink (S5) (Fig. 4C) with an onset latency of 17.5 ms
and a peak latency of 19 ms appeared between 200 and 300 mm. The
four sinks (S1, S2, S3 and S4) detected with 100 ms pulses continued
to be present at the same locations and latencies but with higher
amplitudes. In the case displayed in Fig. 4C, sink S4 appeared spread
out indicating the possible presence of other sinks with even longer
latencies and high thresholds. As was the case with 30- and 100-ms

stimulation pulses, no sinks were detected below the retinorecipient

layers.
A cluster analysis of onset latencies of the sinks obtained from ve
animals distinguished ve separate sinks that appeared at 2.9 (n = 5),
5.2 (n = 3), 6.3 (n = 4), 9 (n = 3) and 14.3 (n = 3) ms and had
corresponding durations of 1.4, 1.5, 2.3, 3 and 3.3 ms. Onset latency
and duration covaried, in such a way that sinks with longer latencies
lasted longer (r = 0.92). In three of the ve animals some initial sinks
were found with a latency that gradually increased with depth at an
average rate of 6.7 6 2.5 ms/mm (mean 6 SD, n = 5).
The histological data indicated that every sink in every animal was
located above layer 7. The sinks had an average vertical span of
130 6 76 mm (mean 6 SD, n = 18) and a range between 40 and
300 mm. In all cases the sink with the lowest onset latency was also
the most supercially located. We normalized each set of sink depths
by dividing by the depth of the rst sink and tested the null hypothesis
that secondary sinks, those with latencies greater than 4 ms, were
located at the same depth as the rst one. The null hypothesis was
rejected (Student's t-test, at P < 0.01) thus indicating that the rst sink
is more supercially located than the sinks with longer latencies. The
average relative depth of all secondary sinks is 1.44 6 0.25
(mean 6 SD, n = 13) thus indicating that, as a group, secondary sinks
are deeper than the rst.

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

52 J. C. Letelier et al.

FIG. 4. Colour maps of CSDs evoked by electrical stimulation. Colour-coded representation of CSD proles obtained by recording TFPs at 30-mm separation.
Sinks appear as red-yellow, sources appear as deep blue. The regular blue-green bands are a side-effect of the CSD calculation. (A) At low stimulus intensity, a
prominent sink with a peak latency of 5.5 ms lasting 7 ms appeared at 300 mm (S). At the very surface a small amplitude and short latency sink can be distinguished
(P) reecting the massive and synchronous activation of the fastest retinal axons. (B) With 100 ms pulses new sinks appear and a clearer spatiotemporal resolution
is obtained. The S sink triggered by 30 ms pulses is resolved in two components. The rst component (S1) has a peak latency of 4.2 ms and the second (S2) 6.2 ms.
The second component reaches the 400 mm level. Two additional sinks (S3 and S4), with onset latencies of 7 and 12 ms, can be distinguished in the 300400 mm
region. The sink P is now clearly visible at the surface. (C) With 300-ms pulses a massive recruitment of sink/source pairs mostly located in layers 57 is evident.
Five postsynaptic sinks can be distinguished between 200 and 400 mm (S1, S2, S3, S4 and S5) with peak latencies of 4, 5.3, 7, 12 and 19 ms, respectively.

All sinks had a corresponding source located below them, thus

forming a sink/source pair. In only two animals a source/sink/source
triplet was detected and in both cases the superior source was much

less intense than the inferior source. The mean extent of the sources
was 188 6 96 mm (n = 18), and correlated with the mean extent of
sinks (r = 0.80). The neutral zone separating each sink from its

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

CSD analysis in pigeon optic tectum

53

FIG. 5. Colour maps of CSDs evoked by two ash intensities. The sink/source prole obtained by visual stimulation was highly variable among different animals.
At low ash intensity (A), the presence of many sinks in the supercial as well as the deep tectal layers can be observed. The earliest sinks have a latency of
24 ms. At higher ash intensity (B), more sink/source pairs appear and the earliest sinks decrease their latency to 20 ms. Compare the presence of sinks below
the retinorecipient layers with Fig. 4. (Colour map as in Fig. 4)

corresponding source was 60 6 45 mm (mean 6 SD, n = 18). The

temporal course of each source matched the sink's prole.
Near the tectal surface, the visually evoked CSD proles appeared
qualitatively similar to electrically evoked CSDs. Below the
retinorecipient layers, visual stimulation triggered large sinks that
were never observed with electrical stimulation. These visually
evoked sinks exhibited more variability than the CSD pattern
obtained by electrical stimulation (Fig. 5). Despite this variability,
the following invariant features were always present: (i) appearance
of early sinks in the retinorecipient area; (ii) at least some of these
sinks were followed, after extinction, by strong sources located at the
same depth (see the succession of sinks/sources located at 300 mm in
Fig. 5); (iii) appearance of late polysynaptic sinks below the
retinorecipient area, especially at 800 mm (equivalent to layer 9).

Multiunit activity evoked by visual and electrical stimulation

In experiments with visual stimulation, multiunitary visually driven
activity was recorded as soon as the electrode entered the surface of
the tectum. Many observations indicated that these responses came
from retinal bers and/or their terminal arborizations on the tectum.
Firstly, the activity was composed of many potentials of different
sizes responding synchronously, as expected from the recordings of
the densely packed branches of ganglion cell terminals. Secondly, the
responses to the `on/off' of spots of lights or to the movement of
small bars presented on the receptive eld were very brisk, and with
no accommodation to repetitive stimulation. Thirdly, the neural
activity had no spontaneous discharge, in agreement with direct
recordings of avian ganglion cells or their axons (Miles, 1972;
Hughes & Pearlman, 1974). Fourthly, the receptive eld of these

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

54 J. C. Letelier et al.
and 40), selectivity for motion stimulus and strong habituation to
repetitive stimulation; all suggest that they originate from tectal cells.
In particular, the frequency of these recordings coincided with the
crossing of the electrode through the tectal cellular layers (especially
layers 68, 10 and 13). Light ashes were able to trigger these tectal
spikes, at least in a fraction of the units recorded.
In the case of electrical stimulation, multiunitary activity in phase
with the negative component of TFPs was recorded at the level of
retinorecipient layers. This indicates that the recordings were taken
from the tectal area activated by the stimulus. Below those layers no
spikes, time-locked to stimulus onset, were found, even at the level of
layers 10 or 13, where well-resolved recordings of isolated spikes are
common because of the large size of the neurons. These ndings
correlate with differences in the CSD proles obtained for the two
types of stimulation, as explained in the Discussion.

Discussion
Analysis of TFPs and CSD

FIG. 6. (A) Flash-evoked compound action potential recorded in the optic

tract. The compound action potential recorded in the optic tract shows a
regular wave with negative peaks separated by a constant interval of 8 ms. (B)
Visually evoked local eld potential (LFP) and simultaneous multiunit activity
(MUA) in the supercial tectum. The gure shows that negative components
in the LFP are synchronous with the ring of MUA. The response
characteristics of MUA indicate that they correspond to retinal axons or their
terminals. Note that the main spike in MUA res during each negative phase
of the LFP.

multiunitary responses was small, always < 3, and changed its

location in space only minimally as the electrode advanced in depth.
This small change in receptive eld position was attributed to drifts in
the eye or to a tangential component of the penetration. All these
elements indicated that retinal axons or their terminals were the main
contributors to the multiunitary activity recorded supercially. In
simultaneous recordings, obtained by applying low- and high-pass
lters to the same signal, the multiunitary activity occurred in phase
with the negative TFP components. Furthermore, in cases where
isolation of single spikes was possible, it was observed that the same
unit red a burst of spikes repetitively during each negative phase.
This synchronous ring correlated with the periodic temporal prole
of the compound action potential to a ash recorded in the optic tract
(see Discussion and Fig. 6A and B). Thus, the main negative
components of the TFPs triggered by the ash are at least partially
explained by the repetitive ring of some ganglion cells.
In addition to the multiunitary activity recorded in the supercial
layers, single units, especially between 700 and 1000 mm, were
frequently recorded. The characteristics of these responses were:
presence of spontaneous discharge, large receptive elds (between 3

Tectal eld potentials evoked by short electrical pulses (30 ms) exhibit
a single negative peak at the most supercial tectal layers. This peak
reverses its polarity around 300400 mm, and becomes a positive
wave toward the deeper layers. Increasing the duration of the pulses
increases the number of negative components, until four or ve peaks
are obtained with 300 ms pulses. This correlation probably resulted
from the recruitment of additional populations of retinal bers with
progressively slower conduction velocities. The classic results of
Bishop (Bishop et al., 1953) showed that in the optic nerve of cats,
brief pulses of 30 ms stimulate thick myelinated bers with fast
conduction velocities, whereas longer pulses recruit unmyelinated
bers with slower conduction velocities. In pigeons, recordings made
in the optic nerve of the compound action potential while
antidromically stimulating the optic tectum have directly shown that
increasing the duration of the stimulating pulse produces the
recruitment of slower conduction velocity groups, up to ve for the
tectum (Mpodozis et al., 1995). The wave of activity of each
population arriving sequentially in the tectum would then produce a
sequence of postsynaptic potentials recorded as negative components
in the TFPs. Another mechanism that may contribute to the
generation of delayed negative components is the synchronized
activation of postsynaptic intratectal circuits. However, as we argue
next, our CSD analysis and spike recordings do not favour this
interpretation.
The spatio-temporal structure of electrically evoked synaptic
currents shows that a single stimulation pulse delivered to the optic
nerve can trigger many sinks located in the retinorecipient layers and
none below layer 7. The recruitment of longer latency sinks as
stimulus duration increased (Fig. 4) reects both the range of
conducting velocities found among retino-tectal axons (202 m/s)
and the increasingly higher threshold of slow-conducting retinal
axons. Also, the temporal coherence of longer latency sinks is lower
than those of earlier sinks. This time-spreading is correlated with the
wider distribution of conduction velocity among slow-conducting
retinal bers (Clare et al., 1969). Taken together all this evidence
(dependency on stimulus duration, relation between temporal
coherence and onset latency, location inside the retinorecipient
layers) strongly suggests that, with electrical stimulation of retinal
bers, every sink detected in the retinorecipient layers is monosynaptically driven by a specic group of retinal axons. Furthermore,
the absence of sinks between layer 7 and layer 13 correlates well with
the absence of neuronal responses locked to the electrical pulse in that
region. Nevertheless, because one-dimensonal CSD analysis only

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

CSD analysis in pigeon optic tectum

55

FIG. 7. Comparison of sinks triggered by electrical stimulation of the optic nerve in the frog pigeon and cat. Summary of tectal CSDs evoked after electrical
stimulation of the optic nerve in frogs, cats and this study. In frogs, sink A is directly produced by retinal input. Sinks C and E result from intratectal circuits and
sinks B and D are triggered by a mixture of direct and indirect inputs. In cats, sinks A and D are produced by a direct retinal input, sink B by an indirect
retinocortical input and sinks C and E by a mixture of both kinds of inputs. In all three systems retinal inputs are segregated according to conduction velocity. In
pigeons the fastest bers terminate more supercially than the rest of the retinal axons and the electrical activation of the optic nerve does not trigger secondary
sinks as in the frog. All sinks appear to be generated by a direct retinal input and not by an indirect telencephalic input as in the cat.

reects synchronized activations of currents in the vertical dimension,

we can not discard the possibility that electrical stimulation may
mobilize non-synchronous intratectal polysynaptic circuits, or polysynaptic circuits involved in lateral interactions.
The properties of the very early sink p (early latency, short
duration and anatomical localization) indicate that it corresponds to
the transversal current generated by the synchronous and massive
activation of fast retinal axons, as they bend to enter the tectum.
Similar presynaptic sinks have been described in the visual cortex of
the rat (Kenan-Vaknin & Teyler, 1994).
The sources, corresponding to the electrically evoked sinks, were
mostly located below these sinks and extended well into the nonretinorecipient area. This spatial arrangement, valid for all sink/
source pairs, indicates that the dendrites of the postsynaptic cells are
radially orientated and that their length is similar to the extent of the
sources. Many tectal cell classes, especially projection cells, have
their dendrites and axons orientated radially; for example, layer 9
neurons that project to the Isthmi nuclei, layer 10 cells that project to
the ventral nucleus of the lateral geniculate and the principal optic
thalamic nuclei, and lamina 13 neurons that give rise to the tectorotundal projection (Hunt & Kunzle, 1976; Woodson et al., 1995;
Karten et al., 1997). Some of the neurons of this last group are
particularly relevant because they have long dendrites that ramify in
layer 5 (Luksch et al., 1998).
The spatio-temporal structure of sinks evoked by light ashes
shows a clear difference with respect to electrically triggered events
in that supercial and deep sinks are always evoked by light ashes.
As expected from the initial negative TFP component, a prominent
sink appears in the retinorecipient layers. Its latency is at least 25 ms
and depends on ash intensity. At the intensities used in these
experiments, other sinks, with longer latencies, also appeared in the
retinorecipient layers. These supercial sinks are in phase with
respect to the main negative TFP components recorded at the tectal
surface. Some of these supercial sinks probably represent postsynaptic potentials produced by the repetitive ring of retinal bers.
Our multiunit recordings (Fig. 6B) and data from experiments
describing the discharge characteristics of retinal axons to whole-eld

illumination (Duff & Cohen, 1975) indicate that retinal ganglion cells
respond with a train of spikes to a single light ash. This is also
corroborated by direct recordings made in the optic tract, as the
compound action potential evoked by a ash showed a very regular
waveform with four or ve evenly spaced peaks (8 ms) of similar
amplitude (Fig. 6A). Thus, some of the supercial sinks are probably
due to a complex interaction among the bursty structure of spike trains
of retinal ganglion cells, their different conduction velocities and the
nonhomogeneous visual properties of these cells. This complexity of
the retinal input makes it difcult to investigate whether some of the
supercial sinks, especially the later ones, could correspond to
polysynaptic activation (or postsynaptic ring). Although the overall
pattern of supercial sinks was similar in all the preparations
stimulated with light ashes, the precise spatio-temporal pattern of
sinks and sources is more variable than the pattern obtained with
electrical stimulation. Below the retinorecipient layers, light ashes
triggered clear and intense sinks. The most intense sinks appear at
700 mm. In association with these clear sinks, neuronal responses,
time-locked to the ash, appear at the same level, thus providing
evidence that secondary circuits are triggered by light ashes.
Spatial segregation of retinal inputs
Besides the temporal segregation of electrically evoked sinks, our
results also demonstrate spatial segregation. The earliest sink,
produced by fast-conducting axons, is more supercially located
than sinks with longer latencies produced by slower-conducting
axons. This difference indicates that, on average, fast retinal bers
arborize more supercially than slow ones. Recent anatomical
ndings (Karten et al., 1997) indicate that layer 5, the main
retinorecipient layer, contains terminal arborizations of the w-5b
population of ganglion cells that have the nest unmyelinated axons
and would thus have the slowest conduction velocity. In agreement
with these observations, we found that the long latency sinks elicited
by electrical stimulation are commonly located at the level of layer 5.
On average then, most of the activity elicited in the tectum by
electrical stimulation of retinal afferences takes place at that tectal
layer.

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

56 J. C. Letelier et al.
The presence of strong sinks near layer 5 also correlates with the
anatomical observations showing that the more abundant kind of
tecto-rotundal projecting neurons, Type I cells, have dendritic
expansions in layer 5, and that they are probably monosynaptically
connected with the retinal endings at that layer (Karten et al., 1997;
Luksch et al., 1998). Type I neurons project to the anterior and
centralis subdivisions of the nucelus rotundus (Karten et al., 1997).
Extracellular recordings have shown that each of these regions of the
rotundus responds to different visual dimensions (Wang et al., 1993).
Taken together, the aforementioned evidence, the results presented
here and the well-known correlation between conduction velocity
classes and functional types [exemplied by the Y, X, W subsystems
(Bishop et al., 1964; Stone, 1983)] suggest that the retinal ganglion
cells projecting into layer 5b, instead of forming a single denite
functional class, are a heterogeneous collection of ganglion cells that
might contribute to synthesize some of the various visual properties
detected upstream in the ascending tectofugal pathway.
Comparison with previous studies in the pigeon
The TFPs obtained in this study differ from those described in
previous studies of the pigeon's optic tectum where only a single `N'
wave has been described (Holden, 1968a; Stone & Freeman, 1971;
Mori, 1973; Bagnoli et al., 1977; Holden, 1980; Dye & Karten, 1996).
Here we report the existence of more complex waveforms with many
negative components. We found a direct relation between the number
of peaks detected and the duration of the stimulus. At short stimulus
duration, our TFPs showed a single `N-like' negativity whose
features (latency, duration and vertical location of the inversion
point) resemble that of the previous studies. Mori (Mori, 1973) also
reported that increasing the strength of the stimulation produces
multipeaked TFPs. However, in his studies he only used data from
experiments where a single peak was elicited.
Another result of the current study, the absence of evoked action
potentials by electrical stimulation, differs from intracellular recording of tectal cells performed in anaesthetized pigeons (Hardy et al.,
1984; Leresche et al., 1986) and from the work of Holden (Holden,
1968b). In the intracellular studies it was found that 25% of tectal
cells responded to an electrical pulse delivered to the optic nerve with
an EPSP followed by at least one action potential. In this study we
explored intensively below the retinorecipient layers and specically
at and near the output layer 13, and no cellular responses were found.
Besides differences in experimental procedures, two reasons could
explain the difference between this study and previous studies.
Firstly, because the depth of the intracellular responses is not
specied in the original study, it could be that some of the responses
were obtained from the retinorecipient layers. Secondly, a well
known side-effect of cellular impalement by a microelectrode is to
render the cell more excitable because the puncture slightly
depolarizes the cells. Thus, EPSPs that would, under normal
conditions, be subthreshold might generate an action potential. In
any case we feel that this question requires further research.
Comparison with other vertebrates
In the frog, electrical stimulation of the optic nerve elicits ve sinks
located at different tectal levels (Fig. 7). The rst sink to appear (sink
A) is located at the bottom of the retinorecipient layers (layer 8) and
is produced by fast, myelinated retinal bers (Witpaard & Keurs,
1975). The second and fourth sinks to appear (B and D) are located
more supercially (at retiniorecipient layer 9) and appear to be a
mixture produced by the activation of slower retinal axons and of
polysynaptic intratectal circuits. The third and fth sinks (C and E)
are both located outside the retinorecipient layers, and can be

exclusively attributed to nonretinal postsynaptic intratectal circuits

(Nakagawa et al., 1997).
Similarly, ve tectal (collicular) sinks have been described after
the electrical stimulation of the optic nerve in the cat. Sink A, located
at the lower part of the stratum griseum supercialis (SGS) is solely
produced by the activation of fast Y retinal bers (30 m/s), while sink
D, produced exclusively by the activation of slow W bers (54 m/s),
is located at the upper part of the SGS. Sinks C and E, located at the
middle part of the SGS, are produced by the activation of fast W
retinal bers (107 m/s) mixed with indirect activation of corticocollicular inputs. Finally, sink B is solely produced by an indirect
corticocollicular input. No synaptic currents attributed to the
activation of intracollicular circuits were found (Freeman & Singer,
1983).
If we compare our results with these studies, we nd important
similarities and differences (Fig. 7). Our results indicate that, as in the
frog and the cat, the sinks produced by activation of retinal synapses
are spatially segregated according to the conduction velocity of the
incoming bers. However, unlike the situation found in frogs and
cats, in the pigeon all the sinks resulting from the electrical
stimulation of the optic nerve appear to be monosynaptically driven
by the retinal inputs. Our results are similar to results with cats in that
we found no evidence of massive activation of intratectal circuits. In
the cat, some of the more delayed sinks are produced by the rebound
of activity generated in the visual cortex. We think that in our case
this is improbable, as it has been shown that the stimulation of the
Wulst (the avian equivalent of the visual cortex) produces evoked
potentials not in the supercial but in the deep tectal layers (Bagnoli
et al., 1977).
Summary and conclusions
Our data indicate that the sink and source pattern of the supercial
layers, for both electrical and visual stimulation, are mostly
monosynaptically generated, and mainly reect the wave of activity
that these stimuli trigger in the optic nerve. Electrical stimulation
generates a wave of activity composed by the single ring of bers
belonging to most of the velocity groups that go to the tectum. Visual
stimulation, on the other hand, triggers a sequence of bursts in bers
from only some of these velocity groups. Electrical stimulation
reveals a laminar segregation of retinal input, with fast-conducting
axons terminating more supercially than slow ones. We also show
that light ashes are more efcient than electrical stimulation in
mobilizing intratectal circuits, as they clearly trigger current
generators and action potentials in deep tectal layers.

Acknowledgements
We thank Elisa Sentis for technical assistance. This research was supported by
grant #1950687 from the Chilean Scientic Council (CONYCIT) to J.C.L.

Abbreviations
CSD, current source density; SGS, stratum griseum supercialis; TFP, tectal
eld potential.

References
Bagnoli, P., Francesconi, W. & Magni, F. (1977) Visual Wulst inuences on
the optic tectum of the pigeon. Brain Behav. Evol., 14, 217237.
Benowitz, L.I. & Karten, H.J. (1976) Organization of the tectofugal visual
pathway in the pigeon: a retrograde transport study. J. Comp. Neurol., 167,
503520.
Binggeli, R.L. & Paule, W.J. (1969) The pigeon retina: Quantitative aspects of
the optic nerve and ganglion cell layer. J. Comp. Neurol., 137, 118.

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757

CSD analysis in pigeon optic tectum

Bishop, P.O., Jeremy, D. & Lance, J.W. (1953) The optic nerve properties of a
central tract. J. Physiol. (Lond.), 121, 415432.
Bishop, P.O., Levick, W.R. & Williams, W.O. (1964) Statistical analysis of
the dark discharge of lateral geniculate neurones. J. Physiol. (Lond.), 170,
598612.
Clare, M.H., Landau, W.M. & Bishop, G.H. (1969) The relationship of optic
nerve ber groups activated by electrical stimulation to the consequent
central postsynaptic events. Exp. Neurol., 24, 400420.
Cowan, W.M., Adamson, L. & Powell, T.P.S. (1961) An experimental study
of the avian visual system. J. Anat., 95, 545563.
Duff, T.A. & Cohen, D.H. (1975) Optic chiasm bers of the pigeon: discharge
characteristics in response to whole eld illumination. Brain Res., 92, 145
148.
Dye, J.C. & Karten, H.J. (1996) An in vitro study of retinotectal transmission
in the chick: Role of glutamate and GABA in evoked eld potentials. Visual
Neurosci., 13., 747758.
Freeman, B. & Singer, W. (1983) Direct and indirect visual inputs to
supercial layers of cat superior colliculus: a Current Source-Density
analysis of electrically evoked potentials. J. Neurophysiol., 49, 10751091.
Frost, B.J. & DiFranco, D.E. (1976) Motion characteristics of single units in
the pigeon optic tectum. Vision Res., 16, 12291234.
Frost, B.J. & Nakayama, K. (1983) Single visual neurons code opposing
motion independent of direction. Science, 220, 744745.
Golarai, G. & Sutula, T.P. (1996) Bilateral organization of parallel and serial
pathways in the Dentate Gyrus demonstrated by Current-Source Density
Analysis in the Rat. J. Neurophysiol., 75, 329341.
Hardy, O., Leresche, N. & Jassik-Gerschenfeld, D. (1984) Postsynaptic
potentials in neurons of the pigeon's optic tectum in response to afferent
stimulation from the retina and other visual structures: an intracellular
study. Brain Res., 311, 6574.
Hardy, O., Leresche, N. & Jassik-Gerschenfeld, D. (1985) Morphology and
laminar distribution of electrophysiologically identied cells in pigeon's
optic tectum: an intracellular study. J. Comp. Neurol., 233, 390404.
Holden, A.L. (1968a) The eld potential prole during activation of the avian
optic tectum. J. Physiol. (Lond.), 194, 7590.
Holden, A.L. (1968b) Types of unitary response and correlation with the eld
potential prole during activation of the avian optic tectum. J. Physiol.
(Lond.), 194, 91104.
Holden, A.L. (1977) Concentric receptive elds of pigeon ganglion cells.
Vision Res., 17, 545554.
Holden, A.L. (1980) Field potentials evoked In the avian optic tectum by
diffuse and punctiform luminous stimuli. Exp. Brain Res., 39, 427432.
Hughes, C.P. & Pearlman, A.L. (1974) Single unit receptive elds and the
cellular layers of the pigeon optic tectum. Brain Res., 80, 365377.
Hunt, S.P. & Brecha, N. (1984). The avian optic tectum: a synthesis of
morphology and biochemistry. In Vanegas, H. (ed.), Comparative
Neurology of the Optic Tectum. Plenum, New York, pp 619648.
Hunt, S.P. & Kunzle, H. (1976) Observations on the projections and intrinsic
organization of the pigeon optic tectum: an autoradiographic study based on
anterograde and retrograde axonal and dendritic ow. J. Comp. Neurol.,
170, 153172.
Hunt, S.P. & Webster, K.E. (1975) The projection of the retina upon the optic
tectum of the pigeon. J. Comp. Neurol., 162, 433.
Iniguez, C., Gayoso, M.J. & Carreres, J. (1985) A versatile and simple method
of staining nervous tissue using Giemsa dye. J. Neurosci. Meth., 13, 7786.
Jassik-Gerschendfeld, D. & Guichard, J. (1972) Visual receptive elds of
single cells in the pigeon optic tectum. Brain Res., 40, 303317.
Jassik-Gerschenfeld, D., Guichard, J. & Tessier, Y. (1975) Localization of
directionally selective and movement sensitive cells in the optic tectum of
the pigeon. Vision Res., 15, 10371038.
Karten, H.J., Cox, K. & Mpodozis, J. (1997) Two distinct populations of tectal

57

neurons have unique connections within the retinotectorotundal pathway of

the pigeon (Columba livia). J. Comp. Neurol., 387, 449465.
Karten, H.J. & Hodos, W. (1967). A Stereotaxic Atlas of the Brain of the
Pigeon (Columba Livia). Johns Hopkins Press, Baltimore.
Kenan-Vaknin, G. & Teyler, T.J. (1994) Laminar pattern of synaptic activity
in rat primary visual cortex: comparison of in vivo and in vitro studies
employing the current source density analysis. Brain Res., 635, 3748.
Leresche, N., Hardy, O., Audinat, E. & Jassik-Gerschenfeld, D. (1986)
Synaptic organization of inhibitory circuits in the pigeon's optic tectum.
Brain Res., 365, 383387.
Luksch, H., Cox, K. & Karten, H.J. (1998) Bottlebrush dendritic endings and
large dendritic elds: motion-detection neurons in the tectofugal pathway. J.
Comp. Neurol., 369, 399414.
Miles, F.A. (1972) Centrifugal control of the avian retina. III. Effects of
electrical stimulation of the isthmo-optic tract on the receptive eld
properties of retinal ganglion cells. Brain Res., 48, 115129.
Mitzdorf, U. (1985) Current source-density method and application in cat
cereberal cortex: investigation of evoked potentials and EEG phenomena.
Physiol. Rev., 65, 3791.
Mori, S. (1973) Analysis of eld response in the optic tectum of the pigeon.
Brain Res., 54, 193206.
Mpodozis, J., Letelier, J.C., Concha, M.L. & Maturana, H. (1995) Conduction
velocity groups in the retino-tectal and retino-thalamic visual pathways of
the pigeon (Columba livia). Int. J. Neurosci., 81, 123136.
Nakagawa, H., Miyazaki, H. & Matsumoto, N. (1997) Principal neuronal
organization in the frog optic tectum revealed by a current source density
analysis. Visual Neurosci., 14, 263275.
Nicholson, C. & Freeman, J. (1975) Theory of current source-density analysis
and determination of conductivity tensor for anuran cerebellum. J.
Neurophysiol., 38, 356368.
Ramon y Cajal, S. (1911). Histologie Du Systeme Nerveux de L'Homme et Des
Vertebres. Maloine, Paris.
Reiner, A., Brecha, N. & Karten, H.J. (1979) A specic projection of retinal
displaced ganglion cells to the nucleus of the basal optic root in the chicken.
Neuroscience, 4, 16791688.
Reiner, A. & Karten, H.J. (1982) Laminar distribution of the cells of origin
descending tectofugal pathways in the Pigeon (Columba livia). J. Comp.
Neurol., 204, 165187.
Reperant, J. & Angaut, P. (1977) The retinotectal projection in the pigeon: an
experimental optical and electron microscope study. Neuroscience, 2, 119.
Stone, J. (1983). The Y/X/W Classication of Cat Retinal Ganglion Cells.
Plenum Press, New York.
Stone, J. & Freeman, J.A. (1971) Synaptic organization of the pigeon's optic
tectum: a Golgi and current source-density analysis. Brain Res., 27, 203
221.
Uchiyama, H., Yamamoto, N. & Ito, H. (1996) Tectal neurons that participate
in centrifugal control of the quail retina: a morphological study by means of
retrograde labeling with biocytin. Visual Neurosci., 13, 11191127.
Vaknin, G., DiScenna, P. & Teyler, T. (1988) A method for calculating current
source density (CSD) analysis without resorting to recording sites outside
the sampling volume. J. Neurosci. Methods, 24, 131135.
Wang, Y.C., Jiang, S. & Frost, B.J. (1993) Visual processing in pigeon nucleus
rotundus: Luminance, color, motion and looming subdivisions. Visual
Neurosci., 10, 2130.
Witpaard, J. & Keurs, H.E.D.J.T. (1975) A reclassication of retinal ganglion
cells in the frog, based upon tectal endings and response properties. Vision
Res., 15, 13331338.
Woodson, W., Shimizu, T., Wild, J.M., Schimke, J., Cox, K. & Karten, H.J.
(1995) Centrifugal projections upon the retina: an anterograde tracing study
in the pigeon (Columba livia). J. Comp. Neurol., 362, 489509.

2000 European Neuroscience Association, European Journal of Neuroscience, 12, 4757