How moving visual stimuli modulate the activity of the substantia nigra pars reticulata.
ABSTRACT The orientation of spatial attention via saccades is modulated by a pathway from the substantia nigra pars reticularis (SNr) to the superior colliculus, which enhances the ability to respond to novel stimuli. However, the algorithm whereby the SNr translates visual input to saccade-related information is still unknown. We recorded extracellular single-unit responses of 343 SNr cells to visual stimuli in anesthetized cats. Depending on the size, velocity and direction of the visual stimulus, SNr neurons responded by either increasing or decreasing their firing rate. Using artificial neuronal networks, visual SNr neurons could be classified into distinct groups. Some of the units showed a clear preference for one specific combination of direction and velocity (simple neurons), while other SNr neurons were sensitive to the direction (direction-tuned neurons) or the velocity (velocity-tuned neurons) of the movement. Furthermore, a subset of SNr neurons exhibited a narrow inhibitory/excitatory domain in the velocity/direction plane with an opposing surround (concentric neurons). According to our results, spatiotemporally represented visual information may determine the discharge pattern of the SNr. We suggest that the SNr utilizes spatiotemporal properties of the visual information to generate vector-based commands, which could modulate the activity of the superior colliculus and enhance or inhibit the reflexive initiation of complex and accurate saccades.
HOW MOVING VISUAL STIMULI MODULATE THE ACTIVITY OF THE
SUBSTANTIA NIGRA PARS RETICULATA
A. BERÉNYI,aP. GOMBKÖTO?,aÁ. FARKAS,a
Z. PARÓCZY,aZ. MÁRKUS,aR. G. AVERKIN,a,b
G. BENEDEKaAND A. NAGYa*
aDepartment of Physiology, Faculty of Medicine, University of Szeged,
Dóm tér 10, H-6720 Szeged, Hungary
bBay Zoltán Foundation for Applied Research, BAYGEN, Derkovits
fasor 2, H-6726 Szeged, Hungary
Abstract—The orientation of spatial attention via saccades is
modulated by a pathway from the substantia nigra pars re-
ticularis (SNr) to the superior colliculus, which enhances the
ability to respond to novel stimuli. However, the algorithm
whereby the SNr translates visual input to saccade-related
information is still unknown. We recorded extracellular sin-
gle-unit responses of 343 SNr cells to visual stimuli in anes-
thetized cats. Depending on the size, velocity and direction of
the visual stimulus, SNr neurons responded by either in-
creasing or decreasing their firing rate. Using artificial neu-
ronal networks, visual SNr neurons could be classified into
distinct groups. Some of the units showed a clear preference
for one specific combination of direction and velocity (simple
neurons), while other SNr neurons were sensitive to the di-
rection (direction-tuned neurons) or the velocity (velocity-
tuned neurons) of the movement. Furthermore, a subset of
SNr neurons exhibited a narrow inhibitory/excitatory domain
in the velocity/direction plane with an opposing surround
(concentric neurons). According to our results, spatiotempo-
rally represented visual information may determine the dis-
charge pattern of the SNr. We suggest that the SNr utilizes
spatiotemporal properties of the visual information to gener-
ate vector-based commands, which could modulate the ac-
tivity of the superior colliculus and enhance or inhibit the
reflexive initiation of complex and accurate saccades. © 2009
IBRO. Published by Elsevier Ltd. All rights reserved.
Key words: visual response map, saccade, artificial neuronal
Elementary sensorimotor reactions such as startle reac-
tions, the orienting reflex or reflexive saccades are con-
trolled by forebrain structures, including one of the main
output structures of the basal ganglia, the substantia nigra
(SN) (Pollack, 2001; Afifi, 2003). Nigrotectal connections
may serve as an important route for forebrain control over
elementary sensorimotor reactions of the midbrain (Gray-
biel, 1978; Behan et al., 1987). The role of the primate SN
in visually-guided saccades was extensively studied by
Hikosaka and Wurtz (1983). They described the activity of
substantia nigra pars reticularis (SNr) in behaving mon-
keys, reporting that these units decreased firing in re-
sponse to stationary visual stimuli used as saccade tar-
gets. Their work established the saccade-initiation theory
(Wurtz and Hikosaka, 1986), which assumes that suspen-
sion of tonic inhibitory GABAergic input by the SNr to the
superior colliculus (SC) is a permissive step toward a subse-
quent saccadic eye movement. Their findings, with respect to
response properties of SNr cells, were later extended by the
observation of increased activity of some SNr cells in re-
sponse to visual sensory stimulation (Magariños-Ascone et
al., 1994; Handel and Glimcher, 1999) and modulation of SNr
responses by moving visual stimuli (Schwarz et al., 1984;
Nagy et al., 2005a,b, 2006). The latter is not surprising, since
movement is an important natural behavioral stimulus, and
motion-dependent forecasting is needed to accurately track
and target moving stimuli.
The present study examines the nature of the encoding
algorithm by which the SNr maintains its diversified func-
tion, by testing the relationship between the response
characteristics of single SNr cells and the physical proper-
ties of moving stimuli. We hypothesize that the SNr may
supply specific information about moving stimuli to the SC,
which promotes the correct guidance of sight. Specifically,
we wish to know whether these neurons have the ability to
encode specific features of stimulus movement in their
action potential trains, information which may help to orient
attention toward the forecasted position of the stimulus.
We also tested whether excitatory and/or inhibitory re-
sponses are stimulus-dependent, and whether they are
present in separate neuronal populations. To answer these
questions, we recorded and analyzed the responses of a
large number of visually-active SNr neurons in anesthe-
tized cats. Based on to the visual response characteristics
of these neurons, we offer an expanded hypothesis con-
cerning the role of the SNr in the control of saccade initi-
ation and accompanying visuomotor processes.
Animal preparation and surgery
The experiments were carried out on 10 adult cats of both sexes,
weighing from 2.5 to 4.0 kg. All procedures were carried out to
minimize the number and the suffering of the animals. They followed
the European Communities Council Directive of November 24, 1986
(86/609 ECC) and the National Institutes of Health Guidelines for the
Care and Use of Animals for Experimental Procedures. The experi-
mental protocol had been accepted by the Ethical Committee for
Animal Research at the Albert Szent-Györgyi Medical and Pharma-
ceutical Center of the University of Szeged.
Anesthesia was initiated with ketamine hydrochloride (Calyp-
sol, 30 mg kg?1, i.m., Richter, Budapest, Hungary). After cannu-
lation of the femoral vein and the trachea, the animals were placed
*Corresponding author. Tel: ?36-62-545-869; fax: ?36-62-545-842.
E-mail address: email@example.com (A. Nagy).
Abbreviations: AEV, anterior ectosylvian visual area; DAergic, dopa-
minergic; KS, Kolmogorov–Smirnov; SC, superior colliculus; SN, sub-
stantia nigra; SNr, substantia nigra pars reticularis.
Neuroscience 163 (2009) 1316–1326
0306-4522/09 $ - see front matter © 2009 IBRO. Published by Elsevier Ltd. All rights reserved.
in a stereotaxic head holder. The wound edges and pressure
points were treated generously with procaine hydrochloride (Pro-
kain, 1%, TEVA, Debrecen, Hungary). Anesthesia was continued
with halothane (Narcotan, 1.6% during surgery and 1.0% during
recordings, Zentiva, Praha, Czech Republic; the minimum alveo-
lar concentration (MAC) level of halothane being held at 1 and 0.5,
respectively), and was maintained for 3 to 5 days. The depth of
anesthesia was monitored by regular inspection of the pupil size
on the non-treated side, and the examination of electrocortico-
gram and electrocardiogram recordings. The animals were then
immobilized with gallamine triethiodide (Flaxedyl, 20 mg kg–1, i.v.,
Sigma, St. Louis, MO, USA). During the experiment, a solution
containing gallamine (8 mg (kg h)–1), glucose (Glucosum 40%, 10
mg (kg h)–1, Pannonpharma, Pécs, Hungary) and dextran (Rheo-
macrodex 10%, 50 mg (kg h)–1, Baxter, Platting, Germany) in
Ringer’s solution (B. Braun, Melsungen, Germany) was infused
continuously at a rate of 3 ml (kg h)–1. Atropine (atropinum sulfu-
ricum, 0.1%, 0.2 ml, s.c., Egis, Budapest, Hungary) and ceftriax-
one (Rocephin, 40 mg (kg day)–1i.m., Roche, Budapest, Hungary)
were administered. The end-tidal CO2level and the rectal tem-
perature were monitored continuously and kept constant at 3.8–
4.2% and 37–38 °C, respectively.
Recording and stimulation
Electrophysiological single-cell recordings were performed extra-
cellularly via parylene-insulated tungsten microelectrodes (AM
System Inc., Carlsborg, WA, USA, 2 M?). Vertical penetrations
were made into the SN between Horsley–Clark coordinates ante-
rior 4–7 mm, lateral 4–6 mm, at a stereotaxic depth between 4
and 7 mm. Individual action potentials were selected with the help
of a data acquisition system (SciWorks Datawave, Datawave
Technologies, Berthoud, CO, USA). Putative GABAergic SNr neu-
rons were selected via the duration of their action potentials and
their spontaneous firing rate (Ungless et al., 2004).
At the end of each experiment, the animal was deeply anesthe-
tized with pentobarbital (Euthanyl, 200 mg kg–1, i.v., Bimeda-MTC,
Cambridge, Canada) and transcardially perfused with 4% parafor-
maldehyde solution. Brains were removed, sliced into coronal sec-
tions of 50 ?m, and stained with Neutral Red. The positions of the
recorded neurons were localized on the basis of the tracks of
the electrode penetrations, and their depths were related to the
surface of the SN. The position of the uppermost neuron that could
be distinguished as belonging to the SN based on firing rate and
recording depths was used to define the dorsal nigral surface.
Visual receptive fields of the neurons were estimated subjec-
tively by listening to the neuronal responses to the movements of
a light spot generated by a handheld lamp. To quantitatively
characterize the motion sensitivity of SNr neurons, we chose a
very simple stimulus, which provides a high contrast ratio, and
whose parameters are easy to quantify. Specifically, moving spots
were projected onto a tangent screen (LCD projector, refresh rate
100 Hz; resolution 1280?720 pixels; contrast ratio: 1300:1; re-
sponse time: 8 ms) centered on the area centralis and positioned
at a distance of 57.3 cm from the eye of the animal. In agreement
with previous reports, tested visual fields covered almost the
whole visual field of the contralateral eye. Spots of two different
diameters were used: 1° or 5°. A standard stimulus set of 80
stimulus parameter combinations was used to test the visual respon-
siveness of each neuron. Specifically, we moved each contrast spot
in eight different directions within the projected space (0–315° in 45°
increments, where 0° corresponded to straight up direction), with five
different movement velocities (5, 20, 40, 80 and 160 °/s). For each
recorded neuron, each stimulus combination was presented at least
10 times. For each trial control records were made for 1000 ms
before presenting the stimulus (the prestimulus time), and then the
moving stimulus was presented for an additional 1000 ms (the peri-
stimulus time). The number and temporal distribution of the action
potentials recorded during visual stimulations were stored for off-line
analysis (sampling rate: 20 kHz) and were visualized as peristimulus
Analysis of the visual responsiveness of the SNr
The data were analyzed with MATLAB®software (The Mathworks,
Inc., Natick, MA, USA). The duration of the recording from each
cell is proportional to the chance of losing it during the recording
session (and also with the instability of the recording). Conse-
quently, the high number of different stimulus parameters tested
forced us to test only a limited number of repetitions for each
different stimulus. The responses of the SNr neurons to visual
stimuli in anesthetized cats were usually weaker than those of
other visually responsive structures in the visual system. More-
over, the low number of cases made t-test comparisons between
the pre- and peristimulus periods vulnerable to sudden bursts. For
these reasons, this statistical procedure was poorly suited for this
work. To obtain a parameter that reliably estimates the effect of
the applied stimulus on the activity of a neuron based on a
relatively small number of recorded trials, we tested the responses
in overlapping narrow time windows with the help of the Kolmog-
orov–Smirnov (KS) test. While all of the prestimulus periods dur-
ing the whole recording of a neuron contained spontaneous ac-
tivity, we handled them as a common dataset for spontaneity. This
dataset was tested for stability and homogeneity. Those record-
ings in which more than 5% of trials exceeded the ?1 SD range of
mean spontaneous activity were considered to have inhomoge-
neous spontaneous activity (bursting, fluctuations, etc.), or to be a
result of unacceptable isolation during recording, and were ex-
cluded from the analysis. With the help of this strict rule, we were
able to minimize false detection of response fluctuations. Two
hundred sequences of 200 ms were randomly selected from the
dataset, and the firing rates of each 200-ms-long segment were
calculated (Fig. 1A). In a second step, the firing rates for each
stimulus combination were estimated separately. The firing rate of
each 100-ms-long sequence (with 50% overlap) within the peris-
timulus period was compared with the previously created dataset
(which represented the spontaneous activity) by using the KS test
(Fig. 1B–D). We defined those 100 ms-long sequences of the
peristimulus time as responses for which the KS test demon-
strated a significant difference from the spontaneous activity.
These “significant” segments are marked with colored bars in Fig.
2, which denotes the responses of an SNr neuron to various visual
stimuli. To quantify the strength of a response, we summed the
absolute values of the net firing rates (the difference of the stim-
ulated firing rate and the spontaneous firing rate) during the sig-
nificant sequences of the peristimulus intervals. We tested our
estimation method on numerous negative control recordings (on
recordings without stimulation), and also on previous well-charac-
terized neuronal data. The KS test, combined with the calculation
of net firing rates, proved to be an effective tool for estimation
of the strength of a response, because it eliminates the distur-
bances caused by non-consequent burst-like activity among
trials, and also gives a reliable quantification of the strength.
To visualize the stimulus preference of each neuron, two-dimen-
sional color-coded maps were constructed, where the axes repre-
sented the independent stimulus parameters, and the color scale
denoted the responses to the different parameter combinations.
Classification of response properties
Based on visual inspection of the response characteristics, we
classified the SNr neurons into 10 different tuning types (for details
concerning the preference types, see the Results section). To
have an objective, reproducible method with which to sort the
response maps into hypothesized classes, we used an artificial
neural network containing 40 joints as an input layer and 10 joints
as an output layer (representing the 10 different hypothesized
A. Berényi et al. / Neuroscience 163 (2009) 1316–13261317
Fig. 1. Analysis of the visual responsivity of SNr neurons. (A) The left panel depicts the raster plot of the spontaneous activity of a SNr neuron across
480 recorded trials, each 1000 ms long. The 200 randomly-selected, 200-ms-long segments which served as a representative dataset for the
spontaneous firing rate of the investigated cell are indicated by red lines. The distribution function of this dataset can be seen in the right panel, where
the abscissa defines the firing frequency ranges, and the ordinate the frequencies each range. (B, C) Two 100-ms-long segments of the neuronal
activity during visual stimulation. The left panel shows the peristimulus time-histogram (PSTH) of a neuron. Each PSTH includes a prestimulus and
a peristimulus period; the time scales are presented on the abscissas (ms). The duration of visual stimulation is denoted by the thick horizontal black
line above the PSTHs, and ordinates demonstrate the cumulated spike counts in each 25-ms-wide bin. Below the PSTHs, the raster plots of the
activities of the 10 trials are delineated. The unshaded part denotes the 100-ms-long segment of the response from which the distribution function in
the right panel is calculated. The KS test indicates that these distributions in parts b and c are not significantly different from the distribution of
spontaneous activity (part a, right panel). (D) A part of the PSTH which contains a visually-induced neuronal firing frequency increase; in this segment,
the distribution of the firing rates differs significantly from that of the spontaneous activity. For interpretation of the references to color in this figure
legend, the reader is referred to the Web version of this article.
A. Berényi et al. / Neuroscience 163 (2009) 1316–13261318
classes), together with three hidden layers, each with 400 neu-
rons. This network was automatically generated by a standard
built-in function of the MATLAB software. The number of neurons
within the different layers was adjusted empirically, to achieve the
fastest, learning performance. The output layer was adjusted to
have a log-sigmoid transfer function, and thus the response value
of each output component ranged between 0 and 1. The system
was embedded with resilient back-propagation as a self-learning
algorithm. With this method, a properly trained network can cor-
rectly recognize inputs it has not seen before. (Because of the
high number of possible variations within each group, a hypothet-
ical definition of the classes is unavoidable; more objective unsu-
pervised learning algorithms, such as the self-organizing map,
cannot be used for this classification task.) This generalization
ability was used to classify the response characteristics. The
resilient back-propagation technique is widely used for automated,
objective classification tasks due to its high internal stability (Ser-
pen and Corra, 2002; Grip et al., 2003; review: Lotte et al., 2007).
Above a certain complexity, neural networks of this kind show a
rather stable performance (Palaniappan, 2006). Our choice con-
cerning the number of neurons and hidden layers resulted in
effective recognition performance, with acceptable computational
To train the network, we generated artificial response char-
acteristics (40–100 of each tuning type) and presented them to
the network over several hundred epochs until the recognition
error limit criterion (mean squared error, ?10?5) was met. After
the training session, the real data for each recorded SNr neuron
were presented, and the network provided 10 values representing
the probability of fit into the 10 different types of tuning character-
istics. The schematic buildup of this artificial neural network is
outlined in Fig. 3. For each automated classification result, we
calculated the confidence value as:
where Vcis the calculated confidence value, while P1and P2
denote the probability of the first and second most probable class
type, respectively. We accepted a result as well-classified if the
probability of the primary class was ?0.5, and its confidence value
was ?0.5. If a response map did not meet this characterization
criterion, then it was visually inspected. If the subjective catego-
rization was in agreement with the automated result, the neuron
was included within the analysis; otherwise we categorized it as
Visual responsivity of the SNr neurons
The visual responses of 312 well-categorized SNr neu-
rons, with monophasic spikes with waveform lengths of ?1
ms indicating they were likely to be GABAergic (Grace and
Fig. 2. Visual responses of a SNr neuron. The 5?8 peristimulus time-histograms (PSTHs) represent the responses of a SNr GABAergic neuron during
the corresponding stimuli, with the same conventions as in Fig. 1. A significant increase in firing rate in a specific period in response to a given stimulus
condition is indicated by a red bar, and a significant decrease by a blue bar. The direction and velocity of the movements of the applied stimuli are
marked on the common abscissa and ordinate, respectively. In the lower left corner the corresponding response map can be seen as an inset.
The axes denote the same direction and velocity values as the major figure’s main axes. Inhibition and excitation are marked with proportionally dark
blue and red colors, respectively.
A. Berényi et al. / Neuroscience 163 (2009) 1316–13261319
Bunney, 1983; Ungless et al., 2004), were analyzed. The
spontaneous activity of these neurons was high, with a mean
of 27 spikes/s (SD?13 spikes/s, range: 17–76 spikes/s). The
31neuronsintheunclassifiedgroup showed rather stochas-
tic response maps, without any regular pattern. Interest-
ingly many of these neurons had fluctuating responses,
containing both excitatory and inhibitory segments. These
single units were excluded from the further analysis.
The visual receptive fields of the classified neurons
resembled those from our previous findings, since they
covered most of the contralateral visual hemifield including
the area centralis (Nagy et al., 2005a). The majority of the
visually active neurons responded optimally to high veloc-
ities (?40 °/s), but poorly to stationary visual stimulation.
No particular movement directions were preferred by a
majority of the SNr neurons across the population. The
spontaneous neuronal activity and velocity preference dis-
played a weak negative correlation; single units with higher
spontaneous firing rate preferred lower stimulus velocities
(R??0.12, P?0.025). We found no such correlation con-
cerning the direction sensitivity.
The stimulus conditions which evoked the greatest
change in neuronal activity were regarded as preferred
conditions for the cells. There were 132 (42%) cells with
dominant excitatory responses (an increase in their firing
rate during stimulation). Their average firing rate in re-
sponse to the preferred conditions was 34 spikes/s
(SD?17 spikes/s, range: 19 to 84 spikes/s), a 21% aver-
age increase relative to the spontaneous activity. For the
180 (58%) neurons with dominant inhibitory responses (a
decrease in their firing rate during stimulation), the mean
firing rate in response to the preferred conditions was 21
spikes/s (SD?11 spikes/s, range: five to 68 spikes/s), 18%
less than the mean spontaneous activity. Comparison of
the spontaneous activity of the cells with the visual re-
sponses observed under the preferred conditions revealed
significant differences for both the excitatory (P?0.01) and
the inhibitory (P?0.01) SNr neurons (Wilcoxon rank-sum
Velocity and direction tuning of the SNr visual
The major finding of this study is that a large proportion of
the visual SNr neurons were not exclusively inhibitory or
excitatory in nature, but could display either excitatory
(increased activity) or inhibitory (decreased activity) visual
responses, depending on the stimulus parameters. These
SNr cells were uniform in the sense that they were able to
respond with a definite increase in activity, but only under
certain stimulus conditions (Fig. 2). These response char-
acteristics were utilized to generate a map for each cell,
and from these maps the neurons were classified into five
categories (Figs. 4 and 5) with the help of an artificial
neural network (see Experimental Procedures).
A majority of the neurons (61%, n?190), referred to as
simple cells, exhibited a simple stimulus preference, dis-
playing significant changes in activity in response to well-
defined stimulus combinations. These cells revealed a
roughly circular region with narrow directional and ve-
locity tuning in its best response zone (Fig. 4A, B),
although minor regions with weaker responses were
present. Eighty-one (26%) of these simple neurons re-
sponded predominantly with excitation (increased activity),
while 109 (35%) underwent a definite inhibition (i.e. a
decreased activity during visual stimulation; Fig. 5).
Fifty-seven (18%) of the recorded SNr cells, referred to
as direction-sensitive neurons, responded to stimuli mov-
ing in a particular direction (Fig. 4C, D). In this case, the
region of best response displayed a linear organization
aligned along a specific direction. Seventeen (5.5%) of
these cells exhibited an increase in activity in response to
the preferred stimulus, and 40 cells (13%) showed a de-
crease (Fig. 5). No correlation was found between the
directions evoking excitation and those evoking inhibition.
Fig. 3. Scheme of the artificial neural network used for response profile classification. The figure demonstrates the input dataset containing the
response map of a recorded neuron, the layers of the artificial neural network, and the response characteristic profile produced by the network. The
color coding and the conventions are the same as in Fig. 4. For a detailed explanation of the tuning of artificial neural layers, see the text.
A. Berényi et al. / Neuroscience 163 (2009) 1316–13261320
Within the limits of the small sample, all directions occurred
with the same frequency. No neurons were found in the
SNr that preferred two opposing directions of movements.
Twenty (6.4%) of the recorded cells were observed to
be velocity-sensitive neurons (Fig. 4E, F). These cells
responded to stimuli moving at a certain velocity, irrespec-
tive of their direction, which mapped as a linear region of
best response aligned with a specific velocity. Seven
(2.2%) of the responses were excitatory and 13 (4.2%)
inhibitory (Fig. 5). The distribution of the preferred veloci-
ties was not characteristic; all velocities occurred with sim-
Both the velocity- and direction-tuned classes dis-
played maps that had additional areas of weaker modula-
tion. Often the response areas adjacent to the main re-
sponse were of the opposite sign. This organization was
even more striking in the remaining 45 neurons (14.4%),
referred to as concentric neurons. These cells exhibited a
specific pattern of complex response characteristics (Fig.
4G, H). Twenty-eight (8.9%) cells responded to a specific
Fig. 4. Examples of visual response tuning maps. Each stimulus preference map shows the responsiveness of an SNr GABAergic cell with a specific
type of preference. The abscissa denotes the eight different directions of stimulus movement, and the ordinate the different velocities. The strength
of the response to each specific stimulus parameter combination is color-coded; inhibitory responses (firing frequency decreases during stimulation)
are marked in blue, and excitatory responses (increase in firing rate) in red. The intensities of the colors are proportional to response strengths. For
better visibility, the color codes of the preference maps are normalized and smoothed via a bicubic spline technique. The neuronal responses to the
applied visual stimuli were classified into 4?2 distinct classes: (A, B) simple excitatory and inhibitory, (C, D) direction-sensitive excitatory and inhibitory,
(E, F) velocity-sensitive excitatory and inhibitory and (G, H) concentric neurons. For further details concerning the preference classes, see the text.
A. Berényi et al. / Neuroscience 163 (2009) 1316–13261321
stimulus condition with an increase in activity, and to even
slightly different neighboring conditions (both velocity and
direction) with a definite decrease (Fig. 5). In other words,
these neurons exhibited a simple response surrounded by
an inhibitory domain in the velocity/direction plane. Con-
versely, 17 (5.5%) cells responded to a specific stimulus
condition with a decrease in activity and to neighboring
conditions with a definite increase (Fig. 5).
The neurons detailed above were distributed randomly
in the SNr. No spatial clustering of the neuronal stimulus
preference types was observed. Similarly, we did not find
any correlation between the location of the cells and their
velocity or direction preference.
Stimulus size modulation of the responses
To address the question of how the size of the stimulus
affects the response characteristics of SNr neurons, light
spots of 1° or 5° in diameter were used to elicit visual
responses from 139 of the recorded neurons. Stimulation
with the 5° light spot led to a mean decrease of 18% in the
excitatory response and a mean increase of 40% in the
inhibitory response. Thus, the excitatory response to the 1°
spot was significantly stronger than that to the 5° spot
(Mann–Whitney test, P?0.01). In contrast, the inhibitory
response was significantly stronger (P?0.01) to the 5° spot
than to the 1° spot. It was noteworthy that the overall
responsivity remained the same for each cell, irrespective
of the stimulus size; only the ratio of the excitatory and
inhibitory responses changed. With the smaller spot as
visual stimulus, 57% of the responses observed within the
overall population were excitatory and 43% were inhibitory.
The larger stimulus made the response character maps
more ambiguous, making the automated classification of
these maps far less reliable than those of the small stim-
ulus. Further investigations with more stimulus diameters
are needed to decide whether the response class of each
neuron is independent of stimulus size.
This study highlights the excitatory and inhibitory effects of
moving visual stimuli on SNr neurons in anesthetized cats.
This is a suitable model for investigating visual information-
processing in the SN, since it lacks the numerous direct
and indirect non-sensory influencing factors (e.g. reward
prediction, unexpectedness and motor processes) present
during behavioral paradigms (Sato and Hikosaka, 2002;
Dommett et al., 2005; Hikosaka, 2007). However, anes-
thesia has a direct influence both on the spontaneous
excitability and on the responsivity of the recorded neurons
(Villeneuve and Casanova, 2003). In addition to the gen-
eral depression, feedback information originating from the
motor executor system is also absent from the investigated
circuitry. Experiments carried out on awake animals usu-
ally report a lower proportion of excitatory type responses
than those on anesthetized animals. This may be because
the presence of higher spontaneous activity during awake
conditions decreases the ability of excitatory SNr inputs to
generate significant activity changes. By contrast, inhibi-
tory inputs would have the opportunity for greater modu-
lation of the SNr cell activity. The result would be a shift in
balance toward inhibition under awake conditions. With the
introduction of the sliding KS test into our analysis, the
detection of presumed weakened responses was possible.
However we have to emphasize the high computational
cost of this analysis in contrast to the classic methods. A
further disadvantage of this algorithm is the loss of the
temporal distribution pattern, which may also carry impor-
tant information. Below, we speculate on the role of the
dualistic behavior of the SNr neurons, and propose a se-
quence of information coding with which the SNr may
control the visuomotor functions of the SC.
Physiological properties of the recorded SNr
On the basis of their neurotransmitters, most of the neu-
rons of the SNr can be classified mainly into dopaminergic
(DAergic) and GABAergic classes, although a few non-
DAergic–non-GABAergic cells remain, whose transmitters
have not yet been clarified. Ficalora and Mize (1989) dem-
onstrated that the nigrotectal tract consists of the axons of
GABAergic neurons. We selected presumed GABAergic
cells from the neuronal population during recording by
considering two properties: (1) these cells have narrow
spike forms (?1 ms) in extracellular recordings, and can
therefore be reliably differentiated from DAergic cells
which produce broad spikes of over 2 ms that usually
appear as spike doublets or triplets (Grace and Bunney,
1983; Ungless, 2004) and (2) GABAergic cells have been
shown indirectly to be fast-firing type II cells, in contrast
with the slow-firing type I DAergic cells (Guyenet and
Fig. 5. Distribution of visual response characteristics of 312 neurons.
The abscissa denotes the hypothesized 10 stimulus preference
classes, while the ordinate shows the frequency of occurrence of each
class. The percentage distributions are indicated above each bar.
Abbreviations: SiE, simple excitatory; SiI, simple inhibitory; DiE, direc-
tion-sensitive excitatory; DiI, direction-sensitive inhibitory; OrE, orien-
tation-sensitive excitatory; OrI, orientation-sensitive inhibitory; VelE,
velocity-sensitive excitatory; VelI, velocity-sensitive inhibitory; EsI,
concentric neuron—excitation surrounded by inhibition; IsE, concen-
tric neuron—inhibition surrounded by excitation.
A. Berényi et al. / Neuroscience 163 (2009) 1316–1326 1322
Aghajanian, 1978). Based on electrode track reconstruc-
tions, the SNr neurons investigated in our study were
located in the area of origin of the nigrotectal neurons
(Beckstead et al., 1981; Jiang et al., 2003). Thus, their
physiological properties and location indicate that the re-
corded neurons were likely GABAergic nigrotectal cells.
In general, most of the SNr neurons we studied exhib-
ited sensitivity to moving visual stimulation. Earlier studies
led to the categorization of GABAergic SNr neurons into
two classes. Most of the neurons responded to visual
stimuli with strong inhibition, while a smaller proportion
increased their firing rate (Handel and Glimcher, 1999;
Basso and Wurtz, 2002; Comoli et al., 2003). A main
finding of the present study was that these neurons can be
either excited or inhibited, depending on stimulus condi-
tions. In our preliminary results larger stimuli elicit more
intense inhibition in the SNr, while strongly decreasing
excitatory responses, though the amount of total re-
sponses remained similar.
Based on our results, we propose a new classification
of SNr neurons. In addition to the classes with narrowly
tuned neurons displaying velocity and direction sensitivity,
we have found neuronal classes which showed broad
band tuning to one of these stimulus properties. These
neurons may serve as stand-alone direction or velocity
detectors, and may reflect a higher level of organization.
Perhaps they receive multiple inputs from different simple
cells and thus integrate their preference maps. This would
also explain the existence of concentric preference maps,
where the simple response conditions are surrounded by
inverse responses. These neurons would disinhibit or
block their target SC neurons, only if the movement of the
stimulus precisely fit their preference. As a result, they
would produce a vectorial saccade command for a specific
direction and distance, consistent with previous evidence
for complex direction and velocity tuning of SNr GABAergic
neurons (Nagy et al., 2005a).
Similarities of stimulus preference between the SNr
units and neurons of other visual structures
The source of the complex direction and velocity re-
sponses observed in SNr is currently unknown. Previous
studies stimulating the anterior ectosylvian visual area
(AEV) with simple geometric forms suggested that it also
contains a high proportion of direction-selective units
(Mucke et al., 1982; Benedek et al., 1988; Hicks et al.,
1988). Direction-selective cells are also common in areas
17 (Hammond and Andrews, 1978), 18 (Rose and Blake-
more, 1974), and 21b (Tardif et al., 2000) and in the
posteromedial lateral suprasylvian area (PMLS) (Blake-
more and Zumbroich, 1987), but are less common in areas
19 (Bergeron et al., 1998) and 21a (Wimborne and Henry,
1992; Dreher et al., 1993). Orientation selectivity, in which
neurons are sensitive to movements along one axis,
usually with inverse response to the opposing directions,
is a characteristic feature of AEV units, but was not
found in this investigation of the SNr. Inhibitory re-
sponses are a characteristic feature of cells in area 21b
(Tardif et al., 2000) and also cells in the deep layers of
SC (Dreher and Hoffman, 1973). Detecting neurons with
velocity and direction sensitivity (analogous to those
discovered earlier in the SC and other structures along
the extrageniculo–extrastriatal pathway (Benedek et al.,
1996; Nagy et al., 2003; Waleszczyk et al., 2007)) sug-
gests that the SNr receives strong modulation from this
part of the visual system.
Visual afferents supplying the SNr
We did not detect any clustering or correlation between the
position, direction or velocity preference of cells with dif-
ferent characters in the SNr. This supports the notion that
the SNr is not retinotopically arranged or systematically
mapped in any more complex fashion. Anatomical studies
indicate two sets of neural circuits that may transmit visual
information from the retina to the SN. A schematic repre-
sentation of these anatomical connections is provided by
Fig. 6. Visual information presumably reaches the SNr
through the conventional corticostriatal route (Saint-Cyr et
al., 1990; Norita et al., 1991). Visually active SNr neurons
may also receive visual input through direct and indirect
tectonigral pathways (Tokuno et al., 1994; Comoli et al.,
2003). The SC also projects to the pedunculopontine nu-
cleus (Redgrave et al., 1987) and the subthalamic nucleus
(Tokuno et al., 1994; Coizet et al., 2009), and both of these
nuclei make direct contact with neurons in the SN (Lokwan
et al., 1999). It has been suggested that basal ganglia
circuits including the subthalamic nucleus could be a
source of visual information for the SN cells with excitatory
responses (Jiang et al., 2003). Further, the SC sends
strong visual efferents to the suprageniculate nucleus of
the thalamus (Katoh et al., 1995), which in turn provides
visual information to the caudate nucleus (Harting et al.,
2001), thereby forming a tecto-thalamo-striatal route. Al-
though the SN receives predominantly inhibitory inputs
from the striatum, an excitatory striatonigral pathway (Ro-
dríguez et al., 2000) may also transmit visual information to
the nigral neurons.
Functional role of SNr GABAergic neurons
It is unclear whether the visually responsive SNr neurons
take part in sensory information processing, or are instead
components of the reverberating motor circuitry of the
basal ganglia. The need for integration of sensory informa-
tion and motor commands suggests the latter, but direct
evidence can only be gained from awake, behaving animal
experiments. The convergence of different sensory (Nagy et
al., 2005b, 2006), oculomotor (Sato and Hikosaka, 2002;
Hikosaka and Wurtz, 1983) and somatomotor (Schultz,
1986) information onto SNr neurons is well supported, but
this does not mean that every sensory neuron has a role in
motor functions. Based on previous studies, we conclude
that saccade-related neurons that display visual activity
are more likely to project to the SC than those without
sensory activity (Hikosaka and Wurtz, 1983).
A shift of attention toward novel stimuli in the surround-
ing space requires an analysis of the movement parame-
ters of the object by the saccade control system. Detection
of novelty may be provided by DAergic SN cells (Martin
A. Berényi et al. / Neuroscience 163 (2009) 1316–1326 1323
and Waszczak, 1994, 1996; Rice et al., 1997; Radnikow
and Misgeld, 1998; Comoli et al., 2003; Dommett et al.,
2005; Redgrave and Gurney, 2006). Thus, pars compacta
cells may directly or indirectly confer the ability to appro-
priately modulate saccadic movements onto SNr cells. The
fact that we found an equally represented preference for
every stimulus direction and velocity combination suggests
that, despite the absence of a topographic organization in
the SNr, some kind of functional organization may be
present. Wurtz and colleagues (Wurtz and Hikosaka,
1986; Basso and Wurtz, 2002) postulated a higher-order
topographic organization between the visual stimulus site
and the center of the receptive fields. We suggest that
each SNr neuron might be connected to corresponding SC
cells in such a way that they can activate or block specific
saccades (McIlwain, 1990). The SC, on the other hand,
receives a retinotopic, highly-ordered input both from the
retina (Graybiel, 1975; Ogawa and Takahashi, 1981; Beck-
stead and Frankfurter, 1983) and from the cortical visual
areas (McIlwain, 1977; Berson and McIlwain, 1983; Ber-
son, 1988; Harting et al., 1992). These two projections
converge onto SC neurons, and furnish its retinotopic or-
ganization (McIlwain, 1990). The simple SNr cells re-
corded here may activate or inhibit specific saccades that
would drive the eye toward a position where a moving
stimulus is predicted to be when the saccade is carried out
(Jiang et al., 2003). Moreover, Kaneda and colleagues
(2008) recently showed that the ipsilateral nigrotectal pro-
jection also targets GABAergic interneurons in the SC. By
modifying the balance between the effect exerted directly
on the projection neurons of the SC and through this
indirect connection, the SNr can adjust both the spatial and
temporal aspects of the activity of SC motor neurons.
The functional relation between the SNr and SC in
pursuit eye-movements and saccade generation is known,
but the code behind the information transmission remains
controversial. Two current theories about the meaning of
population activity in the SC (Van Gisbergen et al., 1987;
Lee et al., 1988; Van Opstal and Van Gisbergen, 1989)
suggest that the activity of each ensemble determines a
vector of eye movement with a definite amplitude and
direction. This means that the SC expresses an integrated
signal that takes into account both the actual eye position
(McIlwain, 1990), and the expected future position. These
might be integrated within the SC, or possibly within an
external structure that serves as an integrative center. The
inhibitory nigrotectal neurons presumably help to sharpen
the collicular response map by disinhibiting some tectal
units when they decrease their own firing rate, while sup-
pressing other tectal units when they increase their firing
rate. The capacity of the SNr signal to modify the saccade
direction was verified recently by Liu and Basso (2008).
They found that electrical activation of the GABAergic SNr
cells profoundly modified the amplitude and direction of
saccades in behaving paradigms, and the stimulation in-
fluenced both cell types in the SC. Li et al. (2006) con-
cluded that as a further step of integration, the mutual
inhibition among the populations of SC neurons will shape
the sum gross response profile of the SC, forming a dy-
namic balance between inhibition and excitation caused by
the SNr. This output signal (a dynamic sum of the “mini-
vectors” represented by the single units of the active pop-
ulation) is suitable for driving brainstem oculomotor cen-
ters (Anderson et al., 1998). The vectorial response pro-
files of both the buildup and burst cells of the SC suggest
that they are driven by two different sources: the incoming
feedback signal reflecting the actual eye position (or the
difference between the site of current attention) and infor-
mation on the position (or expected position) of the point of
interest. The input information is integrated in both the SNr
and the SC, in multiple consecutive steps.
Fig. 6. Visual and oculomotor connections of the SNr. Connections within the ascending tecto-fugal visual system are marked with solid arrows, while
the oculomotor circuitry is denoted with dashed arrows. Note that the ascending tecto-fugal pathway and the oculomotor pathway overlap each other.
Abbreviations: SCs, Sci, SCd, superior colliculus superficial, intermediate and deep layers respectively; PPT, pedunculopontine-tegmental nucleus;
STN, subthalamic nucleus; Sg, suprageniculate nucleus; CN, caudate nucleus; FEF, frontal eye field; IVA, insular visual area.
A. Berényi et al. / Neuroscience 163 (2009) 1316–13261324
To summarize, this paper provides further understanding
of how visual information may modulate the activity of the
SNr neurons. The spatiotemporally represented visual in-
formation may determine the sensorimotor integrative
function of the SNr. We suggest that the SNr could control
the activity of the SC through direct nigrotectal connec-
tions, and could enhance or inhibit the reflex initiation of
saccades to moving targets.
Acknowledgments—The authors express their gratitude to G. D.
Molnár and K. Hermann for their valuable technical assistance, to
P. Liszli for his expert help, to A. Peto ˝ for the data collection, and
to G. Mochol, D. Wójcik and S. Łe ˛ski for the computational help.
This work was supported by OTKA-NKTH/Hungary grant 68594
and OTKA/Hungary grant PD 75156. A.N. is a János Bolyai Re-
search Fellow of the Hungarian Academy.
Afifi A (2003) The basal ganglia: a neural network with more than
motor function. Semin Pediatr Neurol 10:3–10.
Anderson RW, Keller EL, Gandhi NJ, Das S (1998) Two-dimensional
saccade-related population activity in superior colliculus in mon-
key. J Neurophysiol 80:798–817.
Basso MA, Wurtz RH (2002) Neuronal activity in substantia nigra pars
reticulata during target selection. J Neurosci 22:1883–1894.
Beckstead RM, Edwards SB, Frankfurter A (1981) A comparison of the
intranigral distribution of nigrotectal neurons labeled with horserad-
ish peroxidase in the monkey, cat, and rat. J Neurosci 1:121–125.
Beckstead RM, Frankfurter A (1983) A direct projection from the retina
to the intermediate gray layer of the superior colliculus demon-
strated by anterograde transport of horseradish peroxidase in
monkey, cat and rat. Exp Brain Res 52:261–268.
Behan M, Lin CS, Hall WC (1987) The nigrotectal projection in the cat:
an electron microscope autoradiographic study. Neuroscience
Benedek G, Mucke L, Norita M, Albowitz B, Creutzfeldt OD (1988)
Anterior ectosylvian visual area (AEV) of the cat: physiological
properties. Prog Brain Res 75:245–255.
Benedek G, Fischer-Szatmári L, Kovács G, Perényi J, Katoh YY
(1996) Visual, somatosensory and auditory modality properties
along the feline suprageniculate-anterior ectosylvian sulcus/insular
pathway. Prog Brain Res 112:325–334.
Bergeron A, Tardif E, Lepore F, Guillemot JP (1998) Spatial and
temporal matching of receptive field properties of binocular cells in
area 19 of the cat. Neuroscience 86:121–134.
Berson DM (1988) Convergence of retinal W-cell and corticotectal
input to cells of the cat superior colliculus. J Neurophysiol 60:
Berson DM, McIlwain JT (1983) Visual cortical inputs to deep layers of
cat’s superior colliculus. J Neurophysiol 50:1143–1155.
Blakemore C, Zumbroich TJ (1987) Stimulus selectivity and functional
organization in the lateral suprasylvian visual cortex of the cat.
J Physiol 389:569–603.
Coizet V, Graham JH, Moss J, Bolam JP, Savasta M, McHaffie JG,
Redgrave P, Overton PG (2009) Short-latency visual input to the
subthalamic nucleus is provided by the midbrain superior collicu-
lus. J Neurosci 29:5701–5709.
Comoli E, Coizet V, Boyes J, Bolam JP, Canteras NS, Quirk RH,
Overton PG, Redgrave P (2003) A direct projection from superior
colliculus to substantia nigra for detecting salient visual events. Nat
Dommett E, Coizet V, Blaha CD, Martindale J, Lefebvre V, Walton N,
Mayhew JE, Overton PG, Redgrave P (2005) How visual stimuli
activate dopaminergic neurons at short latency. Science 307:
Dreher B, Hoffmann KP (1973) Properties of excitatory and inhibitory
regions in the receptive fields of single units in the cat’s superior
colliculus. Exp Brain Res 16:333–353.
Dreher B, Michalski A, Ho RH, Lee CW, Burke W (1993) Processing of
form and motion in area 21a of cat visual cortex. Vis Neurosci
Ficalora AS, Mize RR (1989) The neurons of the substantia nigra and
zona incerta which project to the cat superior colliculus are GABA
immunoreactive: a double-label study using GABA immunocyto-
chemistry and lectin retrograde transport. Neuroscience 29:567–
Grace AA, Bunney BS (1983) Intracellular and extracellular electro-
physiology of nigral dopaminergic neurons—1. Identification and
characterization. Neuroscience 10:301–315.
Graybiel AM (1975) Anatomical organization of retinotectal afferents in
the cat: an autoradiographic study. Brain Res 96:1–23.
Graybiel AM (1978) Organization of the nigrotectal connection: an
experimental tracer study in the cat. Brain Res 143:339–348.
Grip H, Ohberg F, Wiklund U, Sterner Y, Karlsson JS, Gerdle B (2003)
Classification of neck movement patterns related to whiplash-as-
sociated disorders using neural networks. IEEE Trans Inf Technol
Guyenet PG, Aghajanian GK (1978) Antidromic identification of dopa-
minergic and other output neurons of the rat substantia nigra. Brain
Hammond P, Andrews DP (1978) Orientation tuning of cells in areas
17 and 18 of the cat’s visual cortex. Exp Brain Res 31:341–351.
Handel A, Glimcher PW (1999) Quantitative analysis of substantia
nigra pars reticulata activity during a visually guided saccade task.
J Neurophysiol 82:3458–3475.
Harting JK, Updyke BV, Van Lieshout DP (1992) Corticotectal projec-
tions in the cat: anterograde transport studies of twenty-five cortical
areas. J Comp Neurol 324:379–414.
Harting JK, Updyke BV, Van Lieshout DP (2001) Striatal projections
from the cat visual thalamus. Eur J Neurosci 14:893–896.
Hicks TP, Benedek G, Thurlow GA (1988) Organization and properties
of neurons in a visual area within the insular cortex of the cat.
J Neurophysiol 60:397–420.
Hikosaka O (2007) Basal ganglia mechanisms of reward-oriented eye
movement. Ann N Y Acad Sci 1104:229–249.
Hikosaka O, Wurtz RH (1983) Visual and oculomotor functions of
monkey substantia nigra pars reticulata. I–IV. J Neurophysiol 49:
Jiang H, Stein BE, McHaffie JG (2003) Opposing basal ganglia
processes shape midbrain visuomotor activity bilaterally. Nature
Kaneda K, Isa K, Yanagawa Y, Isa T (2008) Nigral inhibition of
GABAergic neurons in mouse superior colliculus. J Neurosci
Katoh YY, Benedek G, Deura S (1995) Bilateral projections from the
superior colliculus to the suprageniculate nucleus in the cat: a
WGA-HRP/double fluorescent tracing study. Brain Res 669:
Lee C, Rohrer WH, Sparks DL (1988) Population coding of saccadic
eye movements by neurons in the superior colliculus. Nature
Li X, Kim B, Basso MA (2006) Transient pauses in delay-period activity
of superior colliculus neurons. J Neurophysiol 95:2252–2264.
Liu P, Basso MA (2008) Substantia nigra stimulation influences mon-
key superior colliculus neuronal activity bilaterally. J Neurophysiol
Lokwan SJ, Overton PG, Berry MS, Clark D (1999) Stimulation of the
pedunculopontine tegmental nucleus in the rat produces burst
firing in A9 dopaminergic neurons. Neuroscience 92:245–254.
A. Berényi et al. / Neuroscience 163 (2009) 1316–13261325
Lotte F, Congedo M, Lécuyer A, Lamarche F, Arnaldi B (2007) A
review of classification algorithms for EEG-based brain-computer
interfaces. J Neural Eng 4:R1–R13.
Magariños-Ascone C, García-Austt E, Buño W (1994) Polymodal sen-
sory and motor convergence in substantia nigra neurons of the
awake monkey. Brain Res 646:299–302.
Martin LP, Waszczak BL (1994) D1 agonist-induced excitation of
substantia nigra pars reticulata neurons: mediation by D1 recep-
tors on striatonigral terminals via a pertussis toxin-sensitive cou-
pling pathway. J Neurosci 14:4494–4506.
Martin LP, Waszczak BL (1996) Dopamine D2, receptor-mediated
modulation of the GABAergic inhibition of substantia nigra pars
reticulata neurons. Brain Res 729:156–169.
McIlwain JT (1977) Topographic organization and convergence in
corticotectal projections from areas 17, 18, and 19 in the cat.
J Neurophysiol 40:189–198.
McIlwain JT (1990) Topography of eye-position sensitivity of saccades
evoked electrically from the cat’s superior colliculus. Vis Neurosci
Mucke L, Norita M, Benedek G, Creutzfeldt O (1982) Physiologic and
anatomic investigation of a visual cortical area situated in the
ventral bank of the anterior ectosylvian sulcus of the cat. Exp Brain
Nagy A, Eördegh G, Norita M, Benedek G (2003) Visual receptive field
properties of neurons in the caudate nucleus. Eur J Neurosci
Nagy A, Eördegh G, Norita M, Benedek G (2005a) Visual receptive
field properties of excitatory neurons in the substantia nigra. Neu-
Nagy A, Paróczy Z, Norita M, Benedek G (2005b) Multisensory re-
sponses and receptive field properties of neurons in the substantia
nigra and in the caudate nucleus. Eur J Neurosci 22:419–424.
Nagy A, Eördegh G, Paróczy Z, Márkus Z, Benedek G (2006) Multi-
sensory integration in the basal ganglia. Eur J Neurosci 24:
Norita M, McHaffie JG, Shimizu H, Stein BE (1991) The corticostriatal
and corticotectal projections of the feline lateral suprasylvian cortex
demonstrated with anterograde biocytin and retrograde fluorescent
techniques. Neurosci Res 10:149–155.
Ogawa T, Takahashi Y (1981) Retinotectal connections within the
superficial layers of the cat’s superior colliculus. Brain Res
Palaniappan R (2006) Utilizing gamma band to improve mental task
based brain-computer interface design. IEEE Trans Neural Syst
Rehabil Eng 14:299–303.
Pollack AE (2001) Anatomy, physiology, and pharmacology of the
basal ganglia. Neurol Clin 19:523–534.
Radnikow G, Misgeld U (1998) Dopamine D1 receptors facilitate
GABAA synaptic currents in the rat substantia nigra pars reticulata.
J Neurosci 18:2009–2016.
Redgrave P, Gurney K (2006) The short-latency dopamine signal: a
role in discovering novel actions? Nat Rev Neurosci 7:967–975.
Redgrave P, Mitchell IJ, Dean P (1987) Descending projections from
the superior colliculus in rat: a study using orthograde transport of
wheatgerm-agglutinin conjugated horseradish peroxidase. Exp
Brain Res 68:147–167.
Rice ME, Cragg SJ, Greenfield SA (1997) Characteristics of electri-
cally evoked somatodendritic dopamine release in substantia nigra
and ventral tegmental area in vitro. J Neurophysiol 77:853–862.
Rodríguez M, Abdala P, Obeso JA (2000) Excitatory responses in the
“direct” striatonigral pathway: effect of nigrostriatal lesion. Mov
Rose D, Blakemore C (1974) An analysis of orientation selectivity in
the cat’s visual cortex. Exp Brain Res 20:1–17.
Saint-Cyr JA, Ungerleider LG, Desimone R (1990) Organization of visual
cortical inputs to the striatum and subsequent outputs to the pallido-
nigral complex in the monkey. J Comp Neurol 298:129–156.
Sato M, Hikosaka O (2002) Role of primate substantia nigra pars
reticulata in reward-oriented saccadic eye movement. J Neurosci
Schultz W (1986) Activity of pars reticulata neurons of monkey sub-
stantia nigra in relation to motor, sensory, and complex events.
J Neurophysiol 55:660–677.
Schwarz M, Sontag KH, Wand P (1984) Sensory-motor processing in
substantia nigra pars reticulata in conscious cats. J Physiol
Serpen G, Corra J (2002) Training simultaneous recurrent neural
network with resilient propagation for static optimization. Int J Neu-
ral Syst 12:203–218.
Tardif E, Lepore F, Guillemot JP (2000) Spatial properties and direc-
tion selectivity of single neurons in area 21b of the cat. Neuro-
Tokuno H, Takada M, Ikai Y, Mizuno N (1994) Direct projections from
the deep layers of the superior colliculus to the subthalamic nu-
cleus in the rat. Brain Res 639:156–160.
Ungless MA, Magill PJ, Bolam JP (2004) Uniform inhibition of dopa-
mine neurons in the ventral tegmental area by aversive stimuli.
Van Gisbergen JA, Van Opstal AJ, Tax AA (1987) Collicular ensemble
coding of saccades based on vector summation. Neuroscience
Van Opstal AJ, Van Gisbergen JA (1989) A model for collicular effer-
ent mechanisms underlying the generation of saccades. Brain
Behav Evol 33:90–94.
Villeneuve MY, Casanova C (2003) On the use of isoflurane versus
halothane in the study of visual response properties of single cells
in the primary visual cortex. J Neurosci Methods 129:19–31.
Waleszczyk WJ, Nagy A, Wypych M, Berényi A, Paróczy Z, Eördegh
G, Ghazaryan A, Benedek G (2007) Spectral receptive field prop-
erties of neurons in the feline superior colliculus. Exp Brain Res
Wimborne BM, Henry GH (1992) Response characteristics of the cells
of cortical area 21a of the cat with special reference to orientation
specificity. J Physiol 449:457–478.
Wurtz RH, Hikosaka O (1986) Role of the basal ganglia in the initiation
of saccadic eye movements. Prog Brain Res 64:175–190.
Supplementary data associated with this article can be found, in
the online version, at 10.1016/j.neuroscience.2009.07.031.
(Accepted 15 July 2009)
(Available online 21 July 2009)
A. Berényi et al. / Neuroscience 163 (2009) 1316–1326 1326