Responses of Class R3 Retinal Ganglion Cells of the Frog to Moving Configurational Bars: Effect of the Stimulus Velocity

Comp. Biochem. Physiol. Vol. 119A, No. 1, pp. 387–393, 1998 Copyright  1998 Elsevier Science Inc. All rights reserved.

ISSN 1095-6433/98/$19.00 PII S1095-6433(97)00440-6

Responses of Class R3 Retinal Ganglion Cells of the Frog to Moving Configurational Bars: Effect of the Stimulus Velocity Christophe Beauquin and Fre´de´ric Gaillard Laboratoire de Neurophysiologie, De´partement Neurosciences, CNRS, U.M.R. 6558 Faculte´ des Sciences, 40 Avenue du Recteur Pineau, F-86022 Poitiers, France ABSTRACT. Discrimination of ‘prey’ (bars elongated in the direction of movement; W- or H-orientation) and ‘non-prey’ (bars perpendicular to the direction of movement; A- or V-orientation) stimuli in freely moving amphibians is velocity-invariant. Whether or not this phenomenon is present in cells belonging to a general decision making neuronal process remains questionable. Present investigations report the effect of the angular velocity of the stimulus on the discrimination function of class R3 (transient ON-OFF) retinal ganglion cells. The main conclusions of this work are the following: (1) irrespective of the angular velocity, class R3 neurons always prefer vertically (A-) to horizontally (W-) oriented stripes as long as the stimulus length remains inferior to the receptive field size; (2) this preference for small A-stimuli is best expressed when stimuli are moved at V 5 7.6°/s; (3) a preference reversal is induced by stripes longer than the receptive field via a dual process involving both spatial and temporal mechanisms; (4) this preference reversal is velocity-dependent: the longer the bar, the faster the velocity should be. comp biochem physiol 119A;1:387–393, 1998.  1998 Elsevier Science Inc. KEY WORDS. Frog, retina, ON/OFF ganglion cells, stimulus orientation, velocity

INTRODUCTION For the last twenty years, response characteristics of retinal and tectal neurons in amphibians have been extensively investigated using moving bars elongated either in the direction of movement (W- or H-stimuli), or perpendicular to the direction of movement (A- or V- stimuli), instead of square or disk stimuli. Such electrophysiological studies (6,10,12) led progressively to the description of many new neuronal subclasses, each of them being characterized by a specific ‘form/contrast’ function representing its ability to discriminate prey (W-)- and non-prey (A-)-like targets (29). Moreover, since these ‘configurational’ stimuli were able to induce normal visuo-motor reactions in freely moving animals (4), they proved to be powerful tools in defining the role of each cell category in a general decision-making neuronal process (5,6,7). However, because the discrimination between W- and A- stimuli by toads (Bufo bufo) and frogs (Rana temporaria) was reported to be velocity-invariant (2,6,8), these pioneer neurophysiological studies were all performed at a unique,

standard angular velocity (V 5 7.6°/s). The resulting data were promptly disputed (19,21,25,26). These investigations showed, first, that freely behaving salamanders prefer horizontal (W-) stimuli at low velocities, then vertical (A-) stimuli at high velocities; and second, that many tectal neurons (mainly class T5 neurons), both in salamanders and in toads, express a clear velocity-dependent preference reversal. The effect of the stimulus velocity on the discrimination function of retinal ganglion cells (class R1–R5) was never investigated clearly. Ewert (6) reported only that the configurational preference of class R3 neurons in toads is not altered within the range 5 # V # 18°/s. This study focused on the behavior of class R3 (transient OFF-ON) ganglion cells because these cells (i) provide a major input to class T5 neurons in frogs and toads (28); and (ii) were recently shown to be a useful reference for our studies concerning the functional properties of the ipsilateral retino-tecto-isthmotectal units (1). MATERIALS AND METHODS

Address reprint requests to: Fre´de´ric Gaillard, Laboratoire de Neurophysiologie, De´partement Neurosciences, CNRS, U.M.R. 6558, Faculte´ des Sciences, 40 avenue du Recteur Pineau, F-86022 Poitiers, France. Tel. (33)-05-4945-38-53; Fax (33)-05-49-45-40-14; E-mail: [email protected] poitiers.Fr Received 12 October 1996; accepted 23 July 1997.

Experiments were performed on adult water frogs (Rana esculenta) kept in a vivarium under a normal nycthemeral rhythm 6–14 days before use. After immobilization with an injection of d-tubocurarine (0.5–0.6 mg i.m.) and a local

C. Beauquin and F. Gaillard

388

anaesthesia of the area to be operated on (0.3–0.5 ml Xylocaı¨ne 1% sub-cutaneously), the skull was opened and the optic tecta were surgically exposed. The brain was protected from drying by applying mineral oil. The eyelids were cut off and the corneas were dampened with a drop of normal Ringer. Frogs were then wrapped in moist gauze to ensure an oxygen supply via the skin for the duration of the experimental session. The non-stimulated (left) eye was always covered by an opaque occluder. After surgical preparation, the animals were placed on a micromanipulator platform facing an X-Y plotter used as a stimulation screen (homogeneous white background; luminance Lb 5 7.4 cd/m2). The plotter was set vertically 25 cm away from the animal. The pen holder of the plotter was modified to carry stimulating targets. Electrophysiological investigations were restricted to the superficial layers of the rostro-lateral part of the right optic lobe. This region is known to receive inputs from temporal retinal areas that project to the frontal visual field (15). Action potentials in response to visual stimuli were recorded extracellularly with 2M-NaCl filled micropipettes (o.d. < 5 µm; impedance < 2.5 MΩ at 1000 Hz) and fed onto a classical electrophysiological Neurolog amplifier (Digitimer; cut off frequency 5 300 Hz-3 kHz; gain 5 3 20 000). The amplifier output was connected to an oscilloscope for direct observation and to a computer (Macintosh IIci, Apple) equipped with a multifunction I/O Lab-NB board (National Instruments). Stimulus control, data acquisition, and analysis were all performed using a custom-made computer program (24) developed with the graphical programming system LabVIEW  (V 2.2.1; National Instruments). The depth of the recorded units was calculated from readings on the microelectrode advance driver, where the electrical contact between the electrode and pia mater is reference zero. Once a single unit was isolated, the outer limits of its excitatory receptive field (ERF) were drawn on the screen by manually moving a small 2 3 2° black square. The location of the ERF centre was defined in a Cartesian co-ordinated system where the horizontal and the vertical axes correspond respectively to the projections of the inter-ocular plane and of the midsagittal plane of the animal (1). After a rapid qualitative study for classification, stimulus-response relationships were established quantitatively. These studies concerned (i) the unit response to black bars of constant width (l 5 2°) but of increasing lengths (2° , L , 20°) moved through the ERF at constant angular velocities (V 5 3°/s; V 5 7.6°/s or V 5 15°/s); and (ii) the unit response to black stripes of fixed dimensions (2 3 4°; 2 3 6°; 2 3 8° and 2 3 16°) moved through the ERF at increasing angular velocities (0.76 , V , 45°/s). These bars were oriented (10,27) either in the direction of movement (Fig. 2; Wconfiguration) or perpendicular to the direction of movement (Fig. 2; A-configuration). The contrast between stimuli and background was held constant [|C| 5 (Lb 2 Ls)/

(Lb 1 Ls) < 0.75]. Each stimulus was presented 1–3 times per single session and in a pseudo-random order. Successive presentations were separated by a 90–120s interval to prevent neuronal habituation. For technical reasons, all the stimuli were moved vertically upwards. It is worth noticing that most of the visual neurons in anurans, including class R3 retinal ganglion cells, respond slightly better (although not enough to be statistically significant (11)) to vertical than to horizontal movements. Unless otherwise specified, discharge frequencies in each unit were calculated as usual, from the first to the last recorded spike. Data were then used to examine the ability of class R3 units to discriminate Aand W-oriented bars of equal dimension using a ‘contrastlike’ formula D (W,A ) 5 [R (W ) 2 R (A)]/[R (W ) 1 R(A)] where R (W ) is the discharge frequency for a given W-oriented stimulus and R (A) is the discharge frequency for an equivalent but A-oriented stimulus (9). Differences between the responses were tested for significance by means of the Student’s t-test. RESULTS Visual units investigated in the present study (N 5 121) were all considered as typical class R3 retinal units on the basis of their responses to usual qualitative visual tests (16,22). They were mainly located between 250 and 450 µm below the tectal surface. Receptive fields (determined by moving manually a 2° black square) were about 4.6° 6 0.9° in diameter. These units were not spontaneously active. They responded well to black targets 1–2° in diameter moved on a uniform white background in any direction and at any angular velocity. They did not habituate to moving stimuli presented at a rate of about F 5 1/s. They did not respond to stationary, movement-gated stimuli. They always responded briefly to sharp changes in the background illumination. In half of the units, the OFF response (latency < 50ms) was judged stronger than the ON response (latency < 100ms). Visual units showing ambiguous (‘R3like’) functional properties were discarded. Eighteen of these typical R3 units were stimulated with W- and A-stimuli moved at V 5 7.6°/s, the basic angular velocity. Results are shown in Fig. 1A (open and filled circles). In short, responses to W-stimuli remain stable (29 , R (W ) , 34 imp./s.) as the length of the bar is increased from L 5 2° to L 5 20°. On the other hand, the firing rate increases up to a maximum of R (A ) < 42 imp./s, and then decreases continuously when A-stimuli are used. The optimal stimulus length computed from the usual area/length function [R (A) 5 k(L) 6 (a1,a2).log L; imp./s; (3,10)] is reached for L(opt) < 5.5°. Responses obtained with A- and W-stimuli of equal dimension are statistically different (p # 0.01; two-tailed t-test) except for L 5 8° and L 5 12°. The corresponding discrimination curve (Fig. 1B) shows a negative peak (D (W,A ) 5 20.13) for L 5 6°, then becomes

Velocity Dependence in Frog Retina

FIG. 1. Effect of the velocity on the configurational sensitiv-

ity of class R3 retinal neurons (N 5 55). (a) The average firing rate (6sem) is plotted against the bar length (degrees). Data obtained at 3 different angular velocities: V 5 3°/s (N 5 23), V 5 7.6°/s (N 5 18), V 5 15°/s (N 5 14). Open symbols: W-stimuli; Filled symbols: A-stimuli. (b) For bars of equal dimension, the discrimination factor D (W,A ) 5 [R( W ) 2 R( A )]/[R( W ) 1 R( A )] was calculated and plotted against the bar length.

positive as the stimulus length is increased. The preference reversal between W- and A-stimuli is observed for L(rev) < 10°. Present data are similar to previous ones (1,27). Twenty three further units were stimulated with W- and A- stimuli of increasing length moved at V 5 3°/s (i.e. about one half the preceding velocity). Corresponding curves are shown in Fig. 1A (open and filled triangles). The main results can be summarized as follows. Discharge rates to W-stimuli remain constant (R (W ) < 16 imp./s), independent of the length of the bar. Responses to A-stimuli show

389

a slight increase up to R (A ) < 19 imp./s for L < 4–6° and then a regular decrease. The optimal stimulus length computed from the usual area/length function was found to be L(opt) < 5°. In this experimental situation, responses to Aand W-stimuli of equivalent dimension were found to be statistically different (p # 0.01; two-tailed t-test) only for bars superior to 12° in length. The corresponding discrimination curve (Fig. 1B) levels off in the negative phase (D (W,A) < 20.07; L < 6°), then becomes strongly positive when the stimulus length is increased. The preference reversal between A- and W-stimuli is observed for L(rev) < 7°. Finally, fourteen units were stimulated with W- and Astimuli of increasing length moved at V 5 15°/s (i.e. twice the routine velocity). Corresponding curves are shown in Fig. 1A (open and filled squares). For this angular velocity, the relationship between the average firing rate of class R3 ganglion cells and the length of W-stimuli is clearly variable. It shows a net peak (R (W ) 5 83.5 6 10 imp./s) for bars 4° in length. Average firing rates do not differ statistically except for this specific value (p # 0.05; two-tailed t-test). In response to A-stimuli, the average firing rate shows a rounded plateau for 4 , L , 8° (differences between the corresponding values are not statistically different; p . 0.05; two-tailed t-test) and then a continuous decrease. The optimal target dimension computed from the usual area/length function was found to be L(opt) < 6.0°. Responses elicited by A- and W-stimuli of equivalent dimension were found to be statistically significant only for bars superior to 16° in length. The corresponding discrimination curve (Fig. 1B) shows a very flat negative phase (D (W,A ) < 20.06; L 5 6°), then becomes steadily positive when the stimulus length is increased. The preference reversal between A- and W-stimuli is observed for L(rev) < 12°. From these experiments, it can be stated that increasing the angular velocity (from 3 to 15°/s) of moving A- and W-stimuli modifies only slightly the remarkable L(opt) value (from 5° to 6°), but induces a noticeable shift in the point of preference reversal L(rev) (from 7° to 12°) as well as a strong flattening of the discrimination curve. It can also be reported that the maximum preference for A-stimuli is obtained when these stimuli are specifically moved at V 5 7.6°/s, whereas the strongest preference for W-stimuli is expressed at best when V 5 3°/s. These statements are supported by results (Figs. 2,3) obtained in one particularly successful unit. This unit (i) was located about 430 µm below the tectal surface; (ii) displayed typical properties of class R3 ganglion cells; and (iii) followed the velocity of a stimulus according to a power function [R 5 k(v). Vc; imp./s; k(v) 5 8.7; (c) 5 0.82; 0.76 , V , 45°/s; 2° black square; (16)]. Comparison of the discharge rates to both the leading and trailing edges of long (L . 6°) W-stimuli suggested furthermore that this unit was OFFdominated. Whatever the velocity, the ON response was about one half the OFF response (R (ON) /R (OFF ) < 0.5–0.6;

390

C. Beauquin and F. Gaillard

FIG. 2. Typical discharge profiles recorded in one particu-

larly successful class R3 retinal ganglion cell when stimulated with 6° and 20° W- and A-oriented bars moved at V 5 3°/s (left column) and V 5 15°/s (right column). For each recording session, the original (upper row) and the corresponding digitalized (lower row) spike trains are shown. Time scale (thick lines) 5 1s.

Fig. 2). This class R3 unit was then stimulated with Wand A-oriented bars of increasing length moved at the three specific angular velocities used in this work. Typical recordings are presented in Fig. 2 and the related experimental curves are shown in Fig. 3A, B. These curves are similar in shape to those presented in Fig. 1, the only differences being the absolute values of the noteworthy points, mainly because the ERF of this cell was found to be rather small (< 3.5°). In this particular unit, L(opt) remains constant (around 4°) while L(rev) increases from 6° to 8° as the stimulus velocity is increased. The maximum negativity of D (W,A ) is again obtained at V 5 7.6°/s. The strongest preference for W-stimuli is also expressed when V 5 3°/s. In order to get a more extensive view of the effect of the stimulus velocity on the discrimination sensitivity of class R3 neurons, D (W,A ) values obtained Fig. 1B for each length were plotted against the experimental angular velocities and approximated by curve fitting for a very low (V 5 0.76°/s) and a rather high (V 5 25°/s) velocity. The best fit was always obtained with a second order polynomial regression. Results are presented in Fig. 4A, the curve obtained for V 5 7.6°/s being taken as the reference. Our estimation indicates: (i) for a very low stimulus velocity (V 5 0.76°/s), the discrimination factor is negligibly small for bars inferior to the ERF, then becomes strongly positive for bars longer than the ERF diameter; (ii) for a rather high velocity (V 5 25°/ s), the discrimination curve shows a strong positive phase

FIG. 3. Quantitative stimulus-responses relationships de-

rived from the preceding recordings. (A) Effect of W-stimuli (open symbols) and A-stimuli (filled symbols) of increasing length on the unit discharge. (B) Corresponding discrimination curves.

for stimuli smaller than the ERF, the reversal point being obtained for L < 12°. Class R3 cells would thus undoubtedly prefer long W-oriented stripes at low velocity and small Woriented stripes at high velocity. These estimations have been compared to physiological data obtained on 66 class R3 units (4.5° , ERF , 5.5°). These units were stimulated with A- and W-stimuli of constant lengths 2 3 4° (N 5 15), 2 3 6° (N 5 15), 2 3 8° (N 5 13) and 2 3 16° (N 5 23) moved between 0.76 and 45°/s (velocity function studies; see Methods section). D (W,A ) values were then calculated as usual. As a result, curves (Fig. 4B) obtained for our standard angular velocities (V 5 3, 7.6 and 15°/s) look essentially similar to those presented earlier (Figs. 1B, 3B), the only differences being the absolute values of the discrimination factor. The maximum

Velocity Dependence in Frog Retina

391

oriented stimuli smaller than the ERF. Values of D (W,A ) remain near zero, becoming negative for bars longer than the ERF. DISCUSSION

FIG. 4. Effect of the angular velocity of the stimulus on the

discrimination factor: expected and physiological data. (A) The discrimination values obtained Fig. 1B were plotted as a function of the angular velocity and first approximated by curve fitting for two unused velocities, V 5 0.76°/s and V 5 25°/s. Resulting D(W,A ) values were then graphed as usual. (B) Class R3 ganglion cells (N 5 66) were stimulated with both A- and W-stimuli of different lengths moved at increasing velocities (0.76 , V , 45°/s; see text). D(W,A ) values were calculated for bars of equal dimensions and replotted as usual. Curves resemble strongly those obtained Fig. 3B. Note that preference reversal occurs only when stimuli are longer than the ERF.

negativity for D (W,A ) is obtained with bars < 4° in length (L/ERF < 0.8–0.9) independent of the angular velocity. The reversal preference increases from L < 5° for V 5 3 to L < 8 for V 5 15°/s. The maximum negativity is obtained with V 5 7.6°/s. The highest preference for long Woriented bars is obtained at the lowest angular velocity (V 5 0.76°/s). However, moving stimuli at V 5 35°/s do not produce the expected sharp preference reversal for W-vs A-

The major question in neuroethology is whether the recognition of prey/predator-like objects results from the activity of dedicated ‘feature detector’ neurons, or from the interaction of multiple neurons each processing a different aspect of the stimulus. Since the pioneer work of Lettvin et al. (20), a considerable amount of data concerning the functional properties of retinal and tectal neurons in amphibians has been collected (6,16). In the past few years, the extensive use of moving bars having a specific configuration with respect to the direction of movement led to the description of various neuronal subtypes, some of them being especially sensitive to ‘worm-like targets’ [T5(1)-T5(2) neurons; (12)]. Whether this neuronal specificity remains invariant with the experimental conditions (mainly the angular velocity of the stimulus) remains an important question. Some authors (19,25,26) have reported that a large population of the tectal visual neurons recorded in both anurans and salamanders reverse their preference for configurational stimuli at medium or high angular velocities. In Salamandra salamandra as well as in Hydromantes italicus, some tectal neurons were found to show a clear preference inversion between A- (V-) and W- (H-) oriented bars (2 3 8° in length) for 6 # V # 20°/s (19,25). A similar preference inversion between squares and W-(H-) stimuli was observed in about one third of tectal neurons of Bufo bufo (26). By contrast, behavioral experiments have consistently shown that the preference for ‘worm-’ vs ‘antiworm-like’ targets in these animals was independent of the stimulus velocity [within a large range: 3.5 # V # 64°/s, (4,8,13)]. Ewert and colleagues (13) therefore estimated that, ‘‘this [neuronal] velocity-dependent preference inversion was of a limited value from a neuroethological point of view,’’ and consequently performed their neurophysiological investigations at a unique angular velocity [V 5 7.6°/s; (6,12,27)]. They never demonstrated clearly that the discrimination curves of the investigated neurons were velocity-invariant. Our results clearly show that increasing the angular velocity of bars oriented either in the direction of movement (W-stimuli) or perpendicular to the direction of the movement (A-stimuli) undoubtedly affects the shape of the discrimination curve of class R3 retinal ganglion cells. Whereas the maximum negative value of D (W,A ) is always found for stimuli 5–6° in length (i.e. for stimuli approximately equal to the ERF size), the point of preference reversal is shifted toward larger values and the slope of the positive phase of the discrimination curve is flattened when the stimulus velocity is increased. Moreover, class R3 neurons exhibit a net preference inversion between A- and W-oriented bars in specific experimental conditions. When bars

C. Beauquin and F. Gaillard

392

either inferior or equal to the ERF (L # 5°; Fig. 4B) are used, the configurational preference is preserved, irrespective of the angular velocity of the stimulus (at least within the range 0.76 # V # 35°/s). In these conditions, R (A) is always superior to R (W ) and the corresponding D (W,A) values always remain negative. Consequently, A-oriented stripes are always preferred to W-oriented stripes. However, when bars longer than the ERF are used, the situation is slightly more complex: W-oriented bars are preferred up to a critical angular velocity for which a preference reversal occurs. This critical angular velocity may be related to the bar length (Figs. 3B, 4B): the preference reversal is found at V < 3– 5°/s for L 5 6°, at V < 10–15°/s for L 5 8° and at V < 15–20° for L 5 16°. These values compare favourably with those reported above (19,25). Our data can be explained rather simply considering the spatial and temporal properties of the constitutive elements of the receptive field of class R3 neurons. First, the receptive field of retinal neurons in amphibians is made (in a first approximation) of a central excitatory zone (ERF) surrounded by a purely inhibitory area (IRF). Secondly, class R3 neurons respond essentially to the extension of a given stimulus perpendicular to the direction of movement (10,16,27). Thirdly, it was assumed some years ago (3,16,18) that the time constant of the elements responsible for lateral inhibition in the retinal network is greater than that of the excitatory elements. Finally, the inhibitory process is delayed by about 40–80ms with respect to the central excitatory process (23). With bars inferior to the ERF, a preference reversal will never occur simply because, in this experimental situation, the neuronal activity depends essentially on the spatial summation of the elements forming the ERF. The leading edge of a given A-oriented bar will always activate more elements of the ERF than a similar, but W-oriented, bar. As a consequence, D (W,A ) values will be always negative. These values will only decay (become less negative) with increasing angular velocities, the stimuli being progressively less efficient to activate the elements of the ERF. If we assume a time constant of about 50–100ms for the response of class R3 neurons (14,17), D (W,A) values for small bars (L , ERF) would be null for V < 50–100°/s. At such a velocity, responses of class R3 neurons usually become saturated (14). An unsuspected result of this study is, however, the strong and statistically significant preference observed for such small bars (L 5 4–6°, Fig. 1B) specifically moved at V 5 7.6°/s. The presence of an experimental artefact can be ruled out. This phenomenon was observed in our two sets of experiments (Figs. 1,4) performed one year apart and using two distinct protocols. Rather, it seems to be clearly related to some peculiar temporal characteristics of the elements forming the ERF. One possibility would be that OFF and ON responses elicited respectively by the leading and the trailing edge of the 2° A-stimuli would be added to-

gether for this specific angular velocity, thus producing an increase in the rate of the neuronal discharge. When bars longer than the ERF are used, inhibitory elements forming the surrounding IRF are strongly stimulated, mainly by the leading edge of A-oriented stripes. Low angular velocities (V , 5°/s) allow these inhibitory mechanisms to develop fully so that neuronal responses induced by long (L 5 12–16°) A-stimuli are systematically inferior (< 30%; Fig. 1A) to those produced by standard 2 3 2° targets [see also (16,18)]. Responses to similar but W-oriented bars being equal to those produced by the reference 2° target, D (W,A ) is positive. At high angular velocities (for instance V . 15°/s), the power of the surrounding inhibition on the central excitation decreases (i) because of the slow response dynamics of the elements forming the IRF, and (ii) because this inhibition is delayed more and more in relation with the response of the central excitatory elements. Responses to A-oriented bars are progressively ‘disinhibited’ and consequently D (W,A ) becomes negative. The main conclusions of this work are the following: (1) irrespective of the angular velocity, class R3 neurons always prefer vertically (A-) to horizontally (W-) oriented stripes as long as the stimulus length remains inferior to the ERF size; (2) this preference for small A-stimuli is best expressed when stimuli are moved at V 5 7.6°/s; (3) a significant preference inversion for such stimuli could never occur because the neuronal response is directly linked to the mechanisms of spatial summation within the excitatory field; (4) preference reversal is only induced by stripes longer than the ERF via a dual process involving both spatial and temporal mechanisms; (5) this preference reversal is velocitydependent: the longer the bar, the faster the velocity should be. Therefore, present data ask two questions for further experiments: (i) ‘‘is the ‘velocity-dependent’ classification of the tectal neurons valid whatever the bar length ?’’ (13), and (ii) ‘‘how is the configurational sensitivity of tectal neurons involved in prey-catching behavior when stimuli are moved at a high velocity (V 5 20–40°/s) known to induce maximum orienting responses in toads (8)?’’ The authors wish to thank Miss G. McDermott for reviewing the English.

REFERENCES 1. Beauquin, C.; Poindessault, J-P.; Gaillard, F. Responses of ipsilateral retino-tectal (type I1) units of the frog (Rana esculenta) to moving configurational bars. Comp. Biochem. Physiol. 111A:561–568;1995. 2. Burghagen, H.; Ewert, J-P. Influence of the background for discriminating object motion from self-induced motion in toads Bufo bufo (L.). J. Comp. Physiol. 152:241–249;1983. 3. Butenandt, E.; Gru¨sser, O-J. The effect of stimulus area on the response of movement detecting neurons in the frog’s retina. Pflu¨gers Arch. Ges. Physiol. 298:238–293;1968. 4. Ewert, J-P. Quantitative Analyse von Reiz-Reaktionsbezie-

Velocity Dependence in Frog Retina

5.

6.

7. 8. 9.

10. 11.

12.

13. 14.

15. 16. 17.

hungen bei visuellen Auslo¨sen der Beutefang-Wendereaktion der Erdkro¨te (Bufo bufo L.). Pflu¨gers Arch. 308:225–243;1969. Ewert, J-P. Neuroethological analysis of the innate releasing mechanism for prey-catching behavior in toads. In: Ewert, J-P.; Capranica, R.R.; Ingle, D.J. (eds). Advances in Vertebrate Neuroethology. New York: Plenum Press; 1983:413– 475. Ewert, J-P. Tectal mechanisms that underlie prey-catching and avoidance behaviors in toads. In: Vanegas H. (ed). Comparative Neurology of the Optic Tectum. New York: Plenum Press; 1984:247–416. Ewert, J-P. Neuroethology of releasing mechanisms: Prey catching in toads. Behav. Brain Sci. 10:337–405;1987. Ewert, J-P.; Arend, B.; Becker, V.; Borchers, H-W. Invariants in configurational prey selection by Bufo bufo (L.). Brain Behav. Evol. 16:38–51;1979a. Ewert, J-P.; Borchers, H-W.; Wietersheim A.v. Question of prey feature detectors in the toad’s Bufo bufo (L.) visual system: A correlation analysis. J. Comp. Physiol. 126:43–47; 1978. Ewert, J-P.; Hock, F.J. Movement-sensitive neurones in the toad’s retina. Exp. Brain Res. 16:41–59;1972. Ewert, J-P.; Krug, H.; Scho¨nitz, G. Activity of retinal class R3 ganglion cells in the toad Bufo bufo (L.) in response to moving configurational stimuli: Influence of the movement direction. J. Comp. Physiol. 129:211–215;1979b. Ewert, J-P.; Wietersheim, A.v. Musterauswertung durch Tectum- und Thalamus/Praetectum-Neurone im visuellen System der Kro¨te Bufo bufo (L.). J. Comp. Physiol. 92:131– 148;1974. Finkensta¨dt, T.; Ewert, J-P. Processing of area dimensions of visual key stimuli by tectal neurons in Salamandra salamandra. J. Comp. Physiol. 153:85–98;1983. Gaillard, F.; Corbacho, F.; Arbib, M-A. A model of the physiological responses of ON-OFF anuran ganglion cells. In: Paugam-Moisy, H.; Royet, J-P.; Zighed, D.A. (eds). Le Neuromime´tisme. Paris: Herme`s; 1995:155–170. Gaze, R.M.; Jacobson, M. The projection of the binocular visual field on the optic tectum of the frog. Q.J. Exp. Physiol. 47:273–280;1962. Gru¨sser, O-J.; Gru¨sser-Cornehls, U. Neurophysiology of the anuran visual system. In: Llinas, R.; Precht, W. (eds). Frog Neurobiology. Berlin: Springer Verlag; 1976:297–385. Gru¨sser, O-J.; Finkelstein, D.; Gru¨sser-Cornehls, U. The ef-

393

18. 19. 20. 21.

22. 23. 24.

25. 26. 27. 28.

29.

fect of the stimulus velocity on the response of movement sensitive neurons of the frog’s retina. Pflu¨gers Archiv. 300: 49–66;1968. Gru¨sser-Cornehls, U. Neurophysiological properties of the retinal ganglion cell classes of the cuban treefrog, Hyla septentrionalis. Exp. Brain Res. 73:39–52;1988. Himstedt, W.; Roth, G. Neuronal responses in the tectum opticum of Salamandra to visual prey stimuli. J. Comp. Physiol. 135:251–257;1980. Lettvin, J-Y.; Maturana, H.R.; McCulloch, W.S.; Pitts, W.H. What the frog’s eye tells the frog’s brain. Proc. Inst. Radio Eng. 47:1940–1951;1959. Luthardt, G.; Roth, G. The relationship between stimulus orientation and stimulus movement pattern in the prey-catching behavior of Salamandra salamandra. Copeia 1979:442–447; 1979. Maturana, H.R.; Lettvin, J.Y.; McCulloch, W.S.; Pitts, W.H. Anatomy and physiology of vision in the frog. J. Gen. Physiol. 43:129–175;1960. Nye, P.W.; Naka, K-I. The dynamics of the inhibitory interaction in a frog receptive field: A paradigm of paracontrast. Vision Res. 11:377–392;1971. Poindessault, J-P.; Beauquin, C.; Gaillard, F. Stimulation, data acquisition, spikes detection and time/rate analysis with a graphical programming system: An application to vision studies. J. Neurosci. Meth. 59:225–235;1995. Roth, G. Responses in the optic tectum of the salamander Hydromantes italicus to moving prey stimuli. Exp. Brain Res. 45:386–392;1982. Roth, G.; Jordan, M. Response characteristics and stratification of tectal neurons in the toad Bufo bufo L. Exp. Brain Res. 45:393–398;1982. Schu¨rg-Pfeiffer, E.; Ewert, J-P. Investigation of neurons involved in the analysis of gestalt prey features in the frog Rana temporaria. J. Comp. Physiol. 141:139–152;1981. Tsai, H.J.; Ewert, J-P. Edge preference of retinal and tectal neurons in common toads (Bufo bufo) in response to wormlike moving stripes: The question of behaviorally relevant ‘position indicators.’ J. Comp. Physiol. 161:295–304;1987. Wietersheim, A.v.; Ewert, J-P. Neurons of the toad’s (Bufo bufo L.) visual system sensitive to moving configurational stimuli: A statistical analysis. J. Comp. Physiol. 126:35–42; 1978.