HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Perlman, I. Articles by Alpern, M. Search for Related Content PubMed PubMed Citation Articles by Perlman, I. Articles by Alpern, M.

Field sensitivity action spectra of cone photoreceptors in the turtle retina

Ido Perlman, Aviran Itzhaki, Husam Asi and Mathew Alpern

MS 7741 Received 23 December 1997; accepted after revision 21 May 1998.

	ABSTRACT

Top Abstract Introduction Methods Results Discussion References

1. The Stiles two-colour increment threshold technique was applied to turtle cone photoreceptors in order to derive their field sensitivity action spectra.

2. Photoresponses of cone photoreceptors were recorded intracellularly. Flash sensitivities were calculated from small amplitude (< 1 mV) responses. The desensitizing effects of backgrounds of different wavelengths were measured and the background irradiance needed to desensitize the cone by a factor of 10 (1 log unit) was defined as threshold. The reciprocals of these thresholds were used to construct the field sensitivity action spectrum.

3. The field sensitivity action spectra of long-wavelength-sensitive (L) and medium-wavelength-sensitive (M) cones depended upon the wavelength of the test flash used to measure them. This excludes the possibility that turtle cones can function as single-colour mechanisms in the Stiles sense.

4. In fourteen L-cones, the average wavelength of peak sensitivity of the field sensitivity action spectrum was 613·7 ± 7·7 nm for the 500 nm test and 635·6 ± 9·6 nm for the 700 nm test. For six M-cones, these values were 558·5 ± 6·8 and 628·8 ± 10·6 nm for the 500 and 700 nm tests, respectively.

5. Two physiological mechanisms are suggested as contributing to the dependency of the field sensitivity action spectrum upon test wavelength. One is based upon the transmissivity properties of the coloured oil droplets, while the other hypothesizes excitatory interactions between cones of different spectral type.

6. Computer simulations of the field sensitivity action spectra indicate that both mechanisms are needed in order to account for the dependency of the field sensitivity action spectrum upon the wavelength of the test flash.

	INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

Colour perception starts with photon absorptions in at least two different spectral types of cone photoreceptors but depends on a complex network of neuronal interactions at every level of the visual system. Already in second-order neurons, the horizontal cells of cold-blooded vertebrates (e.g. turtle and fish), colour-coded photoresponses can be measured (Kamermans & Spekreijse, 1995; Piccolino, 1995).

Turtle cone photoreceptors represent an extreme case in which the light-induced voltage responses differ from those expected from the absorption spectra of the visual pigments. The action spectra of cone photoreceptors, determined from intracellular recordings in the eyecup preparation, are affected by the transmission properties of the coloured oil droplets located in the cone inner segments (Baylor & Hodgkin, 1973; Schneeweis & Green, 1995), by negative feedback pathways from horizontal cells (Fuortes et al. 1973; Perlman et al. 1994) and possibly by excitatory interactions between cones belonging to different spectral types (Normann et al. 1984, 1985; Itzhaki et al. 1992).

The Stiles two-colour increment threshold technique has been used to circumvent post-receptoral neuronal interactions in order to derive the action spectra of the fundamental colour mechanisms (Stiles, 1939, 1949). In this procedure, the background irradiance needed to desensitize the response to a fixed test stimulus by a predetermined criterion (10-fold), is defined as threshold. The relationship between threshold and background wavelength is the field sensitivity action spectrum that has been used to define the spectral properties of the colour mechanism (termed the mechanism) underlying perception of a stimulus of specific wavelength.

In this study, we applied the Stiles two-colour increment threshold technique to turtle cone photoreceptors and derived their field sensitivity action spectra for 700 and 500 nm test flashes. Our goal was to verify whether the light-induced voltage response of a single cone represented a fundamental colour mechanism in the Stiles sense, and, if so, to verify whether the field sensitivity action spectrum was related to the absorption spectrum of the visual pigment. We found that the photoresponses of a single turtle cone did not obey the principle of univariance and did not represent a fundamental colour mechanism in the Stiles sense.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Preparation

Photoresponses of cone photoreceptors were recorded intracellularly from the eyecup preparation of the turtle Mauremys caspica. The turtle was decapitated with a guillotine and the head pithed. The eyecup was prepared according to procedures that have been previously described in detail (Baylor & Hodgkin, 1973; Itzhaki & Perlman, 1984). Briefly, the eye was hemisected with a razor blade and the anterior part was discarded. After removing the vitreous humour, the posterior eyecup was placed in a chamber, surrounded by moist cotton to minimize retinal dehydration. A continuous flow of 95 % O₂-5 % CO₂ mixture, saturated with water vapour, was directed above the preparation. All experiments were performed in accordance with institutional guidelines.

Data acquisition

Microelectrodes were pulled (P87, Sutter Instruments, Novato, CA, USA) from capillary tubing and filled with 3 M potassium acetate solution. Microelectrode resistance was 150-300 M. The signal recorded by the microelectrode was amplified (Almost Perfect Electronics, Basel, Switzerland) and then displayed on an oscilloscope screen and monitored by a pen recorder (Gould 2200, Cleveland, OH, USA). The amplified signal was continuously digitized and stored on the hard disk of a personal computer equipped with a Labmaster DMA data acquisition board (Scientific Solutions, Solon, OH, USA).

Optical system

Two light channels originated from a single light source (250 W tungsten halogen filament). One beam served for test stimuli and the other for continuous background illumination. The intensity and spectral content of each beam were independently controlled by sets of neutral density and narrow-band interference filters. The size of the retinal area illuminated by each of the light beams was independently determined by a set of apertures. The duration of the light stimuli and the inter-stimulus interval were controlled by the computer, which operated an electronic shutter (Vincent Electronics, Rochester, NY, USA). The irradiances of the test and background channels were measured with a PIN 10 photodiode (United Detector Technology, Orlando, FL, USA) and calculated in photons s^-1 µm^-2.

Procedure and analysis

Cones were identified by the following criteria (Baylor & Hodgkin, 1973; Fuortes & Simon, 1974; Perlman et al. 1994): (1) they responded with graded hyperpolarization to bright light stimuli of any wavelength; (2) maximal photoresponse was smaller than 25 mV; (3) receptive field sizes were about 150 µm in diameter. The spectral type of cones (long (L)-, medium (M)- or short (S)-wavelength-sensitive) was defined by their responsiveness to light stimuli of different wavelengths.

Field sensitivity action spectra were measured as closely as possible to the method used to measure the mechanisms in human observers (Stiles, 1939, 1959). Test flashes (50 ms duration) of wavelength that covered a small (110 µm in diameter) retinal area, centred on the impaled cone, were used for measurements of flash sensitivity. Flash sensitivities were calculated from small amplitude (< 1 mV) photoresponses that were within the linear range of the cones. In order to improve signal/noise ratio, these responses were obtained by averaging twenty responses elicited by identical flashes delivered at a rate of 1 Hz. Flash sensitivity was calculated by dividing the amplitude of the photoresponse by the quantal content of the light stimulus, and is given here in units of µV photon^-1 µm². Two wavelengths were used for test flashes: 500 and 700 nm in order to separate the medium- and long-wavelength-sensitive visual pigments. In experiments on S-cones, 430 nm test flashes were also used to study the turtle equivalent of ₁ or ₃ (human blue mechanisms). Large diameter (2800 µm) background lights of different wavelengths (µ), spanning the entire visible range (400-700 nm) and of different intensities, were used to desensitize the cones and to construct the sensitivity-background irradiance curves (equivalent to Stiles' threshold vs. irradiance curves). From these curves, the background irradiances needed to desensitize the cone by a factor of 10 relative to the dark-adapted state were obtained.

Stability of recordings

In satisfactory experiments, the cone remained reasonably stable as judged by frequent measurements of dark-adapted flash sensitivity and by continuous monitoring of dark membrane potential. Figure 1 illustrates one such experiment on an L-cone that was studied for more than 2 h. The data points describe dark-adapted flash sensitivities to 700 and 500 nm tests and their ratios. As expected from the action spectra of turtle L-cones (Baylor & Hodgkin, 1973; Perlman et al. 1994; Schneeweis & Green, 1995), the dark-adapted sensitivity to 700 nm was higher than for 500 nm by more than 1 log unit. In this particular L-cone, the sensitivity to 700 nm was slightly more stable than that to 500 nm, causing a slight reduction in the 700 nm/500 nm sensitivity ratio. In order to account for such variability, the desensitizing effect of each background was calculated relative to the geometrical mean (algebraic mean of log sensitivity) of the dark-adapted sensitivities measured immediately before and after the background was applied.

		View larger version [in this window] [in a new window]
Figure 1. Stability of intracellular recordings of one L-cone Dark-adapted flash sensitivities were measured with 700 and 500 nm test lights ( and , respectively) at different times throughout the experiment. The ratio between these two sensitivities () showed a slight decline with time. In this particular L-cone, intracellular recording was made for more than 2 h.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Long-wavelength-sensitive cones (L-cones)

Figure 2 illustrates a Stiles two-colour increment threshold experiment in a turtle L-cone. The desensitizing effects of two background lights (700 and 560 nm) on the photoresponses elicited by test flashes of 700 and 500 nm are shown. The dark-adapted sensitivities of the L-cone to these two test flashes differed by about 1 log unit. In order to allow comparison between the effects of each background on the two test flashes, the light-adapted sensitivities were normalized to the dark-adapted values. Accordingly, the desensitizing effects of the backgrounds are described in Fig. 2 as normalized sensitivities.

The desensitizing effects of the two background lights (700 and 560 nm) on the 500 nm and the 700 nm tests are shown in Fig. 2A and B, respectively. At first glance, the normalized sensitivity-background irradiance curves of the L-cone reveal a striking similarity to Stiles' threshold versus irradiance (t.v.i.) curves (Stiles, 1949, 1959). The curves for the two different background wavelengths have a similar shape, and only differ in their location along the log background irradiance axis. However, when the data have been replotted in a manner to compare the desensitizing effects of each background wavelength on both test wavelengths (Fig. 2C and D), a clear deviation from Stiles' psychophysical experiments becomes evident. According to Stiles' displacement rules, once the t.v.i. curves for different test wavelengths but for the same background wavelength have been displaced vertically to account for differences in the dark-adapted sensitivity, they should coincide if a single-colour mechanism underlies the generation of the photoresponses. This is clearly not the case for the t.v.i. curves of the turtle L-cone. The red (700 nm) background is more efficient in desensitizing the 700 nm test flash than the 500 nm test flash (Fig. 2C), while the green (560 nm) background is more effective in desensitizing the L-cone when a 500 nm test flash is used compared with the 700 nm test flash (Fig. 2D).

		View larger version [in this window] [in a new window]
Figure 2. The desensitizing action of 700 and 560 nm background lights on an L-cone measured with two test wavelengths: 700 and 500 nm For each background (b/g) irradiance, the sensitivity was normalized relative to the dark-adapted sensitivity as described in the Methods section. Log sensitivity losses by the 700 nm (circles) and 560 nm (squares) backgrounds are described for the 500 nm test (open symbols) and 700 nm test (filled symbols) in A and B, respectively. The same data are replotted as the desensitization of both test wavelengths by the 700 nm background (C) and by the 560 nm background (D). The data points were fitted by eye (dotted curves) to the Stiles curve (Wyszecki & Stiles, 1982).

It was previously shown that the degree of desensitization of single turtle cones, measured with an intracellular microelectrode, by a background light could be quantified by a relationship similar to that used to describe human psychophysical data (Normann & Anderton, 1983; Burkhardt, 1994). Therefore, the data describing the desensitizing effects of the 700 and 560 nm backgrounds on the 700 and 500 nm test flashes (Fig. 2) were fitted by eye to the Stiles standard template (Wyszecki & Stiles, 1982) (dotted curves in Fig. 2). The deviations of data points from the curves are assumed to reflect measurement errors.

Measurement errors make judgement decisions in curve fitting with few background levels difficult. On the other hand, to achieve sensitivity measurements at four different background irradiances at several different background wavelengths requires stable intracellular recordings for about 2 h. In preliminary experiments, we studied several L-cones (n = 7) with many levels (7-11) of different irradiances for a given background wavelength in order to convince ourselves that the sensitivity vs. background irradiance curve for turtle cones could be described by the Stiles' template. We dealt with the above dilemma by reducing the number of background irradiances to four (n = 4), then to three (n = 5) and finally to two and even one (n = 6). In the latter cases, we attempted to use backgrounds that reduced the sensitivity by about 0·6-0·7 log units compared with the dark-adapted case.

The fitted t.v.i. curves were used to derive the field sensitivity. We followed Stiles' procedure and defined the field sensitivity as the reciprocal of the background irradiance that reduced the sensitivity to the test flash by 1 log unit (a factor of 10), relative to the dark-adapted sensitivity. For the L-cone described in Fig. 2, the irradiances of the 560 and 700 nm backgrounds needed to desensitize the photoresponse to the 700 nm test flash by a factor of 10 are 6·41 log photons (560 nm) s^-1 µm^-2 and 6·01 log photons (700 nm) s^-1 µm^-2, respectively. The corresponding values for the 500 nm test are: 5·92 log photons (560 nm) s^-1 µm^-2 and 6·32 log photons (700 nm) s^-1 µm^-2.

Several backgrounds of different wavelengths, spanning the visible spectrum (400-700 nm), were used to desensitize the L-cone photoresponse elicited by the 700 and 500 nm test flashes. Normalized sensitivity vs. background irradiance curves, similar to those shown in Fig. 2, were plotted and fitted to Stiles' function. From these curves, the background irradiances needed to depress the dark-adapted sensitivity by a factor of 10 were derived. The reciprocals of these irradiances define the field sensitivities and are plotted in Fig. 3 as a function of background wavelength. These are the field sensitivity action spectra of the L-cone for 500 and 700 nm tests.

Neither of these action spectra is similar to the absorption spectrum of a visual pigment, probably due to the transmission properties of the oil droplet located in the inner segment of the L-cone (Baylor & Hodgkin, 1973; Lipetz, 1985; Schneeweis & Green, 1995). In order to define the field sensitivity action spectra, we fitted a parabola to the bathochromic (long-wavelength) band of the spectrum that included the region of maximum sensitivity (Mitchell & Rushton, 1971) using a root mean square minimizing routine. The computer algorithm allowed specification of the shortest background wavelength at which curve fitting started in order to minimize the contribution of the oil droplets and to obtain the smallest root mean square. The wavelength for starting parabola fitting was usually 500-560 nm for L-cones and 500-520 nm for M-cones. For the L-cone shown in Fig. 3, the cut-off wavelength was 500 nm for both 700 nm and 500 nm tests. From the curve fitting, the wavelength of peak sensitivity (µ_max) was obtained. The parabolic fitting to the field sensitivity data shown in Fig. 3 indicates that the field sensitivity action spectrum measured with the 500 nm test flash differs substantially from the one measured with the 700 nm test flash. The wavelengths of peak sensitivity are separated by 25 nm: 607 nm for = 500 nm and 632 nm for = 700 nm.

		View larger version [in this window] [in a new window]
Figure 3. Field sensitivity action spectra of the L-cone of Fig. 2 measured with test wavelengths of 500 nm (left) and 700 nm (right) These spectra were constructed from the reciprocals of the background irradiances needed to reduce the flash sensitivity to 500 and 700 nm stimuli by a factor of 10. The main band of each spectrum was fitted to a parabolic function (dotted curves) in order to obtain the wavelength of peak field sensitivity, µmax.

Field sensitivity data were measured in a total of twenty L-cones. However, in only fourteen of these was a sufficient number of background wavelengths studied to allow construction of field sensitivity action spectra. In each of these fourteen L-cones, the wavelength of peak sensitivity was shorter for the field sensitivity action spectrum obtained with the 500 nm test flash compared with that obtained with the 700 nm test. The median wavelengths of peak sensitivity were 612·5 and 633·0 nm for the field sensitivity action spectra measured with the 500 and 700 nm tests, respectively. The difference between the medians - 20·5 nm - is significantly different from zero (P < 0·001) using a non-parametric sign test.

Middle-wavelength-sensitive cones (M-cones)

The application of Stiles' two-colour increment threshold technique to a turtle M-cone is shown in Fig. 4. The desensitizing effects of 700 nm (Fig. 4A) and of 500 nm (Fig. 4B) backgrounds on 700 and 500 nm test flashes are compared. The 700 nm background is clearly more effective in desensitizing the 700 nm test compared with its effect on the 500 nm test, while the 500 nm background exerts a stronger desensitizing action on the 500 nm test compared with the 700 nm test. These data were fitted to Stiles' template (see Fig. 4) in order to determine the background irradiance needed to desensitize the responses of the M-cone to the two test wavelengths by 1 log unit. For the M-cone in Fig. 4, the irradiances of the 700 nm background needed to desensitize the 700 and 500 nm tests are 4·25 and 6·81 log photons (700 nm) s^-1 µm^-2, respectively. The corresponding values for the 500 nm background are 5·89 and 4·65 log photons (500 nm) s^-1 µm^-2 for the 700 and 500 nm tests, respectively. Thus the turtle M-cone also does not obey the displacement rules of Stiles.

		View larger version [in this window] [in a new window]
Figure 4. The desensitizing action of 700 nm (A) and 500 nm (B) background (b/g) lights on flash sensitivity of an M-cone measured with 700 and 500 nm tests ( and , respectively) The dotted curves describe Stiles' function that was fitted to the data points by eye.

The background irradiances needed to reduce the M-cone sensitivity to 700 and 500 nm tests 10-fold were determined for backgrounds of different wavelengths in order to construct the field sensitivity action spectra shown in Fig. 5. The major band (region of maximum sensitivity) in each of these spectra was fitted to a parabolic function with the following cut-off wavelengths: 500 nm for the spectrum using the 500 nm test and 540 nm for the spectrum of the 700 nm test. The wavelengths of peak sensitivity are 564 and 647 nm for the field sensitivity action spectrum measured with 500 and 700 nm test flashes, respectively.

		View larger version [in this window] [in a new window]
Figure 5. Field sensitivity action spectra of the M-cone of Fig. 4 measured with test wavelengths of 500 nm (left) and 700 nm (right) These spectra were constructed from the reciprocals of the background irradiances needed to reduce the flash sensitivity to 500 and 700 nm stimuli by a factor of 10. The main band of each spectrum was fitted to a parabolic function (dotted curves) in order to obtain the wavelength of peak field sensitivity, µmax.

A total of eight M-cones were studied with Stiles' two-colour increment threshold technique. In most (n = 5), four levels of irradiance were studied for each background wavelength. Two cones were studied with three levels of irradiance and two with only one to two levels. In six out of these eight M-cones, enough background wavelengths were studied to allow construction of the field sensitivity action spectra. The median wavelengths of peak sensitivity (558·5 and 624 nm for the field sensitivity action spectra measured with 500 and 700 nm tests, respectively) differ significantly (P < 0·005) when tested by a non-parametric sign test.

Short-wavelength-sensitive cones (S-cones)

S-cones are sparse in the turtle retina (Kolb et al. 1988) and hence are rarely encountered. In this study, field sensitivity data were obtained from three S-cones, but none of these was held long enough to allow construction of field sensitivity action spectra. The irradiances of different background wavelengths needed to desensitize the responses to different test wavelengths by a factor of 10 compared with the dark-adapted state are listed in Table 1 for the three S-cones. It should be noted that in certain cases, the background irradiance available in our optical system was too weak to achieve 1 log unit of desensitization. In these cases the maximal available irradiance is listed in Table 1 as an estimated minimum value.

The data in Table 1 clearly indicate that the field sensitivity of S-cones depends upon test wavelength. In general, the photoresponse elicited by a given test wavelength was more sensitive to background wavelengths close to it. Thus in two S-cones (nos 1 and 2), it was possible to depress the sensitivity to the 700 nm test by a factor of 10 using 700 nm background, while the sensitivity to the 430 nm test could not be reduced by 1 log unit even with the brightest available 700 nm background. In contrast, medium- and short-wavelength backgrounds were more effective in desensitizing the responses to 430 nm than to 700 nm (S-cone no. 2). In S-cone no. 3, the 560 nm background was similarly effective in desensitizing the response to 540 and 450 nm tests, but was not bright enough to induce a 10-fold desensitization of the 700 nm test.

Table 1. Irradiances of backgrounds (b/g) of different wavelengths needed to desensitize the responses of three S-cones to different test wavelengths by a factor of 10

L-cones and M-cones contrasted

Field sensitivity data for large diameter (2800 µm) background lights of different wavelengths were measured in a total of twenty L-cones and eight M-cones using small diameter (110 µm) test flashes of either 500 or 700 nm. In order to allow comparison between different cones studied in different experiments, the field sensitivity data for each cone were normalized in a manner to maintain the relative differences between field sensitivity data obtained for the two test wavelengths. For each L-cone, all the field sensitivity data were normalized to the value measured for the 700 nm test using 700 nm background light, while the field sensitivity data of each M-cone were normalized to the value obtained with the 500 nm test using 500 nm background light. The normalized field sensitivity data were averaged and plotted (means ± S.D.) in Fig. 6 as a function of background wavelength for L-cones and M-cones. Mean field sensitivity data for each spectral type of cone are separated according to the test wavelength used to measure them: 700 and 500 nm. The continuous curves in Fig. 6 are the action spectra of 'the isolated L-cone' and 'the isolated M-cone,' which were measured previously with the same photostimulation system used in the present study (Perlman et al. 1994). These are the best representations of the action spectra of L- and M-cones in the eyecup of the turtle Mauremys caspica reflecting only the absorption spectra of the visual pigments and the transmission properties of the oil droplets.

The field sensitivity action spectrum of M-cones measured with the 500 nm test fits the action spectrum of 'the isolated M-cone' well, while that measured from M-cones with the 700 nm test fits the action spectrum of 'the isolated L-cone' well (Fig. 6B). The field sensitivity action spectra of L-cones (Fig. 6A) are less clear. The one measured with the 700 nm test is similar to the action spectrum of 'the isolated L-cone', but the spectrum obtained with the 500 nm test fits neither the L-cone nor the M-cone.

		View larger version [in this window] [in a new window]
Figure 6. Mean (± S.D.) field sensitivity data of twenty L-cones (A) and eight M-cones (B) Before averaging, the field sensitivity data of each L-cone were normalized to the value obtained with 700 nm background using the 700 nm test, and the data of each M-cone were normalized to the field sensitivity measured with 500 nm background using the 500 nm test. The continuous curves are the action spectra of 'the isolated M-cone' and 'the isolated L-cone', derived experimentally from the eyecup of the turtle Mauremis caspica (Perlman et al. 1994).

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The findings described here exclude the possibility that the light-induced voltage responses of cone photoreceptors in the turtle Mauremys caspica reflect the action of a single mechanism. Rather, the photoresponses elicited in single M- and L-cones reflect interactions of at least two mechanisms characterized by different spectral properties. These findings indicate that in the turtle retina, interactions between spectral mechanisms occur at the photoreceptor layer, while in human observers such interactions have been postulated to occur in post-receptoral loci (Pugh, 1976; Pugh & Mollon, 1979; Pugh & Kirk, 1986).

The linear range photoresponses that were elicited in the dark-adapted state by test stimuli of different wavelengths had similar forms (not shown here) and could be made indistinguishable by adjusting their intensities. However, coloured backgrounds induced selective desensitization and affected the photoresponses in a wavelength-dependent manner. Therefore, if we adopt the statement 'The signals from each cone depend only upon the rate it is effectively catching quanta; it does not depend upon the associated wavelength...' (Naka & Rushton, 1966), as an operational definition for the Principle of Univariance, then the effects of chromatic adaptation, described here, also indicate that cone photoreceptors in the retina of the turtle Mauermys caspica violate this principle. However, Naka & Rushton noted that the Principle of Univariance was applicable to a more molecular, fundamental process: photon absorption. Indeed, measurements of photocurrents in monkey cones demonstrated univariant responses to stimuli of different wavelengths, which produced equal photon absorption (Baylor et al. 1987). This leaves us with the conclusion that along the pathway transforming the photocurrent of a turtle cone photoreceptor into a voltage response, additional input is added with spectral properties that differ from those of the initial event, thereby upsetting the univariant character of the photoresponse. Since synaptic output of a cone depends upon its voltage response to light stimuli, visual information processing in post-receptoral levels of the turtle visual system must assume a non-univariant cone output.

Mechanisms underlying the field sensitivity action spectra of turtle cones

The data described in Fig. 6 clearly indicate that the field sensitivity action spectrum of turtle cone photoreceptor depends upon the wavelength of the test flash used to measure background desensitization. In this section, known properties of turtle cone photoreceptors are used in order to create a computer simulation of the Stiles two-colour increment threshold experiment. These simulations are helpful in an attempt to predict the cellular mechanisms underlying our findings.

A computer program has been written to derive the background irradiance needed to fulfil the following relationship:

S LA() = (1/10)S DA (), (1)

where S _LA() and S _DA() are, respectively, the light- and dark-adapted flash sensitivities of a cone photoreceptor to test flash of wavelength . This is the Stiles criterion for deriving the field sensitivity.

In order to formulate a relationship between light- and dark-adapted flash sensitivities, test flash wavelength, background irradiance and background wavelength, the photoresponse of a single cone has to be expressed in mathematical terms. In turtle cone photoreceptors, coloured oil droplets are located in the inner segments and filter the incident light before it reaches the site of photon absorption, the outer segment. The transmission properties of the oil droplets and the optical pathways of the incident light may induce wavelength-dependent effects in the field sensitivity action spectrum (D. A. Baylor, personal communications). This suggestion is based upon two assumptions: (1) the oil droplets in the inner segments of L- and M-cones contain pigment at high density and act as high-pass filters (Liebman & Granda, 1975; Lipetz, 1984); (2) excitation and adaptation in the cone outer segment are localized, as previously demonstrated in rod photoreceptors (Hemila & Reuter, 1981; Lamb et al. 1981; Cornwall et al. 1990).

In Fig. 7, the relationships between the pathways of the incident light and cone outer segments are illustrated in a schematic manner. The incident light can be roughly divided into two pathways. A fraction of the incident light bypasses the oil droplet and excites only the tip of the outer segment (a in L-cones and b in M-cones). The remaining incident light passes through the oil droplets and mainly excites the base of the outer segments. The degree to which the incident light divides between these two pathways depends upon many factors, including the angle between the light beam and the long axis of the photoreceptors, light scatter and reflections within the retina, and therefore may vary between experiments. In fact, variation in the action spectra of cones belonging to one spectral type have been attributed to differences in the relative contributions of light going through the oil droplets and light scattering around them (Baylor & Hodgkin, 1973; Ives et al. 1983; Schneeweis & Green, 1995). Due to the transmission properties of the oil droplets (Liebman & Granda, 1975; Lipetz, 1984, 1985), short-wavelength light is completely absorbed and can only excite the tip of the cone outer segment by bypassing the oil droplet, while long-wavelength light is transmitted through the oil droplet, and can therefore also excite the base of the outer segment. A mathematical derivation of the cone photoresponse and its dependence upon the oil droplet and the pathway of the incident light is given in the Appendix.

		View larger version [in this window] [in a new window]
Figure 7. Schematic illustration of optical and neuronal contributions to the photoresponses of turtle L- and M-cones The incident light is divided into two major pathways (dashed lines). A fraction (a in L-cones and b in M-cones) bypasses the oil droplet and reaches the tip of the cone outer segment, while the rest of the light passes through the oil droplet and mainly excites the base of the outer segment. The light-induced voltage response of each cone is the summation of two contributions (continuous arrows); the major one reflecting the effects of photon absorption in the cone outer segments (R in L-cones and G in M-cones) and a smaller one originating in cones of different spectral type. A fraction p of the L-cone response is transmitted to the M-cone response, while a fraction q of the M-cone response is transmitted to the L-cone.

The photoresponses of turtle cones, recorded with intracellular microelectrodes, depend upon photon absorption in their own outer segments, but may be also affected by lateral interactions between neighbouring cones of different spectral types as shown schematically in Fig. 7. This suggestion is based upon physiological findings that demonstrated short-wavelength input to turtle L-cones (Normann et al. 1984, 1985) and long-wavelength input to S-cones (Itzhaki et al. 1992), and upon anatomical observations of teleodendria radiating from the cone pedicles and terminating on neighbouring cones regardless of spectral type (Kolb & Jones, 1985; Ohtsuka & Kawamata, 1990). According to this possibility, the recorded photoresponse of turtle cones (R () for L-cones and G () for M-cones) is given by:

R () = R () + qG (), (2)

and

G () = pR () + G (). (3)

Since the transmission properties of the oil droplets cannot be neglected, R () and G () in these equations are, respectively, the responses of 'the isolated L-cone' and 'the isolated M-cone' and reflect the absorption spectra of the visual pigments and the transmission properties of the oil droplets. These are calculated from eqns (A6) and (A7), respectively, which are derived in the Appendix. The parameters p and q determine the extent of excitatory interactions between cones of different spectral types; p is the fraction of the L-cone response that is transmitted to the M-cone, while q is the fraction of the M-cone response that is transmitted to the L-cone. The above equations have been simplified by neglecting quadratic terms of interactions (p × q, p ² and q ²), which are assumed to be too small to affect the simulations. Interactions between cones of the same spectral type are abundant in the turtle retina (Baylor, 1974; Detwiler & Hodgkin, 1979), but are ignored in the model since they are included in the primary response of each cone type (R () and G ()), assuming a homogeneous responsivity of these cones within the area illuminated by the test flash (110 µm in diameter). The latter assumption is valid for the turtle Mauremys caspica where only one type of oil droplet (red) was found for L-cones, and one type (orange-yellow) for M-cones (Kolb et al. 1988).

In order to decide which of the possibilities described above (optical or neuronal interactions) contributed most to the experimentally measured field sensitivity action spectra, we varied the parameters a, b, c, d, p and q (as defined in Fig. 7 and in the Appendix) of the models until the simulated field sensitivity action spectra best fitted (as judged by eye) the experimental data. The simulated field sensitivity action spectra were derived by calculating the total photoresponse of an L-cone and an M-cone from eqns (2) and (3) in conjunction with eqns (A6) and (A7) (see Appendix), and then applying Stiles' criterion for constructing the field sensitivity action spectrum (eqn (1)).

The contribution of the optical considerations alone to the field sensitivity action spectra was assessed by eliminating neuronal interactions (p = 1 and q = 0 for L-cones; p = 0 and q = 1 for M-cones) and varying only the parameters a, b, c and d. The simulated field sensitivity action spectra are compared in Fig. 8 with the experimental data for L-cones (Fig. 8A and C) and M-cones (Fig. 8B and D). The simulated field sensitivity action spectra for the 700 and 500 nm tests are shown as continuous and dashed curves, respectively. The experimental data, reproduced from Fig. 6, are shown as means ± S.D. for the field sensitivity action spectra measured with 700 and 500 nm tests. Two sets of parameters were used: one to obtain the best fit (as judged by eye) of the simulated action spectra to the L-cone data (Fig. 8A and B) and the other set was needed to better describe the M-cone data (Fig. 8C and D).

The simulations of the field sensitivity action spectra using only optical considerations fitted the experimental data measured in L-cones reasonably well (Fig. 8A), but failed completely with regard to the field sensitivity action spectrum determined with the 700 nm test in M-cones (Fig. 8D). Furthermore, the sparse data obtained from S-cones also oppose the optical model as the sole explanation for the findings presented here. The S-cones in turtle retinae contain colourless oil droplets (Lipetz, 1985; Ohtsuka, 1985; Kolb et al. 1988) that have negligible optical density in the range 450-700 nm. Therefore, the oil droplets in S-cones are not expected to affect the Stiles two-colour increment threshold experiment. In contrast to this prediction, the field sensitivity data, obtained from the few successful recordings from S-cones, were dependent upon the wavelength of the test flash (Table 1). The 700 nm test was more sensitive to long-wavelength backgrounds compared with tests of shorter wavelengths. Similarly, the 430 nm test was desensitized to a greater extent compared with the 700 nm test when short-wavelength backgrounds were used.

		View larger version [in this window] [in a new window]
Figure 8. Testing the contribution of optical considerations alone to the field sensitivity action spectra of L-cones (A and C) and M-cones (B and D) Experimental field sensitivity data, obtained with 700 and 500 nm tests ( and ×, respectively) were redrawn from Fig. 6. Simulated spectra were calculated for 700 and 500 nm tests (continuous and dashed curves, respectively) using eqns (A6) and (A7). Two sets of parameters (a, b, c, d) were used for the simulation: one that best fitted the L-cone data (A and B) and the other that better fitted the M-cone data (C and D). The goodness of fit was judged by eye. The parameters for the L-cones (p, a, c) and for the M-cones (q, b, d) are listed.

Since the optical considerations could not satisfy the experimental data, we added excitatory interactions between cones of different spectral type as another cellular mechanism that affected turtle cone photoresponses and tested this new optical-neural interactions model. For this model, flash sensitivities of L- and M-cones were calculated from eqns (2) and (3) in combination with eqns (A6) and (A7) (see Appendix), and then the field sensitivity action spectra were calculated using eqn (1). The best fit (as judged by eye) of the simulated field sensitivity action spectra to the experimental field sensitivity data is shown in Fig. 9 for L-cones (Fig. 9A and C) and for M-cones (Fig. 9B and D). Three different sets of parameters (a, b, c, d, p, q) were used for these simulations. Two of these were used in order to obtain the best fit (as judged by eye) of the simulated field sensitivity action spectra to the data of L-cones (Fig. 9A) and M-cones (Fig. 9B). It should be noted that in each case, the values of a and b were kept equal and those of c and d were kept equal, assuming that single L-cones and M-cones were characterized by similar properties. This assumption does not consider excitatory inputs from double cones that may be characterized by different properties. The third set of parameters was chosen in order to maintain symmetry of the system. Here, the same set of parameters was used to fit the field sensitivity data of L-cones (Fig. 9C) and M-cones (Fig. 9D). As evident in Fig. 9, the fit between the experimental data and the simulated field sensitivity action spectra is reasonable.

		View larger version [in this window] [in a new window]
Figure 9. Fitting the experimental field sensitivity data of L-cones (A and C) and M-cones (B and D) to simulated field sensitivity action spectra that were derived from the optical-neural interactions model Experimental field sensitivity data, obtained with 700 and 500 nm tests ( and ×, respectively) were redrawn from Fig. 6. Simulated spectra were calculated for 700 and 500 nm tests (continuous and dashed curves, respectively) using eqns (2) and (3) in combination with eqns (A6) and (A7). In A and B, different sets of parameters were chosen to obtain the best fit (as judged by eye) of the simulated field sensitivity action spectra to the L-cones and M-cones, respectively. In C and D, one set of parameters was used for both types of cones. The parameters for the L-cones (p, a, c) and for the M-cones (q, b, d) are listed.

		View larger version [in this window] [in a new window]
Figure 10. The contributions of the oil droplets and optical pathways of incident light to the simulated field sensitivity action spectra of an L-cone (A, B, C) and an M-cone (D, E, F) The spectra for the 700 and 500 nm tests (continuous and dashed curves, respectively) were derived by calculating the background irradiances needed to desensitize the cones by a factor of 10. Equations (A6) and (A7) were used to calculate the sensitivity (response for incident photon) of the L-cone and M-cone, respectively, assuming complete localization of excitation and adaptation (c = d = 0). In A and D, only a small fraction (a = b = 0·05) of the incident light bypasses the oil droplets, and most of it passes through the oil droplet. The reverse situation is shown in C and F, where most of the light bypasses the oil droplet (a = b = 0·95). The spectra in B and E represent the case where the incident light divides evenly between the two pathways (a = b = 0·5).

		View larger version [in this window] [in a new window]
Figure 11. The degree to which the tip and base of the outer segments act as separate parts also affects the simulated field sensitivity action spectra as shown for an L-cone (A, B, C) and an M-cone (D, E, F) The spectra for the 700 and 500 nm tests (continuous and dashed curves, respectively) were derived using eqns (A6) and (A7) for the case in which 30 % of the incident light bypasses the oil droplets (a = b = 0·3). In A and D, the tip and the base of the outer segments are almost completely isolated (c = d = 0·05), while in C and F, these two parts behave almost like one unit (c = d = 0·95). The spectra in B and E represent an intermediate case (c = d = 0·5).

In the simulations shown in Fig. 9, we assumed that the contribution of L-cones to the photoresponse of M-cones was larger than the contribution of the M-cones to the photoresponse of L-cones. This basically reflects the relative distribution of L-cones and M-cones as described in the retina of the turtle Pseudemys scripta elegans (Kolb & Jones, 1987) and Mauremys caspica (Kolb et al. 1988). If double cones in the retina of the turtle Mauremys caspica like those in other types of turtles contain long-wavelength-sensitive visual pigment (Lipetz, 1985; Ohtsuka, 1985), then the ratio of L-cones to M-cones is larger than 3 : 1. Therefore, even if the interactions between one L-cone and one M-cone are symmetrical, each M-cone receives inputs from more L-cones than each L-cone from M-cones.

The curves in Fig. 9 indicate that very little interaction is needed to account for the experimental field sensitivity action spectra. If 3 % (p = 0·03) of the L-cone response is transmitted to the M-cone and 0·8 % (q = 0·008) of the M-cone photoresponse is transmitted to the L-cone, the experimental field sensitivity data can be reasonably fitted by the computer simulation. Such a small degree of interaction between cones of different spectral type can only be detected with measurements of small amplitude photoresponses. When a bright long-wavelength stimulus elicits a maximal response in L-cones (about 20 mV), only 0·6 mV of the M-cone response can be accounted for by excitatory input from L-cones (20 × 0·03). This can be significant if the same stimulus elicits a small response (< 1 mV) in the M-cone outer segment. However, when the stimulus intensity is increased further, the absolute contribution of the L-cones to the M-cones does not change since the L-cones are saturated. Since the response arising from photon absorption in the M-cone's outer segment increases, the relative contribution of the excitatory input from the L-cones to the M-cone response decreases. For these reasons, only sensitivity measurements from small amplitude photoresponses are sensitive enough to detect these excitatory interactions between cones of different spectral types, while measurements of large photoresponses for the construction of intensity-response curves may miss such interactions.

It should be noted that assuming excitatory interaction only between cones of similar spectral type and variability in the transmission properties of the oil droplets within each class of cones (Schneeweis & Green, 1995) cannot account for our field sensitivity findings. This possibility is an extension of the optical considerations but is only relevant to the L-cones in which variability has been demonstrated (Lipetz, 1984, 1985; Ohtsuka, 1985). As discussed above, optical considerations alone can account for the field sensitivity data of L-cones even with a homogeneous population of oil droplets (Fig. 8); however, the main problem is with the field sensitivity action spectra of M-cones.

The data described here indicate that when a red stimulus is applied, the outputs of the L-cones and of the M-cones (and probably also of the S-cones) reflect a single visual mechanism: that of the L-cones. This implies that regardless of any type of neuronal interactions, when a dim red stimulus is used, the signals in all the channels conveying retinal output to the central nervous system in the turtle originate in L-cones. When a green (500 nm) test flash is used, the retinal output channels convey more complex information. Some reflect pure M-cone excitation, while others reflect a mixture of M-cones and L-cones.

The photoresponses of 'isolated' L-cones and M-cones in the turtle retina

The incident light impinging upon the photoreceptors from the vitreal side as in the in vivo situation, is separated into two pathways (Fig. 7). One portion of the light goes through the oil droplet before reaching the outer segment and the other bypasses the oil droplet and reaches the tip of the outer segment. In order to simplify the analysis, it is assumed that a fixed fraction (a for L-cones and b for M-cones) of the incident light, regardless of wavelength, bypasses the oil droplet and reaches the tip of the cone outer segment. Since a mathematical model is being developed here for background desensitization, the photoresponses defined in the following equations are within the linear range of the cones.

The photoresponse of 'the isolated L-cone' (no neuronal interactions) that is elicited by a light stimulus of wavelength in the dark-adapted state can be expressed as the sum of two contributions:

R _DA() = aR () + (1 - a) D_r () R (). (A1)

R _DA() is the photoresponse of the dark-adapted L-cone and R () is the absorption spectrum of the long-wavelength visual pigment. D_r () is the transmissivity spectrum of the red oil droplet.

An analogous equation can be written for the M-cone:

G_DA() = bG () + (1 - b) D_y () G (), (A2)

where G () is the absorption spectrum of the medium-wavelength visual pigment and D_y () is the transmission properties of the yellow oil droplet in the M-cone inner segment.

Intracellular recordings from cone photoreceptors in the turtle retina indicated that the reduction in the flash sensitivity to a given wavelength by a background light followed a Weber-Fechner-type relationship (Baylor & Hodgkin, 1974; Normann & Anderton, 1983; Copenhagen & Green, 1987; Burkhardt, 1994):

S _LA()/S _DA() = 1/(1 + I (µ)/I_o (µ)). (A3)

In this equation S _LA() and S _DA() are the calculated flash sensitivities (responses to 1 photon) to test light measured respectively in the light- and dark-adapted states. I (µ) is the irradiance of the background field of wavelength µ and I_o (µ) is the background irradiance of that wavelength that reduces the sensitivity to test light by a factor of 2, relative to the dark-adapted value.

If background adaptation in the cone outer segment is localized as suggested by previous studies in rod photoreceptors (Hemila & Reuter, 1981; Lamb et al. 1981; Cornwall et al. 1990), then the tip and the base of the outer segment should be dealt with as two separate parts of the same cone, contributing independently to the total photoresponse. Assuming a linear summation, the total photoresponse of 'the isolated L-cone', R _LA(), during illumination with a background of wavelength µ and irradiance I (µ) is:

R _LA() = a R ()/(1 + a I (µ)/I_or (µ)) +

(1 - a) D_r () R ()/(1 + (1 - a) D_r (µ) I (µ)/I_or (µ)), (A4)

where the first term on the right describes the adaptation of the tip of the outer segment and the second term describes the adaptation of the base of the outer segment. R () is the dark-adapted absorption spectrum of the long-wavelength visual pigment and I_or (µ) is the irradiance of background µ needed to reduce the dark-adapted response by a factor of 2.

Using similar considerations, an analogous equation is derived for the M-cone:

G_LA() = b G ()/(1 + b I (µ)/I_og (µ)) +

(1 - b) D_y () G ()/(1 + (1 - b) D_y (µ) I (µ)/I_og (µ)). (A5)

Equations (A4) and (A5) are based on the assumption that light bypassing the oil droplet excites only the tip of the outer segment and is desensitized only by background light that bypasses the oil droplet, while test light that goes through the oil droplet excites only the base of the outer segment and is desensitized only by background light that goes through the oil droplet. This complete separation is unreasonable considering the small dimensions of the cone outer segments. Thus the tip of the outer segment can also be desensitized by background light that has passed directly through the oil droplet and long-wavelength test that goes through the oil droplet also excites the tip of the outer segment and can be desensitized by short-wavelength backgrounds that bypass the oil droplet. In order to account for these effects, an additional factor, c for L-cones and d for M-cones, is introduced. This factor indicates the degree to which light (test or background) that goes through the oil droplets can reach the outer segment tip and cause excitation and desensitization. With these considerations, eqns (A4) and (A5) become

R _LA () = a R ()/(1 + a I (µ)/I_or (µ) + c (1 - a) D_r (µ) I (µ)/I _or (µ)) +

(1 - a) D_r () R ()/(1 + c a I (µ)/I_or (µ) + (1 - a) D_r (µ) I (µ)/I_or (µ)),

(A6)

and

G_LA () = b G ()/(1 + b I (µ)/I_og (µ) + d (1 - b) D_y (µ) I (µ)/I_og(µ)) +

(1 - b) D_y () G ()/(1 + d b I (µ)/I_og (µ) + (1 - b) D_y (µ) I (µ)/I_og (µ)).

(A7)

In eqns (A6) and (A7), the first term on the right describes how a stimulus that excites only the tip of the outer segment is desensitized by backgrounds of any wavelength that bypass the oil droplet and also by long-wavelength background lights that pass directly through the oil droplets. The second term describes the excitation of the outer segment base and tip by test light that passes directly through the oil droplet and can be desensitized by long-wavelength backgrounds that pass through the oil droplet and by backgrounds of any wavelength that bypass the oil droplet.

In order to derive a quantitative description of the field sensitivity action spectra and to appreciate their dependence upon the oil droplets and the optical pathways, the following assumptions have been made. (1) The absorption spectra of the long- and medium-wavelength sensitive photopigments in the turtle Mauremys caspica are the same as those in the turtle Pseudemys scripta elegans and can be described by the action spectra measured in the isolated turtle retina (Schneeweis & Green, 1995). (2) In the retina of the turtle Mauremys caspica, L-cones contain only red oil droplets and M-cones contain only yellow oil droplets (Kolb et al. 1988). (3) The transmissivity spectra of the oil droplets are similar to those of the red and yellow oil droplets in other species of fresh water turtles. (4) The values of I_or (µ) and I_og (µ), the irradiances of background µ needed to desensitize the long-wavelength and middle-wavelength visual pigments by a factor of 2, are inversely proportional to the absorption spectra of the visual pigments (Wyszecki & Stiles, 1982). With these assumptions, field sensitivity action spectra were derived by calculating the irradiance of background µ needed to desensitize the L- and M-cones for 700 and 500 nm tests by a factor of 10 (eqn (1)).

Figures 10 and 11 show the simulated field sensitivity action spectra for 700 and 500 nm test flashes for an L-cone and an M-cone. The contribution of the oil droplets and the separation of the incident light into two pathways (factors a and b) to the field sensitivity action spectra is shown in Fig. 10. These spectra were calculated with c = d = 0, i.e. excitation and desensitization are localized and the tip and base act like two separate entities. When most of the incident light passes through the oil droplets (a = b = 0·05), the field sensitivity action spectra strongly depend upon the test wavelength (Fig. 10A and D). It looks similar to the absorption spectra of the visual pigment for the 500 nm test that excites only the outer segment tip and is strongly attenuated at short-wavelength backgrounds when a 700 nm test is used. The two field sensitivity action spectra become more similar to each other as the fraction of incident light bypassing the oil droplets increases until for a = b = 0·95 the two spectra have a similar pattern (Fig. 10C and F).

The degree to which the tip and base are separated (factors c and d) also affects the field sensitivity action spectra of L- and M-cones, as shown in Fig. 11. The field sensitivity action spectra for 700 and 500 nm tests were calculated for a = b = 0·3, i.e. 70 % of the incident light goes through the oil droplets and 30 % bypasses them. When the tip and the base are almost completely separated (c = d = 0·05), the simulated field sensitivity action spectra exhibit strong dependency upon the test wavelength (Fig. 11A and D), which is mainly evident for short-wavelength backgrounds. As the values of c and d are increased, the differences between the two spectra decrease until for c = d = 0·95, the outer segment behaves like one unit and the field sensitivity action spectrum is independent of the test wavelength (Fig. 11C and F).

	REFERENCES

Top Abstract Introduction Methods Results Discussion References

Baylor, D. A. (1974). Lateral interaction between vertebrate photoreceptors. Federation Proceedings 33, 1074-1077	[Medline]
Baylor, D. A. & Hodgkin, A. L. (1973). Detection and resolution of visual stimuli by turtle photoreceptors. The Journal of Physiology 234, 163-198	[Medline]
Baylor, D. A. & Hodgkin, A. L. (1974). Changes in time scale and sensitivity in turtle photoreceptors. The Journal of Physiology 242, 729-758	[Medline]
Baylor, D. A., Nunn, B. J. & Schnapf, J. L. (1987). Spectral sensitivity of cones of the monkey Macaca fascicularis. Journal of Physiology 390, 145-160.
Burkhardt, D. A. (1994). Light adaptation and photopigment bleaching in cone photoreceptors in situ in the retina of the turtle. Journal of Neuroscience 14, 1091-1105	[Abstract]
Copenhagen, D. R. & Green, D. G. (1987). Spatial spread of adaptation within the cone network of turtle retina. The Journal of Physiology 393, 763-776	[Abstract]
Cornwall, M. C., Fein, A. & MacNichol, E. F. J. (1990). Cellular mechanisms that underlie bleaching and background adaptation. Journal of General Physiology 96, 345-372	[Abstract]
Detwiler, P. B. & Hodgkin, A. L. (1979). Electrical coupling between cones in turtle retina. The Journal of Physiology 291, 75-100	[Abstract]
Fuortes, M. G. F., Schwartz, E. A. & Simon, E. J. (1973). Colour dependence of cone responses in the turtle retina. The Journal of Physiology 234, 199-216	[Medline]
Fuortes, M. G. F. & Simon, E. J. (1974). Interactions leading to horizontal cell responses in the turtle retina. The Journal of Physiology 240, 177-198	[Medline]
Hemila, S. & Reuter, T. (1981). Longitudinal spread of adaptation in the rods of the frog's retina. The Journal of Physiology 310, 501-528	[Abstract]
Itzhaki, A., Malik, S. & Perlman, I. (1992). The spectral properties of short wavelength (blue) cones in the turtle retina. Visual Neuroscience 9, 235-241.	[Medline]
Itzhaki, A. & Perlman, I. (1984). Light adaptation in luminosity horizontal cells in the turtle retina: role of cellular coupling. Vision Research 24, 1119-1126	[Medline]
Ives, J. T., Normann, R. A. & Barber, P. (1983). Light energy intensification by cone oil droplets: electromagnetic considerations. Journal of the Optical Society of America 73, 1725-1731.
Kamermans, M. & Spekreijse, H. (1995). Spectral behavior of cone-driven horizontal cells in teleost retina. In Progress in Retinal and Eye Research, ed. Osborne, N. N. & Chader, G. J., vol. 14, pp. 313-360. Elsevier Science Ltd, Oxford, UK.
Kolb, H. & Jones, J. (1985). Anatomic pathways for excitatory connections between red and green cones in the turtle retina. Journal of Neurophysiology 53, 304-317.
Kolb, H. & Jones, J. (1987). The distinction by light and electron microscopy of two types of cones containing colorless oil droplets in the retina of the turtle. Vision Research 27, 1445-1458	[Medline]
Kolb, H., Perlman, I. & Normann, R. A. (1988). Neural organization of the retina of the turtle Mauremys caspica: a light microscope and Golgi study. Visual Neuroscience 1, 47-72	[Medline]
Lamb, T. D., McNaughton, P. A. & Yau, K.-W. (1981). Spatial spread of activation and background desensitization in toad rod outer segments. The Journal of Physiology 319, 463-496	[Medline]
Liebman, P. A. & Granda, A. M. (1975). Super dense carotenoid spectra resolved in single cone oil droplets. Nature 253, 370-372	[Medline]
Lipetz, L. E. (1984). Pigment types, densities and concentrations in cone oil droplets of Emydoidea blandingle. Vision Research 24, 605-612.
Lipetz, L. E. (1985). Some neuronal circuits of the turtle retina. In The Visual System, ed. Fein, A. & Levine, J. S., pp. 107-132. Alan R. Liss, Inc., New York, USA.
Mitchell, D. E. & Rushton, W. A. H. (1971). The red/green pigments of normal vision. Vision Research 11, 1045-1056	[Medline]
Naka, K. I. & Rushton, W. A. H. (1966). S-potentials from colour units in the retina of fish (Cyprinidae). The Journal of Physiology 185, 536-555	[Medline]
Normann, R. A. & Anderton, P. J. (1983). The incremental sensitivity curve of turtle cone photoreceptors. Vision Research 23, 1731-1733	[Medline]
Normann, R. A., Perlman, I. & Daly, S. J. (1985). Mixing of color signals by turtle cone photoreceptors. Journal of Neurophysiology 54, 293-303	[Medline]
Normann, R. A., Perlman, I., Kolb, H., Jones, J. & Daly, S. J. (1984). Direct excitatory interactions between cones of different spectral types in the turtle retina. Science 224, 625-627	[Medline]
Ohtsuka, T. (1985). Relation of spectral types to oil droplets in cones of turtle retina. Science 229, 874-877	[Medline]
Ohtsuka, T. & Kawamata, K. (1990). Telodendrial contact of HRP-filled photoreceptors in the turtle retina: pathways of photoreceptor coupling. Journal of Comparative Neurology 292, 599-613	[Medline]
Perlman, I., Itzhaki, A., Malik, S. & Alpern, M. (1994). The action spectra of cone photoreceptors in the turtle (Mauremys caspica) retina. Visual Neuroscience 11, 243-252	[Medline]
Piccolino, M. (1995). The feedback synapse from horizontal cells to cone photoreceptors in the vertebrate retina. In Progress in Retinal and Eye Research, ed. Osborne, N. N. & Chader, G. J., vol. 14, pp. 141-196. Elsevier Science Ltd, Oxford, UK.
Pugh, E. N. (1976). The nature of the 1 mechanism of W.S. Stiles. The Journal of Physiology 257, 713-747	[Abstract]
Pugh, E. N. & Kirk, D. B. (1986). The mechanisms of W.S. Stiles: An historical review. Perception 15, 705-728.	[Medline]
Pugh, E. N. & Mollon, J. D. (1979). A theory of the 1 and 3 color mechanisms of Stiles. Vision Research 19, 293-312	[Medline]
Schneeweis, D. M. & Green, D. G. (1995). Spectral properties of turtle cones. Visual Neuroscience 12, 333-344	[Medline]
Stiles, W. S. (1939). The directional sensitivity of the retina and the spectral sensitivities of rods and cones. Proceedings of the Royal Society B 127, 64-105.
Stiles, W. S. (1949). Increment thresholds and the mechanisms of color vision. Documenta Ophthalmologica 3, 138-163.
Stiles, W. S. (1959). Color vision: the approach through increment threshold sensitivity. Proceedings of the National Academy of Sciences of the USA 45, 100-114.
Wyszecki, X. & Stiles, W. S. (1982). Color Science: Concepts and Methods, Quantitative Data and Formulae, 2nd edn. John Wiley & Sons Inc., New York, USA.

Acknowledgements

We would like to thank Dr David M. Schneeweis who kindly supplied us with turtle cone action spectra that he measured in the isolated retina and with the transmissivity spectra of the cone oil droplets. We thank Dr Dennis A. Baylor for his suggestion of the optical model and Dr D. G. Green, Dr E. N. Pugh Jr, Dr S. K. Shevel and Dr D. C. Hood for their comments and suggestions on earlier versions of this manuscript. This research was partially supported by a grant from the Basic Science Foundation, Israel Academy of Sciences and Humanities (to I. P.) and by grants EY 00197 from the National Eye Research Institute and Senior Scientist Investigator Award from the Research to Prevent Blindness (to M. A.). M. Alpern was from the Kellogg Eye Center, University of Michigan, Ann Arbor, MI, USA (died May, 1996).

Corresponding author

I. Perlman: The Bruce Rappaport Faculty of Medicine, Technion-Israel Institute of Technology and the Rappaport Institute, PO Box 9649, Haifa 31096, Israel.

Email: iperlman{at}techunix.technion.ac.il

Author's present address

H. Asi: Department of Visual Sciences, Institute of Ophthalmology, University of London, London, UK.

This Article Abstract Full Text (PDF) Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Perlman, I. Articles by Alpern, M. Search for Related Content PubMed PubMed Citation Articles by Perlman, I. Articles by Alpern, M.