Dark adaptation recovery of human rod bipolar cell response kinetics estimated from scotopic b-wave measurements

Abstract

We recorded ganzfeld scotopic ERGs to examine the responses of human rod bipolar cells in vivo, during dark adaptation recovery following bleaching exposures, as well as during adaptation to steady background lights. In order to be able to record responses at relatively early times in recovery, we utilized a ‘criterion response amplitude’ protocol in which the test flash strength was adjusted to elicit responses of nearly constant amplitude. In order to provide accurate and unbiased measures of response kinetics, we utilized a curve-fitting procedure to fit a smooth function to the measured responses in the vicinity of the peak, thereby extracting both the time-to-peak and the amplitude of the responses. Following bleaching exposures, the responses exhibited both desensitization and accelerated kinetics. During early post-bleach recovery, the flash sensitivity and time-to-peak varied according to a power-law expression (with an exponent of 6), as found in the presence of steady background light. This light-like phenomenon, however, appeared to be set against the backdrop of a second, more slowly recovering ‘pure’ desensitization, most clearly evident at late post-bleach times. The post-bleach ‘equivalent background intensity’ derived from measurements of flash sensitivity faded initially with an S2 slope of ∼0.24 decades min⁻¹, and later as a gentle S3 tail. When calculated from kinetics, the results displayed only the S2 slope. While the recovery of rod bipolar cell response kinetics can be described accurately by a declining level of opsin in the rods, the sensitivity of these cells is reduced further than expected by this mechanism alone.

In a recent study of dark adaptation of the human scotopic ERG b-wave, we estimated the level of ‘equivalent background intensity’ that appeared to be experienced by the rod bipolar cells, as a function of post-bleach time (Cameron et al. 2006). For large bleaches, though, we were surprised to find that the concept of equivalence to background illumination did not hold, since the responses obtained post-bleach were not directly comparable to those found during light adaptation (Cameron et al. 2006, p. 523). This led us to propose that, following larger bleaches, the rod bipolar cells might experience both a purely desensitizing effect that disappeared slowly and a phenomenon resembling exposure to real light.

However, those experiments had not been designed to probe the precise kinetics of the b-wave responses, and nor had they been designed to examine the responses at relatively early post-bleach times. Here we adopt a ‘criterion response amplitude’ approach (rather than our previous fixed flash strength approach) that permits good resolution of the response kinetics even at early times after the bleach. In addition, we develop a means of accurately estimating the time-to-peak of the flash response. We employ these approaches to monitor the changes in kinetics of the post-bleach responses, and thereby to calculate the recovery of what we refer to as the ‘kinetics-equivalent’ background intensity.

These experiments confirm our earlier suspicion that the rod bipolar cell response kinetics are to some extent disconnected from response sensitivity following large bleaches, and show that a very slowly recovering desensitization is present in addition to the conventional phenomena that are presumed to arise from the presence of unregenerated opsin within the rod outer segments.

Some of our findings have recently been reported in abstract form (Cameron & Lamb, 2007, 2008).

Methods

The methods and equipment used for recording the ERG b-wave were similar to those described by Cameron et al. (2006), modified to enable accurate examination of the response time-to-peak and with flash strengths adjusted so as to provide a criterion response amplitude.

Subjects

The subjects were four of the authors, aged 32–59 years, who had normal vision apart from minor errors of refraction. Ethical approval for the study was obtained from the Australian National University's Human Research Ethics Committee, and the work was conducted in accordance with the tenets of the Declaration of Helsinki. Prior to participation, subjects provided informed, written consent.

ERG recordings

The ERG was recorded from one eye (nominated by the subject, and thereafter the same for all experiments) using a ‘DTL’ conductive nylon-fibre electrode. Two drops of 1% tropicamide were instilled into the test eye to fully dilate the pupil (diameter ∼7 mm), and the subject was then dark-adapted for ∼25 min. During experiments, a video image of the test eye was captured on tape.

Ganzfeld illumination

Full-field light stimuli were delivered within a ganzfeld apparatus (described in Cameron et al. 2006). Light intensities were measured with an IL-1700 photometer (International Light), fitted with a radiometric barrel and scotopic (Z-CIE) filter. Intensities were measured in units of cd m⁻² (for steady illumination) and cd m⁻² s (for flashes), and have been converted to Td and Td s through multiplication by the pupil area in mm². Throughout this paper, all intensities are given in scotopic units.

Flash stimuli

Test flashes were produced with a blue LED (λ_max= 470 nm). The flash strength was set by adjusting the duration (range: 20–4000 μs) and magnitude (3–300 mA) of the LED current.

Background illumination

Background illumination was provided by a second blue LED (λ_max= 470 nm), driven at 20 mA with pulse width modulation at a repetition rate of 50 Hz. The luminous intensity of the background could be increased in 0.5 log unit steps from 5 × 10⁻⁴ to 1.5 cd m⁻² by setting the pulse width from 1 to 3000 μs, in a 1–3–10 sequence.

Bleaching illumination

Bleaches were delivered with up to 12 ultra-bright white LEDs, long pass filtered at 520 nm (HT101, Lee Filters) to remove blue light. The maximum luminous intensity that could be delivered was ∼10 000 cd m⁻².

Estimation of bleaching levels

Dark adaptation recovery of the b-wave was examined following bleaches of a wide range of strengths. For one subject (A.M.C.), with whom we conducted our most detailed investigation, we employed bleaching exposures of six levels, while for the remaining three subjects, three levels were used. For most bleach levels (except near-total), the exposure duration was kept brief (21 ± 1.6 s, mean ±s.e.m.). For near-total bleaches, however, we used a longer exposure duration (68 ± 8.8 s) at the maximum luminous intensity available (∼10 000 cd m⁻²). Even for the longer exposure durations used here, we estimate that only a small quantity of the pigment (on the order of a few percent) would have been able to regenerate during the bleaching step. Thus, the final calculated bleach levels (see below) were unlikely to be greatly influenced by the parameters of pigment regeneration.

To achieve a required bleach level, we nominated an exposure duration, measured the subject's dilated pupil diameter, calculated the required intensity, and set the ganzfeld illumination to this intensity by specifying the number and current of the LEDs to be activated at bleach onset. During bleach delivery, the mean LED intensity was noted. Subsequently, we estimated the mean pupil area during the exposure based on pupil diameter measurements obtained from the video recording, with allowance for any occlusion of the pupil by the eyelids. The ERG results are averaged from 2 to 3 experiments performed at each nominal bleaching level, and hence the bleach intensities and magnitude values reported herein are average values for each level.

The bleach calculations were performed by numerically solving eqn (A12) of Mahroo & Lamb (2004). The parameters chosen were: Q_e (= 1/σ) = 10^6.7 Td s, v = 0.088 min⁻¹, and K_m= 0.185. The value of the bleaching constant, Q_e, is slightly lower than the values used previously (10^6.8 Td s, by Cameron et al. 2006; 10^6.85 Td s, by Thomas & Lamb, 1999) and the justification for this change is given in Results.

Stimulus timing and data acquisition

Stimulus timing and data acquisition were controlled by a custom program ‘MatERG’, which ran under Matlab (The MathWorks, Inc., Natick, MA, USA). ERG signals were filtered from 0.16 Hz to 500 Hz, and sampled at 5 kHz. Subsequent data analysis was performed offline using the same program. Artefact rejection was based on the criteria developed by Paupoo et al. (2000).

Criterion response amplitude protocol

During the course of each experiment, the test flash strength was adjusted so as to elicit a b-wave with amplitude close to a criterion level; the criterion level chosen was usually that amplitude elicited under dark-adapted conditions by a flash of 0.04 Td s, typically 25 μV (see Results). Rather than attempt to set the flash strength solely interactively during a dark adaptation experiment, we made estimates in advance of the flash strengths that would be required at different post-bleach times. Initial predictions were made from previous results for test flashes of fixed strength (Cameron et al. 2006), and then refined during the course of our experiments. Likewise the flash strengths required in the presence of backgrounds were provisionally estimated from published light adaptation results for flashes of fixed strength (Cameron et al. 2006), and then refined from the measurements that we made. On rare occasions, if the response appeared either too large, or too small, the flash strength was modified during the experiment.

Estimation of time-to-peak and amplitude of flash responses

In order to obtain an objective estimate of the time-to-peak of the flash response, we fitted a smooth curve to the upper region of the averaged response, in the vicinity of the peak. We chose the Poisson function

(1a)

(1b)

which is sufficiently asymmetric to describe the responses of both photoreceptor (Fuortes & Hodgkin, 1964; Baylor et al. 1979) and of bipolar cells (Ashmore & Falk, 1980). Here n_P is the Poisson exponent. The first unscaled form, eqn (1a), uses a time constant, τ. The second form, eqn (1b), has been normalized to a peak amplitude of r_peak which occurs at a time-to-peak of t_peak.

We used least-squares fitting to extract the three parameters n_P, t_peak and r_peak for each averaged response. We restricted the fitting to the region of time over which the response exceeded 40% of maximal; we tested other values for the lower limit, from 20% to 70% of maximal, and found only minor variations in the fitted values of t_peak and r_peak.

Normalization of sensitivity measurements: fitting of eqn (3)

The absolute amplitude of the criterion responses varied somewhat on different experimental days, presumably due to differences in the scleral placement of the recording electrode (e.g. for flashes of ∼0.04 Td s, range 10–30 μV). To allow for such variations, and enable comparison of results collected in different sessions, we normalized the flash sensitivity measurements (from Fig. 4 onwards) to unity. The dark adaptation measurements have been normalized to the mean value of the pre-bleach and fully recovered sensitivities (recorded over 4 min and 4–8 min, respectively). For the light adaptation measurements, we fitted a Weber Law expression (eqn 3) to the entire set of results for a given subject, using the least-squares criterion; parameter values for the half-desensitizing background intensity (I₀) and Weber exponent (n_W) were yolked across all experiments for that subject, while the vertical scaling (S_D) was fitted uniquely for each experiment. The results for each experiment were then normalized by dividing by the value obtained for S_D.

Figure 4. Relation between b-wave flash sensitivity and time-to-peak.

A and B, flash sensitivity as a function of time-to-peak during light adaptation by steady backgrounds (grey circles), and during recovery from small bleaches (▵, 2.6%, and •, 6.3%). C and D, sensitivity as a function of time-to-peak following large bleaches (▵ 29%○, 57%, and ♦, 99%). The results for the moderate bleach (12%) were intermediate between the results for small and large bleaches. The post-bleach measurements of sensitivity were normalized as described in Methods. For this figure, the light-adapted measurements of sensitivity were normalized to the value found on the ∼0.02 Td background (and the ‘dark-adapted’ measurement was removed), to enable direct comparison with the post-bleach measurements. The power-law description of eqn (2) is thus plotted as , where S_Dim and t_Dim, are the sensitivity (0.85/1) and time-to-peak (121 ms) obtained on a dim background (∼0.02 Td). Subject A.M.C.

Estimation of rod signal

The scotopic b-wave response to dim flashes primarily reflects the activity of rod bipolar cells, at least up to the test flash strength at which the b-wave is preceded by a perceptible a-wave component (Pugh et al. 1998; Robson & Frishman, 1999; Pang et al. 2007). At higher flash strengths, the contribution of other retinal cells (e.g. rods and/or cone on-bipolar cells) may begin to significantly influence the records. Arguably, most of the additional activity is contributed by the rod signal. We therefore sought to estimate the influence of the rod signal on our criterion amplitude measurements, both during post-bleach recovery, and during light adaptation by steady backgrounds. For post-bleach recovery, we predicted the rising phase of the rod signal using the recovery of maximal a-wave amplitude a_max, as specified by eqn (9) of Thomas & Lamb (1999), in conjunction with the rising phase kinetics of the Lamb & Pugh (1992) model (see eqn (6) of Thomas & Lamb, 1999). For all cases we used an amplification constant A = 4.3 s⁻² and a delay time t_d= 2.5 ms. To estimate prebleach and fully recovered rod responses, we used a_max=−170 μV. Post-bleach responses were modelled using the same equation, but with the number of photoisomerizations, Φ, varied according to the flash strength delivered, and the a_max for each post-bleach interval specified according to eqn (9) of Thomas & Lamb (1999), with values for the parameters τ_a and c_a estimated from Fig. 7 of that paper. For measurements obtained during light adaptation, we could have modelled the rising phase of the rod response using the same approach, except with a_max specified by eqn (8) of Thomas & Lamb (1999). Instead we described the complete time course of the rod response, using eqn (6a) of Friedburg et al. (2001), with Φ set according to the flash strength delivered on each background, and with the inactivation parameters t₀ and τ_rec set for each background intensity based on their Figs 9 and 10.

Figure 7. Comparison of the equivalent background intensity values obtained from measurements of sensitivity and kinetics.

Plot of the equivalent background intensities derived from the post-bleach measurements of sensitivity (e.g. Fig. 6A) and of kinetics (e.g. Fig. 6B). The symbols plot average results (±s.e.m.) for all four subjects, at two bleaching levels: ∼6.3 ± 0.1% (•) and ∼96 ± 1.3% (♦). The plotted symbols have been averaged following binning across intensity into successive ∼0.5 log unit spanning increments; points occurring above or below the plotting limits were removed. The results for each bleach level were fitted to the raw (unaveraged) points over the indicated range (continuous lines), by log transforming the points, then applying standard linear regression, with a fixed slope of 1. This fitting was used to determine the log transformed y-intercept (c), plotted here as 10c.

Results

Our experiments were designed to examine the ‘equivalent background’ experienced by rod bipolar cells following bleaching, through measurement of the kinetics of the ERG b-wave flash response. We begin by illustrating our approach to quantifying response kinetics (estimation of time-to-peak), and describing our choice of criterion response amplitude. We then illustrate post-bleach recoveries as well as light-adapted responses, which we use to extract estimates of the equivalent background intensity. Finally, we compare the recovery of equivalent background determined from the response kinetics with that determined from the more conventional approach of analysing flash sensitivity.

Estimation of b-wave kinetics: time-to-peak of the flash response

Figure 1A and B illustrate ERG b-wave responses from two subjects at a series of five dim flash strengths, and demonstrate our fitting procedure for the estimation of response time-to-peak (and amplitude). The smooth curves are least-squares fits of the Poisson expression, eqn (1b), when fitted over the region from 40% maximal and upwards, indicated in red; the blue parts of the curves plot the continuation of the same expression below the region of fitting. Inspection suggests that the red curves provide a good fit, and that they should provide a reliable (and unbiased) estimate of the time-to-peak. Likewise they provide reliable estimates of response amplitude, although the amplitudes could instead have been approximated without fitting the time course (Cameron et al. 2006). For the arrowed responses, the best-fitting value of the Poisson exponent n_P in eqn (1a) was typically 10–14; during light or dark adaptation the required value was sometimes lower (8–9).

Figure 1. Choice of criterion response amplitude.

To decide which criterion level of amplitude to use, we began by attempting to locate the b-wave's linear operating range by recording ERG responses to a series of dim test flashes, successively doubled in strength (from ∼3.9 × 10⁻³ to ∼0.063 Td s). Panels A and B present results for two subjects (A.M.C. and T.D.L., respectively). The results of A are unfiltered averages, previously published (in filtered form) in Fig. 2B of Cameron et al. (2006) (121 ± 6.4 traces per flash strength). Those of B are from a new experiment (73 ± 1.8 traces per flash strength). In C and D, the responses in A and B have been converted into units of flash sensitivity. The flash sensitivity records for the two dimmest flashes (grey traces) are more accelerated and exhibit a different pattern of response compression than the remaining traces, possibly due to a contribution from sensitive inner retinal activity. The peak flash sensitivities have been indicated for all but the two dimmest flash strengths using the same symbol scheme applied in A and B. During the experiment illustrated in B (with T.D.L.), we also recorded responses to a number of brighter flashes. The complete set of amplitude (E) and time-to-peak (F) measurements obtained for this subject are plotted in the lower two panels. In E, the curve indicates the Naka–Rushton saturation function (, r_max= 110 μV, and Q₀= 0.12 Td s), and the blue line is its linear asymptote.

The records from Fig. 1A and B have been transformed into flash sensitivities in Fig. 1C and D by dividing by the corresponding flash strength delivered. The responses to the two dimmest flashes are shown in grey (without their fitted curves), on the assumption that they are likely to be influenced more prominently by sensitive inner retinal activity, e.g. the scotopic threshold responses, pSTR (Korth et al. 1994; Frishman et al. 1996a) and nSTR (Frishman & Sieving, 1995), reviewed, for example, in Robson & Frishman (1999). Although a very dim background (∼0.02 Td) was applied during collection of the families to reduce the influence of sensitive inner retinal activity (Frishman et al. 1996b; Naarendorp et al. 2001; Saszik et al. 2002), it is entirely plausible that this activity was not eliminated. Thus the two dimmest flash responses (lower two traces, Fig. 1A and B), and the flash sensitivities obtained from these records (grey traces, Fig. 1C and D), may contain significant contributions from both rod bipolar cells and sensitive inner retinal (amacrine/ganglion) cells.

Choice of criterion response amplitude

Our selection of an appropriate response amplitude for the ‘criterion amplitude’ approach was guided by two main considerations: the degree to which the b-wave could be regarded as exhibiting (a) linearity of amplitude, and (b) invariance of kinetics (see for example, Fig. 6, Naarendorp et al. 2001; Fig. 5, Robson et al. 2004). Figure 1E plots estimates of response amplitude obtained by Poisson fitting for subject T.D.L. The curve is the classic Naka–Rushton saturation function, while the blue line indicates its linear asymptote. For the purposes of the fitting, the amplitude measurements for the two dimmest flashes were ignored (on the grounds that they probably included substantial STR components). From this description we found that the response to the third flash strength (∼20 μV) deviated ∼8% from linearity, while the response to the fourth (∼30 μV) diverged ∼24%. However, the response to the third flash strength provided a slightly shorter estimate of the time-to-peak (∼116 ms), than did the response to the fourth (∼120 ms) (Fig. 1F), possibly due to noise, or a residual contribution from sensitive inner retinal activity.

We eventually settled upon the response to the fourth flash strength (∼0.03 Td s), which typically elicited a response amplitude of 20–30 μV (arrowed in Fig. 1A and B). Despite deviation from strict linearity of amplitude, the response was still able to provide a reasonable approximation of the flash sensitivity, and in line with the principal goal of the investigation, provided the best available monitor of ‘invariant’b-wave response kinetics (with minimal intrusion from other components). As described in the Methods, the flash strength used under other conditions (light adaptation or dark adaptation) was adjusted to elicit as nearly as possible the same response amplitude.

Responses obtained using the criterion response amplitude approach: dark adaptation and light adaptation

Figure 2 shows representative responses obtained using the criterion response amplitude approach from subject A.M.C.; the left column (Fig. 2A) is for dark adaptation following a near-total bleach, while the right column (Fig. 2B) is for light adaptation to a series of background intensities. In Fig. 2A, the uppermost trace was obtained at late times during recovery. As the post-bleach recordings were obtained on a dim background (∼0.02 Td) to minimize sensitive inner retinal activity, this trace is commensurate with the second trace from the top in Fig. 2B. The lowermost traces are responses under ‘substantially adapted’ conditions, i.e. at early post-bleach times (Fig. 2A), or on a bright background (Fig. 2B). The peak of each fitted response is indicated by an open circle.

Figure 2. Criterion amplitude b-wave responses.

A, criterion amplitude responses recorded during dark adaptation recovery following a bleach of 97% (post-bleach time increases from bottom to top). The responses represent averages obtained during consecutive 2 min intervals following bleach delivery, but to avoid overcrowding, a few of the intervals have been omitted. The post-bleach time (in bold), and flash strength delivered (in regular font) are noted next to each average. B, comparison with responses obtained in darkness, and on eight backgrounds of progressively higher intensity (top-bottom). The intensity of the steady background illumination (in bold), and flash strength (regular font) are indicated. In both panels A and B, each of the responses was averaged from up to 49 traces. The dashed curves shown at the bottom of each panel plot the predicted rod photoreceptor contribution at the indicated post bleach time (panel A), or background intensity (B) based on the flash strength delivered. Note that when comparing responses of panels A and B produced by flashes of similar strength, the size of the a-wave (i.e. the visible region of the rod signal), is virtually identical. Likewise, the rising phase of the rod photoreceptor signal, which was modelled as described in Methods for the whole range of flash strengths, was quite comparable in the two cases. Subject in both experiments, A.M.C.

At a relatively early time of 10–12 min after the large bleach (Fig. 1A, bottom), the response time-to-peak was shortened by a factor of ∼2 from its prebleach or final post-bleach value of ∼120 ms; at this same time the response was desensitized by ∼200-fold, as shown by the fact that the test flash had been increased from 0.03 to 6.7 Td s. At successively later times after the bleach (i.e. proceeding upwards in the left panel), the time-to-peak of the flash response steadily slowed and the sensitivity steadily increased. The average values for sensitivity and for time-to-peak, obtained from two experiments delivering a near-total bleach to this subject, are plotted against post-bleach time subsequently in Fig. 5A and B (filled diamonds).

Figure 5. Flash sensitivity and time-to-peak following bleaching exposures, and in the presence of backgrounds of increasing intensity.

Panels A and B plot measurements of b-wave flash sensitivity and time-to-peak, before and up to 35 min after bleaching exposures of six strengths (▵, 2.6%; •, 6.3%; ⋄, 12%; ▵, 29%; ○ 57%; and ♦, 99%), for subject A.M.C. The grey vertical lines show the timing of bleach delivery. Panels C and D plot b-wave flash sensitivity and time-to-peak, measured during light adaptation (grey circles) for the same subject. In both A and C the measurements of sensitivity have been normalized, as described in Methods. In C, the reduction in sensitivity during light adaptation has been described by Weber's Law (eqn (3), continuous black curve). In D the acceleration of kinetics found during light adaptation has been fitted both with a Scaled Weber curve (eqn (4), continuous grey curve), and with an empirical description (eqn (5), continuous black curve). In A, the continuous curves plot the I_equiv predictions of eqn (6b), substituted into Weber's Law (eqn 3). In B, the set of curves show the I_equiv predictions of eqn (6a), substituted into the empirical t_peak relation (eqn 5). In both cases, additional parameters were included to allow for the presence of the dim steady background during the recordings. We have attempted to correct the measurements in C and D in order to remove contributions from the rod signal (○). This was performed by subtracting the predicted rod signal from the records (see Fig. 2), and then using the Poisson curve to fit r_peak and t_peak to the subtracted waveforms. Based on this estimate, it would appear that the rod signal has little effect on the measurements except perhaps at the two brightest backgrounds. Given that the proportional contribution of the rod signal to the measurements is likely to be reasonably similar during dark and light adaptation (see Fig. 2), the equivalent background derived from the measurements should accurately reflect events in rod bipolar cells.

In the right-hand column of Fig. 2, the set of traces obtained on steady backgrounds of progressively greater intensity appear broadly similar to the traces obtained in the left-hand column at progressively earlier post-bleach times. The average values for sensitivity and time-to-peak, obtained from five experiments employing a wide range of backgrounds with this subject, are plotted against background intensity subsequently in Fig. 5C and D.

Criterion amplitude responses of the kind illustrated in Fig. 2 have been replotted in Fig. 3 after conversion to sensitivity, by dividing by flash strength (as in Fig. 1C and D). The bottom row (Fig. 3E and F) plots the same two experiments as in Fig. 2, for a near-total bleach and for light adaptation, while the panels above plot experiments at bleach levels from 2.5% upwards. For the bleaching experiments (Fig. 3A–E), the traces have been colour-coded as follows: black, prebleach; coloured, over consecutive 2 min intervals; grey, late after the bleach. For each trace, the peak of the fitted response is again indicated by open circles.

Figure 3. Flash sensitivity families obtained by scaling the measured criterion amplitude responses by the flash strength delivered.

Panels A–E plot flash sensitivity families recorded at a range of times during recovery from bleaching. The calculated bleach strengths were 2.5, 6.1, 12, 30 and 97%. The time interval during which each average trace was measured is indicated using colour coding (see legend key). Some of the post-bleach averages have been omitted in these panels. Panel F shows a representative flash sensitivity family during light adaptation by steady background illumination. The results shown in panels E and F are derived from the criterion amplitude responses of Fig. 2. The maximum b-wave sensitivity varied somewhat over the six experiments, presumably due to differences in the scleral location of the recording electrode. By contrast, the maximum time-to-peak was reasonably consistent (at ∼120 ms), whether measured prebleach, or at late post-bleach times (A–E), or for the purposes of comparison, on the 0.02 Td background (second average from top, F). Subject A.M.C.

Relationship between peak sensitivity and time-to-peak

Visual inspection of the circles denoting the response peaks in each panel of Fig. 3 shows that there is a powerful correlation between sensitivity and time-to-peak, during dark adaptation and light adaptation. The nature of this relationship is examined in Fig. 4, where normalized sensitivity (S/S_D) is plotted against time-to-peak (t_peak) in linear coordinates (upper row) and in logarithmic coordinates (lower row).

In the case of photoreceptor light adaptation, it has long been known that time-to-peak and sensitivity are typically related by the power-law expression

(2a)

where n_A is the power-law exponent of adaptation (see Fuortes & Hodgkin, 1964, eqn 11; Baylor & Hodgkin, 1974, eqn 21). When normalized to the dark-adapted state, with dark-adapted time-to-peak t_D and dark-adapted sensitivity S_D, eqn (2a) becomes:

(2b)

In the left column of Fig. 4, the grey symbols plot the relationship we found for light adaptation of the scotopic b-wave responses, while the black curve (in Fig. 4A) and black line (in Fig. 4B) plot eqn (2) with an exponent of n_A= 6. For comparison, previous studies have found exponents of 6.8 in Limulus photoreceptors (Fuortes & Hodgkin, 1964), 3–5 in turtle cones (Baylor & Hodgkin, 1974), and 2.5 in toad rods (Baylor et al. 1980).

Very interestingly, a power-law relation with the same exponent also provides a good description of much of the range of the post-bleach measurements in Fig. 4, although it would appear that the results for small and large bleaches exhibit a difference in scaling. For this figure, measurements obtained with one subject (A.M.C.) have been averaged according to bleach level; each set of symbols combines results from two or three experiments (of the sort illustrated in Fig. 3), in which bleaches of similar magnitude were delivered. In the left column, the measurements following small bleaches are indicated by open triangles (2.6%) and filled circles (6.3%), and these points fall marginally below the theoretical curve and line used to describe the light adaptation measurements. In the right column, the measurements following larger bleaches are indicated by filled triangles (29%), open circles (57%), and filled diamonds (99%). The grey curve and grey line replicate those plotted in the left column, and are clearly well above the points. Over most of the recovery period following large bleaches, the points are well described by this power-law relation with the same exponent, although the relationship needs to be scaled down by a factor of ∼2 (black curve and line). Then, at late stages in recovery, the points rise almost vertically.

These findings would be explicable if, during the early stages of recovery from large bleaches, the rod bipolar cells experience not only an equivalent background that closely resembles real light, but also a compressive effect that roughly halves the sensitivity (for responses with given kinetics). Then, during the final stage of recovery, when the kinetics have fully recovered, only the compressive effect remains, and that also gradually disappears. For the recoveries from small bleaches shown in the left column, it is possible that a weaker compressive effect occurs as well.

Post-bleach recovery of b-wave flash sensitivity and kinetics

Measurements of post-bleach recovery of sensitivity and of time-to-peak are gathered in the left column of Fig. 5, following bleaches of six strengths (2.6, 6.3, 12, 29, 57 and 99%). As in Fig. 4, the points for each bleach level were obtained by averaging results from either two or three experiments.

As has been reported previously, the recovery of sensitivity (Fig. 5A) proceeds with a characteristic time course, with larger bleaches inducing a larger initial desensitization and rightward shift of recovery (Cameron et al. 2006). Our results in Fig. 5B for response kinetics suggest that similarly, following bleaching exposures, the time-to-peak recovers with a common time course, with larger bleaches inducing greater initial shortening and the rightward displacement of a common curve.

Comparison of Fig. 5A and B shows that complete recovery of the time-to-peak occurs earlier than complete recovery of sensitivity. Thus, for the largest bleach (filled diamonds), complete recovery of t_peak occurred by ∼25 min post-bleach, at which time the sensitivity remained ∼2-fold desensitized, and with complete recovery taking a further 5–10 min; this observation corresponds to the finding of the final nearly vertical region of the points in Fig. 4C for t_peak > 120 ms.

Light-adapted sensitivity and time-to-peak

In order to transform our measurements of post-bleach time-to-peak and sensitivity into ‘equivalent background intensities’, it is first necessary to describe the dependence of the two parameters, t_peak and S, on the intensity of real backgrounds, i.e. to measure light adaptation. Averaged measurements from five experiments of the kind illustrated in Figs 2B and 3F are plotted in the right column of Fig. 5.

Sensitivity

The light-adapted sensitivity, shown in Fig. 5C by the grey symbols, is well described by the continuous black curve, which plots the conventional Weber Law expression:

(3)

where I is the background intensity, I₀ is the half-desensitizing intensity, and n_W is the Weber exponent. In order to estimate these three parameters, we used least-squares fitting, as described in Methods. The values obtained for this subject (A.M.C.), and for the other three subjects, are given in Table 1.

Table 1.

Parameters describing light-adapted sensitivity and time-to-peak

	Sensitivity, S		Time-to-peak, tpeak
	Weber Law, eqn (3)		Scaled Weber, eqn (4)		Empirical, eqn (5)
Subject	I0	nW	tD	nA	tD	tslope	I1	I2
A.M.C.	0.119	0.90	127.3	6.0	122.5	34.1	0.148	21.5
M.J.P.	0.105	0.97	126.3	6.4	119.3	38.8	0.235	21.3
R.R.	0.094	0.90	122.2	6.1	118.1	29.5	0.087	37.0
T.D.L.	0.106	0.96	122.8	7.1	118.5	39.1	0.234	12.7
Mean	0.106	0.93	124.7	6.4	119.6	35.4	0.176	23.1

Time-to-peak

The dependence of time-to-peak on adapting background intensity is plotted in Fig. 5D by the grey symbols. In describing this behaviour analytically, we attempted two approaches. First, we transformed the Weber expression (eqn (3)) according to the previously determined power-law relation between sensitivity and time-to-peak (eqn (2b)), to give:

(4)

where the parameters are as defined in eqns (2b) and (3). Second, we devised an empirical relation:

(5)

where t_D is again the dark-adapted time-to-peak, t_slope describes the steepness of the change in time-to-peak with intensity, I₁ defines the intensity at which the time-to-peak begins shortening substantially, and I₂ defines the intensity beyond which it approaches an asymptotic lower limit.

To fit eqn (4) (grey curve), we used values for n_W and I₀ already determined for the decline of sensitivity, and then used least-squares fitting to determine t_D and n_A. To fit eqn (5) (black curve) we used least-squares fitting to determine all four parameters. Both curves provide a reasonable description of the light-adapted measurements of t_peak. The parameter values determined for the subject (A.M.C.) illustrated in Fig. 5, and for the other three subjects, are given in Table 1.

Inverse expressions for conversion to equivalent background intensity

Using the relationships just derived, measurements of post-bleach sensitivity and time-to-peak (of the kind illustrated in Fig. 5A and B) can be converted to ‘sensitivity-equivalent’ and ‘kinetics-equivalent’ background intensities. In doing so, we require the inverse of each of the expressions eqns (3)–(5). In addition, we need to account for the fact that (in order to minimize the contribution of sensitive inner retinal activity) the dark adaptation experiments were always performed in the presence of a continuous very dim background of intensity I_dim, together with the fact that the sensitivity measurements in the bleaching experiments have been normalized to the sensitivity, S_dim, on that dim background. The respective inverse relations can then be written as the following three expressions for equivalent background intensity, I_equiv. The first of these is the ‘Crawford transform’ given previously as eqn (5) in Cameron et al. (2006) using the Weber law:

(3′)

The second derives from the scaled Weber formulation:

(4′)

And the third derives from the empirical relation:

(5′)

Decay of equivalent background intensity following bleaching

Figure 6 plots the decay of equivalent background intensity for subject A.M.C. derived from the data in the left column of Fig. 5 by transformation using the inverse expressions (3′)–(5′). For Fig. 6A, the ‘sensitivity-equivalent background’ has been calculated from the data in Fig. 5A using the standard Crawford transform given in eqn (3′). For Fig. 6B and C the ‘kinetics-equivalent background’ has been calculated from the data in Fig. 5B by transformation with eqns (4′) and (5′).

Figure 6. The post-bleach equivalent background intensity calculated from measurements of flash sensitivity and kinetics.

A, the post-bleach flash sensitivity results (Fig. 5A) were substituted into eqn (3′) to derive the ‘sensitivity-equivalent’ background intensities. In this panel, and in panels B and C below, values above or below the vertical limits have been plotted at those limits. The black curves plot eqn (6b), with Ψ_S2= 0.24, and Ψ_S3= 0.06 decades min⁻¹, and the values for I_S2(0), k₂ and I_S3(0) reported in Table 2. B, the post-bleach measurements of time-to-peak (Fig. 5B), were substituted into eqn (5′) to obtain the ‘kinetics equivalent’ background intensities. The straight lines plot the single exponential decay of eqn (6a), with Ψ_S2= 0.24 decades min⁻¹. C, comparison of the two methods for deriving the ‘kinetics equivalent’ background from the time-to-peak measurements: eqn (5′) (from panel B above, grey points) and eqn (4′) (black points).

For the sensitivity-equivalent backgrounds plotted in Fig. 6A, the behaviour is broadly similar to that presented by Cameron et al. (2006, Fig. 8), although by using the criterion response amplitude protocol, we have been able to extend the measurements to higher values of equivalent background intensity (i.e. to earlier times) than previously. On the other hand, since the minimum test flash strength used in the current study was dimmer, the measurements were somewhat noisier at late times, so the lower limit has also been increased (from 0.01 to 0.02 Td). The kinetics-equivalent backgrounds plotted in Fig. 6B were derived using the empirical description of eqn (5′). The measurements plotted in Fig. 6C were obtained using two approaches: scaled Weber (eqn (4′), black symbols) and empirical (eqn (5′), grey symbols). In general, the values from the scaled Weber-based description (eqn 4′) declined more steeply, especially at late times, possibly because eqn (4) fits less well at dim backgrounds (see Fig. 5D).

Figure 8. Predictions for opsin regeneration and comparison with the kinetics-equivalent background.

Panel A shows the predicted free opsin concentration in rods following six bleaching strengths (2.6, 6.3, 12, 29, 57 and 99%), based on eqn (A6) of Mahroo & Lamb (2004) (ν= 0.088 min⁻¹ and K_m= 0.185). A horizontal line has been plotted at the arbitrary level of 0.2% opsin remaining. In panel B, the post-bleach time at which the opsin theory curves intersect this criterion level is plotted using grey crosses. The black circles in panel B are derived from panel B of Fig. 6, and plot the time taken for the kinetics-equivalent background to reach a criterion level of 0.1 Td. Other symbols (⋄, M.P.; ▵, R.R.; □, T.D.L) show results for the other three subjects obtained in the same way, except that the criterion was set at 0.8 Td for T.D.L.

For the kinetics-equivalent intensity derived from eqn (5′), and plotted in Fig. 6B, we found that for each bleach level the points were well-described by a straight line in these semi-logarithmic coordinates. The slope of 0.24 decades min⁻¹ corresponds to the ‘S2’ component of recovery reported previously (Lamb, 1981; Lamb & Pugh, 2004; Cameron et al. 2006), and indicates the presence of only a single exponentially decaying component described by:

(6a)

where Ψ_S2 is the slope of component S2 and I_S2(0) is its initial magnitude.

In Fig. 6A it is clear that a straight line is not sufficient to describe the points, and the curves plot the sum of two exponentially decaying components, termed ‘S2’ and ‘S3’, according to the equation given below, revised from eqn (6) of Cameron et al. (2006):

(6b)

where Ψ_S2 and Ψ_S3 are slopes for each component, set to 0.24 and 0.06 decades min⁻¹, respectively. The term k₂I_S2(0) represents the initial magnitude of the S2 component: k₂ is a scaling factor for component S2, estimated from the sensitivity rather than from the kinetics, while I_S2(0) is exactly as used in eqn (6a). I_S3(0) denotes the initial magnitude of component S3.

In the three other subjects, each studied at three bleaching strengths, the results were qualitatively closely similar to those in Fig. 6. The parameters obtained from fitting eqn (6) are listed in Table 2; in all subjects the slopes of components S2 and S3 were close to 0.24 and 0.06 decades min⁻¹.

Table 2.

Bleach exposures (Td s) and calculated bleach strengths (%), together with parameters for describing the post-bleach decay of the equivalent background derived from flash sensitivity (S) and kinetics (t_peak) (eqn 6) (see also Fig. 6A and B)

Subject				∼2%	∼6%	Approx. ∼10%	∼30%	∼60%	∼100%
A.M.C.	•	Bleach	(Td s)	2.5 × 103	5.9 × 103	1.2 × 104	3.1 × 104	7.4 × 104	5.8 × 105
			(%)	2.6%	6.3%	12%	29%	57%	99%
		IS2(0)	(Td)	0.4	2.2	8.6	52	4.7 × 102	9.9 × 103
		k2		1.2	1.3	2.0	2.3	2.5	2.5
		IS3(0)	(Td)	0.06	0.11	0.22	0.47	1.1	2.5
M.J.P.	⋄	Bleach	(Td s)	—	6.3 × 103	—	3.3 × 104	—	3.6 × 105
			(%)	—	6.1%	—	30%	—	93%
		IS2(0)	(Td)	—	2.9	—	91	—	1.2 × 104
		k2		—	1.0	—	2.0	—	2.0
		IS3(0)	(Td)	—	0.11	—	0.46	—	1.2
R.R.	▵	Bleach	(Td s)	—	5.9 × 103	—	2.9 × 104	—	3.6 × 105
			(%)	—	6.4%	—	28%	—	96%
		IS2(0)	(Td)	—	1.5	—	47	—	3.9 × 103
		k2		—	1.2	—	2	—	2.5
		IS3(0)	(Td)	—	0.08	—	0.38	—	1.6
T.D.L.	□	Bleach	(Td s)	—	5.7 × 103	—	3.2 × 104	—	3.2 × 105
			(%)	—	6.3%	—	32%	—	94%
		IS2(0)	(Td)	—	8.7	—	4.8 × 102	—	1.1 × 105
		k2		—	0.8	—	0.8	—	1.0
		IS3(0)	(Td)	—	0.07	—	0.43	—	1.1

For eqn (6), Ψ_S2 and Ψ_S3 were set at 0.24 and 0.06 decades min⁻¹. Symbol scheme as applied in Fig. 8.

Our conclusions from these observations are firstly that component S3 of dark adaptation recovery, as measured using the b-wave of the scotopic ERG, is a feature only of the sensitivity measurements and not of the time-to-peak measurements, and secondly that the sensitivity measurements yield an estimate for component S2 that is scaled up from that obtained by measuring the time-to-peak.

Relationship between sensitivity- and kinetics-equivalent background intensities

The interrelationship between the two measures of equivalent background intensity, extracted by the sensitivity-equivalence approach and the kinetics-equivalence approach, is examined in Fig. 7 for two bleach levels. In this figure, the equivalent background values have been averaged from results obtained with all four subjects. The filled circles plot the estimates extracted following small bleaches of ∼6%, and are well described by the black straight line that represents exact equality. The filled diamonds are for near-total bleaches; in double logarithmic coordinates they also exhibit unit slope, but the relationship is displaced upwards by a factor of ∼2 (grey line). Note that most of the data in this figure represents component S2, though there is some contribution from component S3 over the lower intensity range for the large bleach.

Comparison of observed kinetics-equivalent background with predictions of model of opsin regeneration

Considerable evidence supports the notion that, for psychophysical dark adaptation, the time course of recovery of the S2 component reflects the reconversion of opsin (bleached rhodopsin) to rhodopsin (reviewed in Lamb & Pugh, 2004). We now examine the corresponding situation for the kinetics-equivalent background extracted from measurements of response time-to-peak.

Figure 8A plots the predictions for the fraction of opsin remaining unregenerated, following bleaches of the six strengths that we employed, calculated according to the theory of Lamb & Pugh (2004) and Mahroo & Lamb (2004):

(7)

where Ops(t) is the fraction of opsin remaining as a function of time t, B is the fractional bleach, v is the initial rate of regeneration after a total bleach and K_m is the semi-saturation constant. Note that eqn (7) is implemented using the Lambert W function, W(x), as discussed in Lamb & Pugh (2004) and Mahroo & Lamb (2004). For Fig. 8A we used v = 0.088 min⁻¹ and K_m= 0.185; these parameter values were within the range of values previously used to describe opsin's regeneration to rhodopsin based on measurements from retinal densitometry (Fig. 10, Lamb & Pugh, 2004). In terms of these parameters, the slope of the S2 region is obtained from Lamb & Pugh (2004) eqn (15) as:

(8)

or 0.245 decade min⁻¹, represented by the parallel line region of the curves in Fig. 8A, and matching the decay time course of the kinetics-equivalent background in Fig. 6B.

The rightward displacement of the theoretical curves with increasing bleach strength also appears comparable with experiment (when the bleaching constant Q_e, is set at 10^6.7 Td s, discussed below). In order to make this comparison, we selected a criterion level of unregenerated opsin in Fig. 8A, together with a criterion level of equivalent background in Fig. 6B. The absolute values of these criterion levels are not very important (provided they are within the S2 region), but the relative levels need to be chosen so as to give corresponding times. We chose 0.2% unregenerated opsin in Fig. 8A in conjunction with 0.1 Td in Fig. 6B. Thus, the grey crosses in Fig. 8B represent the times at which the opsin regeneration theory curves intersect the dashed horizontal line (Fig. 8A); the grey curve in Fig. 8B shows this same prediction as a continuous function of bleach level, based on eqn (A17) of Mahroo & Lamb (2004):

(9)

with the parameters above.

Likewise, the filled circles in Fig. 8B plot the intersection of the straight lines in Fig. 6B with the 0.1 Td criterion level (dashed horizontal line). The remaining symbols were obtained for the results from the other three subjects in exactly the same way, except that for subject T.D.L. (open squares) the criterion level was set to 0.8 Td.

The lateral displacement of the criterion equivalent background intensities (Fig. 8B) is dependent upon the bleach strength, as calculated from eqn (A12) of Mahroo & Lamb (2004), using three parameters (v, K_m and Q_e). However, further analyses revealed that the only parameter value to appreciably influence the calculated bleach strength, and hence the horizontal position of the criterion equivalent background points in Fig. 8B, was the value used for the bleaching constant, Q_e (range trialled: 10^6.4–10⁷). The other two parameters (i.e. v and K_m) had little effect on the results, at least for the range of values that we trialled (v = 0.086–0.094; K_m= 0.17–0.2).

The best alignment of the criterion background data points to the grey curve of eqn (9) could be achieved using Q_e= 10^6.7 Td s, which provides a means for verifying the value of the bleaching constant applicable under our experimental conditions (see Methods). This value is slightly lower than applied previously (i.e. 10^6.8, Cameron et al. 2006). However, by choosing to analyse the kinetics-equivalent background in Fig. 8B, instead of the sensitivity-equivalent background (as per Fig. 10, Cameron et al. 2006), we have been able to separate the ‘light-like’ S2 component from the influence of component S3.

Based on the fact that the grey curve is able to provide a reasonable description of the experimental measurements, this suggests that for all four subjects the post-bleach recovery of rod bipolar cell flash kinetics can be described as being due to the fading of an equivalent light that arises from the presence of unregenerated opsin.

Discussion

These experiments have confirmed the notion, gained in our preliminary experiments (Cameron et al. 2006), that ERG b-wave responses exhibit accelerated response kinetics and reduced flash sensitivity following bleaches. In addition, we have shown that for a range of times in early post-bleach recovery, the sensitivity varies as the 6th power of the time-to-peak, and that a power function with the same slope (∼6) describes the relation found on backgrounds. Thus although the response families obtained following bleaching exposures are by no means identical to those recorded on backgrounds, during post-bleach recovery there exists a distinct region in which b-wave responses are both accelerated and desensitized, resembling exposure to real light of declining intensity.

The flash presentation protocol adopted for the present experiments differs from our previous fixed strength approach (Cameron et al. 2006), and involved varying the flash strength in order to produce small ‘criterion amplitude’ responses. While this approach is technically more demanding, in requiring advance estimation of the approximate test flash strength to be delivered at different post-bleach times, it has the advantage of enabling better resolution of response kinetics than is feasible with flashes of fixed strength. In addition, despite the criterion level having been set slightly above the linear response amplitude range, this should not influence the values of equivalent background intensity calculated from measurements of sensitivity (see below), since light adaptation was estimated using a similar protocol (see Cameron et al. 2006 for supporting evidence).

In order to obtain an unbiased and reasonably reliable estimate of the time-to-peak (and amplitude) of the flash responses in the presence of noise, we developed the approach of fitting an empirical curve to the recordings, in the vicinity of the peak. While previous investigations have used the criterion amplitude approach to study the dark adaptation recovery of b-wave flash sensitivity (e.g. Dowling, 1963; Green et al. 1975), to our knowledge, ours is the first study to have examined the recovery of response kinetics. In addition, we know of no other studies to have used empirical fitting as a means to estimate the time-to-peak and thereby quantify the b-wave kinetics.

The results that we obtained for post-bleach sensitivity, and for the extracted ‘sensitivity-equivalent’ background intensity following bleaches, were closely similar to those that we reported recently using the fixed flash strength approach (Cameron et al. 2006), in that they showed recovery that could be described by the fading of an equivalent background comprising two components, referred to in the literature as S2 and S3. However, in order to fit the results of the present study, we have revised our original description (i.e. eqn (6), Cameron et al. 2006), to the form given in eqn (6b). Using eqn (6b), we have again been able to show that the sensitivity-equivalent background declines initially at ∼0.24 decades min⁻¹, then at late times at ∼0.06 decades min⁻¹, closely matching the time course of components S2 and S3 seen in psychophysical dark adaptation (Lamb, 1981; Jackson et al. 1999; Lamb & Pugh, 2004).

This new expression, however, may provide an improved description of some of our other findings, such as the lower scaling for the relation between sensitivity and time-to-peak in Fig. 4C and D compared with Fig. 4A and B, and the higher scaling of the sensitivity-equivalent background in Fig. 7 following large bleaches. The magnitude of this ‘scaling’ phenomenon following large bleaches, remains roughly a constant factor of ∼2, at least until the time course of the response has virtually recovered.

In terms of the results plotted in Fig. 4C, one can think of the trajectory of post-bleach recovery comprising a movement rightwards and upwards along the black curve, until the kinetics are fully recovered (at ∼25 min after a near-total bleach), and thereafter a final vertical trajectory as the sensitivity recovers without further change in time-to-peak. In terms of Fig. 6A, the vertical positioning of the ‘S2’ component after large bleaches would seem to be elevated by a factor of ∼2 as a result of the extra desensitization. In this same panel, it is apparent that the elevation is accompanied by a late tail phase, indicating a final slow change in sensitivity that is unrelated to any change in kinetics. Our new model could perhaps better capture these results than our previously published version, because it incorporates an ‘S2’ component scaling factor, k₂.

The major new finding of our study is that the post-bleach recovery of the ERG b-wave flash response kinetics can be explained in terms of a single component of equivalent background that declines with an S2 slope of ∼0.24 decade min⁻¹. This is shown first by the results in Fig. 6B (and similar experiments, not illustrated, for the other three subjects), and is also supported by the good correspondence between experiment and theory in Fig. 8B. On the well-established view that the S2 component is caused by the presence of unregenerated opsin in the rod outer segments, we conclude that following bleaching exposures the response time course in rod bipolar cells provides a reflection primarily of the residual opsin level in the rods.

In contrast, the mechanism responsible for the ‘purely desensitizing’ S3 component remains unclear. We have previously speculated that S3 might be caused by the massive release of all-trans retinoid by the rod outer segments following large bleaches, possibly by the closure of ion channels in the photoreceptors or even in the bipolar cells themselves (Cameron et al. 2006). Currently, we have no additional information to either support or refute this possibility.

Our findings may provide an insight into the postreceptoral correlates of adaptation within the overall rod system (i.e. during dark and light adaptation). Having shown that the post-bleach recovery of b-wave flash sensitivity can be described in terms of an equivalent background that fades with values reported for S2 and S3 (Cameron et al. 2006; and this paper), we are now seeking to discover precisely how these findings correlate with dark adaptation at higher levels of visual system, by measuring the recovery of visual threshold psychophysically, in a series of experiments with the same subjects (Ruseckaite et al. in preparation).

Our current findings may also provide an explanation for the postreceptoral basis of adjustments in the rod system's temporal resolution, which is longest under dark-adapted conditions in order to maximize the number of photon absorptions registered, but which becomes shorter as the rod system adjusts to higher scotopic levels of ambient illumination (i.e. from conditions of starlight to twilight) (Barlow, 1958; Conner, 1982; Sharpe et al. 1988), or after exposure to bright light (Friedburg et al. 1998). This improvement of temporal resolution, in concert with regulation of sensitivity and changes in spatial summation, is likely to play a key role in the avoidance of saturation, and provide the basis of effective vision over the rod system's 10⁶-fold operating range.

Although there are clear differences in the way in which the rod system's temporal resolution can be measured at the cellular level, and behaviourally, experiments by Friedburg et al. (1998) (Fig. 7) suggest that following exposure to a large bleach (∼66% of the rhodopsin), the temporal summation of visual responses (determined from the ‘critical duration’, t_c; see Fig. 1, Krauskopf & Mollon, 1971), is initially accelerated, and that over the course of the next ∼10 min, recovery of summation at threshold appears mediated by the cone system. After ∼10 min post-bleach, the rod system appears to have recovered sufficiently to begin mediating summation, and thereafter, summation times steadily lengthen, probably as a single component. Reminiscent of our results in Fig. 6B for the 60% bleach, the temporal summation returns to dark-adapted levels by ∼20 min post-bleach, while visual threshold remains elevated for a further ∼10 min. Thus, based on the findings of Friedburg et al. (1998), and the data presented in this study, the S3 component of (desensitization-) equivalent background that is found in psychophysical experiments (Lamb, 1981) may represent a pure desensitization present in the rod system rather than a phenomenon closely equivalent to light.

Conclusions

The calculated post-bleach ‘equivalent background’ estimated for ERG scotopic b-wave responses declines with values reported for components S2 and S3 of psychophysical dark adaptation. In rod bipolar cells S2 can be tracked in isolation from S3 by examining the post-bleach recovery of response kinetics. The S2 behaviour is likely to be due to quenching of opsin activity within the rods by regeneration to rhodopsin. Thus, much of the dark adaptation behaviour of the overall scotopic visual system appears to be set by events measurable at the first synapse of rod vision.