Spectral sensitivity. (A) Average spectral sensitivity functions (SSFs) of P. xuthus photoreceptors; the plot is modified with permission from Arikawa (2003); here, error bars represent standard error of the mean. Green-sensitive, red-sensitive and broadband photoreceptors are the short visual fibers (SVFs) terminating in the lamina. (B) SSFs of individual LMCs are shown as thin gray traces; two dissimilar SSFs, a narrow and a wide one, are shown in green and yellow–green, respectively. The thick black trace is the average SSF; here and in all following figures, error bars represent standard deviation.

Light response properties of the stained LMCs. (A–C) Voltage responses to 250 ms light pulses (horizontal lines). For each LMC, responses at two light intensities are presented: at an intensity where the LMC produced the highest amplitude off-spike [black traces, usually at neutral density (ND)2], and at a 10-fold lower light level (gray traces). (A) Recordings from the L1/2 cell shown in Fig. 1A,B. (B) Recordings from the L1/2 cell shown in Fig. 1C,D; the arrow indicates the off-spike amplitude. (C) Recordings from the L3/4 cell shown in Fig. 1F,G. (D) Distribution of off-spike amplitudes from 41 spiking LMCs; the off-spike amplitude threshold was 7.2 mV (see Materials and Methods).

Spiking in LMCs. (A) Typical responses of a spiking LMC to a 250 ms light stimulus at different light intensities (ND5–0). Horizontal bars (see also B and E) indicate the light stimulus; dashed gray lines (see also B) indicate the resting potential. (B) Responses of a different LMC to 10 ms pulses. (C) Dependence of hyperpolarization on light intensity for responses to the 10 ms stimulus over a 10⁵ range of intensities (I; I₀, intensity of a non-attenuated stimulus). (D) Dependence of off-spike amplitudes on light intensity; small sub-threshold off-transients in dim light are shown as white squares, while ‘regular’ computer-detected spikes are shown as gray circles. Data in C and D were obtained from the LMC in B. (E) Sustained spiking during 4 s constant light stimulation in the dark.

Responses of LMCs to current injections in the dark. (A) Responses to a series of 250 ms current pulses; the stimulation protocol is shown above. Here and elsewhere, 0 mV indicates the resting potential. Inset to the right shows a magnified part of the response after the end of the current pulses. (B) Responses to a series of 500 ms current pulses (shown above), in −0.1 nA increments, followed by a 0.2 nA pulse. Inset to the right shows a magnified part of the response at the start of the depolarizing pulses. (C) Voltage responses to 250 ms current pulses (shown above) in the dark. Inset to the right shows oscillatory voltage responses after the end of strong hyperpolarizing current pulses. (D) Voltage responses to a stimulation protocol (shown above) consisting of 500 ms current pulses, in −0.4 nA increments, followed by a series of 500 ms current pulses, in 0.1 nA increments. Inset to the right shows membrane potential changes at the onset of the depolarizing testing pulses.

Effects of hyperpolarization on spiking. (A) A spiking LMC was first hyperpolarized with a −0.3 nA injection and then additionally stimulated with 500 ms current pulses (shown above). (B) A spiking LMC was stimulated with 100 ms light pulses (green box above) of the same intensity during hyperpolarization by current pulses to different membrane potentials. (C) A non-spiking LMC was stimulated by 20 ms light pulses during hyperpolarization by current pulses.

Responses to naturalistic stimulation. Sections of responses (1 s duration) of a spiking (blue) and a non-spiking (red) LMC to 15 s naturalistic stimuli (NS; black trace above) obtained at the same light level, ND1. Responses were de-trended.

Spiking cell responses to NS at different illumination levels. (A) Responses (6 s duration) to the same NS (shown above) at four light levels. The superimposed gray traces are trend lines obtained by low-pass filtering the response traces at a 3 dB cutoff frequency of 1 Hz; vertical bars above each trace indicate spikes. (B) Normalized dependency of mean spike rate on light intensity during 15 s of naturalistic stimulation are shown for seven LMCs; numbers in the key indicate the maximal number of spikes in each cell.

Spikes accurately code stimulus features. (A) A 3 s section of a spiking LMC responses to NS at ND2; spikes (blue vertical bars above the trace) were detected using our computer-based algorithm. (B) A corresponding raster plot detailing positions of spikes (vertical bars) during five trials; spikes from the trace in A are shown in ‘trial 2’. (C) Spike-triggered averages were obtained using five spiking cells stimulated with the same NS; error bars are s.d. The delay between stimulus onset and the LMC voltage response was determined by cross-correlating the stimulus and voltage-response traces. The values in colour are average lags, indicating the onset of spike.