Figure 8. Microsaccadic rhabdomere contractions and refractory photon sampling improve visual resolution of moving objects.

(A) High-speed videos showed fast lateral R1-R7 rhabdomere movements to blue/green flashes, recorded under far-red illumination that Drosophila barely saw (Wardill et al., 2012). (B) Rhabdomeres moved inside those seven ommatidia (up-right: their pseudopupil), which faced and absorbed the incident blue/green light, while the others reflected it. Rhabdomeres moved frontwards 8–20 ms after a flash onset, being maximally displaced 70–200 ms later, before returning. (C) Movements were larger and faster the brighter the flash, but slower than R1-R6s’ voltage responses. (D) Movements followed R1-R6s’ logarithmic light-sensitivity relationship. Concurrently, given the ommatidium optics (Stavenga, 2003b; Gonzalez-Bellido et al., 2011), R1-R6s’ receptive fields (RFs) shifted by 0.5–4.0^o. (E) Rhabdomeres moved along the eye’s horizontal (red) axis, with little vertical components (black), adapting to ~ 30% contractions in ~ 10 s during 1 s repetitive flashing. (F) Moving ommatidium structures. Cone and pigment cells, linking to the rhabdomeres by adherens-junctions (Tepass and Harris, 2007), formed an aperture smaller than the rhabdomeres’ pseudopupil pattern. Rhabdomeres moved ~ 2 times more than this aperture, and ~ 10 times more than the lens. (G–H) Simulated light inputs and photoreceptor outputs for the classic theory and new ‘microsaccadic sampling’-hypothesis when two dots cross a R1-R6’s RF (i) front-to-back at saccadic speeds. (G) In the classic model, because the rhabdomere (ii) and its broad RF (i) were immobile (ii), light input from the dots fused (iii), making them neurally unresolvable (iv). (H) In the new model, with rhabdomere photomechanics (ii) moving and narrowing its RF (here acceptance angle, ∆ρ, narrows from 8.1^o to 4.0^o), light input transformed into two intensity spikes (iii), which photoreceptor output resolved (iv). (I) New predictions matched recordings (Figure 8—figure supplement 1). Details in Appendixes 7–8.

Figure 8—figure supplement 1. Microsaccadic sampling hypothesis predicts realistic voltage output to two bright dots crossing a R1-R6’s receptive field in saccadic speeds.

(A) Receptive fields of four R1-R6 photoreceptors, measured with 25 light point stimulator (Appendix 4). (B) Microsaccadic sampling hypothesis predicts that because the rhabdomeres move photomechanically, the photoreceptors’ receptive fields move in the opposite direction and narrow transiently (acceptance angles, ∆ρ, change from 8.2 to 9.5^o to 3.5–4.5^o). (C) The resulting light input for each tested photoreceptor was predicted from its measured receptive field in (A) by the microsaccadic sampling hypothesis. (D and E) These inputs then drove our biophysically realistic R1-R6 model, predicting the photoreceptor voltage output, which was compared to the corresponding real recordings. The simulated R1-R6 output closely resembled the recorded R1-R6 output of the same cells to saccadic two dot stimuli.