bioRxiv preprint doi: https://doi.org/10.1101/083691; this version posted June 1, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1 Seeing through moving eyes - microsaccadic information sampling provides 2 Drosophila hyperacute vision 3 4 Mikko Juusola1,2*‡, An Dau2‡, Zhuoyi Song2‡, Narendra Solanki2, Diana Rien1,2, David Jaciuch2, 5 Sidhartha Dongre2, Florence Blanchard2, Gonzalo G. de Polavieja3, Roger C. Hardie4 and Jouni 6 Takalo2 7 8 1National Key laboratory of Cognitive Neuroscience and Learning, Beijing, Beijing Normal 9 University, Beijing 100875, China 10 2Department of Biomedical Science, University of Sheffield, Sheffield S10 T2N, UK 11 3Champalimaud Neuroscience Programme, Champalimaud Center for the Unknown, Lisbon, 12 Portugal 13 4Department of Physiology Development and Neuroscience, Cambridge University, Cambridge CB2 14 3EG, UK 15 16 *Correspondence to: [email protected] 17 ‡ Equal contribution 18 19 Small fly eyes should not see fine image details. Because flies exhibit saccadic visual behaviors 20 and their compound eyes have relatively few ommatidia (sampling points), their photoreceptors 21 would be expected to generate blurry and coarse retinal images of the world. Here we 22 demonstrate that Drosophila see the world far better than predicted from the classic theories. 23 By using electrophysiological, optical and behavioral assays, we found that R1-R6 24 photoreceptors’ encoding capacity in time is maximized to fast high-contrast bursts, which 25 resemble their light input during saccadic behaviors. Whilst over space, R1-R6s resolve moving 26 objects at saccadic speeds beyond the predicted motion-blur-limit. Our results show how 27 refractory phototransduction and rapid photomechanical photoreceptor contractions jointly 28 sharpen retinal images of moving objects in space-time, enabling hyperacute vision, and explain 29 how such microsaccadic information sampling exceeds the compound eyes’ optical limits. These 30 discoveries elucidate how acuity depends upon photoreceptor function and eye movements. 31 32 INTRODUCTION 33 The acuity of an eye is limited by its photoreceptor spacing, which provides the grain of the retinal 34 image. To resolve two stationary objects, at least three photoreceptors are needed for detecting the 35 intensity difference in between. To resolve two moving objects is harder, as vision becomes further 36 limited by each photoreceptor’s finite integration time and receptive field size (Srinivasan & Bernard, 37 1975; Juusola & French, 1997; Land, 1997). 38 Nevertheless, animals - from insects to man - view the world by using saccades, fast 39 movements, which direct their eyes to the surroundings, and fixation intervals between the saccades, 40 during which gaze is held near stationary (Land, 1999). Because of photoreceptors’ slow integration- 41 time, saccades should blur image details and these are thought to be sampled when gaze is stabilized. 42 Thus, information would be captured during fixations whilst during saccades animals would be 43 effectively blind. This viewpoint, however, ignores fast photoreceptor adaptation, which causes 44 perceptual fading during fixation (Ditchburn & Ginsborg, 1952; Riggs & Ratliff, 1952), reducing 45 visual information. Therefore, to maximize information and acuity, it is plausible that evolution has 46 optimized photoreceptor function in respect to visual behaviors and needs. 47 We have now devised a suite of new experimental and theoretical methods to study this 48 question both in time and over space in Drosophila R1-R6 photoreceptors. The Drosophila compound 49 eyes are composed of ~750 seemingly regular lens-capped modules called the ommatidia, which 50 should provide the fly a panoramic visual field of low optical resolution (Barlow, 1952; Land, 1997). 51 Each ommatidium contains eight photoreceptor cells (R1-R8), pointing to seven different directions. 52 The ultraviolet and blue-green-sensitive outer photoreceptors, R1-R6, initiate the motion vision 53 pathway, whilst the central R7 and R8, which lie on top of each other, detect different colors from one 54 direction (Wardill et al., 2012). Owing to the eye’s neural superposition principle, R1, R2, R3, R4, R5 1 bioRxiv preprint doi: https://doi.org/10.1101/083691; this version posted June 1, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1 and R6, each from a separate neighboring ommatidium, also point to the same direction. By pooling 2 their output for synaptic transmission, the photoreceptor spacing (spatial resolution) effectively 3 matches the ommatidium spacing (average interommatidial angle, Δφ = 4.5o (Götz, 1964; Land, 1997; 4 Gonzalez-Bellido et al., 2011) but the signal-to-noise ratio of the transmitted image could improve by 5 √6 (van Steveninck & Laughlin, 1996; Zheng et al., 2006). 6 Here we show how evolution has improved Drosophila vision beyond these classic ideas, 7 suggesting that light information sampling in R1-R6 photoreceptors is tuned to saccadic behavior. 8 Our intracellular recordings reveal that R1-R6s capture 2-to-4-times more information in time 9 than previous maximum estimates (Juusola & Hardie, 2001a; Song et al., 2012; Song & Juusola, 10 2014) when responding to high-contrast bursts (periods of rapid light changes followed by quiescent 11 periods) that resemble light input from natural scenes generated by saccadic viewing. Biophysically- 12 realistic model simulations suggest that this improvement largely results from interspersed “fixation” 13 intervals, which sensitize photoreceptors - by relieving their refractory microvilli (Song et al., 2012; 14 Song & Juusola, 2014; Juusola et al., 2015) - to sample more information from phasic light changes. 15 Remarkably, over space, our intracellular recordings, high-speed microscopy and modeling 16 further reveal how photomechanical photoreceptor contractions (Hardie & Franze, 2012) work 17 together with refractory sampling to improve spatial acuity. We discover that by actively modulating 18 light input and photoreceptor output, these processes reduce motion blur during saccades and 19 adaptation during gaze fixation, which otherwise could fade vision (Ditchburn & Ginsborg, 1952; 20 Riggs & Ratliff, 1952; Land, 1997). The resulting phasic responses sharpen retinal images by 21 highlighting the times when visual objects cross a photoreceptor’s receptive field, thereby encoding 22 space in time (see also: Ahissar & Arieli, 2001; Donner & Hemilä, 2007; Rucci et al., 2007; Kuang et 23 al., 2012a; Kuang et al., 2012b; Franceschini et al., 2014; Viollet, 2014). Thus, neither saccades nor 24 fixations blind the flies, but together improve vision. 25 Incorporation of this novel opto-mechano-electric mechanism into our ‘microsaccadic 26 sampling’-model predicts that Drosophila can see >4-fold finer details than their eyes’ spatial 27 sampling limit – a prediction directly confirmed by optomotor behavior experiments. By 28 demonstrating how fly photoreceptors’ fast microsaccadic information sampling provides hyperacute 29 vision of moving images, these results change our understanding of insect vision, whilst showing an 30 important relationship between eye movements and visual acuity. 31 32 RESULTS 33 These results establish that Drosophila exploit image motion (through eye movements) to see spatial 34 details, down to hyperacute resolution. A fly’s visual acuity is limited by how well its photoreceptors 35 resolve different photon rate changes, and their receptive field sizes. However, because each 36 photoreceptor’s signal-to-noise ratio and receptive field size adapt dynamically to light changes, 37 acuity also depends upon the eye movements that cause them. To make these relationships clear, the 38 results are presented in the following order: 39 1st We show that photoreceptors capture most visual information from high-contrast bursts, and 40 reveal how this is achieved by refractory photon sampling and connectivity (Figures 1-5). 41 2nd We show that saccades and gaze fixations in natural environment results in such high-contrast 42 bursts, implying that eye movements work with refractory sampling to improve vision 43 (Figure 6). 44 3rd We demonstrate that photoreceptors contract to light in vivo and explain how these 45 microsaccades move and narrow their receptive fields (Figures 7-8) to sharpen light input and 46 photoreceptor output in time. 47 4th Collectively, these dynamics predict that Drosophila see finer spatial details than their 48 compound eyes’ optical resolution over a broad range of image velocities (Figure 9), and we 49 verify this by optomotor behavior (Figure 10). 50 Videos 1-4 and Appendixes 1-10 explain in detail the new ideas, methods, experiments and theory 51 behind these results. 52 53 Breaking the code by coupling experiments with theory 54 To work out how well a Drosophila R1-R6 photoreceptor can see the world, we compared 55 intracellular recordings with realistic theoretical predictions from extensive quantal light information 2 bioRxiv preprint doi: https://doi.org/10.1101/083691; this version posted June 1, 2017. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. 1 sampling simulations (Appendixes 1-3), having the following physical