Logo Logo
Hilfe
Kontakt
Switch language to English
Internal structure of the fly elementary motion detector
Internal structure of the fly elementary motion detector
Flies use visual motion information for flight control, stabilization and object tracking. However, information about local motion such as direction and velocity is not explicitly represented at the level of the retina but must be computed by subsequent motion detection circuitry. The output of these circuits can be recorded in large, direction-selective lobula plate tangential cells, that integrate over hundreds of elementary motion detectors. The computational structure of these detectors is best described by the Reichardt model, where the signals from two neighboring photoreceptors become multiplied after one of them has been delayed. However, the neural correlate of the Reichardt Detector, i.e. the identity, physiology and connectivity of the constituting cells, has escaped further characterization due to technical difficulties in recording from these small neurons. In this thesis, I investigated the internal structure of the fly motion detection circuit by a combination of electrophysiology, computer simulations and mathematical modeling. First, I studied the effect of the mean luminance on motion detection. I found that the response strength of lobula plate tangential cells strongly depends on stimulus contrast but barely changes as a function of mean luminance. Adaptation to a new mean luminance follows an exponential decay with a time constant of several hundred milliseconds. I next investigated the structural consequences of splitting the visual input into ON and OFF components, as recently discovered in the fruit fly. The original Reichardt Detector can be refined by incorporating these findings, giving rise to two alternative structures. The 4-Quadrant-Detector consists of four independent subunits of the Reichardt type, correlating ON with ON, OFF with OFF, ON with OFF and OFF with ON signals. In contrast, the 2-Quadrant-Detector consists of two subunits only, that correlate ON with ON and OFF with OFF signals. In order to distinguish between these two models, I first stimulated flies with apparent motion stimuli consisting of a sequence of two brightness steps at neighboring locations, while recording the motion detector output in lobula plate tangential cells of the blow fly. I found strongly direction-selective responses to ON-ON and OFF-OFF sequences, but also to ON-OFF and OFF-ON sequences. At first sight, these results seem to support the 4-Quadrant-Detector. However, I showed with simulations and an analytical treatment that the 2-Quadrant-Detector, when equipped with an appropriate preprocessing stage, is capable of reproducing such responses as well. Based on predictions from model simulations, I designed a new stimulus protocol consisting of a sequence of short brightness pulses instead of steps. For such stimuli, the 2-Quadrant-Detector does not produce significant responses to ON-OFF and OFF-ON sequences, in contrast to the 4-Quadrant-Detector. The corresponding recordings cannot be reconciled with the 4-Quadrant-Detector but are in good agreement with the 2-Quadrant-Detector. I therefore conclude that the internal structure of the y elementary motion detector consists of two non-interacting subunits for detecting ON and OFF motion, respectively. These results mark an important step in the ongoing dissection of the insect motion detection circuit by providing an updated model that better matches the structure and physiology of the corresponding neural hardware.
Not available
Eichner, Hubert
2012
Englisch
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Eichner, Hubert (2012): Internal structure of the fly elementary motion detector. Dissertation, LMU München: Fakultät für Biologie
[thumbnail of Eichner_Hubert.pdf]
Vorschau
PDF
Eichner_Hubert.pdf

6MB

Abstract

Flies use visual motion information for flight control, stabilization and object tracking. However, information about local motion such as direction and velocity is not explicitly represented at the level of the retina but must be computed by subsequent motion detection circuitry. The output of these circuits can be recorded in large, direction-selective lobula plate tangential cells, that integrate over hundreds of elementary motion detectors. The computational structure of these detectors is best described by the Reichardt model, where the signals from two neighboring photoreceptors become multiplied after one of them has been delayed. However, the neural correlate of the Reichardt Detector, i.e. the identity, physiology and connectivity of the constituting cells, has escaped further characterization due to technical difficulties in recording from these small neurons. In this thesis, I investigated the internal structure of the fly motion detection circuit by a combination of electrophysiology, computer simulations and mathematical modeling. First, I studied the effect of the mean luminance on motion detection. I found that the response strength of lobula plate tangential cells strongly depends on stimulus contrast but barely changes as a function of mean luminance. Adaptation to a new mean luminance follows an exponential decay with a time constant of several hundred milliseconds. I next investigated the structural consequences of splitting the visual input into ON and OFF components, as recently discovered in the fruit fly. The original Reichardt Detector can be refined by incorporating these findings, giving rise to two alternative structures. The 4-Quadrant-Detector consists of four independent subunits of the Reichardt type, correlating ON with ON, OFF with OFF, ON with OFF and OFF with ON signals. In contrast, the 2-Quadrant-Detector consists of two subunits only, that correlate ON with ON and OFF with OFF signals. In order to distinguish between these two models, I first stimulated flies with apparent motion stimuli consisting of a sequence of two brightness steps at neighboring locations, while recording the motion detector output in lobula plate tangential cells of the blow fly. I found strongly direction-selective responses to ON-ON and OFF-OFF sequences, but also to ON-OFF and OFF-ON sequences. At first sight, these results seem to support the 4-Quadrant-Detector. However, I showed with simulations and an analytical treatment that the 2-Quadrant-Detector, when equipped with an appropriate preprocessing stage, is capable of reproducing such responses as well. Based on predictions from model simulations, I designed a new stimulus protocol consisting of a sequence of short brightness pulses instead of steps. For such stimuli, the 2-Quadrant-Detector does not produce significant responses to ON-OFF and OFF-ON sequences, in contrast to the 4-Quadrant-Detector. The corresponding recordings cannot be reconciled with the 4-Quadrant-Detector but are in good agreement with the 2-Quadrant-Detector. I therefore conclude that the internal structure of the y elementary motion detector consists of two non-interacting subunits for detecting ON and OFF motion, respectively. These results mark an important step in the ongoing dissection of the insect motion detection circuit by providing an updated model that better matches the structure and physiology of the corresponding neural hardware.