Motion detection by the mammalian retina has been thought to rely largely on the intrinsic biophysics of the dendrites of starburst amacrine cells (SACs). Sebastian Seung and colleagues have combined new machine-learning techniques with crowdsourcing through the Eye Wire brain-mapping game to redraw the wiring diagram for amacrine cells and bipolar cells. Their results show that direction selectivity is established at the presynaptic level - in the spatiotemporal inputs to the amacrine cells - identifying neural circuits rather than intrinsic properties of SACs as the key to direction selectivity. This new model brings the mouse retina closer in certain respects to the Reichardt motion detector characteristic of insect vision.
展开▼
机译：哺乳动物视网膜的运动检测被认为主要依赖于星爆无长突细胞（SAC）树突的固有生物物理学。 Sebastian Seung及其同事通过Eye Wire脑映射游戏将新的机器学习技术与众包相结合，以重绘无长突细胞和双极细胞的接线图。他们的研究结果表明，方向选择性建立在突触前水平上（在无长突细胞的时空输入中），将神经回路而非SAC的固有特性识别为方向选择性的关键。这种新模型在某些方面使鼠标视网膜更接近昆虫视觉的Reichardt运动检测器特性。
展开▼
