The ability of the retina to adapt to changes in mean light intensity and contrast is well known. Classically, however, adaptation is thought to affect gain but not to change the visual modality encoded by a given type of retinal neuron. Recent findings reveal unexpected dynamic properties in mouse retinal neurons that challenge this view. Specifically, certain cell types change the visual modality they encode with variations in ambient illumination or following repetitive visual stimulation. These discoveries demonstrate that computations performed by retinal circuits with defined architecture can change with visual input. Moreover, they pose a major challenge for central circuits that must decode properties of the dynamic visual signal from retinal outputs.

中文翻译：

---

灵活的神经硬件支持 Retina 中的动态计算

视网膜适应平均光强度和对比度变化的能力是众所周知的。然而，传统上认为适应会影响增益但不会改变由给定类型的视网膜神经元编码的视觉模式。最近的研究结果揭示了挑战这一观点的小鼠视网膜神经元中意想不到的动态特性。具体而言，某些细胞类型会随着环境照明的变化或重复视觉刺激而改变它们编码的视觉模式。这些发现表明，由具有定义架构的视网膜电路执行的计算可以随着视觉输入而改变。此外，它们对必须解码来自视网膜输出的动态视觉信号的特性的中央电路构成了重大挑战。