摘要视觉行为学评估能够在动态、量化和任务依赖的条件下，反映眼科疾病动物模型中从视觉感知到行为输出的全链路功能状态，是连接神经环路机制与临床表型的重要研究工具。近年来，随着刺激控制、自动化识别和多模态融合等技术的发展，视觉行为学评估方法在检测灵敏度、数据一致性及跨实验室可比性方面均取得显著提升。根据反应特征，可将视觉行为学评估大致分为自发反应范式与操作性任务范式。前者包括视动反应、视动性眼震反应、光暗偏好测试和突现视觉刺激响应等，依托先天性反射或本能行为，适于早期功能筛查和病程监测；后者包括Morris水迷宫、视觉梯形水迷宫、二选一强迫选择任务、触摸屏判别系统及虚拟现实导航平台等，可在严格控制的刺激与反馈条件下，解析图形识别、对比敏感、空间导航与感知决策等高级功能。多项研究证实，这些方法的功能学指标与视网膜结构、电生理参数及干预疗效具有良好相关性，且这些功能学指标已在基因治疗、干细胞移植及药物干预等研究中得到验证。当前的挑战主要在于跨实验室重复性不足、结果易受非特异因素干扰及任务参数缺乏统一标准。未来，应加强任务流程与刺激参数的标准化建设，推动与在体成像、电生理等技术的多模态联合，实现从刺激输入、神经环路处理到行为输出的因果推断，并在任务设计中与临床功能评估构念对齐，建立跨物种、跨阶段的功能比较框架，从而进一步提升临床前研究的机制解释力与临床转化价值。

abstractsBehavioral assessment of visual function can dynamically and quantitatively capture the full-chain functional state from visual perception to behavioral output in ophthalmic disease animal models, serving as a crucial tool for linking neural circuit mechanisms with clinical phenotypes. In recent years, advances in stimulus control, automated recognition, and multimodal integration have markedly improved the sensitivity, data consistency, and inter-laboratory reproducibility of these methods. Based on response characteristics, visual behavioral assessments can be broadly categorized into spontaneous response paradigms and operant task paradigms. The former include optomotor response, optokinetic reflex, light/dark transition test, and looming visual stimuli, which rely on innate reflexes or instinctive behaviors and are suitable for early functional screening and disease monitoring. The latter include the Morris water maze, visual water task, two-alternative forced choice task, touchscreen-based discrimination systems, and virtual reality navigation platforms, which, under strictly controlled stimulus and feedback conditions, enable detailed analysis of higher-order functions such as pattern recognition, contrast sensitivity, spatial navigation, and perceptual decision-making. Numerous studies have demonstrated that functional indicators derived from these methods correlate well with retinal structure, electrophysiological parameters, and therapeutic outcomes, and these functional indicators have been validated in gene therapy, stem cell transplantation, and pharmacological intervention studies. Current challenges include insufficient reproducibility across laboratories, susceptibility of results to nonspecific confounding factors, and the lack of standardized task parameters. Future directions should focus on standardizing task protocols and stimulus parameters, promoting multimodal integration with in vivo imaging and electrophysiology, achieving causal inference from stimulus input through neural circuit processing to behavioral output, and aligning task design with clinical functional assessment constructs. Establishing cross-species, cross-stage comparative frameworks will further enhance the mechanistic interpretability and translational value of preclinical research.