观察学习—通过观察其他动物而习得行为的能力—一度被认为仅限于脊椎动物。近年来熊蜂等无脊椎动物也展示了很强的观察学习能力，如仅通过短短几分钟对其他蜂从特定刺激取食的观察，熊蜂即可形成对该刺激的显著偏好。然而观察学习的关键过程是怎样的，以及其在熊蜂相对简单神经环路上的实现机制尚不明确。我们前期研究展示，计算模型可帮助理解蜂类多种认知能力的神经环路实现机制。预实验结果显示，基于深度学习的动物连续追踪新技术可成功量化熊蜂在自由活动时的动态行为过程。据此我们推测：基于连续追踪的行为过程分析可能揭示熊蜂观察学习的关键时间和空间因素，基于熊蜂视角的计算模型研究可能提示其视觉与蘑菇体神经环路实现观察学习的机制。本项目拟以追踪行为细节为切入点，通过熊蜂观察学习行为实验和计算建模方法，阐明熊蜂观察学习的关键过程，探索其神经环路层面的编码方式，为理解观察学习的一般神经基础提供新思路。

Learning by observation – the ability of acquiring behavior by witnessing another animal – is once considered the preserve of vertebrate species. Yet recent years some invertebrates such as bumblebees also demonstrate impressive observational learning capacities: after only a few minutes watching other bees feeding on certain stimuli, the naive observer shows significant preference to those stimuli, but little is known about what processes are crucial for the development of the observational learning during this observational phase, nor do we have a good understanding on how observational learning could be implemented in the miniature brain circuit of the bumblebees'. Our previous work show that computational models can be helpful in understanding the neural network level accounts of multiple cognitive feats. Our preliminary result on this project verified the plausibility of analyzing bumblebee free-moving behavioral processes based on continuous tracking. Taken together, we here hypothesize that behavioral analysis on continuous tracking data can reveal the key temporal and spatial factors of bumblebee observational learning, and the computational modeling of the bee visual and mushroom body circuit with realistic visual inputs from a bumblebee's perspective, can provide possible neural implementation of observational learning. Combining continuous tracking, behavioral experiments and computational modeling, the proposed project aims to clarify the behavioral processes of bumblebee observational learning, to explore possible neural circuit mechanism underpinning this capacity, and to provide insights for further understanding the neural basis of observational learning.