Background

Skinner-box systems are fundamental in behavioural research. They are objective, reliable and can be used to carry out procedures otherwise impossible with manual methodologies. Recently, jumping spiders have caught the interest of scientists for their remarkable cognitive abilities. However, inquiries on their learning abilities are still few, since we lacked a proper methodology capable of overcoming the inherent difficulties that this family poses when carrying out a conditioning protocol.

New method

In this paper, a new, automated, open-source Skinner-box, intended for the study of jumping spiders is presented. The system is 3d printable, cheap, fully open-source; is controlled with a Raspberry Pi Zero by a Python script. Since spiders are too lightweight to activate large physical object, the SPiDbox employs photo-sensors.

Results

To validate the methodology, 30 Phidippus regius underwent a training procedure for a simple discrimination task to validate the effectiveness of the system. The spiders managed to learn the task, establishing the effectiveness of the SPiDbox.

Comparison with existing methods

This automated training appears to be more reliable and effective than traditional methodologies. Moreover, its highly scalable, as many SPiDboxes could be used in parallel.

Conclusions

The SPiDbox appears to be an effective system to train jumping spiders, opening up the possibility to study learning in increasingly more complex tasks, possibly extending our understanding of jumping spiders’ cognitive abilities.

中文翻译：

---

SPiDbox：用于研究跳跃蜘蛛的开源“ Skinner-box”系统的设计和验证。

背景

Skinner-box系统是行为研究的基础。它们是客观，可靠的，可用于执行手动方法无法执行的程序。最近，跳跃蜘蛛以其卓越的认知能力引起了科学家的兴趣。但是，关于他们的学习能力的询问仍然很少，因为我们缺乏能够克服该家庭在执行调节协议时所固有的困难的正确方法。

新方法

在本文中，提出了一种新的，自动化的，开源的Skinner-box，旨在研究跳跃的蜘蛛。该系统是3d可打印的，廉价的，完全开源的；由Raspberry Pi Zero由Python脚本控制。由于蜘蛛的重量太轻，无法激活大型物理对象，因此SPiDbox采用了光电传感器。

结果

为了验证该方法，对30个菲比斯州进行了简单的歧视任务训练程序，以验证系统的有效性。蜘蛛程序设法学习了任务，从而确定了SPiDbox的有效性。

与现有方法的比较

这种自动化培训似乎比传统方法更为可靠和有效。此外，它具有高度可扩展性，因为可以并行使用许多SPiDbox。

结论

SPiDbox似乎是训练跳跃蜘蛛的有效系统，为在越来越复杂的任务中学习学习提供了可能性，可能会扩展我们对跳跃蜘蛛认知能力的理解。