Exploring Mechanisms of Cognitive and Behavioural Development in Humans and Machines

在人类的发育过程中，通过使用简单的能力作为学习更复杂的能力的基础，可以学习质量上的新行为。这种由发展提供的能力的逐步提高，可能对于使成人获得更高水平的认知和行为能力的过程变得容易控制至关重要。同样，开发可以使更智能的机器，特别是机器人的创造变得容易。目前设计机器人控制器的方法依赖于控制机器人行为的规则的预先规定。然而，这种预定义依赖于大量的设计工作或较长的搜索和测试时间。这使得预先指定对于复杂的机器人或执行复杂任务或在复杂环境中操作的机器人来说是困难的。此外，由此产生的控制系统往往只适用于它被开发来解决的特定问题，并且在任何不可预见的情况下是脆弱的。这些问题可以通过向机器人提供适应其遵循的规则并自主创建控制行为的新规则的能力来解决。因此，开发方法可以为机器人提供一种发现行为方式的战略，并可能是生产能够在现实世界中成功运行的机器人的实际必要性。因此，该项目旨在探索生物学中发育过程的神经机制，以便提取适当的计算机制，这些机制可以用于在机器人中实施类似的发育过程。实现这一目标不仅将为创造更具适应性和自主性的机器人提供适当的机制，而且还将为人类发展提供有价值的见解。

During the developmental process in humans, qualitatively new behaviours are learnt by using simple abilities as a basis for learning more complex ones. This progressive increase in competence, provided by development, may be essential to make tractable the process of acquiring the higher-level cognitive and behavioural abilities observed in adults. Similarly, development could make tractable the creation of more intelligent machines, and specifically robots. Current methodologies for designing robot controllers depend on the prespecification of the rules governing the robot's behaviour. However, such predefinition relies on a significant design effort or on prolonged search and testing time. This makes prespecification intractable for a complex robot or one performing a complex task or operating in a complex environment. Furthermore, the resulting control system tends to be applicable only to the particular problem it was developed to solve and is brittle in the event of any unforeseen circumstances. These problems could be resolved by providing the robot with the ability to adapt the rules it follows and to autonomously create new rules for controlling behaviour. Developmental approaches could thus provide robots with a strategy for discovering how to behave and may be a practical necessity for producing robots capable of operating successfully in the real world. This project thus aims to explore the neural mechanisms underlying the developmental process in biology in order to extract appropriate computational mechanisms that could be used to implement a similar developmental process in a robot. Achieving this objective will not only provide mechanisms appropriate for creating more adaptive and autonomous robots but will also provide valuable insights into human development.