| Abstract |
Having just arrived in the department, I will begin by very quickly introducing myself. Afterward, I will present my research on interval timing (TI) in neural networks. TI is the capacity of counting the passing of time. To implement TI, it is necessary to possess an internal clock to count the passing time, but also a memory to store how much time has passed. Multiple models have been proposed to explain how it can be achieved in the human brain, but TI has also been observed in organisms without neural system, i.e. implemented only through the internal dynamics of the chemical interactions present in the system. The research I will present relies on evolutionary robotics to evolve a neural controller (CTRNN) implementing TI relying only on internal dynamics, i.e. no synaptic plasticity is provided. The results show that dynamics only is sufficient to implement a general TI system capable of measuring the duration of a stimulus from 1s to 5s, and to reuse this information later on to complete a robotic task. |