**4. Conclusion**

In this paper, we implement a classic self-learning paradigm in rats: associative memory learning (fear conditioning) using a mobile robot and a neuromorphic system


### **Table 4.**

*Comparison of scale and association capability with other state-of-the-art works.*

(Loihi chip) in an online learning scenario. In specific, we use a mobile robot as the substitute for the rats in fear conditioning experiments. Two signal pathways are assigned for conditional and unconditional stimuli. In our experiments, vibration signals emulate unconditional stimuli, while brightness of lights is assigned as conditional stimuli. Originally, the mobile robot only moves when it detects vibration signals. After providing these two signals at the same time several times, the robot performs a movement when light signals are present alone. The detections of lights and vibrations are implemented with Integrate and Fire Neurons. In addition, the movement of the robot is controlled by the specific-designed response neurons. The signal pathway modification during associative memory learning is implemented with Hebbian learning. Compared to other state-of-the-art works, our work successfully reproduces the fear conditioning of rats in a real-world scenario with no labeled data and pretraining process.
