Deep learning, relying on large datasets, has made significant advancements through annotated data for training. However, this reliance restricts its feasibility in fields like planetary robotics. In contrast, animals learn through interaction with their environment, forming associations between events and objects, a process known as associative learning. By emulating this biological learning method, we can potentially overcome the challenges of data scarcity in deep learning. Most current implementations of associative memory are limited to small-scale simulations and offline environments. This study, however, takes a bold step forward, exploring the application of associative learning in a real-world setting using a cognitive-inspired robotic system and neuromorphic hardware, specifically Intel’s Loihi chip. This chip, designed to emulate the structure and function of the human brain, is ideally suited for our research. It can replicate fear conditioning without the need for pretraining or labeled datasets. In our practical demonstration, the cognitive-inspired robot learns to associate a light stimulus with a vibration stimulus, as evidenced by its movement responses. Synaptic weights are adjusted using Hebbian learning principles during this associative learning process. Integrating Intel’s Loihi chip into our system allows visual signal processing through specialized neural assemblies, significantly enhancing the robot’s learning capabilities.