Emulating the sense of touch is fundamental to endow robotic systems with perception abilities. This work presents an unprecedented mechanoreceptor-like neuromorphic tactile sensor implemented with fiber optic sensing technologies. A robotic gripper was sensorized using soft and flexible tactile sensors based on Fiber Bragg Grating (FBG) transducers and a neuro-bio-inspired model to extract tactile features. The FBGs connected to the neuron model emulated biological mechanoreceptors in encoding tactile information by means of spikes. This conversion of inflowing tactile information into event-based spikes has an advantage of reduced bandwidth requirements to allow communication between sensing and computational subsystems of robots. The outputs of the sensor were converted into spiking on-off events by means of an architecture implemented in a Field Programmable Gate Array (FPGA) and applied to robotic manipulation tasks to evaluate the effectiveness of such information encoding strategy. Different tasks were performed with the objective to grant fine manipulation abilities using the features extracted from the grasped objects (i.e., size and hardness). This is envisioned to be a futuristic sensor technology combining two promising technologies: optical and neuromorphic sensing.