The processing and transmission of information in a biological brain is assured by the neurons, which communicate together or process information through the emission of electrical impulses (called action potential). For this reason, a potential difference occurs both between neurons and between the inner and outer membranes of an individual neuron. Due to this, the biological neural network is constantly subjected to intricate electromagnetic conditions. In this study, a non-polynomial memristor is employed to simulate the impact of electromagnetic radiation on an improved Rulkov neuron. From the theoretical investigations carried out, the system shows both stable and unstable dynamics depending on the value of one of its specific parameters. Results from numerical analyses uncover that the improved Rulkov neuron under electromagnetic radiation can exhibit rich and diversified dynamics such as self-excited and hidden attractors, evolving from regular to chaotic motions. In addition, for certain specific values of its parameters, interesting and intriguing features such as antimonotonicity (i.e. bubble like bifurcation) and offset boosting are revealed. Furthermore, physical results obtained from breadboard experiments carried out using an Arduino Due microcontroller board corroborates with numerically obtained findings. The improved Rulkov neuron is finally exploited to generate random bits and NIST test is performed to confirm the unpredictable features of the generated bits.