AbstractResearch and development of memristor‐based neuromorphic networks has entered the stage of large‐scale application. However, the understanding of the adopted memristors is still vague, meaning a thorough study of their local activity is urgently necessary. Here, an electrode tip is created to focus the electrical field on the active locality, where the NbO quantum dot (QD) self‐assembles in the middle of the Nb‐doped AlNO film. The QD size can fluctuate under an appropriately controlled electron beam or electrical field, suggesting reversible nucleation behavior. The QD system periodically pulses when the input is enhanced to a co‐tunneling threshold or to one that activates the oscillation of the QD size. These two mechanisms are attributed to the local activity accounting for the information‐energy conversion of the QD system. This study presents a more precise and low energy‐consumption approach for neuromorphic computing by manipulating the local activity.