Active materials inspired from biological systems are a paradigmatic model to study non-equilibrium statistical mechanics. A common method to access their non-equilibrium statistics is to measure the fluctuation distribution of a tracer particle embedded in the material. Usually, discrete motor-induced events cannot be characterized in this way due to the multitude of active processes taking place simultaneously during measurement.

By decoupling the structural evolution of the system from its fluctuations, we are able to measure the statistics of such discrete events showing the appearance of distinct shoulders in the van Hove correlation function at high motor concentration. We do so by tailoring a model active system based on cytoskeleton proteins to maintain approximately steady-state dynamics over several hours. The shoulders' appearance coincides with a transition from steady-state dynamics to slowly evolving dynamics as the motor concentration increases. We further demonstrate how these discrete active events accumulate at longer lag times to a broadened Gaussian distribution, and compare it to simple minimal simulations that capture the main features of our experiments.