Sensory systems have evolved to extract information from the continuous stream of external stimulation to the brain. For the auditory domain, the activity in the auditory cortex is thought to play a key role in representing simple and abstract stimulus features. In this respect, the perception of rhythmic patterns in a sound requires a complex integration of temporal auditory information, essential to inform perceptual decisions. However, the particular link between rhythmic stimulus features, neuronal activity and perceptual behavior remains elusive. In our work, we address this relationship in three steps: (i) We characterize a large library of pulsed auditory stimuli with a defined set of rhythmic features. (ii) We use statistical modeling to investigate how the different stimulus features relate to stimulus-evoked activity in the mouse auditory cortex. (iii) We test perceptually-guided behavior in mice and humans, using a decision-making task with auditory pulsed stimuli. Our analyses reveal that activity patterns in the auditory cortex can be explained by a specific set of stimulus features, covering rhythm-related aspects of the pulsed stimuli, such as the rate and clustering of pulses. Moreover, we estimate the perceptual relations between pairs of stimuli by their features, allowing us to predict perceptual decisions over the course of learning. Interestingly, the feature weights in humans and mice are highly similar and thus appear to capture general perceptual dimensions that are evolutionary conserved. Together, we present a quantitative framework to predict and interpret the representational and perceptual similarities of pulsed auditory stimuli in mice and humans.