Neurosymbolic artificial intelligence is a growing field of research aiming to combine neural networks learning capabilities with the reasoning abilities of symbolic systems. We introduce a formalism for supervised multi-label classification informed by prior knowledge. We build upon this formalism to re-frame three abstract neurosymbolic techniques based on probabilistic reasoning. We then evaluate experimentally and compare the benefits of all three techniques across model scales on several informed classification tasks. Finally, we discuss the computational complexity of probabilistic reasoning, which is of cardinal importance to assess the scalability of probabilistic neurosymbolic techniques.