Traditional process mining techniques are constrained by the use of a single case identifier. To address this, object-centric process mining was developed. Current object-centric process models like object-centric Petri nets are restricted by representational biases, for example concerning the concrete distribution of objects to their succeeding activities (abstracted by places) or optionally involved object types (abstracted using variable arcs). This paper presents four key contributions to object-centric process discovery, to overcome the mentioned representational biases. First, we define object-centric causal nets, allowing for the modeling of complex process behavior. Second, we introduce a baseline algorithm for mining these nets from object-centric event logs. Third, we provide a publicly available implementation. Fourth, we evaluated the implementation quantitative and qualitatively, demonstrating the algorithm’s applicability across diverse event logs.