Over the past decades, the hippocampal formation has undergoneextensive study leading researchers to identify a vast array ofcells with functional properties (place cells, splitter cells,etc). In the present work, we aim at investigating whether theactivity of those cells derive from the anatomy of the hippocampalformation and the properties of neuron or derive instead from theactual behavior of the animal. To do so, we simulated an agentnavigating inside an 8-shaped track, making alternating choices(T-maze alternating task). We designed a random network, based onthe reservoir computing paradigm, that process distance-basedsensors and output a direction change (constant speed). Despiteits simplicity, the model successfully solved the task whilebearing no structural similarity with the hippocampalformation. We subsequently followed the comprehensive and recentreview on splitter cells by Duvelle et al., and applied the exactsame analysis onto the activity on the cells composing ourmodel. We were able to identify splitter cells (as well as placecells, head direction cells and decision cells) and confirms asignificant portion of the theoretical hypotheses of Duvelle etal. regarding splitter cells. Beyond these results, this workstrongly suggests that the activity of such cells originates fromthe actual behavior as opposed to any structural or anatomicalorigin: any model doing the same task will exhibit the same cellactivity. From a broader point of view, this work questions theepistemic role of such cells in our understanding of thehippocampal formation.