We　studied　bursting　patterns　underlying　bifurcation　phenomena　and　chaotic　spiking　in　a　computational　leech　heartbeat　model.　We　observed　the　gradient　physical　properties　of　the　ISI　series　and　amplitude　(shift　of　the　membrane　potential)　when　the　parameter　g(leak)　was　mildly　changed　and　found　different　bistable　areas.　The　resulting　computation　implies　that　(i)　classification　of　the　intensity　of　the　input　information　is　feasible　in　this　regime,　(ii)　a　neuron＇s　working　level　can　be　marked　by　its　range　in　a　typical　bifurcation,　and　(iii)　there　are　invisible　triggers　underlying　subtle　mechanisms　in　the　model.