Reaction currents in chemical networks usually increase when increasing their driving affinities. But far from equilibrium the opposite can also happen. We find that such negative differential response (NDR) occurs in reaction schemes of major biological relevance, namely, substrate inhibition and autocatalysis. We do so by deriving the full counting statistics of two minimal representative models using large deviation methods. We argue that NDR implies the existence of optimal affinities that maximize the robustness against environmental and intrinsic noise at intermediate values of dissipation. An analogous behavior is found in dissipative self-assembly, for which we identify the optimal working conditions set by NDR.[3]. Close to equilibrium, such response is severely constrained [4].Since currents are proportional to affinities, á ñ = J R , the response R must be positive to ensure positivity of the entropy production 1 S = á ñ = J R 0 2 . Far from equilibrium, instead, á ñ J need not be linear in thus making R not only dependent on the entropy production. Kinetic aspects become relevant [5], thus opening the way to regimes of negative differential response (NDR) [6]. This counterintuitive, yet common phenomenon has been found in a wealth of physical systems after its first discovery in low-temperature semiconductors [7]. Examples are particles in crowded and glassy environments [8][9][10][11][12], tracers in external flows [13,14], hopping processes in disordered media [15,16], molecular motors [17,18], polymer electrophoresis in gels [19], quantum spin chains [20], graphene and thermal transistors [21,22]. The shared feature underlying all these systems is a trapping mechanism arising by (e.g. energetic, geometric, topological) constraints on the system states [23].Here, we show that NDR plays a key role in open chemical reactions networks [24][25][26]. We show for three paradigmatic models-substrate inhibition, autocatalysis and dissipative self-assembly-how it appears in the average macroscopic behavior as well as in the stochastic regime. While the first two are well described core reaction schemes in living organisms [27,28], the latter is currently drawing the attention of chemists [29,30]. Within the scope of these examples we discuss the role of NDR with respect to environmental and intrinsic noise [31][32][33][34]. We first show that the region of marginal stability, i.e. where R;0, ensures robustness against external perturbations (in the affinity) at moderate values of dissipation. We then argue that those systems affected by NDR that are not poised in the region of marginal stability, behave so in order to minimize the dispersion of the current. Such precision is found to be achieved at moderate values of dissipation, yet again. Hence, our findings show that the performance of life-supporting processes does not always increase at larger dissipation rates OPEN ACCESS RECEIVED