This suggests that the time-dependent response, but not the steady-state firing rate, of the output neuron could be used to distinguish different gradient configurations. To test this hypothesis we simulated a ‘rotation’ of the antennas in a gradient field, where the average odor level stays the same, but after the rotation the gradient along the antenna changes from zero either positive or negative. Fig \ref{div-813313}b confirms this hypothesis, showing that there can be a difference in the time-dependent response of the output neuron, even if the overall odor level is the same. Thus, in contrast to having to move and sample the ‘average odor level’ to determine a gradient direction, it is possible that the neural properties enable the animal to detect the gradient direction by simply moving its antennas.