Undoing of firing rate adaptation enables invariant population codes

Neural adaptation supports coding efficiency by tuning responses to prevailing stimulus statistics. However, when information is represented by populations of neurons, adaptation of individual units could degrade behaviorally relevant signals. Here, we investigate how the fly olfactory system implements adaptation to background odors in olfactory receptor neurons (ORNs) and the consequences for combinatorial coding in downstream circuits. We show that adaptation of ORN firing rate is compensated at the axon terminal, where inhibitory presynaptic feedback supports background-invariant calcium responses to increases (but not decreases) in odor concentration. Computational analyses demonstrate that such invariance requires an adaptation strategy in ORNs that shifts response amplitude rather than sensitivity, diverging from efficient coding principles in single neurons. Downstream, postsynaptic projection neurons also exhibit background-invariant responses through presynaptic plasticity involving Unc13 proteins. Thus, the olfactory circuit restores stable population codes by undoing peripheral firing rate adaptation, enabling an asymmetric contrast representation that preserves stimulus identity across backgrounds.