Neurons' coordinated firing generates meso- and macro-scale network activity and oscillations. These, in turn, influence the individual neurons’ firing. An example of this cross-scale modulation is phase precession, where the phase of the theta oscillation, as measured by local field potentials (LFP), governs hippocampal place cell firing. By introducing a new method referred to as state-conditioned decoding, we uncover a different cross-scale hippocampal modulation. Unlike in phase precession, here the LFP’s spectral properties (i.e., instantaneous frequency and amplitude) modulate the firing of neurons with respect to the behavioural context: the differential patterns of firing across behavioural contexts depend on spectral properties in the theta band regardless of phase.We used pre-collected LFP and neurons' spikes from the hippocampus of rats performing an odour-memory task. As a baseline, we ran standard decoding analysis on the spikes to predict task success per time point across all trials (Figure 1A). For state-conditioned decoding, we first detected quasi-stationary states with distinct spectral properties from the LFP using a data-driven model; then, we ran decoding analysis for each state separately (Figure 1B).We detected two states: a low-power-lower theta (7-8 Hz) state and a high-power-higher theta (8-9 Hz) state (Figure 1C). The critical result is that state-conditioned decoding substantially outperformed standard decoding, despite using fewer trials (figure 1D).Our results suggest that theta activity can be split into two states with distinct frequency and power, and that the activation of these influences the firing patterns across neurons with respect to behavioural context.