Fig. 7: Sequential activation of mPFC→dmStr neurons.

a Representative deconvolved ΔF/F traces from 10 example maintenance cells. Cell numbers correspond to the sorting in (b). b Normalized deconvolved neuronal signals during the maintenance period for an example mouse. Left: Mean signals for a randomly selected half of correct trials used as a training set. Middle: Mean signals for the remaining test set of correct trials. Right: Mean signals for mistake trials. In all plots, neurons are sorted according to the signal peak times in the training data. c Relative time shift of response peaks compared to training data for all neurons. Peak times shifted less for test trials (median = 1.0 s, 75th percentile = 2.8 s) compared to mistake trials (median = 1.80 s, 75th percentile = 4.8 s). ***p = 6.62 × 10⁻⁵, two-sided Wilcoxon signed rank test, n = 341 neurons. d Examples of two trials, showing an orderable cell pair [A,B] (left) and a non-orderable pair [A,C] (right). Cell A is active earlier than cell B in most of the trials (BDO = 0.96), whereas cell C is active earlier than cell A in some trials but later in others, resulting in a lower BDO of 0.66. e Left: BDO matrix calculated for the maintenance period for all neurons recorded in an example mouse. Right: BDO matrix for shuffled data. Neurons are sorted by their average BDO. f Average absolute value BDO (±s.e.m.) across all neuronal pairs for six mice (black), calculated for all 10 task phases and compared to shuffled data (green). Source data are provided as a Source Data file.