Correlation-dependent reorganization and receptive field convergence

When a neuron could be recorded stably for sufficient time we verified our hand-plotted estimates of receptive field location using an automated stimulus presentation/response collation system (see Supplementary Methods online). As illustrated by the examples shown in Supplementary Figure 1, the two methods of estimating receptive field location show a close correspondence. We measured the discrepancy between the handplotted and automated receptive field location estimates for all our available experimental data. The range of the distribution of these discrepancies is clearly exceeded by the distances of the estimated receptive field shifts, confirming that these apparent shifts are not the consequence of misestimating receptive field location (Supplementary Fig. 2a). This is further confirmed by the observation that there is no detectable correlation between receptive field shift distance and the discrepancy, for each individual receptive field, between the two methods of receptive field position estimation (Supplementary Fig. 2b).