Figure 9.

Full, non-blank-subtracted imaging is poorly fit by spiking and gamma LFP. A₁–B₃, Single case showing the full (i.e., non-blank subtracted) spiking (A₁) and LFP gamma (A₂), the corresponding predictions (B₁ and B₂), and the measured full imaging response (B₃; same example as in Fig. 2, following the same conventions). Note the prominent positive signal for the blank and low-contrast stimuli (compare with blank-subtracted measured imaging in Fig. 2B₃). Insets depict the optimal HRFs, with fits consistently poorer than for the corresponding blank-subtracted responses (compare Fig. 2B₁–B₃, insets). Note the large residuals (bottom panels), particularly for low-stimulus contrasts, which are comparable in amplitude to predicted signals.