The Peri-Saccadic Perception of Objects and Space
Figure 3
Predicted Source and Shape of Oculomotor Feedback, and Predicted Target Area of Compression
(A) Goodness of fit (pre) for the time course and spatial range of compression with respect to typical properties of cells in oculomotor areas. Unclipped activity and open movement fields lead to a drop in the goodness of fit. A time course which resembles the firing pattern of burst cells is consistent with the data, whereas build-up like activity with a half maximum value around 46 ms prior to saccade requires a damped gain function in the target area to compensate the early distortion.
(B) Effect of open movement fields on the localization of flashed bars in the critical phase from −25 to 0 ms before a 20° saccade.
(C) Predicted shape of the feedback signal in visual space for a 20° saccade. The model with anisotropic magnification predicts a shape that is circumscribed for a particular eccentricity but spreads to different angles with constant eccentricity. For comparison, the model with isotropic magnification produces a round shape with a strong spread of the signal to a broader range of eccentricities.
(D) Comparison of monkey receptive field sizes with the model prediction (Text S2). The line shows the required minimal receptive field size for each layer. Please note, due to the non-linear spatial pooling in the model, the receptive field values are upper bounds and not mean values. The dots indicate the maximal receptive field size for a particular eccentricity in the respective cortical area as reported in the literature. For the area to be consistent with the model the dots should be close to or exceed the constraint given by the model. Layer 1: The receptive field sizes in V4 are close to the minimal receptive field size of L1. Receptive field sizes in MT and TEO are sufficiently large. Layer 2: Both TE and LIP are consistent with the prediction of the model for L2. For larger eccentricities, receptive field sizes in LIP are below the lower limit obtained from the model. However, since the critical stimuli in the data (Figure 2A) which constrain the receptive field size in L2 were all presented at an eccentricity of less than 20° (in the opposite hemifield than the one where the saccade target appeared) we should not exclude LIP.
(E) Effect of small receptive field sizes in L1 and L2 (dashed lines in [D]) on the localization of flashed bars in the critical phase from −25 to 0 ms before a 20° saccade.