Figure 7.

Parameter estimates from GLM2 associated with grid events using different model designs and grid code symmetries. In the right entorhinal cortex, the six-fold symmetrical model provided the best model fit, both for the “aligned > misaligned” contrast and for the single parametric modulation regressor. Calculating the mean grid orientation over two runs, vs. using separate mean grid orientations for each run, made little difference to the parameter estimates. In the left entorhinal cortex, the effect of generating the mean grid orientation over multiple runs vs. separate single runs was more variable and other symmetrical models provided equally good model fits for this hemisphere. All bars show the mean parameter estimate averaged over all voxels within right and left entorhinal cortex, respectively.