Figure 2 - Identifying hexagonally symmetric signals across the whole brain

Contributed by alexandra.constantinescu on June 20, 2016

Collection: Organizing conceptual knowledge in humans with a gridlike code

Description: There are several aspects to note about this map. These points are detailed in the supplementary information of the paper.  First, it is the merger of two maps. Because of the unusual difficulty of imaging the entorhinal cortex (ERH), we ran FMRIB’s FIX tool using a threshold of 20, which substantially improved the signal in this region. However, it also reduced signal in other areas of interest. Therefore, this map is the result of merging FIX-applied ERH data with data from the rest of the brain without applying FIX. Second, the test is in all probability statistically biased. The transformation from F-to-Z that forms part of this test requires a perfect estimation of the autocorrelation. In the case that the autocorrelation is imperfectly estimated, the F-to-Z transform will result in Z-stats that have a positive bias in each individual, resulting in an over-estimation of the group effect. This is the likely reason why the histogram of Z-scores is right-shifted by 1 standard deviation. Note that because of these issues that preclude statistical testing on this map, this map was NOT used for statistical testing in the paper, but only for selecting regions of interest for later analyses. Note also that the analysis that generated this map is independent of the statistical tests. This map asks whether there is 6-fold periodicity when each scan is treated independently. The later statistical analyses test whether the grid angle (phase) from one scan aligns to the grid angle (phase) from another scan. These statistical tests are performed in all of the other figures in the paper. For a detailed discussion of these points, please see the main text and the supplementary information of the paper.

Task View 3D View