Grid Cells for Conceptual Spaces? Nikolaus Kriegeskorte, Katherine R. Storrs  Neuron  Volume 92, Issue 2, Pages 280-284 (October 2016) DOI: 10.1016/j.neuron.2016.10.006 Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 1 How Conjunctive Grid Cells Might Give Rise to 6-fold Modulation in Measures of Bulk Population Response (A) Activity of a single grid cell recorded in rat entorhinal cortex. Gray traces indicate the path of the rat as it freely explored a square enclosure; blue dots indicate the locations at which neural spikes were recorded. (B) Schematic of the location and direction preferences of six idealized conjunctive grid cells (denoted by six color shades). Each cell fires only in particular locations, which form the vertices of a triangular grid (drawn only for the dark purple cell with a 60° preference), and only if the animal is moving in the cell’s preferred direction when it encounters that location (indicated by arrow). In this example, each of the six cells creates a grid of the same orientation and spatial scale but shifted relative to one another. In entorhinal cortex, additional cells create grids at larger and smaller spatial scales, and with other shifts, but all with the same orientation. (C) Firing rates of the six hypothetical grid cells in (B) shown as a function of the animal’s heading direction. The directions preferred by conjunctive grid cells are not continuously distributed but are aligned with one of the six main axes of the cell’s grid axis (indicated by black arrows). (D) Total firing rate summed across the six cells, as a function of heading direction. There is a 6-fold modulation in the population activity, which may be detectable over a much larger population of similar cells using fMRI. Credit for (A): Tor Stensola, CBM/Kavli Institute for Systems Neuroscience. Neuron 2016 92, 280-284DOI: (10.1016/j.neuron.2016.10.006) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 2 “Conceptual Space,” Experimental Design, and Key Results (A) A two-dimensional space defines the neck and leg lengths of a bird silhouette. Some locations in the space are arbitrarily associated with Christmas-themed symbols. During training, participants learn which bird shapes are associated with which symbols. (B) On each trial of the brain scanning experiment, participants watch an initial random bird morph in a random direction for 1 s, then imagine it continuing to morph in that same direction for a further 4 s. To complete the trial, the participants choose which, if any, of the learned Christmas symbol locations would have been encountered along the morph trajectory. (C) A whole-brain analysis identifies voxels in which activation displays a 6-fold modulation as a function of bird morph direction. (D) Within those voxels, a subsequent analysis tests for evidence of a consistent grid angle. Putative grid angle is estimated from one half of the data, and the second half of the data is split into subsets of trials in which the morph direction was aligned (red) or misaligned (gray) with the putative grid angle. In vmPFC and several other areas, signal is higher on aligned than on misaligned trials, indicating a 6-fold modulation with a replicable angle within an individual. (E) Across participants, the strength of the alignment consistency effect predicted performance in the behavioral task. Figure adapted from Constantinescu et al. (2016). Neuron 2016 92, 280-284DOI: (10.1016/j.neuron.2016.10.006) Copyright © 2016 Elsevier Inc. Terms and Conditions