Grid-like Processing of Imagined Navigation

Abstract

Grid cells in the entorhinal cortex (EC) of rodents [1] and humans [2] fire in a hexagonally distributed spatially periodic manner. In concert with other spatial cells in the medial temporal lobe (MTL) [3-6], they provide a representation of our location within an environment [7, 8] and are specifically thought to allow the represented location to be updated by self-motion [9]. Grid-like signals have been seen throughout the autobiographical memory system [10], suggesting a much more general role in memory [11, 12]. Grid cells may allow us to move our viewpoint in imagination [13], a useful function for goal-directed navigation and planning [12, 14-16], and episodic future thinking more generally [17, 18]. We used fMRI to provide evidence for similar grid-like signals in human entorhinal cortex during both virtual navigation and imagined navigation of the same paths. We show that this signal is present in periods of active navigation and imagination, with a similar orientation in both and with the specifically 6-fold rotational symmetry characteristic of grid cell firing. We therefore provide the first evidence suggesting that grid cells are utilized during movement of viewpoint within imagery, potentially underpinning our more general ability to mentally traverse possible routes in the service of planning and episodic future thinking.

Figure 1

Experimental Design and Behavioral Data (A) Trial structure during the imagination block, showing example screen shots for the cue, imagination, and feedback periods. Participants were cued with a single object at the top of the screen and required to rotate such that they were facing toward the remembered location of that object. They then closed their eyes and imagined moving from their current location to the remembered location of the object. Following this, they waited for a jittered period of time (2–6 s) before moving to the object location and pressing a button. The object then appeared in the correct location, and participants had to navigate to it during the feedback period prior to the start of the next trial. The timing for each period was user defined by either pressing a button (during cue, imagination, and object placement) or moving into the object (during feedback). Times (in s) above each period label show the mean time across all trials and participants for each period. (B) A bird’s-eye view of the circular arena with an example path across both imagination blocks for a single participant in black and object locations for the two blocks in red and yellow, respectively. (C) Histogram showing the percentage of trials per 2° of heading angle error for the object placement task for a single participant. (D) Histogram showing percentage of time across both imagination blocks per 15° of heading angle for a single participant. (B–D) Data shown are from the participant with the median heading angle error across all participants (see Figure S1 for data across all participants).

Figure 2

Movement-Related Grid-like Signal (A) Sinusoidal modulation of BOLD response by heading angle with 6-fold rotational symmetry for movement > stationary periods in EC (−21, −12, −36; p < .05 SVC; shown at p < .005 unmasked for display purposes; see Figure S3B for a masked image), from the split-half analysis where grid orientation was estimated on half the data and applied to the other half, separately for movement, stationary, and imagination periods. (B) % signal change from peak shown in (A) for 6-fold rotational symmetry during movement and stationary periods. (C and D) % signal change for peak shown in (A) for 4- and 8-fold rotational symmetries during movement and stationary periods. Note that we saw no effect for 4- or 8-fold symmetries in the entire EC, i.e., the null effect shown here is not specific to the region of interest based on the 6-fold analysis. Error bars show ±1 SE; ^∗∗p < .01; ns, not significant (relative to baseline).

Figure 3

Imagination-Related Grid-like Signal (A) Sinusoidal modulation of BOLD response by heading angle with 6-fold rotational symmetry for imagination > stationary periods in EC (+21, −12, −33; p < .05 SVC; shown at p < .005 unmasked for display purposes; see Figure S3C for a masked image), from the analysis where grid orientation was estimated during all movement periods and applied to the imagination and stationary periods. (B) % signal change from peak shown in (A) for 6-fold rotational symmetry during imagination and stationary periods. Error bars show ±1 SE; ^∗∗∗p < .001; ^∗p < .05 (relative to baseline). (C) Histogram showing the percentage of participants per 5° of angular distance between grid orientations during movement and imagination periods (circular mean across participants = −5.5°).