Jeffery et al. present an inspiring overview of behavioural and neurobiological studies of neural map formation and navigation in three-dimensional environments. Based on their review of the literature, they suggest that three-dimensional space is “bicoded,” meaning that different neural mechanisms are used to map environments in the horizontal and vertical planes. Only the horizontal plane has a metric for distance and direction. The suggested dual coding scheme raises several interesting questions.
The authors propose that three-dimensional space is encoded as a mosaic of contiguous map fragments. These fragments need not be horizontal but can, in principle, have any orientation relative to the Earth's surface. The proposed fragmentation is reminiscent of the decomposition of grid cell maps that takes place in complex environments in two-dimensional space. A few years ago, Dori Derdikman and colleagues measured the effect of environmental compartmentalization on the spatial periodicity in grid cells (Derdikman et al. Reference Derdikman, Whitlock, Tsao, Fyhn, Hafting, Moser and Moser2009). Rats were trained to run in a zigzag fashion through 10 consecutive hairpin-shaped flat corridors in a large square enclosure. The compartmentalization of the recording environment disrupted the global grid pattern observed in the same cells in the absence of the corridors. When internal walls were inserted, separate maps were formed for each corridor. Each time the rat turned from one alley to the next, the grid map was reset, and a new sequence of grid fields unfolded. The internal walls broke the grid map into smaller maps, one for each compartment. A similar pattern of sub-maps is visible in entorhinal neurons during navigation in other multi-alley environments (Frank et al. Reference Frank, Brown and Wilson2000). The sharp transitions at the turning points in these mazes appear to take place simultaneously, in a coherent manner, in grid cells and place cells (Derdikman et al. Reference Derdikman, Whitlock, Tsao, Fyhn, Hafting, Moser and Moser2009). Taken together, the studies suggest that complex real-world environments are not represented by a single universal map. Instead, the brain has multiple contiguous grid cell and place cell maps, each covering a small uninterrupted space within the larger environment (Derdikman & Moser Reference Derdikman and Moser2010). As argued by Jeffery and colleagues, this fragmentation of the cognitive map is probably not limited to the horizontal plane. Fragmentation may be an efficient mechanism for mapping complex environments, both within and across planes.
The fragmentation of the spatial map introduces new issues, however. How are fragments “glued” together at transition points? How does the brain adjust for changes in the slope of the terrain between segments? Is metric information carried over from one fragment to the next? Can animals path-integrate across segments? It is possible, as proposed by the authors, that the integration of the fragments is solved by neural systems and computational algorithms not involving the currently known spatial cell types in the hippocampus or entorhinal cortex. In order to understand how animals navigate in real-world environments, we need to identify the factors that elicit fragmentation and the mechanisms that link one fragment to the next.
We agree that some observations speak in favour of partly specialized mechanisms for encoding space in horizontal and vertical planes, at least in terrestrial animals. This includes the presence of separate cell populations for encoding of pitch and yaw in the lateral mammillary nucleus of the rat (Stackman & Taube Reference Stackman and Taube1998), as well as the finding that head direction cells in rats respond exclusively to variations in the horizontal (yaw) plane, or the plane of locomotion if the surface is tilted (Stackman et al. Reference Stackman, Tullman and Taube2000). The proposed segregation receives further support from a study demonstrating less sensitivity to the vertical dimension in place cells and grid cells in rats (Hayman et al. Reference Hayman, Verriotis, Jovalekic, Fenton and Jeffery2011). As indicated by Jeffery et al., it remains possible that vertical modulation was not expressed in that study because the animals maintained a horizontal orientation during the vertical movement. Moreover, with a predominance of horizontal movements, we cannot exclude that horizontal and vertical dimensions differed with respect to availability of metric cues and that precise fields would emerge also in the vertical direction if sufficient information were available. Jeffery et al. also acknowledge that precise representation of the vertical dimension may require a minimum of prior experience with active movement in three-dimensional environments. Most laboratory rats lack such experience. Collectively, the limited existing data raise the possibility that horizontal and vertical dimensions rely on somewhat segregated mechanisms, but the findings do not rule out that experience contributes to such segregation or that the brain, as a whole, has an integrated representation of volumetric space. More experimental work is clearly needed to settle these issues. The target article of Jeffery et al. takes important steps towards identifying critical questions that must be addressed.
Jeffery et al. present an inspiring overview of behavioural and neurobiological studies of neural map formation and navigation in three-dimensional environments. Based on their review of the literature, they suggest that three-dimensional space is “bicoded,” meaning that different neural mechanisms are used to map environments in the horizontal and vertical planes. Only the horizontal plane has a metric for distance and direction. The suggested dual coding scheme raises several interesting questions.
The authors propose that three-dimensional space is encoded as a mosaic of contiguous map fragments. These fragments need not be horizontal but can, in principle, have any orientation relative to the Earth's surface. The proposed fragmentation is reminiscent of the decomposition of grid cell maps that takes place in complex environments in two-dimensional space. A few years ago, Dori Derdikman and colleagues measured the effect of environmental compartmentalization on the spatial periodicity in grid cells (Derdikman et al. Reference Derdikman, Whitlock, Tsao, Fyhn, Hafting, Moser and Moser2009). Rats were trained to run in a zigzag fashion through 10 consecutive hairpin-shaped flat corridors in a large square enclosure. The compartmentalization of the recording environment disrupted the global grid pattern observed in the same cells in the absence of the corridors. When internal walls were inserted, separate maps were formed for each corridor. Each time the rat turned from one alley to the next, the grid map was reset, and a new sequence of grid fields unfolded. The internal walls broke the grid map into smaller maps, one for each compartment. A similar pattern of sub-maps is visible in entorhinal neurons during navigation in other multi-alley environments (Frank et al. Reference Frank, Brown and Wilson2000). The sharp transitions at the turning points in these mazes appear to take place simultaneously, in a coherent manner, in grid cells and place cells (Derdikman et al. Reference Derdikman, Whitlock, Tsao, Fyhn, Hafting, Moser and Moser2009). Taken together, the studies suggest that complex real-world environments are not represented by a single universal map. Instead, the brain has multiple contiguous grid cell and place cell maps, each covering a small uninterrupted space within the larger environment (Derdikman & Moser Reference Derdikman and Moser2010). As argued by Jeffery and colleagues, this fragmentation of the cognitive map is probably not limited to the horizontal plane. Fragmentation may be an efficient mechanism for mapping complex environments, both within and across planes.
The fragmentation of the spatial map introduces new issues, however. How are fragments “glued” together at transition points? How does the brain adjust for changes in the slope of the terrain between segments? Is metric information carried over from one fragment to the next? Can animals path-integrate across segments? It is possible, as proposed by the authors, that the integration of the fragments is solved by neural systems and computational algorithms not involving the currently known spatial cell types in the hippocampus or entorhinal cortex. In order to understand how animals navigate in real-world environments, we need to identify the factors that elicit fragmentation and the mechanisms that link one fragment to the next.
We agree that some observations speak in favour of partly specialized mechanisms for encoding space in horizontal and vertical planes, at least in terrestrial animals. This includes the presence of separate cell populations for encoding of pitch and yaw in the lateral mammillary nucleus of the rat (Stackman & Taube Reference Stackman and Taube1998), as well as the finding that head direction cells in rats respond exclusively to variations in the horizontal (yaw) plane, or the plane of locomotion if the surface is tilted (Stackman et al. Reference Stackman, Tullman and Taube2000). The proposed segregation receives further support from a study demonstrating less sensitivity to the vertical dimension in place cells and grid cells in rats (Hayman et al. Reference Hayman, Verriotis, Jovalekic, Fenton and Jeffery2011). As indicated by Jeffery et al., it remains possible that vertical modulation was not expressed in that study because the animals maintained a horizontal orientation during the vertical movement. Moreover, with a predominance of horizontal movements, we cannot exclude that horizontal and vertical dimensions differed with respect to availability of metric cues and that precise fields would emerge also in the vertical direction if sufficient information were available. Jeffery et al. also acknowledge that precise representation of the vertical dimension may require a minimum of prior experience with active movement in three-dimensional environments. Most laboratory rats lack such experience. Collectively, the limited existing data raise the possibility that horizontal and vertical dimensions rely on somewhat segregated mechanisms, but the findings do not rule out that experience contributes to such segregation or that the brain, as a whole, has an integrated representation of volumetric space. More experimental work is clearly needed to settle these issues. The target article of Jeffery et al. takes important steps towards identifying critical questions that must be addressed.