| dc.description.abstract | Navigation and the ability to orient oneself in the external environment is fundamental for humans and other animals. The hippocampus and the medial entorhinal cortex (MEC) contain cells that are believed to support spatial processing and form a system for fast high-capacity declarative memory. This includes hippocampal place cells, which fire when an animal is at a particular location, and grid cells in the MEC, which express multipeaked firing fields forming a regularly spaced hexagonal pattern. Grid cells are organised in a small number of distinct modules that have different grid spacing and that are believed to operate as separate sub-networks within the MEC. This thesis aims to contribute to a better insight into mechanisms of spatial processing in place and grid cell circuits in freely moving rats, with a particular focus on coding at the level of large numbers of neurons. We utilise newly developed technologies for large-scale electrophysiology, which allow for simultaneous recording of the activity of hundreds of neurons.
In the first part of this thesis (paper I), we investigate responses to a salient object in single cells and populations of cells in the hippocampus. It is well known that objects influence the firing of hippocampal cells, but the exact nature of hippocampal object representation is not fully understood, and whether there exists a population code for object location remains to be determined. By recording from large numbers of CA1 cells, we identify a small group of cells that express object-vector properties, by firing when the rat is at a fixed distance and direction from the object. Further, we show that whilst most cells change their firing rate or spatial firing pattern in response to the object, there are large variations in the responses of single cells. On the population level, the activity is organised systematically as a function of the rat’s distance to the object. This organisation is distributed across all CA1 cells and does not depend on single-cell properties, highlighting a mechanism in CA1 for coding of distance from the animal to the object that depends on the combined activity of the network.
In the second part of the thesis (paper II and paper III) we investigate coordination of activity across grid cell modules by recording from up to hundreds of grid cells simultaneously. Grid cell activity is known to be tightly coupled within a module, and this coupling is maintained between different environments and across different behavioural states. However, less is known about coordination between the modules and under which conditions grid modules decouple.
In paper II we show that when the environment is compressed along two dimensions by insertion of new walls, most grid cells rescale according to the change in size and shape of the environment, regardless of grid module identity. However, we find that in four out of five rats, the cells belonging to the smallest grid modules exhibit a bimodal distribution of responses, where one group of cells rescales, and the other group does not. This suggests the existence of submodules of grids cells that have similar grid firing properties but respond differently to environmental changes. We also show that compression does not affect grid cell activity in the entire firing rate map. Instead, there are large changes in the cells’ firing patterns near where the wall is inserted, and small changes further away from the inserted wall. This is consistent across grid modules and indicates that grid cell firing is largely influenced by local geometry. Together, our results show that grid modules are mostly functionally coupled during environmental compression.
In paper III, we investigate coordination of grid modules as rats navigate in darkness with minimal information about their position from external cues. We show that grid patterns deteriorate in the dark; however, the spike-rate correlations between inter-module cell pairs are similar in light and in the dark. We further demonstrate, using a likelihood-based approach and a decoding approach, that the inter-module phases are dynamically coupled both in light and in the dark. Collectively, the results show that the states of the different modules are tightly coupled, even when the internal representation of position deviates from the animal’s real position.
Taken together, the work presented in this thesis shows that certain features of cognitive maps might be best understood as emergent properties of large numbers of cells. This emphasises the importance of considering the co-existence of population-level representations with representations on the single-cell level. Further, our results suggest that the hippocampus support a high-dimensional code with dynamic mapping of environmental features. In contrast, the MEC has a robust lowdimensional representation of environments, with high levels of coherence between grid cells. | en_US |
