Hippocampal Place Cells Dynamically Encode Value of Available Goals During Spatial Navigation

Abstract

The hippocampus plays a role in spatial navigation, and place cells (cells that fire selectively for an animal’s location in an environment) are thought to provide an anatomical basis for a cognitive map (a map-like representation of space in the brain). Recently, however, place cells have been found to encode more complex aspects of navigation than simply place; including information about the location or presence of upcoming goals. The present experiments aimed to investigate the involvement of the hippocampus in representing the value of a goal (taking into account the size of the reward and the cost in obtaining it), and its contribution to navigation. Place cells were recorded while rats chose between two goals of differing value, in a spatial decision-making task conducted in a figure-of-eight maze. The value of a goal to an animal was inferred based on the proportion of trials in which that goal was chosen. It was hypothesised that cells would fire differentially in the maze depending on the available rewards, and that cell activity would be modulated by the value of upcoming rewards. It was also hypothesised that place cells representing goal value would dynamically encode changes in value when reward sizes were altered, and that this may influence the animal’s future behaviour. Additionally, it was predicted that a subset of cells would fire “out-of-field”, prior to the animal reaching the reward location those cells represented, possibly indicating the mental projection of the animal to that location. The results showed that when animals are presented with rewards of two different values, more place cells are involved in the representation of the reward of higher value to the animal, as measured by the number of cells with fields in that region, despite there being no difference in firing rate. In the maze’s central stem, many cells showed differential firing based on upcoming goal choice. Additionally, a number of cells showed out-of-field firing, firing differentially in a secondary field with regard to the animal’s next choice, and the goal that cell represented. When reward size in each goal changed every twenty trials of the 100-trial session, animals’ behaviour changed quickly to accommodate new reward values, and mean firing rate in the central stem mirrored the pattern of behaviour. The firing rate in the first pass through the stem following a change of reward size, but before a change in behaviour, was strongly correlated with the subsequent distribution of behaviour, whereas the firing rate for the last pass of each block was not. These results support previous research suggesting that the hippocampus is briefly engaged during route planning, but is not needed during the execution of a determined route. They also show for the first time that place cells dynamically encode goal value during spatial navigation. It is proposed that during spatial decision-making, value information, possibly originating from the prefrontal cortex, is transferred to the hippocampus, where it is incorporated into the spatial map of the environment, enabling the animal to most efficiently direct its behaviour.