Resolving orientation in an environment with conflicting directional information

Several populations of neurons recorded from freely moving rats, such as place cells, head direction cells and grid cells, anchor their spatial receptive fields to polarising cues of the environment. Cue rotation usually leads to a matching rotation of the receptive fields of these neurons. Cue conflict experiments provide evidence that different types of cues in the environment can have differential strength in controlling the receptive fields. We hypothesised that a given cue is more useful to facilitate navigation if it is consistent with other cues in the environment. Based on this hypothesis, we predicted that the amount of control exerted by a cue is affected by how coherent it is with other cues in the environment. To explore the hypothesis, we used a computational model of the head direction cell network as described by K. Zhang. 1996, J Neurosci, 16:2112-26. The activity bubble of the network was updated by angular head velocities calculated from the head direction profiles used for the simulations and the direction represented by the network was determined by the vector sum of the units in the network. Additional units representing the cues from the environment were added to the model. These cue perception units connected to all units of the head direction cell network and the weights of the connections were strengthened or weakened by a competitive Hebbian learning rule. The firing rates of the cue perception units were Gaussian functions of the actual head directions of the simulation. We showed that, similar to the cue control experiments, these cue perception units could reset the activity bubble of the head direction cell network once appropriate weights were established by the learning rule. Next, we created conflicts between the cue perception units with comparable firing rates and demonstrated that the largest coherent set of units exerted stronger cue control over the network. We also explored the role of the perception units whose input to the head direction cell network were too weak to affect the position of the activity bubble, in a cue conflict situation. We found that the weak units among a coherent set of cue perception units also facilitated the set to gain dominance on cue controlling the network during conflict resolution. We showed that, with competitive Hebbian learning rule, the network chose the largest set of coherent cues to orient its activity bubble. This demonstrated how the amount of cue control from one cue could be affected by its relationship with other cues in the environment. We plan to verify the model by implanting unit recording electrodes to the freely moving rats and expose them to different cue conflict situations.