Neuronal computation underlying the generation of a transitive inference in mice

Inferential reasoning entails the ability to form flexible relationships between items or events that were never directly experienced together. Humans and animals can draw on previous experience and combine information from multiple discrete, loosely related items or events into a 'cognitive map'. This internal model of the world generalises the structure of problems onto a common framework of spatial or abstract relational knowledge. We aim to understand the mechanism underlying the generalisation of information enabling adaptive behaviour across related stimuli. We designed a transitive inference task with two parallel groups of odour triplets, while simultaneously recording spiking activity and local field potential (LFP) in multiple hippocampal and cortical subregions implicated in this task using high-density silicon probes. We ask whether the hippocampus (HPC) conveys a conjunctive code, continuously mapping the current specific sensory stimuli coded by lateral entorhinal cortex (LEC) onto the general abstract structure of the task coded by medial entorhinal cortex (MEC). First, we aim to identify neurons selectively active upon presentation of each specific odour. We will examine the temporal relationship between firing patterns of neurons representing different odours to determine whether the sequentiality of odour presentation is maintained in the neural representation. We will also assess the temporal relationship across brain areas participating in the task to determine the directionality of information flow during both the formation and retrieval of the inference. Finally, we will focus on brain oscillations at different frequencies during active and inactive behavioural states to determine their importance in synchronizing activity between representations for individual odours and between brain areas, potentially nesting cognitive short-cuts giving rise to flexible inferences.