Hippocampal circuit dynamics during learning and sleep in freely behaving macaque monkeys

The ability of organisms to learn and form memories based on experience and to use these memories flexibly in different situations is a hallmark of cognitive evolutionary processes. The neural basis for this ability, however, is typically studied in laboratory settings with minimal environmental cues and a restricted behavioral repertoire. These limitations have been shown to affect cognitive performance in memory tasks and neural organization that supports memory. To overcome these constraints, we applied two technological developments that were designed to allow the study of the neural circuit dynamics of memory in macaques with improved ecological validity. First, we designed a 3-D enclosure to study cognition in freely moving macaques in a more naturalistic environment. Monkeys flexibly learned multiple unique sequences of objects-in-context, where both the macaques and the object-sequences were fully situated within that 3-D context. Second, Second, we performed wireless recordings using high-density, chronically implanted linear arrays. Contextual memory formation relies on the spatiotemporal dynamics of the hippocampus; therefore, we recorded across layers of the hippocampal area in CA1 as monkeys performed the task and during post-learning sleep. We evaluated for the first time in primates the properties of different cell groups and their relation to oscillatory dynamics in CA1 of the hippocampus. Active wakefulness and sleep states were characterized by distinct modes of oscillation. Theta (4-10Hz) oscillations dominated sleep and strongly entrained spiking activity of different cell groups. During active wakefulness, supra-theta oscillations (15-80Hz) became prominent and spikes preferentially locked to the gamma band. Cross-frequency coupling analysis revealed that theta and gamma oscillations are decoupled in monkey CA1, unlike that described in rodents. Inhibitory cell groups dynamically reconfigured their phase relationship with the pyramidal cell group in a brain-state-dependent manner, which could allow different cell groups to be allocated to the control of oscillations as a function of frequency. As suggested in rodents, but not yet described in primates, the pyramidal cells from superficial and deep substrata differed in their physiological properties and showed selective interactions with the inhibitory cell groups. This included a tightly time-resolved coupling with those cells that decreased firing rates during the sharp-wave ripple oscillation. Finally, these pyramidal cells were biased to form cell assemblies with members of their own strata, and rarely across strata. These results suggest that monkey CA1's oscillatory dynamics differ from rodents, potentially due to differences in the behavioral repertoire and/or underlying circuit mechanisms; but that some of the underlying regulatory mechanisms such as inhibitory timing and cell assembly clustering among populations, may be conserved. As such, theories of learning and memory from rodents that are dependent on oscillatory bands may not readily translate to primates; however, the microcircuit principles including sublayers in the radial axis of CA1 persist in supporting different behavioral-cognitive implementations of a (partially) conserved underlying circuitry.