Working memory features are embedded in hippocampal place fields

Hippocampal principal neurons display both spatial tuning properties and memory features. Whether this distinction corresponds to separate neuron types or a context-dependent continuum has been debated. We report here that the task-context ("splitter") feature is highly variable along both trial and spatial position axes. Neurons acquire or lose splitter features across trials even when place field features remain unaltered. Multiple place fields of the same neuron can individually encode both past or future run trajectories, implying that splitter fields are under the control of assembly activity. Place fields can be differentiated into subfields by the behavioral choice of the animal, and splitting within subfields evolves across trials. Interneurons also differentiate choices by integrating inputs from pyramidal cells. Finally, bilateral optogenetic inactivation of the medial entorhinal cortex reversibly decreases the fraction of splitter fields. Our findings suggest that place or splitter features are different manifestations of the same hippocampal computation.

Distinct roles of hippocampus and medial prefrontal cortex in spatial and nonspatial memory

In earlier work, patients with hippocampal damage successfully path integrated, apparently by maintaining spatial information in working memory. In contrast, rats with hippocampal damage were unable to path integrate, even when the paths were simple and working memory might have been expected to support performance. We considered possible ways to understand these findings. We tested rats with either hippocampal lesions or lesions of medial prefrontal cortex (mPFC) on three tasks of spatial or nonspatial memory: path integration, spatial alternation, and a nonspatial alternation task. Rats with mPFC lesions were impaired on both spatial and nonspatial alternation but performed normally on path integration. By contrast, rats with hippocampal lesions were impaired on path integration and spatial alternation but performed normally on nonspatial alternation. We propose that rodent neocortex is limited in its ability to construct a coherent spatial working memory of complex environments. Accordingly, in tasks such as path integration and spatial alternation, working memory cannot depend on neocortex alone. Rats may accomplish many spatial memory tasks by relying on long-term memory. Alternatively, they may accomplish these tasks within working memory through sustained coordination between hippocampus and other cortical brain regions such as mPFC, in the case of spatial alternation, or parietal cortex in the case of path integration. © 2016 Wiley Periodicals, Inc.

Speed modulation of hippocampal theta frequency correlates with spatial memory performance

Hippocampal theta rhythm is believed to play a critical role in learning and memory. In animal models of temporal lobe epilepsy (TLE), there is evidence that alterations of hippocampal theta oscillations are involved in the cognitive impairments observed in this model. However, hippocampal theta frequency and amplitude at both the local field potential (LFP) and single unit level are strongly modulated by running speed, suggesting that the integration of locomotor information into memory processes may also be critical for hippocampal processing. Here, we investigate whether hippocampal speed-theta integration influences spatial memory and whether it could account for the memory deficits observed in TLE rats. LFPs were recorded in both Control (CTR) and TLE rats as they were trained in a spatial alternation task. TLE rats required more training sessions to perform the task at CTR levels. Both theta frequency and power were significantly lower in the TLE group. In addition, speed/theta frequency correlation coefficients and regression slopes varied from session to session and were worse in TLE. Importantly, there was a strong relationship between speed/theta frequency parameters and performance. Our analyses reveal that speed/theta frequency correlation with performance cannot merely be explained by the direct influence of speed on behavior. Therefore, variations in the coordination of theta frequency with speed may participate in learning and memory processes. Impairments of this function could explain at least partially memory deficits in epilepsy.