Grant: Discovery Grants Program - Individual

Memory consolidation involves the conversion of memories that depend on the hippocampus (HC) for their formation into efficient, HC--independent, forms, encoded by intrinsic neocortical (NC) synapses. Little is known about how consolidated memories are encoded at the cellular and network levels. Neural firing in HC is driven by spatial context, and this signal is thought to provide NC with a contextual code to aid in memory retrieval. Recently, we made 2 important discoveries about HC--NC interactions. First, a substantial fraction of cells in NC develops selectivity for spatial context, but this plasticity is impaired by HC lesions. Once formed, however, NC representations survive HC loss. This effect is the sine qua non of memory consolidation, and is consistent with our prior predictions that HC output must be stored in NC in order to coordinate memory retrieval across multiple NC regions (the so-called HC "index" theory), and that, under the guidance of HC, NC creates new `feature detectors' for elements of complex experience, leading to more efficient, `schematic' NC representations. These findings open up a new field of memory research, and our long-term program is to clarify how NC representations change under the guidance of HC during the process of memory consolidation. Some of the most important questions to be addressed in the next cycle are: How do external and HC inputs to NC interact to cause tuning of cells to specific locations? Are modality-specific sensory inputs necessary, or is HC output sufficient? Is the sharpening of NC tuning facilitated by 'off-line' memory reactivation driven by HC? Are the newly formed coding units coupled with others to form 'attractors', or are they more like sensory feature detectors that become sharply tuned during early postnatal experience and respond to specific features in a relatively context-independent manner? Is the observed plasticity a result of direct HC--NC synaptic interaction, or are the HC influences cascaded through NC association pathways? Can new spatial representations in NC be created artificially by spatially coupled electrical activation of small populations of NC cells, as has been achieved by others in HC? This would enable clarification of cortical plasticity mechanisms at the network level, and possibly serve as a means of targeted memory trace reactivation Ca2+ imaging and electrophysiological recordings will be made in mice behaving in a VR set-up, in which tactile and visual cues are controlled. Opto- and chemo-genetic manipulations of interneuron excitability will be used to suppress HC output, and hence, to assess its role in the plasticity of NC representations. Our results will open a new dimension of memory research, which will have both basic science and preclinical applicability. They are also of considerable interest to the machine learning community, who increasingly try to exploit knowledge of memory mechanisms in the brain to advance Artificial Intelligence.