Biological neural networks can represent information in the collective action-potential discharge of neurons and store information in the synaptic connections that both comprise the network and govern its function. Our work shows that, in the mammalian hippocampus, the strength and organization of synaptic connections change during learning as a function of their use. Do these experience-induced synaptic changes store information in memory, alter and control network function, or both? And are there neurobiological differences between storing information in hippocampal memory and changing how the hippocampus processes information?
The lecture will review our investigations of the biochemical mechanisms responsible for altered synaptic function in the rodent hippocampus during cognitive behavior.
I will then show that inhibiting crucial molecular mechanisms required to maintain strengthened hippocampal synapses can cause memory loss without abolishing structured hippocampal population activity: place-cell activity remains organized, but the geometry of the population representation is altered. Thus, disrupting a molecular mechanism required for maintaining memory does not simply disrupt hippocampal network function, but instead transforms the organization of its activity.
These findings challenge standard dedicated-coding and engram-cell concepts and instead motivate a dynamical-systems and network-science view of neuronal network function. From this perspective, the consequences of manipulating individual components of a recurrent neural network cannot necessarily be predicted from the properties of those components alone, providing a framework for understanding the paradoxical network responses that can result from precise optogenetic manipulations.
