I am pleased to share with you our findings on the flexibility of hippocampal circuits to rapidly encode arbitrary patterns. biorxiv.org/content/10.110…. A great collaborative effort with @englishdanielf, the uLED team at UM, and Roman Huszar and of course, Gyuri Buzsaki.
@englishdanielf We were inspired by Magee and Burgalossi who showed that stimulating a hippocampal neuron induced place field. We wanted to see if this works with opto and to stimulate ensembles to see if they replay together and with the existing fields– simple enough hypotheses.
@englishdanielf sciencedirect.com/science/articl…
ncbi.nlm.nih.gov/pubmed/28883072
From Bittner et al., 2015
ncbi.nlm.nih.gov/pubmed/28883072
From Bittner et al., 2015
@englishdanielf A quick aside, I think the Bittner work is beautiful and caught the imagine of the field, including my own, since it shows a mechanism for what we imagine the hippocampus to do – change rapidly to encode arbitrary, strong inputs thus imprinting the episodic memory
@englishdanielf This view never felt complete to me as it ignored the preexisting structure on which learning takes place. Priors help us learn and filter what we learn about. We show memory interference and reconsolidation deficits.
@englishdanielf Does the hippocampus really sacrifice all that structure to build a compressible code? Put another way, what we wanted to test was whether the hippocampus can really learn about arbitrary inputs.
@englishdanielf I tried and failed many times to do a simple pattern completion study in CA3 where I stimulate 3 sites (baseline) then 4 sites over and over (encoding, 1000s of times even), then three sites (test), and I never saw activity on the missing site (rows = CA3 neurons).
@englishdanielf I couldn’t imprint. But maybe that’s because the experiments were open loop and the state of the brain differed with each encoding pulse. One day I will publish these null results. I needed to try to Bittner/Magee/Burgalossi/Diamantaki approach.
@englishdanielf Most of the time, stimulation did nothing. I am sure many labs have tried and were disappointed by the stubbornness of the brain. But I also have a suspicion as to why we have been missing this kind of result in the literature, Stark 2012 and Rickgauer 2014 notwithstanding
@englishdanielf i think its because the neurons can sometimes behave in unexpected ways. Some CA1 cells remapped kind of to the stimulation zone
@englishdanielf But some remapped to random other places
@englishdanielf Even some non-stimulated cells remapped. The light delivery was highly focal, so I know that I was not directly depolarizing these cells (250um away).
@englishdanielf And though I only stimulated on one side of the track, I saw remapping on the non-stim side – interesting since the place fields on inbound runs are said to be orthogonal to those on the outbound
@englishdanielf So what is going on here? It began to feel like we just need to inject some energy into the system for it to snap into some new state. And there were hints of that existing state even prior to remapping – or put another way the remapping was not random.
@englishdanielf There were some sporadic spikes in the field-to-be prior to frank place field emergence
@englishdanielf And we could see that the distance between the place fields POST STIM was predictive of the PRE STIM ripple coupling
@englishdanielf So what drove changes in the non-stimulated CA1 cells and what made cells remap to random locations? My previous work with @englishdanielf, we looked closely at the pyramidal control of feedback inhibition, we wanted to see if this synapse was involved
@englishdanielf But we lacked the statistical tools to measure changes in coupling over time. With Roman Huszar and inspired by @AbedGhanbari biorxiv.org/content/10.110… we developed at GLM that tracks the boost in drive from a presynaptic to postsyn. cell above rate fluctuation in the post syn cell
@englishdanielf @AbedGhanbari This method is one of the coolest things to come out of all of this work, in my opinion, and it gives a totally new view of synaptic coupling. The first thing that hit us, was that fluctuations in synaptic pairs covary. There are multiple competing sub-networks.
@englishdanielf @AbedGhanbari So we did some dimensionality reduction and worked in ICA space. stimulation changed how these synaptic pairs covary. We speculate that this could be a mechanism driving the reorganization of place fields – but more experiments, and technological advances are needed to nail this
@englishdanielf @AbedGhanbari If you like this work, please come to my poster. Morning session Oct 20th abstractsonline.com/pp8/#!/7883/pr…
@englishdanielf @AbedGhanbari If you like this way of thinking, hire me! I am currently looking for a position to run a lab with a focus on how memory networks change to assimilate new information, the role of interneuron plasticity, and how patterns of synchrony can go awry
@englishdanielf I don't know why my rasters look so crummy but the other images look OK. same data as in the preprint.
@englishdanielf @AbedGhanbari @ihstevenson we have started putting the short term plasticity back into our dynamic model and we still see these sharp transitions. We communicated back and forth with Abed quite a bit, but I would love to hear your thoughts.
