by Rich, Liaw and Lee (Science 2014)
Where are place fields created in new environments?
Record 253 pyramidal cells from dorsal CA1 on progressively extended track from 3m to 10m to 22m to finallly 48m. Only analyzed neurons that could be isolated in sleep periods flanking experimental sessions.
Modelled field placement per neuron as Poisson process (locations i.i.d., with some fixed rate per neuron); excellent fit for neurons with 6 or more fields. Then asked whether population could also be modeled by something simple. Using Poisson again didn’t work because variance > mean (overdispersed), prompting consideration of a negative binomial model, which can be seen as an infinite mixture of Poisson distributions with different rates. This NB model fit best with parameters \(r= 0.57, p = 0.14\).
Can calculate “recruitment curves” by asking what cumulative fraction of neurons are participating from a given starting point (left). The recruitment curves were identical as a function of distance from starting location (right).
tags: place-fields - hippocampus - CA1 - negative-binomial - science