Spike Distributions and Model Selection


I noticed that in the why models, we compared neurons from different distributions for instance a neuron where spikes were mostly of a uniform distribution to one where the distribution was best described by an exponential distribution.

I would like to know why we compared such neurons instead of comparing say neuron 1(uniformly distributed) and neuron 1(exponentially distributed). I think this was touched in the Faculty Q & A but I unfortunately could not post my question via Zoom.

Thank you,

Apologies if this question already exists and/or this is the wrong place to post(I am still figuring out neurostars).

1 Like


I don’t believe they were comparing neurons per se. Different neurons were not giving rise to the different distributions.
The Inter spike intervals of neurons in general looked like they were coming from an exponential distribution (from the “how” section of the tutorial). They were trying out different distributions to try and explain WHY the ISI of neurons in general look like exponential distribution (vs some other distribution). What is so special about the exponential distribution that nature follows this?

So we compared different distributions and measured their information and we learnt that the exponential is special because given a constraint it maximises the information.

Hope that answers your question.