New Science paper[1] from Mayank Mehta’s group showing evidence for dendritic spikes in behaving animals, twice as many during behavior than during sleep. Not surprising for the HTM-inclined, these would presumably represent either prediction events that don’t lead to active cells, and/or multiple segments simultaneously predicting. I would be very interested to see a synaptic plasticity study measuring STDP with respect to presynaptic spikes, postsynaptic dendritic spikes, and postsynaptic somatic spikes.
Also very interesting, in addition to the large number of dendritic spikes, the study reports large amplitude slow current fluctuations in the dendritic trees, and Mehta speculates this could be evidence of analog computation. What do you think? Evidence for temporal pooling?
Yes, I agree - this makes total sense in the context of HTM Theory. When you have sequences in natural/complex settings you usually have many more possibilities (predictions) than actual events. This should lead to many more dendritic spikes than action potentials. I actually saw evidence of this first reported in [1] where they state: “These data indicate that, at distal dendritic recording sites, spiking occurs at higher rates and with a higher degree of ‘burstiness’ than spiking recorded at the soma…”
I always show this property on my list of “HTM properties that can be experimentally tested”, so it’s nice to see more evidence of this.
This may not be exactly what you are looking for but check out [2].
Could you explain that? Sounds intriguing! We do need mechanisms for sustained firing and plasticity for temporal pooling to happen.
Smith SL, Smith IT, Branco T, Häusser M. Dendritic spikes enhance stimulus selectivity in cortical neurons in vivo. Nature [Internet]. 2013;503:115–20. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24162850
Losonczy A, Makara JK, Magee JC. Compartmentalized dendritic plasticity and input feature storage in neurons. Nature. Nature Publishing Group; 2008;452:436–41.
The study reports large dendritic membrane potential fluctuations that strongly modulate the rate of dendritic spikes. They didn’t measure the input that caused the activity, so we can’t tell whether sustained dendritic spiking events are due to repeated extrinsic input or slow intrinsic currents.
The analog computation hypothesized by Mehta is essentially that the amplitude of dendritic membrane potential might communicate a scalar value to the soma by dendritic spike rate (they measured a sigmoid relationship here). Unless we know whether or not the fluctuations are sustained intrinsically, I don’t think this qualifies as evidence of temporal pooling, it would just be evidence that dendritic spiking rates can be graded.