An interesting new paper providing some insights into synchrony across cortical layers…
The synchronized activity of six layers of cortical neurons is critical for sensory perception and the control of voluntary behavior, but little is known about the synaptic mechanisms of cortical synchrony across layers in behaving animals. We made single and dual whole-cell recordings from the primary somatosensory forepaw cortex in awake mice and show that L2/3 and L5 excitatory neurons have layer-specific intrinsic properties and membrane potential dynamics that shape laminar-specific firing rates and subthreshold synchrony. First, while sensory and movement-evoked synaptic input was tightly correlated across layers, spontaneous action potentials and slow spontaneous subthreshold fluctuations had laminar-specific timing; second, longer duration forepaw movement was associated with a decorrelation of subthreshold activity; third, spontaneous and sensory-evoked forepaw movements were signaled more strongly by L5 than L2/3 neurons. Together, our data suggest that the degree of translaminar synchrony is dependent upon the origin (sensory, spontaneous, and movement) of the synaptic input.
http://www.cell.com/cell-reports/pdf/S2211-1247(16)30594-0.pdf
4 Likes
Thanks for the paper. Since Numenta makes use of bursting, people here might want to look at Figure S2 in supplementary materials, which can be found in this doc
http://www.cell.com/cms/attachment/2058558551/2062038483/mmc2.pdf
1 Like
@Deanhorak - thanks for posting this. Thanks @s.aleksashenko - I couldn’t understand the Figure (admittedly I haven’t spent a huge amount of time on it). Could you summarize the finding and relevance to HTM?
I’m not sure any really profound new insights were derived from this study, but it represents the first time that we’ve been able to observe the flow of information between cortical layers in living, active mice in-vivo.
The findings do suggest that synchrony between various layers plays a role in the animal’s perception and it’s behavior. It suggests that individual layers might interpret activation patterns in different ways while remaining synchronized with each other during various states of rest or movement, with the result of providing the correct output patterns.
We know from earlier studies that while resting, the sensory cortex produces spontaneous regular slow pulses, and that various layers remain synchonized. This study suggests that when movement of the specimen occurs, this synchronization is perturbed, It indicates that the relative state of rest or movement of the animal effects the state of neural firing within layers, even when provided the same stimulus (movement seems to increase firing among nerves in the deeper layers than in those nearer the surface of the cortex).
As far as how this relates to HTM, I would suggest any empirical findings that provide new insights into the relationships between cortical layers is invaluable in an attempt to understand how they work together to produce the results they do. It serves as a touchstone to ensure our theories are consistent with the evidence. In this case, since movement or rest are global (or non-local) states, and yet alter the firing patterns and synchrony between layers in the local column, it indicates that any HTM theory should heavily weigh information from broad states, in the interpretation of activity among cortical layers. Admittedly I have been too busy with my own work lately to take the time to fully understand the inter-layer relationships in the current HTM theories and how they are evolving, but it seemed this study would be of interest in that regard.
2 Likes
Thanks @Deanhorak, that’s very helpful. I totally agree that these sorts of experiments that can detect interactions between layers in live situations are extremely useful. We saw some more work along these lines at Cosyne last year and this year, so hopefully this is a sign of more experiments like these.
1 Like