It is the relative silence between the firing spikes of neurons that shows what they are really up to, a new study suggests.
“The brain appears to use these durations of silence to encrypt information,”
said Dr. Joe Z. Tsien, neuroscientist at the Medical College of Georgia at Augusta University of his new Neural Self-Information Theory.
It’s widely held that neurons generate perceptions, thoughts and actions by emitting electrical pulses called action potentials or spikes. One problem with that standard measure of neuron action is that neurons are essentially always firing at some level and with spontaneous fluctuation, even when it’s not clear what is happening as a result, according to Tsien, Georgia Research Alliance Eminent Scholar in Cognitive and Systems Neurobiology and a corresponding author of the study
Cracking The Neural Code
Tsien uses the analogy of an ocean surface that may look calm compared to a tsunami, but is never truly still. Many scientists have noted that there can also be variation in how even the same neuron responds to the same stimulus or even a quiet, resting state.
Yet, there must be some kind of operating principle that enables us to think and act in real time in the face of this ongoing variability, he says. Brain scientists call the decades-old puzzle cracking the neural code.
Tsien’s team has evidence from monitoring mouse neurons during various activities that the magic happens when you see a group of neurons each entering an atypical state for them – not of firing – but of the relative periods of silence between the firing and entering that period at the same time.
These silent spaces between overt firing are called interspike intervals, and, the neurons having atypical intervals at the same time are part of a clique generating perceptions, actions and thoughts in real time, he theorizes.
“These cells belong to the same group, an assembly. It’s a very general finding about how neuron activity codes information,”
Cliques Of Cells
Applying this new Neural Self-Information Theory, they have identified 15 groups of cell assemblies in the cortex and hippocampus of the brain that work together to enable things like sleep cycles, sensing where you are and how you act in response to things you see and experience.
For example, they studied mice playing a game where a light shines on a wall and the mouse learns that if he pokes a hole in that same spot, rather than four other choices, he will get a food pellet when he returns to where he started. If he doesn’t come back in time or pokes the wrong hole, no food pellet awaits.
“It’s a simple task but highly attention driven, and how the brain executes this task was poorly understood. To identify the cell cliques that help the mouse be successful, you have to find out what each neuron’s interspike intervals looks like when they are out of their normal range of occurrence. Among all the cells you record, you then identify the ones that move into that different state – called a surprisal state – at the same time,”
Tsien said. This time he uses the analogy of a normally chatty individual in an uncharacteristic period of silence.
“That is when these cells start to act as a clique. That is when the neural cliques are coming together to encode a train of thought or a set of actions. If it’s what happens usually, that means it does not carry much information, it’s like a ground state,”
The Neural Self-Information Theory
In information theory, self-information or surprisal is a synonym for the surprise when a random variable is sampled. It is expressed in a unit of information, for example shannons (often called bits), nats, or hartleys, depending on the base of the logarithm used in its calculation. The expected self-information is information entropy and reflects the average surprise or uncertainty associated with sampling a random variable.
The Neural Self-Information Theory was summed up by Tsien is his 2017 paper, as the hypothesis that;
“Neuronal variability operates as the self-information generator and expressor to convey a variable amount of information in the form of silence variability-surprisals. Coordination of these surprisal ISIs in space (across cells) and time can seamlessly give rise to robust real-time cell-assembly code.
Most importantly, this Self-Information Code is completely intrinsic to neurons themselves, with no need for outside observers to set any reference point such as time zeros of stimulation or filtered local field potential oscillation phases. Because the self-information code is operated in the form of the ISI variability-based probability distribution, the downstream neurons can naturally sense these surprisal shift in ISI variability as manifested by sudden deviations from the equilibrium (a form of intracellular memory or variability distribution) of post-synaptic neurons’ biochemical states (i.e., energy production, receptor activation, ion channel open/close state distribution patterns, protein phosphorylation/dephosphorylation rate, receptor insertion/removal, etc.).”
Support for the work came from the NIH, US Army, a NSF ABI Innovation grant, GRA Brain Decoding Project, Shanghai Youth Science and Technology Sail Project, and the Brain Decoding Center grant from the Department of Science and Technology of Yunnan Province.
Meng Li, Kun Xie, Hui Kuang, Jun Liu, Deheng Wang, Grace E Fox, Zhifeng Shi, Liang Chen, Fang Zhao, Ying Mao, Joe Z Tsien
Neural Coding of Cell Assemblies via Spike-Timing Self-Information
Cerebral Cortex, Volume 28, Issue 7, 1 July 2018, Pages 2563–2576, https://doi.org/10.1093/cercor/bhy081
Top Image: Meng Li, et al CC-BY. Seven PRL cell assemblies corresponding to each of the 7 stages of 5CSRT behaviors.