Full Version: Temporal Vs. Rate Coding
The debate in the literature seems to revolve around the significance of temporal coding vs. rate coding, right? Though the debate seems sort of artificial since the brain seems to employ both rate coding and temporal coding. I'd like to hear other people's thoughts on this.
can you quickly explain these types of coding?
yes. Temporal coding assigns importance to the precise timing and coordination of action potentials or 'spikes', whereas rate coding does not assign any such importance and says that it's only the 'mean rate' or 'mean frequency' of spikes that matters. It's sort of a pseudo-debate, though, because both forms of coding are apparently used in the brain, and because we are still ignorant of general principles of neural coding. So I guess it's just a debate for the sake of entertainment.
I guess the question is, what kinds of cognitive properties can be associated exclusively with a particular coding process
| QUOTE |
| what kinds of cognitive properties can be associated exclusively with a particular coding process? |
We don't really know at the moment. It's clear that both synchrony and rate are being employed, as well as higher-order correlations and syn-fire chains, but no-one really knows what's going on. One idea is that the "attentional spotlight" increases synchrony between neuronal activities, thereby rendering them more effective for driving other neurons. Also, rate codes have been fairly conclusively demonstrated in the first-order sensory neurons. But beyond that, we're all in a state of ignorance.
I can recommend three introducing papers to this field. Its a interesting debate between Shadlen & Newsome and W.R. Softky. A pretty nice example for a scientific dialogue.
Shadlen & Newsome published an article (Noise, neural codes and cortical organization) in Current Opininon in Neurobiology (1994) where they pointed out that a system of balanced excitatory/inhibitory inputs may produce spike output with in vivo like properties (e.g. irregularity) in which information may be processed vian rate coding.
Softky answered to this paper in the one of the following issues (Simple codes versus efficient codes) favouring the idea of neuronal coincidence detectors (temporal code).
Last article of this debate was Shadlens answer to Softky (Don't remember the title of the paper but one may find it at Curr. Opin. Neurobiol.)
Recent works expand our view to the question of neural coding. Some workers focused on the question how principal features of neural activity may arise from the temporal evolution of network dynamics. They claim that neural system may arrange themselves in a dterministic chaotic state and that these dynamics may be used for information processing. But a final prove for deteministic chaos in a neural sysrem still remains to be done.
Shadlen & Newsome published an article (Noise, neural codes and cortical organization) in Current Opininon in Neurobiology (1994) where they pointed out that a system of balanced excitatory/inhibitory inputs may produce spike output with in vivo like properties (e.g. irregularity) in which information may be processed vian rate coding.
Softky answered to this paper in the one of the following issues (Simple codes versus efficient codes) favouring the idea of neuronal coincidence detectors (temporal code).
Last article of this debate was Shadlens answer to Softky (Don't remember the title of the paper but one may find it at Curr. Opin. Neurobiol.)
Recent works expand our view to the question of neural coding. Some workers focused on the question how principal features of neural activity may arise from the temporal evolution of network dynamics. They claim that neural system may arrange themselves in a dterministic chaotic state and that these dynamics may be used for information processing. But a final prove for deteministic chaos in a neural sysrem still remains to be done.
one question is whether rate and temporal coding are independent of one another. If rate and synchrony are coupled in neuronal populations, as some studies suggest, then they're not independent and may be conveying the same info. Of course, synchrony is not the only temporal code, but it seems to be what the literature focuses on.
| QUOTE (Unknown @ Apr 01, 05:52 AM) |
| I can recommend three introducing papers to this field. Its a interesting debate between Shadlen & Newsome and W.R. Softky. |
Shadlen and Newsome seem to be the main proponents of rate coding, and Softky, Singer, Abeles, and Reyes seem to be some of the main proponents for temporal coding, but my question is, to what extent are we setting up straw men?
| QUOTE (purkinje @ Jan 12, 06:32 AM) |
| The debate in the literature seems to revolve around the significance of temporal coding vs. rate coding, right? Though the debate seems sort of artificial since the brain seems to employ both rate coding and temporal coding. I'd like to hear other people's thoughts on this. |
There seems to be a lot of evidence for relative temporal coding (i.e. the relative time elapsed between spikes or the sequencing of spikes encodes information regarding the relationship between events or objects).
The idea of a rigid temporal coding, e.g. the way a computer encodes data over a serial connection, seems dubious.
So-called "rate coding" seems likely to be a side-effect of activity except at the sensory and motor edges of the network. At the sensory edge, rate seems to encode stimulus intensity. At the motor edge, bursting is used to drive muscle contractions.
One of the problems that needs to be solved is how does rate coding at the sensory edge (e.g. the retina or pain receptors) get translated into the temporal and synchrony coding that is likely used by the rest of the brain. There must be a mechanism for shifting the time phase of a signal to align it with the rhythms and synchrony patterns in the rest of the brain.
Likely candidates for solving this problem are the "top-down" feedback pathways. Imagine a neuron that is receiving rate coded stimulus from a sensory pathway. This neuron could accumulate a "charge" as a result of repeated input spiking. This neuron is later triggered to fire or not based on top-down feedback signals coming from "higher" areas that are synchronized with the global spiking rhythms of the brain. These global spiking rhythms would be the effects that show up on EEGs (alpha and beta waves as well as the 40Hz frequences implicated in the neural synchrony models).
QUOTE(purkinje @ Jan 12, 2004, 09:32 AM) 
The debate in the literature seems to revolve around the significance of temporal coding vs. rate coding, right? Though the debate seems sort of artificial since the brain seems to employ both rate coding and temporal coding. I'd like to hear other people's thoughts on this.
I vote neither, several studies mid 90's seemed to discredit both in a study of flying bats that reacted to
a single neuron spike ( not a train or series of spikes ) in which the bat changed its direction, further studies found the same behavior in a few insects as well. Neural coding is not well understood. This caused the community quite a stir and then strangely it was back to business as usual. Neuroscience needs a general upgrade to its classical theories.
in very early stages of sensory processing, where background firing rates are very low or nil, it is not uncommon for individual spikes to play a role in neural coding, but in most of the rest of the brain, it is much 'noisier' and higher level codes are more plausible and robust.
QUOTE(lucid_dream @ Apr 13, 2007, 04:05 AM)
in very early stages of sensory processing, where background firing rates are very low or nil, it is not uncommon for individual spikes to play a role in neural coding, but in most of the rest of the brain, it is much 'noisier' and higher level codes are more plausible and robust.
Right, my point was that the experiments were interesting in that the bat reacted to the single spike absense any others as a reference, such that the single spike itself represented information which is very interesting to consider.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.