If we consider the neuron as an information system, its response to the same input actions should be invariable. What does the variability of the studied characteristics of the reaction interneuron to the same stimuli? Could it be that the response of the neuron-detector says only that he at this time excited with a certain level of excitement? Obviously, the neuron is highly adaptive system. However, such systems are not highly accurate. There is a contradiction in the consideration of the neuron at the same time as a highly adaptive and precise information system.
Speaking about "perfect adaptation" seems too naive... Maybe the [valuable] information is not encoded by a single neuron, but rather a group of cells is responsible of its transmission/encrypting. In this case, if the unit itself behave inaccurately (there is no direct/estimated relation with the right result), at a higher level (i.e. swarm intelligence) the correct information is preserved - it may be that the way in which the flow of information is treated within the brain has similarities with error-correcting codes: even if a local misadaptation occurs, neuroplasticity provides an alternative [non-linear] path to what it was meant to [on a higher level].
I agree with Federico. Accuracy is probably rather a property of the whole system (gruop of cells), not a single cell property.
That's what I meant. Cells are adaptive and the studied characteristics of the reaction of the neurons detectors are variable. Then there is the question of the appropriateness of the search of precision mechanisms of the coding within the studied characteristics. But what is the "variability of reaction"? So whether is it inaccurate? Maybe not those characteristics are investigated. No one denies the existence of neural ensembles. But the hypothesis fix "bugs" of reactions at the following levels of processing creates more problems than it answers.
I want to give some explanations. By neuron-sensor can be attached to its spatial and frequency components of the signal.When a signal is applied to the neuron it can output pulses with a repetition rate as a given function of the amplitude of the input signal.This preserves the phase information (time shift). When transmitting signals from neuron to neuron on short relations holds one correspondence between the input and the output of such a chain of neurons. The whole point is treated group of spatial frequency components of the signal of the neural network. At synapses associative network, we remember the spatio-temporal relationships between the individual components of the decomposed signal.Then, using these relations we can call signals from memory, organize cycles, implement other functions. Some considerations can be found in my articles.
Are you saying that within synapses' associative network the "instructions to hold the information (signal)" are preserved?
This reminds me holographics: there is a plate on which are encoded the coordinates (spatial relations) of a projection; the act of is implemented by an incident ray - that "redial" the recorded shape. Similarly, if a 2d layer projects in 3d, a 3d one (brain) could project in 3d+1: if so, signals would be holographically encrypted in neural networks. that is the coding of the information they carry is recorded over a substrate with "less" dimensional properties, in a way that, given an instant (so that there isn't any movement or change regarding the layer), the projection coming from this substrate needs at least one more dimension to be fully represented - a layer that does not vary (constant in [a] time [lapse]) provides a cinematic projection.
In this model, the neuron would be the unit which transmits the informations regarding the correct composition of the signal and its redialing.
Dear Federico, your views are very interesting. It is worth to think about it.
- Badges
- Science topic