3 ms·

the abstract was mentioned in this post https://news.ycombinator.com/user?id=SubiculumCode https://news.ycombinator.com/user?id=SubiculumCode > ...low‐frequenc

• candiodari 8y ago
  Okay but you have to keep in mind that there are multiple waves, and what you might use them for. Think of how a sonar works. Sonars must associate and compare the timing of events, just like neurons. They use as many different frequencies for that as possible (used to be 4-8 frequencies but the state of the art is presumably much more) Let's say I have brainwaves at 0.5Hz, 4Hz, 8Hz and 20Hz wave. And let's assume they are all non-harmonic. If you have an event, and you can quickly inspect the position and phase of all these waves, you now have an extremely accurate and long-period reference for the time the event occurred (think of the individual waves as the numbers on a clock). So the accuracy would be a decent multiple of the highest frequency, so let's say 20-50ms resolution. And the repeat is the multiple of all the wavelengths, so this would give neurons a trivial way to understand or plan events up to 640 seconds away from the current moment, as well as an easy way to refer to the exact moment a prediction will come true, and how different events relate to eachother in time. It's brilliant.
• mattkrause 8y ago
  One theory is that the waves themselves don’t carry much information, but set the stage for other kinds of transmission or computation. In the rodent hippocampus, for example, theta oscillations (3-7ish Hz) don’t tell you much about the animals’ spatial position. The spiking rate of place cells contains some information about where the animal is, but it’s fairly coarse. However, combining spiking and the phase of the slow oscillation (e.g., cell is firing at the trough vs peak) provides much more precise information about where the rat is. A related idea is that oscillations help route information by alternatingly facilitating and suppressing different areas. At one phase of the oscillation, information might preferentially flow from visual areas to higher-level areas. At another, the visual input might be suppressed while memory structures get “their turn to talk.”
  • DoctorOetker 8y ago
    note that my observation is not that the information contained is not "zero usefull information" but "not much usefull informmation". I certainly think a local clock would be usefull, and contains very little information. But any large amount of information density is simply not supported by these observations due to simple wave aspects. A low information signal (for example highly predictable clock, or some other low frequency, high brain volume signal say perhaps a breathing rate not actually the breathing frequency) can still be important. I was only replying to the information processing comment
    • mattkrause 8y ago
      ? I…don't see the argument here. I was just trying to provide some actual examples from neuro.