Noise
NOISE AND DELAYS EXPLAIN WHY SOME GENES OSCILLATE IN ACTIVITY
“unscripted biochemical variations, or noise, combined with time delays in certain biochemical reactions may lead to oscillations in gene regulation that couldn’t otherwise be predicted. Such noise is routinely described by cell biologists who record large phenotypic differences between supposedly identical cells in a single flask of growth medium.â€
…
“The fine-grain fluctuations we see in the genetic regulation within single cells may lead to new insights about variability at the level of the whole organismâ€
Noise is under appreciated. Some speculation on noise…
Noise prevents sharp “edge†effects. Imagine a thermostat with no “noiseâ€. Too cold then turn on, too hot then turn off, the heater would be continually turning on and off. Slop in a steering wheel is another example. If the wheel is too responsive then the driver over corrects.
“Noise†in the genome: Mutations generate new gene alleles. If the new allele significantly improves fitness then it rapidly increases in frequency. If the new allele significantly decreases fitness, then it may disappear. But many mutations won’t have much of an effect either way. Such gene alleles act as genetic “noiseâ€. Such genetic noise produces statistical “outliers†that are extreme phenotypes. (E.g., very high IQ.)
Gene allele “outliers†provide feedback. If the environment changes so that the “outlier†significantly improves fitness then that gene allele frequency increases and the population rapidly adapts. Negative “outliers†would “push†the population away from the “bad†allele. Thus noise makes the genome more robust and stable.
Molecular noise provides the random generator needed to for our body’s immune system to build broad coverage against potential invaders and generate new antibodies to specific molecular targets against invaders that make it past the first immune barrier.
Noise can provide the “mutations†in a Darwinian mechanism such as might occur in thinking and skill learning. (Similar to “heat†in simulated annealing.)
Noise keeps certain neuronal systems healthy. Too little noise can lead to heart attacks or seizures.
Update from Razib: Look in the gnxp files for “noise,” that is the PDF of the paper that the article is based on.
The absence of short-cycle hysteresis in the operation of a thermostat has nothing to do with noise and everything to do with control systems engineering. Generally a thermostat is set so that heating starts at a temperature several degrees lower than the stop temperature. Physical factors can still lead to oscillations if things aren’t set up right, but not rapid ones like you describe.
Slop in a steering wheel, also not an example of noise – rather an example of damping. We can certainly imagine a steering wheel with noise added (random jerks or vibrations) but I doubt there would be any benefit.
The link unfortunately provides few details on the work, which does look interesting – they sort of hint at nonlinear analysis but it’s hard to tell what they’re really up to.
?nothing to do with noise and everything to do with control systems engineering?
(I agree my example of a thermostat depending on noise was poor.)
I?m using a noisy definition of noise. I?m trying to capture a broad abstraction that noise isn?t just unavoidable errors in measurement or production or operation but instead is important for the proper functioning of many systems.
My steering example came from my personal experience working on an old car. A driver needs a little slop so that he gets feedback that the steering wheel is moving before the car wheels actually turn. The needed slop would be related to the brain/muscle feedback neural feedback. I suspect the slop originally came from the rough manufacturing processes in the steering system components. As manufacturing processes improved, the ?noise? had to be designed into the system. So I think an old steering system captures the concept I?m trying to express.
What term would you use for this abstraction?
I suspect that noise in mental categories plays an important role in human intelligence. Too much slop hinders communication. Too little slop hinders abstraction flow across knowledge domains. Translate a concept into another language and then back to the original language. You might get garbage or you might get a creative insight. I suspect something similar happens in our brain as we shift domains, e.g., ?noise? in engineering systems to ?noise? in conversations.
guitar feedback works on this principle.
Noisy definition of noise, I like it :-)
I would still call the steering wheel slop damping. It strikes me as a funny choice of example because the function appears to me to be the *elimination* of noise (small twitches and movements) generated by the driver. With no slop, every little movement turns the wheels, making the driving experience require more attention and constant motor control. Rather OT but Americans and Europeans have rather different tastes in terms of how much slop gives the right feel to the car’s steering…
I agree with your broad insight about the necessity and importance of noise. I wonder whether genetic transcription and error correction is at the theoretical limit of accuracy, or whether it’s demonstrably sloppier so as to hit some fitness maximum. For that matter, I think I’ve seen research that indicates increased mutation rates in response to environmental stress, which would allow a choice (loosely speaking) of appropriate noise level.
Joe O: ?guitar feedback works on this principle.?
Thanks to Razib providing the paper, I now see a connection between guitar feedback and the gene regulatory network.
The distinct molecular reaction delays in the gene regulation feedback loop are similar to the different guitar strings with distinct resonant frequencies. When white noise strikes the guitar strings the resonant frequencies in the noise start the strings vibrating. The energy in the non-resonant frequencies quickly dissipates. Noise in biological systems can act in a similar manner resulting in oscillations in gene regulation.