Biology Reference
In-Depth Information
2.2.4
Probability is not noise
The concept of noise is equally problematical. It is very frequently
confused with probability whereas its meaning is different and closer
to accident. Certain phenomena can be scrupulously described and
predicted by deterministic scientific laws, but when an experiment
is performed there is always a slight discrepancy between the pre-
diction and the observation. This discrepancy is called experimental
background noise. It is due to the fact that one can never perform
a perfect experiment: there is always inaccuracy in the measure-
ment, due to the apparatus used or some disturbance to the exper-
iment from an external factor. The experimental approach consists
of attempting to isolate the phenomena by getting as close as pos-
sible to perfection, but background noise can never be totally elim-
inated. Measurement is always within the framework of a margin of
statistical error. This noise relative to the theoretical result varies
randomly from one experiment to another but, if the phenomenon
studied is deterministic, it does not nevertheless transform it into a
probabilistic phenomenon. Whatever its quantitative importance,
there is always fluctuation which is added to the constant result.
A phenomenon subject to noise therefore has to be differenti-
ated from an intrinsically probabilistic phenomenon. These ideas
are at the heart of fundamental debates in biology. Recent experi-
mental results suggest that gene expression is a probabilistic phe-
nomenon. These results are however interpreted as noise by
numerous authors who think, despite everything, that cell function
is a fundamentally deterministic phenomenon and that there is no
reason to challenge the theory of genetic programming. On the con-
trary, if gene expression and cell function are intrinsically proba-
bilistic, that leads to a major upset in understanding biological
mechanisms and calls into question genetic determinism (see the
following chapters).
To understand the difference between a deterministic mechanism
with noise and an intrinsically random mechanism we might con-
sider the example of a car which has to travel between two towns.
For a deterministic mechanism with noise, a constant speed is set
