Investigation in neurophysiology usually involves measurements of larg
e population-average signals or small sample recordings. There is an u
nderlying assumption that the observations express activity of large g
roups of similarly acting neurons that is the result of a bottom-up sc
enario in which individual cells, via their synaptic interactions, lea
d to the large scale phenomena. The connection between the levels must
be provided by theory, which must also provide the relevant variables
for observation. It is suggested that between the experiment and the
full theory there is a creative, mixed role for simulation: both exper
imental and theoretical. A simulation presents complex dynamics and he
nce is an empirical board for testing theoretical tools, yet its contr
olled behaviour can make predictions about the biological system.