Computation in Neural Nets - Understanding Understanding: Essays on Cybernetics and Cognition - page 41

In table 1 we show that, with two exceptions all logical functions with two
input lines can be computed by a single McCulloch formal neuron, if full
use is made of the flexibility of this element by using various thresholds and
synaptic junctions. For convenience, this table lists, in six different “lan-
guages”, all logical functions for two arguments. The first column gives a
digital representation of Wittgenstein's truth function (second column),
taken as a binary digit number to be read downwards. The third column
gives the appropriate chiastan symbol, and the fourth column shows the
corresponding element with its synaptic junctions and its appropriate
threshold value. The two functions which cannot be computed by a single
element require a network of three elements. These are given in fig. 12 and
are referred to in the appropriate entries. The fifth column shows the same
TABLE 1.