What Is Memory that It May Have Hindsight and Foresight as well? - Understanding Understanding: Essays on Cybernetics and Cognition - page 109

As much as can be said about some features of this elementary four-
element network, I wish to stress again the utter inadequacy of this net to
account for even the most straight-forward cases of conditioned reflexive
behavior in higher animals. The belief harbored perhaps by early reflexol-
ogists, that ultimately such behavior can be reduced to a logical or neural
schema of the sort shown in Fig. 1 has—to my knowledge—been completely
destroyed by the superb work of Jerzy Konorski 15 who showed that, for
instance, in dogs at the first application of the positive conditioned stimu-
lus it elicits a quite distinct “orientation reaction”, i.e., pricking up the ears,
turning the head, etc., while salivation as response is negligible. He goes on
to demonstrate that in almost all experimental set-ups conditioned stimuli
“. . . do not usually possess a single modality, but they supply a number of
cues...”which the animal utilizes and evaluates as to theis significance in
determining future action. Konorski reaches the conclusion that essentially
two principles govern the acquisition of various types of conditioned
reflexes, one, a principle of selection, the other one a principle of insepara-
bility of information from its utilization. Since I consider these principles
of considerable importance in my argument, I shall state them more explic-
itly in Konorski's own words:
(i) Selection
“In solving a given conditioning problem the animal does not utilize all the
information supplied by the conditioned stimuli, but it definitely selects
certain cues, neglecting the other ones.”
(ii) Inseparability
“. . . it is not so as we would be inclined to think according to our intro-
spection, that receipt of information and its utilization are two separate
processes which can be combined one with the other in any way.” Hence:
“Information and its utilization are inseparable constituting, as a matter of
fact, one single process.”
If I may translate these observations into my terminology of before, the
principle of selection becomes a “search for meaning” in the sense that
animal selects those cues—i.e., that information—from which it can opti-
mally draw inferences; while the principle of inseparability becomes a
“recourse to self-reference” in the sense that the animal evaluates the infer-
ences drawn from that information always with regard to its utilization
favorable of its own self.
In search for a minimal network that would exhibit these two principles
of selection and of inseparability of information from its utilization—or of
“search for meaning” and “self-reference”—I came across J.Z. Young's