This article investigates fundamental requirements of an architectural
framework on which social intelligence could be developed and propose
s the self-biased (SB) conditioning scheme as a basis for development
of social intelligence. The objective of SB conditioning is to develop
a set of secondary responses in a social context. SE conditioning req
uires two kinds of components: a set of primary response networks and
an associative memory. A primary response network is unique in this co
nditioning scheme. It includes three types of components: an activatio
n node, an expectation node, and a motor center. This network, by its
expectation node, presupposes an involvement by others, in a social co
ntext, and therefore it can be a basis of social intelligence. A few e
xamples of social intelligence are shown with this SE conditioning sch
eme: ''eagle alarm'' call learning, attachment behavior, and sympathet
ic coupling. The topic of situated personal assistance is also introdu
ced in relation to the proposed scheme.