I give an historical overview of how cognitive (and other) models of users of
interactive IR (IIR) systems have been elicited, constructed and used. Cognitive 
models in the domain of interactive information retrieval (IIR) are understood 
as models that a system (or a person) constructs of a(nother) person's "information
need"; these are called "user models". Context models are models that a system
(or a person) constructs of the conditions that led a(nother) person to engage in
information seeking behavior, various characteristics of that person, and various
aspects of that person's environment, broadly construed. Both types of models
are used in personalization of IIR.

The review begins with discussion of Robert Taylor's 1968 article in which he proposed
four levels of "information need" or "query", and five "filters" according to which 
the librarian and the information seeker identify and clarify (i.e. model) various
aspects of the user, the user's goals, the topic of interest and so on. I then review 
various approaches to understanding of why people engage in information seeking
behavior, and of automatically constructing cognitive and contextual models.
Significant change points include the "cognitive turn" in the early to mid 1970s, 
modeling the human intermediary in the 1980s, and the "interactive turn" in the
1990s. The review concludes with a brief description of, and comments on the current
state of cognitive and contextual modeling in IIR.