180 Cognitive-Experiential Self-Theory of Personality
in order to maintain relationships, she had no way of learning
before therapy that it was unnecessary.
Implications for Research
If there are two different information-processing systems, it
can only be a source of confusion to conduct research as if
there were only one, which is the customary practice. As an
example, given the existence of two different systems, it is
meaningless to investigate “the” self-concept because a per-
son’s self-concept in one system may not conform to the self-
concept in the other system. Moreover, the difference between
the two self-concepts can be of considerable importance in its
own right. The problem of treating the two self-concepts as if
there were only one has been particularly evident in research
on self-esteem, in which individuals are typically classified as
high or low in self-esteem based on self-report questionnaires.
Yet if there aretwoself-concepts, then it is quite possible for
people to be high in self-esteem in one system and low in the
other. For example, a person might be high in self-esteem in
the rational system, as measured by a self-report test, yet low
in self-esteem in the experiential system, as inferred from be-
havior (Savin-Williams & Jaquish, 1981).
There has been much disagreement concerning whether
elevating students’ self-esteem by treating them as successful
no matter what their performance is desirable or undesirable.
In order to resolve this issue, from the perspective of CEST it
is necessary to recognize that high self-esteem at the con-
scious, rational level may coexist with low self-esteem at the
experiential level. It is one thing to teach students to con-
sciously believe they have high self-regard and another to
have them acquire the quiet confidence that comes from feel-
ings of mastery and competence that are a consequence of
real accomplishment. The former in the absence of the latter
can be considered to be no more than self-deception and a po-
tential source of disillusionment in the future. It follows that
not only is it important to examine self-esteem separately in
each of the two systems, but it is equally important to conduct
research on their convergence. What is obviously true of self-
esteem in this respect is equally true of other personality vari-
ables, including basic needs and beliefs.
Although the importance of four basic needs and corre-
sponding beliefs is emphasized in CEST, this is not meant to
imply that lower-level beliefs and needs are not also very im-
portant. Recently, social and personality psychologists have
emphasized midlevel motivational constructs (e.g., Emmons,
1986; Markus & Nurius, 1986; Mischel & Shoda, 1995). It is
assumed in CEST that personality is hierarchically organized,
with broad, basic needs subsuming midlevel motives, which
in turn subsume narrower, situation-specific motives. It would
therefore be desirable to examine the organization of such
needs and beliefs, and to determine in particular the kinds of
relations the different levels establish with each other, as well
as with other variables. It might reasonably be expected that
the lowest-order needs and beliefs are most strongly associ-
ated with situationally specific behaviors, and the higher-order
beliefs and needs are more strongly associated with broad dis-
positions, or traits. The higher-order beliefs and needs can also
be expected to be more weakly but more extensively associ-
ated with narrow behavioral tendencies. Midlevel motives can
be expected to have relations that fall in between those of the
higher- and lower-order needs. A particularly important hy-
pothesis with regard to CEST is that higher-order needs and
beliefs are more resistant to change than are lower-order needs
and beliefs, but should they be changed they have greater ef-
fects on the overall personality structure. Moreover, any major
changes, including positive changes, are disorganizing and
anxiety-producing because of the basic need to maintain the
stability of the conceptual system.
Although considerable research has recently been con-
ducted on midlevel needs that has demonstrated their theoret-
ical importance and predictive value (e.g., Emmons, 1986;
Markus & Nurius, 1986; Mischel & Shoda, 1995), the ques-
tion remains as to how the midlevel needs can best be desig-
nated and measured. The most thorough and compelling list
of midlevel needs to date still appears to be the list proposed
by Henry A. Murray (1938) many years ago. It is interesting
from the perspective of CEST that Murray measured mid-
level needs both explicitly via direct self-report and implic-
itly through the use of the Thematic Apperception Test (TAT;
Murray, 1943). A more psychometrically advanced procedure
for measuring the Murray midlevel needs at the explicit level
has since become available in the form of the Edwards Per-
sonal Preference Schedule (Edwards, 1959).
There is a need for research to further explore the TAT as
a measure of implicit needs and to also examine additional
measures of implicit needs. Included could be older proce-
dures such as word association and sentence completion, as
well as promising new procedures such as priming tech-
niques and subthreshold measures (see Bargh & Chartrand,
in press, for a review of such techniques). It would be inter-
esting to relate the various implicit measures to each other to
determine whether they have enough in common to combine
them into an overall measure. The implicit measure (or mea-
sures) of needs could then be related to explicit measures of
needs, and both could be related to external criteria. Through
such procedures it should be possible to determine in what
ways implicit and explicit measures are similar and different.
It could also be determined whether they contribute in a sup-
plementary way to the prediction of the same variables and