The present study examined whether the cognitive bias for threat is a stabl
e phenomenon in spider phobics and not in nonphobic controls. The rationale
of this study was that the use of emotional bias for individual assessment
in clinical practice requires stability of this phenomenon. In order to as
sess the stability of the emotional bias, the spider Stroop task was admini
stered twice to spider phobics and controls, with a time lag of 3 weeks. In
two experiments, spider phobics (n(exp. 1) = 20; n(exp.) (2) = 33) and con
trols (n(exp. 1) = 24; n(exp. 2) = 25) were selected on basis of a structur
ed interview experiment, subjects anticipated exposure to a real-life spide
r Emotional bias was inferred from color-naming latencies on spider words v
ersus control words. In line with our prediction, the bias for threat was s
table in spider phobic individuals but not in controls. This stability coul
d not be attributed to a general cognitive characteristic because a standar
d Stroop task did not differentiate between the spider phobics and the cont
rols. The results support the view that cognitive processing of threat in a
nxiety is stable. If is proposed to rise both the emotional bias and its st
ability as measures of the constraints on the flexibility of the cognitive
fear-network.