An evolution in theoretical models and methodological paradigms for investi
gating cognitive biases in the addictions is discussed. Anomalies in tradit
ional cognitive perspectives, and problems with the self-report methods whi
ch underpin them, are highlighted. An emergent body of cognitive research,
contextualized within the principles and paradigms of cognitive neuropsycho
logy rather than social learning theory, is presented which, it is argued,
addresses these anomalies and problems. Evidence is presented that biases i
n the processing of addiction-related stimuli, and in the network of propos
itions which motivate addictive behaviours, occur at automatic, implicit an
d pre-conscious levels of awareness. It is suggested that methods which ass
ess such implicit cognitive biases (e.g. Stroop, memory, priming and reacti
on-time paradigms) yield findings which have better predictive utility for
ongoing behaviour than those biases determined by self-report methods of in
trospection. The potential utility of these findings for understanding "los
s of control" phenomena, and the desynchrony between reported beliefs and i
ntentions and ongoing addictive behaviours, is discussed. Applications to t
he practice of cognitive therapy are considered.