Currently there is therefore a need to overcome both the issues surrounding the use of
inappropriate testing populations and the inconsistent and often flawed methodologies that have
been employed in the study of attentional bias. One way to address the first of these issues would
be to induce an attentional bias in a non-clinical population, and then compare findings to
participants from the same population who have had no attentional bias induced. This would be an
ideal way to probe the cognitive mechanisms underlying attentional bias, since all participants would
be free from additional confounds such as the emotional saliency and reward associated with bias-
related stimuli. Such an approach has been attempted, however even these have issues, placing
uncertainty on findings. Yaxely and Zwaan (2005) investigated the development of a smoking-related
attentional bias in both a clinical sample of smokers and a non-clinical sample of non-smokers. They
separated smokers and non-smokers into two groups; one given information that the study was
and naïve smokers. Thus, informed non-smokers and both informed and naïve smokers displayed a
smoking-related attentional bias. The only group that showed no evidence of a smoking-related
attentional bias was the group of naïve non-smokers.
These findings suggest not only that the clinical sample of smokers held a smoking-related
attentional bias, but that the behavioural effect of this was at ceiling before the study information
was provided (Cronbach, 1988). It also suggests that non-smokers can have an attentional bias
induced via information provided before the commencement of testing. However, while this study
highlights that information is sufficient to induce a bias-like effect in a healthy sample, the study was
more social than experimental. There were no strict controls placed on the stimuli, for example they
were not matched for size, contrast or luminance, and it is possible that some of the non-smokers
who were provided with study-related information had relatives who smoked and were therefore
more susceptible to developing a smoking-related attentional bias. Moreover, smoking-related
stimuli are not arbitrary and therefore can trigger an emotional response (Janes et al., 2010). To
investigate the possibility of inducing attentional bias further, the general idea of Yaxely and Zwaan’s
study should be examined in a more tightly monitored laboratory setting alongside an arbitrary
stimulus to add control and remove any potential social/emotional confounds. This will offer a more
precise avenue for the investigation of the cognitive aspects of attentional bias.
An alternative study examined the link between learned associations and the development of an
attentional bias. Unlike Yaxely and Zwaan’s study however, a healthy sample of university students
was used allowing researchers to make direct comparisons between subject groups. Pischek-
Simpson et al. (2009) split a group of university students into two equal groups. One group viewed
neutral and angry faces; however some of the angry faces were paired with a small electric shock
resulting in fear conditioning to induce an attentional bias towards angry faces. The other group
viewed the same images with no accompanying shock. Both groups then completed a dot probe task
who received a shock displayed evidence of an induced attentional bias however, the attentional
bias did not generalise to the angry faces that were not originally accompanied by a shock (Pischek-
Simpson et al., 2009). This study therefore suggests that it is possible to induce an attentional bias in
a normative population. However firstly, due to the use of emotive stimuli (angry faces) there is still
an issue with additional neural processing contributing to the inducement of the bias. Furthermore,
when participants received the dot-probe task, the electrodes that delivered the shocks were still
attached, meaning there was therefore a real chance that an additional shock could be delivered.
This would further add to the additional emotional processing, meaning that the cognitive
mechanisms of attentional bias were not directly investigated.
Pothos and Tapper (2010) did attempt to remove emotional confounds from attentional bias by
inducing a Stroop effect in a healthy sample towards meaningless words in order to investigate if the
amount of automatic associations with a word or the strength of an association causes the Stroop
interference effects. They tested university students across 5 consecutive days to relate single non-
words to either one real word or several related real words. At the end of the training, participants
received a Stroop task containing all words (real and fake), and found that the Stroop effect was
more pronounced for the non-word relating to only one real word (Pothos & Tapper, 2010). Thus,
the Stroop effect is observed more in terms of the strength of a connection, not the quantity of
connections it has with other words. However as previously mentioned, the Stroop test is more of a
measure of semantic interference, not attention. Moreover, since this training had to take at least a
week to be observed shows this suggests that the Stroop paradigm is not a sensitive measure of
attention, which clouds the inferences that can be drawn from this study.