62 COGNITIVE THEORY AND RESEARCH ON ANXIETY
Unfortunately, interpretation of the Stroop findings is hindered by limitations in its
methodology. It is possible that slower color naming could be due to diverting attention
away from threatening words rather than because of enhanced attention to the meaning
of the word (MacLeod, 1999). Also longer reaction times to threatening words could be
due to the interfering effects of an emotional reaction to the word (e.g., startle response),
or because of mental preoccupation with themes related to the word (Bögels & Mansell,
2004). Because of these potential response biases (see Mogg & Bradley, 1999a), probe
detection tests have surpassed the emotional Stroop task as the preferred experimental
paradigm for investigating attentional bias in anxiety.
Dot Probe Detection
The dot probe detection experiment is able to assess hypervigilance for threat in terms
of both facilitation and interference with dot detection without the effects of response
bias (MacLeod, Mathews, & Tata, 1986). In this task a series of word pairs is presented
so that one word is in the upper half and the other word in the lower half of a computer
screen. The trial begins with a central fixation cross presented for approximately 500
milliseconds, followed by a brief presentation (500 milliseconds) of a word pair. On
critical trials a threat and neutral word pair are presented followed by the appearance of
a dot in the location formerly occupied by one of the words. Individuals are instructed
to press a key as quickly as possible when they see the dot. Hundreds of word pair trials
are usually presented with many involving filler neutral– neutral word pairs.
A number of dot probe experiments have demonstrated an attentional threat bias
in clinically anxious patients but not in nonanxious controls. Anxious patients mainly
with a primary diagnosis of GAD exhibit significantly quicker dot probe detection after
physically and socially threatening words (MacLeod et al., 1986; Mogg, Bradley, &
Williams, 1995; Mogg, Mathews, & Eysenck, 1992). Attentional vigilance for threat
has also been found in panic disorder for detection of physically threatening words
(Mathews, Ridgeway, & Williamson, 1996), OCD for contamination words (Tata, Lei-
bowitz, Prunty, Cameron, & Pickering, 1996), and social phobia for negative social
evaluation cues (Asmundson & Stein, 1994). Vassilopoulos (2005), however, found that
socially anxious students showed vigilance for all emotional words (positive and nega-
tive) at short exposure intervals (200 milliseconds) but avoidance of the same word stim-
uli at longer intervals (500 milliseconds). In addition, negative findings have also been
reported, with GAD patients failing to show attentional vigilance for threatening words
or angry faces (Gotlib, Krasnoperova, Joormann, & Yue, 2004; Mogg et al., 1991; see
also Lees, Mogg, & Bradley, 2005, for negative results with high health- anxious stu-
dents).
Researchers have employed a visual dot probe task in which probe detection is
measured to pairs of pictorial stimuli involving angry versus neutral facial expressions
as a more valid representation of social evaluative threat (Mogg & Bradley, 1998).
However, visual dot probe has produced inconsistent results. While some researchers
have reported an initial selective vigilance (quicker probe detection) to angry or hostile
facial expressions at short intervals only (e.g., Mogg, Philippot, & Bradley, 2004), other
researchers failed to find vigilance for threatening or angry faces in analogue or even
clinical social anxiety groups (Gotlib, Kasch, et al., 2004; Pineles & Mineka, 2005),
and others have even reported an opposite finding, with high social anxiety character-