Socially anxious individuals are characterized by an attentional bias for threat-related information. However, this attentional bias to threat may be attenuated [Behav. Res. Ther. 34 (1996) 945] when these individuals are anxious. In the present study, we attempted to increase subjects’ Stroop interference by changing the frequency of words to nonwords. We hypothesized that when the ratio of words to nonwords was low, individuals with social phobia (SPs) would show increased Stroop interference, but only for words related to social threat. Results were consistent with this hypothesis.
To examine the role of attentional bias for socially threatening information in socially anxious individuals, researchers have used the emotional Stroop paradigm (Williams, Mathews, & MacLeod, 1996). In this paradigm, participants are asked to name the color in which emotional words are written, while ignoring the meaning of the words. The finding that socially anxious individuals are slower at color-naming threat-related words compared to neutral words may suggest that these individuals pay particular attention to threat meaning (Hope, Rapee, Heimberg, & Dombeck, 1990; Mattia, Heimberg, & Hope, 1993).
However, Amir et al. (1996) found that individuals with social phobia (SPs) show a reduction in Stroop interference for threat words (i.e., response latencies were faster), but not neutral words, when they are anxious. These researchers interpreted these results to suggest that SPs are able to moderate their attention to threat using strategic (i.e., effortful, conscious, serial; Shiffrin & Schneider, 1977) processes. If socially anxious individuals use strategic processes to reduce their attention allocation to threat, then any manipulation that would encourage relaxing these processes should lead to an increase in Stroop interference for threat words in these individuals.
To examine the malleability of the Stroop interference effect, Tzelgov, Henik, and Berger (1992) showed participants color-incongruent words (e.g., RED printed in green), color-congruent words (e.g., RED printed in red), and color-incongruent nonwords (e.g., WWWW printed in red). There were four groups of participants. Each group saw a total of 192 color-naming trials, but the proportion of color-incongruent words was varied across groups. Participants in the first group saw 96 (50%) color-incongruent words, those in the second group saw 72 (37.5%) color-incongruent words, those in the third group saw 48 (25%) color-incongruent words, and those in the fourth group saw 24 (12.5%) color-incongruent words. Participants who saw fewer color-incongruent words showed more Stroop interference than participants who saw more color-incongruent words. Thus, the decrease in the ratio of color-incongruent words to other stimuli increased Stroop interference. These authors suggested that when the ratio of color-incongruent words to nonwords was low, participants relaxed their effortful (i.e., strategic) processes that may have been responsible for inhibiting the competing meaning of the words. Therefore, subjects may have used strategic processes to limit their automatic attention to color.
The above paradigm is potentially useful in studies of information processing bias in anxious individuals because it may help delineate the role of stimulus set (i.e., ratio of threat to non-threat) and its possible effect on anxiety. To the extent that anxious individuals are sensitive to threat-relevant information, they should be more sensitive to manipulations that increase this bias. Therefore, they should show an increase in Stroop interference for threat when the color-incongruent words to nonwords ratio is low. If this attention allocation is specific to threat, then these changes should be magnified when responding to threat-related words, but not other emotional information.
In the present study, we hypothesized that SPs would show more Stroop interference for social threat, but not positive, words when the ratio of words to nonwords was low than when this ratio was high. To examine this hypothesis, we modified the Tzelgov et al. (1992) study in two respects. First, we used a within-subject design by asking participants to perform the Stroop task in both low and high word ratio conditions to increase power. Second, we used social threat words and positive words allowing us to test the specificity of any effect to threat words.
