Is the face-word task strictly a test of selective attention?

Tasks that have been used in previous studies to address selective attention in major depression have tended to involve the presentation of a neutral element and an emotional item. These stimuli are either presented in the same physical space; as in the emotional Stroop task, where the emotional element (an emotional word) is combined with the neutral component (an ink colour), or in different locations as in the dot probe task, where the emotional stimulus (e.g. an angry face) and neutral item

(e.g. a neutral face) appear at different locations. In these types of studies it is possible to make inferences about differences in the allocation of attention based on patterns of colour-naming times or relative differences in response times to the probes. The face- word variant of the Stroop task that was utilised in the current study was originally designed by Stenberg et al (1998) to investigate the extent to which emotional faces were processed automatically and whether this processing influenced other ongoing cognitive activity. There are, however, a number of similarities between this paradigm and the other tasks that have been used to tap selective attention for emotional material in depression. For example, a number of stimuli in the current study involved a neutral stimulus paired with an emotional item (e.g. a neutral face paired with a positive or negative word, or an emotional face paired with a neutral word). Despite these similarities, there are also clear differences between the face-word task and the other selective attention tasks. For example, many of the compound stimuli presented during the current study consisted of two emotional elements (e.g. a sad face with a positive word). It could be argued that where the face-word pairs were incongruent (e.g. happy face/negative word or sad face/positive word) the task closely resembled the original Stroop task, i.e. in the original Stroop task both elements were colour­ relevant and in the current face-word task both elements were emotion-relevant. It is clear, however, that where the face-word pairs were congruent (happy face/positive word or sad face/negative word) the task was not addressing selective attention but rather emotional priming.

It would seem clear that evidence of slowed valence identification in the incongment conditions could be interpreted as the automatic processing of the to be-ignored stimulus interfering with the processing of the to-be-attended stimulus, in line with

explanations of the traditional Stroop task. Slowed identification of neutral words in the presence of happy faces, as demonstrated by the controls, could be interpreted as a consequence of the participants’ attention being captured by the happy faces, in line with the logic applied in explaining the changes in colour-naming times in studies using the emotional variant of the Stroop task. More rapid identification of words presented in congruent condition would have provided evidence of emotional priming, similar to that demonstrated in Stenberg et al (1998). However, there was no evidence of this phenomenon in the current study. Overall, it can be argued that the face-word task employed in the current study does address elements of selective attention in similar ways to other common attentional tasks. However, it also provides interesting and important information concerning how individuals (depressed & non­ depressed) allocate attention in conditions where two conflicting emotional cues occupy the same spatial location. The data produced by the face-word task is very complex and requires careful analysis in order to understand the findings. In future it would be advisable to try and simplify the design to look separately at emotional priming and selective attention.

3.3,5. Summary and Conclusions (Selective Attention Task)

Taken together, the evidence from the valence identification times and percentage of errors revealed that, contrary to the predictions, depressed individuals did not demonstrate a selective attention bias for sad facial expressions. Conversely, evidence from the identification times revealed that the controls exhibited an attention bias for the happy faces. This supports the notion that depressed individuals are even-handed in their processing, whereas healthy individuals tend to see the world through “rose- tinted” spectacles (e.g. Gotlib et al, 1988). It would seem likely that reduced attention

duration of the patients’ depressive episode. For example, if a depressed person, was in conversation with a significant other, and failed to attend to the conversational partner’s smiling face this could negatively influence the interpretation of the interaction. There was support for the predicted congruence/ incongruence effect on valence identification times. However, it appears that it was depressed participants and not controls that exhibited changes in positive word identification as a function of face valence. Contrary to the findings of Stenberg et al (1998) healthy controls did not exhibit an emotional bias for positive words. Explanations for this non-replication include differences in participant variables (e.g. age, depression) or possibly cultural differences. In line with previous studies (e.g. MacQueen et al, 2000) the patients demonstrated impaired inhibition of distracting “off-task” stimuli, indexed by significantly slowed identification times in circumstances that involved incongment face/word pairs. It is plausible that this relates to a dysfunction of the anterior cingulate in the depressed sample. Overall, depressed individuals were more accurate in their identification of negative words, possibly due to their expertise in processing these words. However, this advantage disappeared when the negative words were presented with incongment (happy) faces, possibly due to the increased cognitive demands of trying to resolve the conflict in meaning between the incongment faces and words.

3.4, Results and Discussion (Recognition Memory Task)