The first hypothesis predicted visuospatial impairments in the left visual field in individuals with depression, reflected as a bias to the right visual field. A statistically significant difference of lateralized attention in patients versus controls, however, was not established. Our second hypothesis predicted a relationship between sad mood, alertness and a left visuospatial attention deficit in depressive patients. Our data supported the view that alertness deficits in depression promote visuospatial impairment in the left hemifield, reflected as a rightward bias of spatial attention, whereas sad mood showed a trend in the same direction. The influence of alertness on visuospatial attention was hypothesized, based on the model of a combined network of alertness and visuospatial attention [15
]. Moreover, the effect of mood - although weak - also supports the hypothesis that sadness inhibits visuospatial attention processing in the right hemisphere and therefore favours left visual field impairment [3
Our patient sample yielded a trend towards left-sided visuospatial deficits, seen as positive values in the number of omissions in two of the three neglect tests (i.e., an attention bias towards the right visual field). Since reaction times revealed no group difference concerning the visual field bias in all three tests, speed-accuracy trade-off could not account for different tendencies in omission scores. Moreover, reaction times must be considered as possible confounds of alertness differences in both groups and should therefore not be interpreted with regard to our hypotheses.
There are, indeed, several reasons that may explain the absence of stronger (and significant) cognitive deficits in the group as a whole. In the present study, mean treatment duration of the patients was longer than two months at testing time. Cognitive disturbances thus may have been reduced by antidepressant medication. When testing acutely depressed patients, a different cognitive profile may be expected. Grant and co-workers administered a neuropsychological test battery to a group of unmedicated middle-aged mildly depressed patients and showed that reaction time was positively correlated to illness severity and that there was a moderate relationship of illness severity and performance deficit in attention shifting [31
]. As our sample comprised mildly to moderate depressed patients, low illness severity may explain the lack of significant differences between depressive patients and healthy controls.
Concerning arousal, whereas depression is commonly associated with low arousal, anxiety is associated with high arousal. On the neural level, the right posterior region is hypothesized to exhibit lower activation in depression and higher activation in anxiety [11
]. According to the spatial attention network model, these different levels of arousal may influence anterior regions in the same way [16
]. Comorbid anxiety in depression may therefore abolish the low arousal and lead to increased activity in the attention network. The fact that depressive patients scored significantly higher in state and trait anxiety than our healthy subjects may have contributed to the relative high values of arousal and consequently may have prevented observing a more pronounced visuospatial attention deficit in the current sample.
Omission scores on the TAP Neglect subtest were predicted by alertness. In the WAF Extinction-Neglect subtest, alertness and sad mood influenced omission scores at a trend-level. Using three different tests and a conservative correction for multiple testing reduced the statistical power in the current study. Furthermore, we noted a degree of variability across the tests.
It has to be taken into account that different visuospatial attention measures differ in sensitivity and recruitment of processing resources. Reviewing several studies on cognitive impairment in depression, Levin and colleagues concluded that the problems arise from using a different strategy in the allocation of resources [32
]. In the present study, this finding may be reflected by a more rightward attention bias in the TAP Neglect and WAF Extinction-Neglect subtests in depressive patients. These two tests comprised a restriction for the time to answer a stimulus, whereas in TAP Visual Scanning the participant was instructed to answer as fast as possible but nevertheless stimulus set changed with the participants' button press. More time could be taken to scan the stimulus set and thus less processing resources may have been needed.
The leftward bias in control subjects observed in two of the three neglect tests may reflect a phenomenon termed pseudoneglect, originally investigated in line bisection tasks and inducing a leftward bias in neurologically healthy subjects. This effect seems to be substantially dependent on the scanning strategy (left-to-right or right-to-left) [33
], and could thereby be another explanation for the leftward bias of depressive patients in the visual scanning task, where a left-to-right scanning strategy is part of the task. The cognitive control of attention distribution is also an important factor in the rehabilitation of visuospatial neglect in stroke patients [34
]. Accordingly, the use of a left-to-right scanning strategy in the subtest TAP Visual Scanning may explain the absence of a comparable trend to a rightward bias seen in the other two tests in depressive patients.
The rather small effect of sad mood on the attention bias may be explained by the restricted validity of the sad mood item of the BDI. When considering the complete BDI score - a well established measure of depressiveness - a stronger effect concerning the relation of attention and depressiveness may have occurred. In our hypothesis, however, we emphasized the influence of sad mood on alertness and spatial attention deficits. Depressiveness includes a wide range of symptoms, including feelings of guilt, reduced drive and suicidal thoughts, thus not only negative emotion as relevant for our hypothesis. A questionnaire measuring negative emotions such as the Positive and Negative Affect Schedule (PANAS) [36
] may be a more useful tool in future studies to investigate the influence of negative emotion on cognitive functioning in depressive disorder.