Results

Controls bisected lines very accurately (table). As expected, the hemineglect group deviated bisection rightwards, although a few patients showed normal or even slight leftward bias, confirming previous observations that bisection does not correlate perfectly with other hemineglect tests (figure).2 ANOVA showed a significant difference between non-neglect patients and controls (F(3,85)=7.85, p<0.0001). Pairwise comparisons showed that both groups of patients with hemifield defects were different from controls (p<0.004 for both hemianopic groups), with both deviating contralaterally. Five of the 16 patients with hemifield defects fell beyond the normal range contralaterally, none ipsilaterally. As a whole, this contralateral deviation was much less than the ipsilateral bias of hemineglect.

By contrast, there was no difference between controls and patients with normal visual fields. No patient with normal visual fields fell outside the normal range.

Contralateral bisection deviation tended to be larger in patients with larger visual field defects (figure). In particular, the six patients with complete macular splitting hemianopia had mean contralateral bisection errors of 6.15% (SD 4.07), whereas the six with sparing of both upper and lower macular regions had mean errors of 1.90% (SD 3.47) (p<0.05). However, these sample numbers and differences are too small to permit conclusions about the quantitative relation of the visual defect to bisection bias.

Discussion

Our prior report10 indicated that hemianopic patients had a contralateral bias to line bisection and the central peak of eye fixation in the ocular scanning preceding bisection. This was found with head fixed conditions, large displays subtending 37° and 45° at a distance of 1.14 m, and the use of a pointer, in seven hemianopic patients. However, that study did not examine patients with lesions and normal visual fields, so the bias may have been a non-specific effect of lateralised brain damage.

This study confirms contralateral bisection bias in a larger sample of 16 hemianopic patients. The bias also occurs under the more “natural” conditions used in neglect studies, with head movements permitted and the use of a pen at a writing table, and for lines subtending smaller visual angles. Also, the effect is not found in patients with hemispheric lesions and normal visual fields. Thus it is a consequence of hemianopia rather than cerebral damage. A possible correlation of contralateral bias with the severity of hemianopia and macular involvement would further support the hypothesis that the visual defect is causative. As most patients with hemineglect have hemianopia, this makes their ipsilateral biases all the more abnormal.

A contralateral hemianopic bias conflicts with predictions from Reuter-Lorenz et al.8 In that study normal subjects shown lines in one hemifield estimated midpoint to be more lateral than its true position. As patients with complete left hemianopia can only view the lines in their right hemifield, this would predict a rightward bisection bias, whereas in fact they biased bisection leftward (contralateral to their lesion). This discrepancy may originate from technical differences between the two studies. They used tachistoscopic presentations of under one second, whereas our patients had unlimited viewing time. They required subjects to judge the position of a mark on the line, whereas our patients made the mark. Perhaps most importantly, they used 2 cm lines, whereas our two studies employed lines an order of magnitude longer. Because the size and even direction of bisection bias in hemineglect depends on line length,3 4 17 biases in other subjects may also vary with this index. Finally, results from normal hemifields may not extrapolate to the intact hemifields of hemianopic patients, as there is accumulating evidence of various alterations in the remaining hemifield in hemianopia.18-21

Why might such a contralateral bisection bias arise in hemianopia? Any visual percept represents not only the physical properties of the stimulus but also the influence of numerous internal factors, including prior visual experience, expectation, and attention, among others. One internal factor that could generate an abnormal bias is a change in attentional distribution. During scanning of a letter array, normal subjects distribute fixations evenly across the display, suggesting a flat distribution of attention in this task.9 Hemianopic patients show a gradient of fixations with increased search in the contralateral hemisphere.9 A similar gradient of eye fixations is seen with line bisection, where hemianopic patients tend to search for the end of the line in their blind hemispace.10 22 23 We hypothesise that, because these patients are aware of their deficit, space on their blind side acquires greater salience, leading to an attentional gradient that is strategically adaptive for their disability. Whether the altered gradient is pathological (in neglect) or adaptive (in hemianopia), the result may be a bias in perceptual judgment in the direction of the gradient, causing ipsilateral bisection bias in hemineglect and contralateral bisection bias in hemianopia.

An alternative explanation is that the hemifield representation of space is not veridical. There is little data and the study of Reuter-Lorenz et al 8 did not examine the peripheral regions where our lines lay. Bowers and Heilman24 found that tactile bisection in one hemispace is biased contralaterally. Whereas this accords with our results, it is not known whether similar bisection errors occur visually. Also, our results are not explained by tangent artifact, as the lines were centred in front of the subjects, and hence the tangent points coincided with the midposition of the lines. Further studies are required to determine whether this contralateral bias is a manifestation of non-veridical spatial representation in hemifields or of an adaptive attentional gradient consequent to hemianopia.