Introduction: In the antisaccade task a subject has to perform a voluntary saccade in darkness towards an eccentric spatial location that is located in the hemifield opposite to a peripheral visual target. Its cortical control has generally been attributed to the frontal lobe. Antisaccade performance requires at least 3 steps, each of them with a specific cognitive impact:

1. the suppression of unwanted reflexive saccades in response to target onset,
2. a shift of visuo-spatial attention from target location across the midline towards the internally-represented saccade goal, and
3. the initiation of a non-visually-guided intentional saccade towards the latter.

These subfunctions might rather be controlled by different parts of the fronto-parietal cortical network of areas involved in saccadic eye movements.

Methods: To address this problem we investigated horizontal antisaccades in 34 patients with focal unilateral lesions of the posterior parietal cortex (PPC), the frontal eye fields (FEF), the supplementary motor area (SMA), or the dorsolateral prefrontal cortex (PFC), and in a control group of 32 age-related healthy adults. Horizontal saccadic eye movements were recorded in darkness using infrared reflection oculography, with the head fixed on a head holder. Subjects fixated a central laser point. As soon as it moved unpredictably to 14 deg of horizontal eccentricity in the left or right hemifield, they had to perform a saccade into the opposite direction.

Results: The patients' deficits consisted of increased error rates of unwanted reflexive saccades toward the visual target, and of prolonged latencies of antisaccades, whereas antisaccade amplitudes showed a high variability in normal subjects and patients and did not differ significantly between the groups. Increased error rates were most pronounced in patients with FEF lesions (p < 0.001), without directional specificity, but were significant also in patients with PPC lesions (p < 0.01) and with PFC or SMA lesions (p < 0.05 or p < 0.1), though to a lesser extent. In contrast, prolonged latencies of antisaccades were most pronounced in patients with PPC lesions, more into the contralateral (p < 0.005) than into the ipsilateral hemifield (p < 0.05). In patients with frontal lobe lesions, only right FEF lesions were associated with prolonged latencies of contraversive antisaccades (p < 0.05), but neither PFC nor SMA lesions. All these deficits were not related to the severity of visual hemi-neglect, indicating that they do not primarily reflect these patients' impairment in shifting attention across the midline, from stimulus location towards the saccade goal.

Conclusion: The cortical control of antisaccades is not restricted to areas of the frontal lobes. Instead, our results demonstrate a specific role of the PPC and - to a lesser extent - of the FEF for the initiation of antisaccades, with directional preponderance for contraversive saccades. In contrast, the whole fronto-parietal network of cortical areas is involved in voluntary suppression of inappropriate reflexive saccades, with predominance of the FEF.