Eye position signals in human saccadic processing

1. We studied saccades to briefly flashed targets in 8 human subjects. The target flash occurred (i) during smooth pursuit ("ramp-flash"), (ii) just before a saccade to another target ("step-flash"), or (iii) during steady fixation ("flash only"). All lights were extinguished after the target flash so that smooth pursuit or saccadic eye movements occurred during the interval of complete darkness between the target flash and the saccade to it. We compared these saccades to those made without intervening eye movement (flash only), and quantified the extent to which the saccadic system compensated for the change in eye position that occurred during the dark interval. 2. Saccades to control flashes were reasonably accurate (mean gain 0.87) and consistent. Compensation for the intervening eye movement in the ramp-flash and step-flash paradigms was highly variable from trial to trial. On average, subjects compensated for 27% of the intervening pursuit eye movement on ramp-flash trials and for 58% of intervening saccadic movement on step-flash trials. 3. Multiple regression analysis showed that the variability did not depend on factors such as variations in underlying saccadic gain, response latency, timing of stimuli or size of the required response. We conclude that this variability is intrinsic to saccadic responses that require the use of an eye position signal. 4. These results show that an eye position signal is available to the saccadic system but that this signal has low fidelity.