Human oblique saccades: quantitative analysis of the relation between horizontal and vertical components

Are the horizontal and vertical components of oblique saccades produced by two separate pulse generators or by a single, vectorial pulse generator? To investigate this question, purely horizontal and vertical ("cardinal") saccades as well as oblique saccades with a meridional direction of +/- 45 deg (horizontal and vertical components of equal size) were recorded in 10 human subjects using a magnetic search coil. The components of oblique saccades were slower than cardinal saccades of comparable size, yet the oblique vector velocity was slightly larger than the velocity along cardinal directions. The onset of the two components was always synchronized, but their times to peak velocity and their durations, although approximately equal on average, frequently were different in individual trials; the component velocities were weakly correlated only. Correspondingly, the trajectory of oblique saccades exhibited various types of curvature which often changed from trial to trial. There was no correlation between curvature and aiming accuracy. These results are discussed in terms of various models of saccade generation. It is suggested that each of the two components is generated by its own local feedback pulse generator; the two components would be coordinated by crosscoupling the two local feedback circuits at the level of their error signals. By contrast, the extraretinal feedback that prepares corrective saccades is apparently not evaluated componentwise but may use a vector representation, since the latency of oblique corrective saccades was a function of the vectorial error magnitude.