Abstract: The purpose of the invention is to determine the topography of retinal reactions with high effectiveness and great certainty. In accordance with the invention, the problem is solved by generating a light-emitting image consisting of surface segments on a surface placed before the eye, and using complete extended m sequences, to brighten and darken the surface segments in each time interval. From the eye reaction signal, with the aid of the cross-correlation function, the reaction functions of the segments are calculated in each time interval, and the quality of the results is evaluated by forming some functions for each time interval. Only reaction signals evaluated as good are added up for the total result. The reaction signal of the eye is monitored for staying within limit values, and when errors are recognized, the m sequences are set back by a number of steps which can be preset, recognized errors in the reaction signal are replaced by the repetition.