Saccadic eye movement was measured in three studies in a tracking task in which increasing stimulus jumping frequency indicates increasing difficulty. In the first study, after taking a benzodiazepine 16 subjects showed decreased saccadic velocity (activation indicator), but their reaction times and anticipatory behaviour (performance indicators) did not parallel this decline, although the saccadic amplitudes did. In the second deactivation study 23 subjects worked on somewhat boring oculomotor tasks for 2 h and showed the same decrease in saccadic velocity but showed almost no performance decline. In the third study a large sample of 254 subjects performed the same short tracking task, which offered the detailed analysis of the correspondence of the saccadic activation indicator to saccadic performance indicators. Almost all correlations were zero correlations. Thus, while performance seems to be practically independent of changing activation, the latter is dependent upon performance from one moment to the next. Stable differences were found between anticipatory and reactive saccades, which indicate an immediate change of saccadic velocity when the response mode changes. Therefore a reformulation of the relationship between performance and activation is proposed. Performance is stabilized by 'concentration' in a two-stage regulation process, i.e., it resists changing activation (for example through fatigue). In the second stage, the 'effort loop', additional brainstem activation is demanded when concentration alone fails to reach the goals that were set by the subject for his performance.