To help understand the order of events that occurs when generating saccades, we simulated and tested two commonly stated decoding models that are believed to occur in the oculomotor system: vector averaging (VA) and center-of-mass. To generate accurate saccades, each model incorporates two required criteria: 1) a decoding mechanism that deciphers a population response of the superior colliculus (SC) and 2) an exponential transformation that converts the saccade vector into visual coordinates. The order of these two criteria is used differently within each model, yet the significance of the sequence has not been quantified. To distinguish between each decoding sequence and hence, to determine the order of events necessary to generate accurate saccades, we simulated the two models. Distinguishable predictions were obtained when two simultaneous motor commands are processed by each model. Experimental tests of the models were performed by observing the distribution of endpoints of saccades evoked by weighted, simultaneous microstimulation of two SC sites. The data were consistent with the predictions of the VA model, in which exponential transformation precedes the decoding computation.