This thesis formulates an optimal estimation approach to the common biophysics problem of visual modeling. More specifically, it deals with the “encoding problem” of biological vision, where, given a stimulus signal, we predict the spike response patterns of the retinal ganglion cells (RGCs) that transmit the signal from the retina to the brain. We optimized a modified version of the generalized linear model , which we labeled mGLM. Analysis confirmed the model’s versatility in capturing a range of cell firing rates and its ability to predict the rate on the 60 Hz scale. The calibrated mGLM parameters were described in the framework of a Kalman filter  that took into account both the pseudorandom input and the measurement noise of the RGC system. Through this formulation, we were able to realize the impact of the measurement noise on the experimental data, an aspect that we believe has not been previously considered in the neural encoding problem.