A range of models of motion processing has been proposed (e.g. motion energy, gradient, Bayesian, ratio) but there is currently no consensus as to how the human visual system computes the speed of a moving image and there is insufficient data to adequately characterise the effects of even a few image parameters such as contrast, luminance and eccentricity upon perceived speed. A series of experiments was conducted in order to inform models of speed encoding and provide insight into the mechanisms underlying spatio-temporal processing in the visual system. Measurements of the ramp after-effect suggest that the after-effect is determined by ramp amplitude rather than gradient and thus offer little support for the existence of gradient motion detectors. However, the findings of luminance-dependent ramp after-effects do provide support for the idea that the after-effect is mediated by ON- and OFF- pathways in the visual system. Measurements of biases in speed perception indicated that speed encoding in the periphery is essentially similar to central vision whilst the results of a further study of speed biases indicated that at lower luminance there is significantly less reduction in perceived speed and greater increase in perceived speed (at low and high speeds respectively) and a concomitant reduction in the speed at which the bias is reversed. This luminance-dependent pattern of results is consistent with ratio-type models of speed encoding but inconsistent with all extant Bayesian models. Overall the results of these experiments offer little support for gradient or Bayesian models. Whilst the results do not rule out other speed-encoding models, only ratio-class models can currently account for the entire pattern of results reported in this series of experiments.