Biological vision processes are at the basis of many studies in the image-processing field. In this context, pre-attentive neural networks developed by S. Grossberg constitute an interesting approach. Pre-attentive networks model the process in biological vision known as emergent perception. They are able to extract meaningful information from the global structure of data rather than from local relationships, yielding to a coherent and complete visual perception, also in case of noisy and incomplete images. The paper evaluates the application of Grossberg's approach to the analysis of scatter plots from nuclear physics experiments. The design and implementation of a pre-attentive neural system developed for this purpose are presented. Simulation results prove the goodness of the approach.