In considering the historical change of a word's meaning, cognitive semantics places the notions of conceptual metaphor, metonymy and synecdoche as essential pillars to its study. However, the main question that arises is why language selects such cognitive abilities but not others. The present paper addresses such a complex query by claiming that the brain hinges on the computational principles of sparseness (or the representational scheme in which only a few units, out of a large population, are used at any time) and entropy (i.e., a measure of the predictability or surprise value of response, see Shannon, 1948) used in order to produce efficient mental representations. This author suggests that such principles take part in neocortical functioning and consequently the analysis of semantic change involves a detailed examination of early visual processing. (Alba Luzondo, Ruiz de Mendoza Group)