It is commonly accepted that tumors arise from cells that escape the homeostatic controls which underlie the healthy histological structure and that cell phenotype is not the result of deterministic biochemical and genetic processes, but rather the stochastic and dynamic outcome of multiple intra- and intercellular regulation networks. This PhD aims to quantitatively study the phenotypic homeostasis of the cell populations and to present an approach to the fundamental question, never heretofore studied, regarding the autonomy versus collective control of cell fate. We studied in the long run, using flow cytometry and in 2D and 3D conditions, the level of expression of CD24 and CD44 of two breast cancer cell lines (SUM149-PT and SUM159-PT). Three phenotypes were isolated (CD24-/CD44+, CD24+/CD44+, CD24-/CD44-), the latter had not previously been documented in the literature. The phenotypic behavior of CD44-low and CD44-high subpopulations has been characterized by assessing their proportion and analyzing the fluorescence map. Thereby, we observed both a periodic behavior of appearance and disappearance of pool of cells characteristics of each cell lines and a phenotypic re-diversification for each subpopulation. Only the resulting population derived from CD24-/CD44- provided the same balance as the original unsorted population. 3D re-diversification process was observed in tumorspheres from CD24-/CD44+ and CD24+/CD44+. The cells CD24-/CD44did not have that potential but nonetheless outlived anoikis. These behaviors suggest that there is an inter-cell coordination regulating the balance of phenotypic proportions. To discover the social rules regulating inter-phenotypic spatial organization, we have set up a reporter of the endogenous variations of CD24 and CD44 and developed a theoretical model of cell interactions. This work has confirmed our hypothesis that inter-cellular social rules are determining the phenotypic expression at both the uni- and multicellular scales.