We  present a model of opinion formation of individuals with respect to a set of products. We show that from the analysis of individual's  opinion we can reconstruct the knowledge networks of preferences and anticipate the judgement about an unseen product.
In geometrical words, one represents agent's preferences as a vector in an hypothetical taste space, whose dimension and base vectors are unknown. A product is represented by a similar vector (in dual space). Agent's opinion on a given product is given by an operation analogous to the scalar product between preferences and properties.
Therefore, products acts like a basis, and opinions as agent's coordinates on such a basis. However, differently from usual geometrical problems, we do not know what the basis is, if it is complete, etc. We assume that all these ``internal'' degrees of fredom are hidden, and that the only observable if Agent's opinion on a set of products.
The main result of our analysis, is the possibility of extracting informations about the hidden degrees of freedom, and in particular the dimensionality of hidden space, from the opinion database. In this way customer's surveys can be used as tools for  studing brain functionality.
This study is also important for establishing the importance of personal informations and the need for their protection.