Yes, if an anonymous visitor has visited your page before, they will be given recommendations based on their cohort and their previous behaviors. If they have not, they will be shown the most recommended pages across all cohorts.
Collaborative filtering is a technique for predicting the interests of an audience in order to make a recommendation. Generally, the audience is placed into cohorts based on previous similar behaviors, in this case, page visits.
Then a new recommendation is made to members in those groups for a page that has not yet been visited.
For example: If Audience Member A visited Page 1 and Audience Member B visited Page 1 and Page 2, we might recommend Page 2 to Audience Member A. This technique is frequently used across many popular personalization algorithms.