Recently, Vigo and Allen (2009) proposed a view of transitive inference as categorization that depends only on discriminative performances, similarity judgments, and the formation of categories of which non-human animals and non-verbal humans are capable. Rather than involving language based statements of premises, a network of simple discriminations provides the prerequisites for the emergence of new discriminations involving the transitive relations among stimuli. This view is illustrated briefly here and considered in the broader contexts of two experimental situations of interest to behavior analysts: repeated discrimination reversal training and symmetry in pigeons.