How do animals actually solve the T maze?

Rats were trained on a reinforced, delayed alternation T-maze task in the presence (cue group) or absence (no-cue group) of salient extramaze landmarks. A surprising finding was that the acquisition and memory performance of the 2 groups did not differ. Manipulations of the extramaze landmarks for the cue group suggested that, although landmarks were used to guide behavior, other sources of information were also used normally. The no-cue group was able to perform the task at above-chance levels even when extramaze, intramaze, and inertial sources of orientation were manipulated. These results suggest that memory performance on the T maze does not rely exclusively on the processing of allocentric spatial relationships in the maze environment.