Configuration of the training environment at each phase. Locations of infrared sensor are shown as teal squares at the nose poke and the reward areas. Feeder locations are indicated by filled circles (empty circles represent possible feeder locations). (A) In phase 1, a small, enclosed space is created ensuring the cue and reward are administered in close proximity. (B) A gap between the arms of the maze and the start box platform is introduced in phase 2. (C) The reward area arms increase in length at phase 3. (D) At phase 5 the central arm is increased in length by 20 cm, the length of the reward area arms is increased to 20 cm. The feeder and sensor locations shift to the end of the reward area arms. (E) At phase 6, the length of the central arm is increased to 50 cm and at phase 7 to 80 cm (F). Also at this phase, the arms reach their maximal length of 40 cm and the distance between the central arm and the reward areas is increased further to 35 cm. Please note that although side arms were present in phases 5 to 7 these are not shown in this figure due to space constraints. Side arms were not utilized during trials (trials are only considered valid when the animal uses the central arm to reach the reward areas) and animals were trained to return to the start box via the central arm between trials. Animals spent only a small fraction of their total time in the maze during a block in the side arms (25th percentile: 1.8%, 50th percentile: 4.7%, 75th percentile: 7.9%), which predominantly occurs at the beginning of a block while the animal briefly explores the maze prior to activating the first trial.