Figure 2
Comparison of scores calculated manually during training, scores based on the number of rewarded pellets, and scores as recorded by the sensors. A rolling average over 3 days was used to calculate the percent of correct scores over the training days (training phase shown in gray). The scores for each animal are plotted separately and also as an average over all animals (thick black line). Scores based purely on sensor readings are unreliable for the first four phases as the tail of the rat could activate the sensors unintentionally. Additionally on the first two days of the first phase reward automatically followed the cue in many trials. These hint trials were excluded from the scoring process however they do occur in the sensor data, resulting in unusually high sensor scores in the first phase. (A) Results from manual scoring. (B) Scoring based on amount of rewards given to the animal. In phase 1, the criteria for a correct trial was set at three or more rewarded pellets, in phase 2 to 5, two or more pellets, and in phase 6 and 7 one or more pellets. (C) Scores based on sensor data from sensors at the reward locations. Activation of a sensor at the reward area corresponding to the cue resulted in a correct trial.
