Fig 1.
The square apparatus used in Exp. 1 and 3 (a) and the longitudinal apparatus used in Exp. 2 and 4 (b). For each apparatus, close up pictures of the tunnels are shown in (c) and (d).
Fig 2.
The training stimuli used in Exp. 1–2 (a) and Exp. 3–4 (b), as well as the test stimuli used for the amodal completion test in Exp. 3 and 4 (c). In this last pair of stimuli, the image on the left induces (in humans and other species) the perception of a full disk, amodally completing behind the dark green occluder, while the image on the right is perceived as an amputated disk placed close to the dark green occluder.
Fig 3.
Overview of the experimental phases, apparatuses and stimuli used in the four experiments.
Exp. 1 and 2 consisted only of a discrimination training, during which fish were trained to discriminate a disk and a cross. In Exp. 3 and 4, during the discrimination training, fish learned to discriminate between a full disk and an amputated disk. In these two experiments, after successfully reaching the learning criterion, the fish also participated to an amodal completion test. In Exp. 1 and 3 a small square apparatus was used, while in Exp. 2 and 4 an I-maze was used.
Table 1.
Final performance (expressed as proportion of first and total choices for the disk) of the two successful fish achieving the learning criterion in Exp. 1.
For each animal the number of trials and sessions needed to reach the criterion (TTC—Trials To Criterion) is reported, as well as the performance level in the last two days of training, for both dependent variables.
Table 2.
Final performance (as proportion of choices for the disk) of the two fish achieving or approaching the learning criterion in Exp. 2.
For each animal the number of trials and sessions needed to reach the criterion (TTC—Trials To Criterion) is reported, as well as the performance level in the last two days of training, for both dependent variables.
Fig 4.
The curve for the dependent variable first choice preference index is plotted on the left (a), while the curve for the total choices preference index is plotted on the right (b). On the y-axis, the level of preference for the full disk (higher values correspond to higher proportion of choices for the full disk). Darker lines represent the performance of the animals that at training were reinforced on the full disk, while the lighter lines represent the performance of animals reinforced on the amputated disk. Mean ± SEM are shown, for each training session. The learning curves of individual fish were anchored to their last training session, before being averaged. This was done to account for the different number of sessions each indfividual needed to reach the learning criterion. Each point represents the performance of a single training session (composed of 10 trials). Subsequent training sessions are ordered from left to right, in chronological order, from the beginning of the training until the achievement of the learning criterion. Data points for which no error bar is displayed represent sessions including one subject (since only one subject needed more than 13 training sessions to complete its training). The number of subjects participating to each training session is reported below the graph, close to the X-axis (the numbers of subjects reinforced on the full- and amputated-disk are reported, respectively, above and below the axis).
Fig 5.
On the y-axis, the level of preference for the full disk displayed in the last two days of training in Exp. 3 (higher values correspond to higher proportion of choices for the full disk).
The results for the two dependent variables, the first choice preference index and the total choices preference index are shown on the left and on the right, respectively. Darker and lighter columns are used to represent the performance of the animals reinforced on the full and on the amputated disk. Mean ± SEM is shown. Asterisks indicate significant departures from chance level. On the left side of the figure, a schematic view of the apparatus is depicted.
Fig 6.
Level of preference for the amodally completing disk displayed during the amodal completion test of Exp. 3 (higher values correspond to higher proportion of choices for the amodally completing disk).
The results for the two dependent variables, the first choice preference index and the total choices preference index are shown on the left and on the right, respectively. Darker and lighter columns are used to represent the performance of the animals reinforced on the full and on the amputated disk. Mean ± SEM is shown. Asterisks indicate significant departures from chance level. On the left, a schematic view of the apparatus is depicted.
Fig 7.
The curves for the first choice preference index and for the total choices preference index are plotted on the left (a) and on the right (b) respectively. On the y-axis, the level of preference for the full disk. Darker and lighter lines represent the performance of the animals reinforced on the full or amputated disk, respectively. Mean ± SEM are shown, for each training session. The learning curves of individual fish were anchored to their last training session, before being averaged. Subsequent training sessions are ordered, chronologically, from left to right. Data points for which no error bar is displayed represent sessions with only one subject (which needed a higher number of training sessions to reach the criterion). The number of subjects participating to each training session is reported below the graph, close to the X-axis (the numbers of subjects reinforced on the full- and amputated-disk are reported, respectively, above and below the axis).
Fig 8.
Results of the last two days of training in Exp. 4.
The first choice preference index is shown on the left (a), while the total choices preference index is shown on the right (b). Darker and lighter columns represent the performance of the animals reinforced on the full and on the amputated disk, respectively. Mean ± SEM is shown. Asterisks indicate significant departures from chance level. On the left, a schematic view of the apparatus.
Fig 9.
Results of the amodal completion test of Exp. 4.
The first choice preference index and the total choices disk preference are shown on the left and on the right, respectively. Darker and lighter columns are used to represent the performance of the animals reinforced on the full and on the amputated disk. Mean ± SEM is shown. Asterisks indicate significant departures from chance level. On the left, a schematic view of the apparatus is depicted.
Fig 10.
Between-experiment comparisons of the level of performance (i.e., proportion of choices for the reinforced stimulus) on the last two days of training.
Graph (a) reports data for the first choice correct responses index, while graph (b) reports data for the total choices correct responses index. Darker columns represent the values for animals trained in the square apparatus (Exp. 1 and 3), while lighter columns indicate the values for animals trained in the long apparatus (Exp. 2 and 4). Within each graph, the two columns on the left represent the values for Exp. 1 and 2 (cross vs. disk discrimination), while the two columns on the right represent Exp. 3 and 4 (full vs. amputated disk discrimination). Mean ± SEM is shown. Asterisks indicate significant departures from chance level.
Fig 11.
Comparison of the performance observed in the amodal completion test in Exp. 3 (training in the square apparatus, darker columns) and in Exp. 4 (training in the long apparatus, lighter columns).
The results for the first choice correct responses index and for the total choices correct responses index are shown on the left and right side of the graph, respectively. Mean ± SEM is shown. Asterisks indicate significant departures from chance level.