A total of 320 trials were performed with the two study dolphins (Table 1), with a minimum of 30 trials per material and animal.

Summary of data collection. Left number of trials run in the experiment, and right number of trials used for the analysis. All columns are the number of Corrects/Incorrects.

Task performance and behaviour

Graph showing the tracking of the swimming path of each individual from head orientation measuring blowhole to snout every 3 ms during the same trial. Above, M1 swimming paths to left and right target showing a curved and side approach. Below, F1 swimming paths to left and right showing an approach to the targets from the front. Green circle shows the correct target (S+) and red circle is the incorrect one.

The male (M1) approached the targets with a curved swim path in all trials, keeping left away from the pool wall. The female (F1) used a frontal approach to the targets in 67.7% of the trials, and a slightly curved route in 33% of the trials.

Performance and target strength of the different materials

The dolphins made minor mistakes when discriminating the plastic or brass spheres from the aluminum sphere but the performance decreased when discriminating steel from aluminum sphere.

The performance of the two dolphins does not show any change over time, as would have been signs of learning.

Scanning behaviour

M1 scanned each target target before deciding on an average of 4.4 times for plastic, 3.3 times for brass and 4.8 times for steel. F1 scanned 3.6 times for plastic, 3.9 times for brass and 5.9 times for steel, increasing proportionally with the difficulty. In both dolphins the highest number of scans corresponds to the most difficult task.