The test arena was opened on the left side and a test bucket – consisting of two shallow grey buckets stacked on top of each other, with a thin slice of carrot placed in the lower one to avoid any olfactory cues – was prepared.

Each horse was pseudorandomly assigned either the left or right side of the arena as the rewarded location, with half of the horses being assigned the left side and half the horses the right side. The rewarded and unrewarded location remained the same for each horse throughout the tests, on both test occasions. The test bucket was placed in the rewarded location and a few slices of carrot were placed in the bucket.

Each test started with a warm-up phase of eight trials. During the first five trials the horse was led into the arena by the owner, passing through the two pairs of orange cones at the entrance. The owner was instructed to stay on the horse’s left side. The owner led the horse up to the test bucket and allowed it to eat the carrots before exiting the arena again. This was repeated a total of five times.

For the final three warm-up trials the owner led the horse by the halter and stopped at the front pair of cones allowing the horse to cross the start line and approach the test bucket alone. If a horse was hesitant to leave the owner’s side the owner was instructed to take another few steps into the arena with the horse before stopping and encouraging the horse to approach the bucket on its own. The distance was then gradually increased over the consecutive two trials until the horse would approach the bucket from the start line. All horses managed to complete the warm-up phase successfully.

For the training phase, the experimenter positioned herself at the entrance to the arena on the left side and the owners were asked to approach the arena as in the last three trials of the warm-up phase, stopping at the front pair of cones and allowing the horse to enter the arena alone. The experimenter measured the time from the moment one of the horse’s front legs crossed the start line until the muzzle was lowered into the bucket.

The position of the bucket was alternated between the rewarded and unrewarded corner of the arena according to a predetermined pseudorandom schedule. It was never placed in the same location more than two times in a row. The approach latencies were recorded by the experimenter and learning criterion was reached when the horse approached the bucket in the rewarded location at least one second faster than the bucket in the unrewarded location, for six consecutive trials.

For the test, the bucket was placed in the ambiguous location, between the rewarded and unrewarded location [Figure 1]. The horse’s latency to approach the bucket in the ambigous location was recorded.