Abstract: Research into higher cognitive abilities of the horse may be limited by developing the adequate experimental design. In this study four pony mares between 8 and 19 years old were included. Three of them reached the criterion to be tested in a new design of matching to sample using a black circle and a cross as visual cues attached to an apparatus. The attention was directed to the question of whether the animals are able to concept formation in a given time period or if their decisions depend on other cues or strategies. After familiarization to the testing area and the test procedure, the animals were given 27 sessions of 20 trials each during 14 weeks. While there was no preference for one of the stimuli used, horses showed a significant left sidedness. None of the mares reached the learning criterion of 80% correct answers in one session. However, the ponies showed procedural learning based on correction runs that were given between incorrect decisions, by then selecting the correct stimulus on the other side of the apparatus. This learning type arose in three individuals in session four, six and eleven, respectively. It is concluded that discrimination tasks may be biased by the involvement of unexpected learning strategies, which complicates the interpretation of such tests and may even mask possible conceptualization capabilities.