Abstract: Complex housing environments such as group housing with automatic feeding or the close contact to humans in sports make high demands on the learning ability of the horse. These learning processes include not only habituation, sensitization and simpler forms of operant conditioning, but also stimulus generalization and possibly some type of concept learning. Studies concerning cognitive abilities in the horse increased in the last decades, but for optimizing housing conditions and horse training, deeper insight into the learning behaviour of this species is necessary. In the present study we used the advantages of a computer based learning device to train 7 Shetland ponies to solve a matching to sample task. With this more complex type of a discrimination task, animals are trained to recognize two out of three presented stimuli as identical. In a first step animals learned to operate the learning device and in further learning steps to recognize and assign geometric symbols (dot, cross, square, and triangle) presented on a LCD screen that were ‘equal’. Four of the 7 ponies could solve the given task by performing over 80% correct responses in two consecutive sessions (p < 0.001). In the subsequent transfer test with new symbols, we found that the ponies were able to transfer the learned rule. In further experiments it should be clarified whether the good learning performance of the ponies in the present study is indeed based on their capability to form an abstract concept of sameness. The present results indicate that ponies possibly posses higher cognitive abilities than so far known.

Abstract: After describing horse-riding as a facility in managing mentally ill patients, a program for chronic schizophrenic in-patients is presented. Clinical experience with this program and also results of a controlled study are reported. The therapeutic value and slope for horse-riding are discussed in relation to different diagnoses.

Abstract: Three horses (Equus caballus) with a history of performing cognitive tasks including discrimination learning, categorization, and concept use were tested to evaluate their long-term memory (LTM) in three experiments. In addition, use of LCD multi-displays for stimulus presentation was incorporated into cognition testing protocol for the first time with horses. Experiment 1 tested LTM for discrimination learning that originally occurred 6 years earlier. Five sets of stimuli were used and the two horses tested showed no decrement in performance on four of the sets; however, both horses did score below chance on one set. Experiment 2 examined long-term categorization recall 10 years after horses had demonstrated the ability to make stimulus selections based on shared characteristics within a given category. The horse tested for LTM after the decade-long interval immediately and consistently applied the previously learned categorization rule to not only familiar but also novel sets of stimuli. Experiment 3 tested another horse for LTM for a relative size concept. This horse had originally demonstrated concept rule use in order to select stimuli based on their relative size to one another. More than 7 years later and without further training, this horse reliably applied the previously established size concept to both familiar and novel sets of stimuli. These findings are the first reports of long-term categorical and conceptual memory in horses and are consistent with observations of domestic and wild horses, which indicate that behavioral and ecological events may be remembered for long periods of time. These studies also demonstrate the adaptive nature of horses with regard to their ability to generalize over several different testing conditions.

Abstract: Abstract&nbsp;&nbsp;We tested whether dogs have a cross-modal representation of human individuals. We presented domestic dogs with a photo of either the owner's or a stranger's face on the LCD monitor after playing back a voice of one of those persons. A voice and a face matched in half of the trials (Congruent condition) and mismatched in the other half (Incongruent condition). If our subjects activate visual images of the voice, their expectation would be contradicted in Incongruent condition. It would result in the subjects` longer looking times in Incongruent condition than in Congruent condition. Our subject dogs looked longer at the visual stimulus in Incongruent condition than in Congruent condition. This suggests that dogs actively generate their internal representation of the owner's face when they hear the owner calling them. This is the first demonstration that nonhuman animals do not merely associate auditory and visual stimuli but also actively generate a visual image from auditory information. Furthermore, our subject also looked at the visual stimulus longer in Incongruent condition in which the owner's face followed an unfamiliar person's voice than in Congruent condition in which the owner's face followed the owner's voice. Generating a particular visual image in response to an unfamiliar voice should be difficult, and any expected images from the voice ought to be more obscure or less well defined than that of the owners. However, our subjects looked longer at the owner's face in Incongruent condition than in Congruent condition. This may indicate that dogs may have predicted that it should not be the owner when they heard the unfamiliar person's voice.