BACK TO QUEBEC CITY 2002 TITLES
Assessing the cognitive
ability of domestic sows from mirror mediated behavior in a modified T-maze
R.M. Brooks1, W.R. Stricklin1
and H.W. Gonyou2
University of Maryland,
College Park, USA; 2Prairie Swine Centre, Saskatoon, Canada
Determination that a sow can mentally conceptualize herself as being outside a confinement stall would have moral implications. However, measuring this ability directly is difficult, if not impossible. Therefore, we attempted to measure the ability of a sow to mentally conceptualize an object being located in a position other than where it is seen by using mirrors. Sows were tested in a modified T-maze for their ability to approach a target visible only in reflection. The modification to the T was “wilted” arms such that a mirror could be placed to reflect an image from the opposite arm of the maze. From an initial 28 sows, 12 were selected for ease of handling and apparent interest in learning to respond to clicker training. Over an 8-week period in five phases, the 12 sows were tested for their ability to approach one of two designated targets to receive a food reward. Throughout the study, sows that did not meet a pre-selected performance standard, displayed a side bias in the maze, or were difficult to handle were eliminated. Phases 1 and 2 involved initial training to target in their home stalls and later inside the maze. Phase 3 involved approaching a single target directly visible within the maze. Phase 4 introduced the mirror, with the target visible in both reflected and direct view. In phase 5, the actual target was removed from the sow’s sight and was only visible in reflection. Sows were rewarded only after approach to the seemingly empty arm (but holding target), ignoring the visually present stimuli of the target in the mirror. Over a 4-week period, 671 trials were recorded (phases 2, 3, 4 and 5) with 38.7 and 61.3% incorrect and correct responses, respectively. Of the original 12 sows, six entered the final two phases involving the mirror. Sows ranged between 50 and 67% correct choices, above levels expected at random in their ability to locate a target using a mirror. While one can conclude that the sow’s behavior differed from chance alone in their ability to locate an object seen only in reflection, one cannot eliminate possibilities such as the sows using spatial learning cues.