Dogs' superior olfactory capabilities and trainability have led to widespread use of dogs for scent detection research and work. Previous research has largely focused on dogs' ability to detect and discriminate single odours, however more research is needed to understand dogs’ ability to use multiple sensory modalities to identify and discriminate odours. Continuing research started by Campbell- Smith (2024), the current study hypothesized, a) dogs are capable of using cross-modal association to match odours to images of shapes and b) the inclusion of an error correction procedure and physical screen border on the touch screen where the images are presented will increase dogs responding accuracy and decrease biases towards specific shapes and screen quadrants, previously observed by Campbell-Smith. Four dogs were trained using shaping and chaining to engage in a conditional discrimination task using an automated scent delivery carousel and touch screen. Challenges faced early on during the initial conditional discrimination sessions prompted the inclusion of a simple visual discrimination task and an errorless learning procedure. The findings were largely consistent with Campbell-Smith, in that all dogs had above chance (50%) accuracy rates overall, but none met the mastery criteria (80%) overall. The correction procedure and physical touch screen border did reduce overall biases in all dogs, but did not increase responding accuracy enough to meet the mastery criteria. This suggests that the procedures used in these experiments may not be optimal for training dogs to accurately match odours and visual stimuli. Alternative procedures should be investigated in the future. The current findings provide more information as to dog's cross-modal identification capabilities, as well as further understanding of dogs’ visual capabilities and a comparison of dogs’ accuracy on conditional discrimination with and without an errorless learning procedure.