Cognitive probes are increasingly being used as an inferred measure of the emotional (and thus welfare) status of the animal. This reflects the bidirectional and interactive nature of emotional and cognitive systems. To date, cognitive paradigms have focused on how the emotional system biases expected outcome of prospective actions within goal-orientated scenarios. Evidence, however, suggests that negative affective state can also modulate attentional mechanisms. Measuring attention alongside other current tests of cognitive bias may provide greater resolution in the measurement of animal welfare. As a starting point for developing cognitive tasks of attentional control, we decided to assess the basic relationship between visual attention and cognitive performance in a farm animal species (sheep). Variation in visual attention and cognitive performance was sought through testing of four different breeds of upland and lowland sheep (Beulah, Bluefaced Leicester, Texel and Suffolk; n = 15/breed) on a visual attention task and a two-choice visual discrimination task (to measure cognitive performance). Cognitive performance and visual attention differed significantly between breeds (F 3,46 = 4.70, p = 0.006 and F3,5o = 6.05, p < 0.001 respectively). The least visually attentive breed of sheep (Blue face Leicester) had the lowest level of cognitive performance and the most visually attentive breed (Suffolk) had the highest level of cognitive performance. A weak but significant relationship between vigilance/fearfulness and visual attention was also observed (t44 = 3.91, p = < 0.001; r2 = 0.23) that appeared to adhere to the Yerkes-Dodson law, with both high and low levels of vigilance/fearfulness having a negative effect on visual attention. These results demonstrate a discernible relationship between visual attention and cognitive performance that provides a basis for further exploring attention systems in the context of changes in animal affective state and thus animal welfare.