Animal welfare can be assessed from the perspective of the biological functioning and health status of animals, in the context of enabling them natural life, and in the perspective based on the affective states of animals. Affective states influence cognitive abilities at the level of judgement, attention and memory, and are also in turn influenced by them. In the present work, we have explored the possibilities for improving poultry welfare as well as improving and developing tests of cognitive bias at three levels. In the first experiment, we investigated the effect of an intermittent light schedule, that was designed to mimic the natural regime with a mother hen. The positive effects of such lighting were faster weight gain, efficient feed conversion ratio, and better behavioural synchronisation during the first days after hatching. In the second experiment, we focused on optimizing a judgement bias test for evaluating the affective states of laying hens kept in different housing systems (enriched cages vs. deep litter pens). After successful training of visual discrimination (pecking to obtain reward) and negative stimuli (not pecking to avoid punishment), the responses of laying hens to ambiguous stimuli with characteristics between two already known, reference stimuli, were tested in a custom-built Skinner box. Optimization of the test focused on the number of types of ambiguous stimuli (Design 1 with one middle stimulus or Design 2 with three ambiguous stimuli, which were a near positive stimulus, middle stimulus, and near negative stimulus) as well as on the reduction of the ratio of reference stimuli to ambiguous stimuli from 2 : 3 to 2 : 1. Despite the adjustments to the test design, we were unable to identify differences in the performance of laying hens from different housing conditions in judgement bias test. However, by adjusting the ratio of stimuli presented, we were able to eliminate a shortcoming of this test, which is a decrease in responses to ambiguous stimuli when they are presented repeatedly, described as a "loss of ambiguity". The judgement bias test was also used to test the effect of tickling as a form of positive tactile stimulation in the third experiment. However, this type of manipulation did not induce detectable changes in the cognitive performance of the laying hens. Similarly, we did not observe differences in the behaviour of laying hens housed in different housing conditions in the fourth experiment, during which the hens were exposed to selected acoustic and visual stimuli in the attention bias test. The absence of any significant differences between hens’ performances in the cognitive bias tests may be because the housing conditions or forms of manipulation used were not perceived by the hens as different, the chosen methodology is not sensitive enough to detect differences induced by the chosen methods of influencing affective states, or the hens are not sensitive enough to the chosen affect-manipulation methods. In the last fifth experiment, we compared spatial learning and memory of laying hens housed in different housing conditions. In a custom-built 8-armed radial maze, hens were trained to visit all arms and obtain a reward placed at their ends with the fewest number of errors. The training was done with either an open central zone or a closed central zone between each of the visits to the arms. The version of training had a significant effect on the learning scores and the number of errors, with hens choosing the optimal serial arm-visiting strategy in training with an open central zone. Hens undoubtedly possess a range of cognitive abilities as well as the ability to subjectively experience emotions, which can serve as indicators of their welfare quality level. In this context, cognitive bias tests represent a promising research tool, but their methodological optimisation is still necessary.