||An important aspect of cognition is whether animals live exclusively in the present or can anticipate the future. Defined as self-control, the ability to choose a large, remote reinforcer over a small, proximate reinforcer available at the same frequency has been examined in a number of species, often proving difficult to demonstrate. We investigated self-control for food in domestic fowl using a standard two-key operant task and an equivalent two-choice return maze (TCRM) task. When hens chose between a 2-s delay to a 3-s feed access (impulsive) and a 6-s delay to a 7-s feed access (self-control), they appeared unable to discriminate in the TCRM but were impulsive in the operant task. We explored reasons for not choosing self-control in the operant task, first by examining the relation between feed access time and actual feed intake. A second operant experiment examined whether failure to show self-control could be attributed to an inability to combine the delay and access (quantity) reward information associated with choices to reach overall predictions of value. New hens chose between a 2-s delay to a 3-s feed access (impulsive) and either a 22-s delay to a 22-s feed access (standard self-control) or a 6-s delay to a 22-s feed access (jackpot self-control). While hens were impulsive in the standard condition, they showed significant and pronounced self-control in the jackpot condition, eliminating the possibility of an absolute cognitive constraint. Impulsive behaviour can instead be explained by temporal discounting: perceived depreciation of reward value as a function of the uncertainty associated with delay. Implications for welfare are discussed.