Gambling is a heterogeneous and complex disorder. Multiple factors may lead to problem gambling, yet one of the most important appears to be the increased presence of cognitive biases or distortions. These biases are thought to precipitate gambling as they can lead to dysfunctional decision making under risk or ambiguity. Modelling these cognitive perturbations in animals can improve our understanding of their neurobiological bases, and potentially stimulate novel treatment options. The first aim of this review is to give a broad overview of some of the cognitive biases that are most commonly associated with gambling. Secondly, we will discuss several animal models that we have developed in which rodent decision-making appears hallmarked by the same cognitive inconsistencies as human choice. In particular, we will discuss two tasks that capture elements of risk and loss averse decision making, and another in which rats appear susceptible to the 'near-miss' effect. To date, findings from both human and non-human studies suggest that these different biases are neuropharmacologically and neurostructurally dissociable, and that dopamine plays a key role in their expression. Lastly, we will briefly discuss areas in both human and animal research where limitations within the field may be hampering a more complete understanding of pathological gambling as a disorder.