The modern AI revolution, kick-started by deep learning architectures (deep-ML), AI researchers have built AIs capable of achieving human-level performance (or, indeed beyond). These AIs are inspired by the structure of the human brain and are created by combining large networks of neurons. Similarly to the human brain, the structure of the network changes over time to facilitate learning. Although powerful learners, at a high-level, deep-ML architectures represent a “black box“: their operations, or how they “think”, is, to a large degree, inaccessible (we don’t know what they know, how they make the decisions they make). Understanding how a deep-ML thinks is akin to understanding how humans think and in my research, I try to apply the tools in methods of Cognitive Science to the problem of understanding (and maybe building better AI).