Excessive checking is reported in non-clinical populations and is a pervasive symptom in Obsessive Compulsive Disorder (OCD). We implemented a free-operant task in humans, previously used in rats, wherein participants can “check” to reduce uncertainty. Participants can press an observing key to ascertain which of two main keys will, if pressed, currently lead to rewards. Over a series of experiments we found that punishment robustly increased observing in non-clinical participants and that observing persisted long after punishment was removed. Moreover, participants appeared insensitive to the initial costs of checking, and a threefold increase in the effort required to observe served to deter participants only to a limited degree. We also assessed observing in OCD patients with no known comorbidities. The patients observed more than control participants and were abnormally insensitive to the introduction of punishment. These findings support the translational value of the task, with similar behaviours in humans and rodents. This paradigm may serve as a unifying platform, promoting interaction between different approaches to analyze adaptive and maladaptive certainty seeking behaviours. Specifically, we demonstrate how seemingly disparate theoretical and empirical approaches can be reconciled synergistically to promote a combined behavioural and cognitive account of certainty seeking.