Despite the fact that the Artificial Intelligence is coming, there are still typically human behaviors to be covered when testing the mobile-dedicated software. In fact, human mobile behaviors same as the way we are using smartphones is very non-deterministic and challenging to follow for any machine.
As the testers, we should keep an eye on specific, but possible interactions that our human users can perform on mobile apps and in mobile web pages.
The cognitive approach I present puts the focus on the fact the users are humans - with their perceptual limitations, memory capacity, attention gaps and other "features and bugs". When testing, we should take into account specific scenarios like using an app when walking, writing message or tapping when driving, losing the smartphone in the coat's pocket and so on.