First of all, as pointed out by others: Mike Mearl's interpretation of the rules is extremely liberal. The rules for passive checks follows the following rule:
A passive check... can represent the average result for a task done repeatedly, such as searching for secret doors over and over again, or can be used when the DM wants to secretly determine whether the characters succeed at something without rolling dice, such as noticing a hidden monster.
Unfortunately the paragraph is vague, when it 'can' represent, and 'can' be used don't confirm what it does represent, nor does it confirm when it can be used, it also doesn't weed out what it does represent, nor does it weed out when it situations where it cannot be used, in this way these are almost merely suggestions. Several conclusions and implications can be made, however!
- First, an implication: A DM can say that throughout the conversation the character is constantly trying to gain insight into what the intentions of the person he is conversing with are. As such, he can use a passive check which represents the average of making the check over an over.
- Second, a conclusion: The word 'or' directly indicates that neither the proposition on what passive checks represent, nor the the proposed usage of them, are required. This would imply that a passive check does not need to be used as a representation of continuous action.
My final conlcusion is this: A RaW reading of the rules allows 5e to be played in a way that passive checks are always in use (so long as you have time to perform the task repeatedly), but a reading of the rules RaW also allows you to determine that passive checks are not always in use. In the end it is up to the DM, but there are rules out there there that (such as ear of deceit) that implies passive checks should not always be used. To begin with, if an ability exists it shouldn't be worthless! If your DM plays things out in a way that a PCs feature is useless, he isn't doing things right.
Finally I want to bring something up that hasn't been brought up, note that this is an equivocation of the previous way I used 'can represent the average result.' If the passive perception represents the mathematic average for a task done repeatedly, the mathematic average for the task done repeatedly would actually more properly be 10.5 + all modifiers that normally apply to the check. 5e rounds down on rational numbers though, so 10 is used. In the case of eye of deceit, 10.5 isn't the average that you will roll on a 20 sided die though. 11.9 technically is. I would personally round this bonus up because of a statistical phenomenon that occurs which makes eye of deceit potentially lose a lot of value when its effects on a 20 sided die roll are converted to an average bonus.