Consider the following contrast:

 

               Date 1

               Alice:     Would you like to get coffee? (implicates) Would you like to go on a date?

               Bob:       I’m sorry, I’m not interested in you romantically.

               Alice:     I didn’t mean as a date.

 

               Date 2

               Alice:     Would you like to go on a date?

               Bob:       I’m sorry, I’m not interested in you romantically.

               Alice:     ?? I didn’t mean as a date.

 

Here’s what didn’t happen.  In Date 1, Bob isn’t uncertain about what Alice is asking.  When Alice claims that her invitation was Platonic, Bob knows that Alice is lying.  Alice also knows that Bob knows that Alice is lying.  Thus, in Date 1, it is common knowledge that Alice’s question implicated, “Would you like to go on a date?”

 

The orthodox conversational model (cf. Stalnaker 1979, 1999, 2002, 2014) makes the following prediction.  When a manifest event occurs, it becomes common ground that the manifest event occurs.  In Stalnaker’s famous example, when a goat walks into the lecture hall, it becomes common ground that a goat has walked into the lecture hall.  Roberts (2002, 2003, 2005) provides evidence from the behavior of noun phrases to corroborate this story.  This means that in Date 1, since it is common knowledge that Alice asked Bob on a date, it is common ground that Alice asked Bob on a date.

 

But this is problematic.  In both Date 1 and Date 2, Alice’s lie, “I didn’t mean as a date,” contradicts information on the common ground.  In Date 2, we get the right result: Alice’s lie sounds atrocious.  But in Date 1, Alice’s lie sounds significantly less degraded—maybe even good. 

 

I call lies like Alice’s lie in Date 1 cases of implausible deniability.  The challenge for this paper is to explain how to reconcile the existence of implausible deniability with the orthodox framework.

 

I trace the issue to a foundational problem within the Gricean program.  I argue that many conversations are not cooperative in a strict sense but are cooperative in a looser and more interesting sense.  In these conversations, interlocutors share the goal of pretending to share each other’s interests.  This innovation predicts not only the implausible deniability of implicatures but also the existence of Gricean implicatures in prima facie uncooperative contexts.  It also provides reason to prefer Lewis’ story about presupposition accommodation over Stalnaker’s account.