This is in Goldstein, L. (2013). Brevity. Oxford: Oxford University Press. http://capitadiscovery.co.uk/brighton-ac/items/1339856. Accessed 22 June 2017.
I like this because it brings questions about intentions and enactive type ideas right together.
3 Intentions and the principles of collaboration
While this account of intentions suggests how speakers might convey information
more economically by recognizing opportunities to overload their communicative
intentions, the account offers important insights into the limits of brevity as well.
Communicative intentions are prototypically collaborative. In communication,
interlocutors use utterances to contribute propositions to conversation, and thereby
to address and resolve open questions, as part of a joint process of inquiry. This
section argues that collaborative intentions generally, and communicative intentions
in particular, are subject to constraints of COHERENCE that limit how tightly
overloaded they can be.
Researchers such as Cohen & Levesque (1991) have argued that intentions
have a distinctive role to play in the deliberations of agents that work together,
because teamwork requires agents to coordinate with one another (Lewis 1969). p. 152
We can point to pervasive analogies between teamwork as it applies in a
cooperative conversation, and teamwork in pursuit of shared practical goals. Let’s
start with understanding. To understand your teammates’ actions, you have to
recognize the intentions with which they act. These intentions involve commitments
not only to action, but also to relevant facts about the circumstances in which the
action is being carried out and about the contributions which the action is going to
make. Recognizing an intention is a process of explanatory inference that can start
from background information about the agent’s action, knowledge, preferences and
goals, but that can also make assumptions to fill in new information about the agent
as well. Reasoning about preferences is particularly important when agents maintain
an open-ended collaborative relationship with one another (Cadilhac, Asher,
Benamara & Lascarides 2011).
Imagine, for example, you are part of a team that’s catering a party. You see
one of your colleagues carrying a full tray of drinks towards a closed door. You
probably conclude that your colleague intends to distribute the drinks to party-goers
in the next room. You’ve used what you already know about your colleague’s
beliefs: you see your colleague moving, you see the full drinks, and it’s obvious that
your colleague is moving purposefully and is aware of the surroundings. You’ve also
used what you already know about your shared goals: drinks must be distributed if
the party is to be a success. At the same time, you’ve made additional assumptions.
Perhaps you were previously unaware that the next room was open to guests, or that
it even existed. But given the intentions you’ve recognized, your colleague must
know about these guests and have the particular goal of serving them.
It’s crucial that intention recognition gives you this new understanding of your
colleague and the ongoing activity. You’ll need it to track the state of the
collaboration and to plan your own contributions to it. Your engagement with each
other means that your colleague’s continued action, like carrying the drinks here,
provides the shared evidence you need to keep coordinating, and relieves you of the
need to explicitly discuss each step of progress in the task. Thomason, Stone &
DeVault (2006) explore this reasoning in more detail. They argue that collaborative
reasoning always involves a shared presumption that team members act
cooperatively and are engaged in tracking each other’s contributions. This recalls
the famous Cooperative Principle and Maxims of Grice (1975), of course.
To use language collaboratively, agents need to recognize the intentions
behind utterances in much the same way p. 153