Home

A home for a research community

“Intention perception” is the ability of humans and other animals to make good guesses about what other creatures intend to do, just by watching them. This may seem trivially easy most of the time, but that is often because we’ve learned that some behaviors are strongly and unambiguously associated with some plans. How many times has it seemed unclear how other drivers at an intersection might turn, or what a teacher is trying to convey, or what players on an opposing team are up to? We all do intention perception subconsciously every few minutes, because it’s necessary for social animals to cooperate.

Since 2009, I’ve been researching this topic with colleagues, and we’ve found just a few other groups worldwide also working on it. The history of its study goes back only to the mid-40s. This site is an effort to facilitate sharing among researchers and convergence toward common notions of what aspects should be studied first, and how to evaluate progress.

Of course, there are also practical applications. For example, whenever you start typing into a search engine, and auto-completion options appear, that’s an effort to nudge your phrasing toward topics the engine may have more relevant responses for. That is, during your typing, you and the machine are working together to understand what your intent is. The speech-driven personal assistants Google Now, Siri, and Cortana similarly try to perceive your intent, or at least steer you toward similar topics they have relevant responses for. Although¬†speech-driven interfaces¬†may be easier to use, or at least feel more natural, the interaction style doesn’t allow for immediate auto-completion-style feedback. Instead it seems we must find ways of making “natural UIs” better able to perceive intent, perhaps by considering more of the context of the interaction (such as the interaction history or location, as search engines already also do). In human conversations, factors such as interpersonal history, concurrent gesture and pose, and location all provide context that provide much of the meaning that the words alone could not convey. My own aim in the practical realm is to understand how people use speech, gesture, and other natural forms to communicate their intent, and to build interfaces using these insights.

If you have something to share here, contact david-at-thisdomain.

rss