Phase 5 · games as experiments
Games that collect data at scale
Bayesian observers and cheaper heuristics are hard to tell apart with lab-sized data. A short browser game that many people play can collect the trials needed. The first game is a playable prototype of the 2013 task itself; three new games, each aimed at an open question from the research map, are in development.
Pick a game
Prototype · the 2013 task as a game
Who Threw It?
A dart lands somewhere on the board, but you only see a blurry smudge. Was it the Pro, whose darts cluster tightly, or the Sleepyhead, whose darts scatter? The blur changes from throw to throw, so the best answer depends on how sure you can be. This is the 2013 task (two categories with different spreads, trial-by-trial noise, ) in two dimensions.
It is a gamified version of the 2013 task, not an extension of it, so it is kept as a prototype. Three new games, each aimed at an open question rather than at the original task, are in development in the other tabs. The prototype stores play data only in your browser; nothing is sent anywhere.
In development · research gap G4
Brain Type
Is being near-Bayesian a stable trait of a person, so that someone who is close to optimal in one kind of task is also close to optimal in others?
In development · perception meets planning
Foggy Four
How do people plan ahead when they cannot see the board clearly, and do they plan over their uncertainty about the board (belief-state planning)?