Skip to main content
. 2013 Oct 16;80(2):312–325. doi: 10.1016/j.neuron.2013.09.007

Figure 1.

Figure 1

Schematic of the Tolman Detour Task

Model-based and model-free decision making in a cartoon of a maze invented by Tolman and Honzik (1930). Left column: the maze has three paths (long, medium, and short), but a boulder can block just the short path (middle; after the subject has found the boulder and comes back to the start) or both short and medium (bottom). Middle column: the model-based system uses a model (thought bubble) of the maze to plan; after discovering the boulder, it knows whether the medium path is open (middle; cyan is best) or blocked (bottom; red is best). Right column: the model-free system learns path lengths based on extensive experience; if no path is blocked, this leads to the optimal choice (top; green is best); when it discovers the boulder by going along the short, green, path, it only knows that this path is blocked and thus tries the medium path (cyan) whether it is viable (middle) or not (bottom) (figure design by Alyssa Dayan).