Go / No-Go
This is a reimplementation of the Go / No-Go task described in [Guitart-Masip et al., 2012].
For this project the graph structure is the following:
environment_graph = {
0: {0: ([1, 2, 3, 4], 0.25), "skip": True},
1: [7, 8], # win - go (action1)
2: [5, 6], # punish - go (action1)
3: [8, 7], # win - nogo (action2)
4: [6, 5], # punish - nogo (action2)
5: [], # Punish low
6: [], # Punish high
7: [], # Win high
8: [], # Win low
Or in graph form:
(Source code
, png
, hires.png
, pdf
)

Task description
In this task, participants are asked to respond to a target in time, or to withhold their response.
First they are presented with a cue, the condition and response-type. There are two conditions: punishment or reward and two response-types go or no-go.
After the cue the participant needs to press space in response to a target stimulus, or to withhold their response.
Depending on their action, the participant receives a probabilistic reward, or probabilistic punishment.
As can be seen in the graph, the task is controlled by the agent_location
or starting_position
in that each cueing condition, has a separate graph
structure.
Marc Guitart-Masip, Quentin J.M. Huys, Lluis Fuentemilla, Peter Dayan, Emrah Duzel, and Raymond J. Dolan. Go and no-go learning in reward and punishment: Interactions between affect and effect. NeuroImage, 62(1):154–166, August 2012. doi:10.1016/j.neuroimage.2012.04.024.