Human Connectome Project Gambling Task

This is a reimplementation of the HCP’s gambling task [Delgado et al., 2000].

For this project the graph structure is the following:

    environment_graph = {
        0: [1, 2],  # go - win
        1: [],  # no go win
        2: [],  # go - no punish
    }

Or in graph form:

(Source code, png, hires.png, pdf)

Task description

In this task, participants are asked to make a gamble: Is the next card larger or less than five?

The participant indicates their choice by pressing either left (lower) or right (larger than five).

After the selection a card is drawn and the participant is presented with the outcome. They gain a reward of 1, if they were correct, 0 if the card is 5 or -0.5 if their choice was incorrect.

Unbeknownst to the participant, the outcome is rigged. The experimenter is able to manipulate the number of wins or losses over time.

In this environment, this is controlled via the condition variable.

The value can be 0 = loss, 1 = neutral, and 2 = win.

In the original experiment and the Human Connectome Project, there are two blocks of the experiment, where participant either win more or lose more. By contrasting the neural activity in the blocks, one can localize brain regions related to punishment or reward.

[DNF+00]

M. R. Delgado, L. E. Nystrom, C. Fissell, D. C. Noll, and J. A. Fiez. Tracking the Hemodynamic Responses to Reward and Punishment in the Striatum. Journal of Neurophysiology, 84(6):3072–3077, December 2000. doi:10.1152/jn.2000.84.6.3072.