Opportunity cost, action selection and response vigor in humans: a behavioral study.

In everyday life, we are constantly confronted to choice situations in which we need to decide whether or not to execute an action, and how to adjust the response vigor, i.e. how fast/strongly to execute that action. These decisions are thought to depend on the value of the expected outcome of the action to be undertaken and on its cost, in terms of effort and time investment. Moreover, how action duration affects choice is supposed to depend on the opportunity cost, i.e. the average value of all the other actions that the agent could be executing instead. However, it remains unclear how these different factors are integrated when human subjects are confronted to real choices. Here we asked human subjects to participate in different versions of a behavioral task involving effort-based decision making and compared their behavior with different model predictions (e.g. Niv et al. 2006). Participants had to squeeze a pressure bulb in different conditions. In the FIXED conditions, the monetary reward, duration and amount of effort required were determined in advance and participants could simply accept or refuse the trials proposed. In the FREE condition, participants could adjust the level of effort invested in order to decrease the duration of the squeezing. This allowed us to evaluate response vigor during the execution of the task. We manipulated opportunity cost by changing the average level of reward and effort required per unit of time. We also measured different psychometric variables, such as skin conductance, pupil diameter and facial EMG activity, during the execution of these tasks. Preliminary results show that human behavior does not follow model predictions. In particular, response vigor is much less affected by the task conditions than predicted. Also, in contrast to model predictions, opportunity cost, here defined as the average level of reward per unit of time, affected more the decision of whether or not to execute an action than response vigor, a result which also contradicts theoretical models. The present study shows that human subjects, when confronted to choices involving the adjustment of effort and time investment as a function of reward values, do not conform to current theoretical models. A new model is thus proposed to take into account this new set of data