Motivation:
In this experiment I want to study if we are able to estimate the cost associated to switching from task A to task B. Task A for instance would be writing a paper and task B analyzing the results from this new experiment I've just run. I know the paper is due like 3 days AGO but I keep working on analyzing those data. I do need to analyze those data some days so I'll get a cool paper later and then a real job but really, is that the best moment to do it? May be it would be better to finish this ##$&%$%$ paper (yes, the one that is now open in word and that I am trying not to look at) and THEN come back to my analysis.
In some cases, there may be a way to estimate the reward associated with each hour spent working on task A or B, and if we were rational we would pick up the task that has the highest reward rate at a given moment. For instance, if I finish this paper now, I can get it in this special issue of Journal Of Procrastination where it is likely to get published AND will be highly visible. If I wait on the other hand, it will get published in a lower profile journal (like Journal Of Hard Work), will get less visibility and I can forget about this Nobel price.
Experiment:
Task A = hitting green targets among red dots
Task B = hitting red targets among green dots
Reward structure:
Black Background = Task 1 - 1$ Task 2 - 0.5$
White Background = Task 1 - 0.5$ Task 1 - 1$
If you've been searching for a red object for a while you can now find it faster than at the beginning: there is a learning curve (Fig. 1: Learning Curve).
Now, each time you switch from Task A to task B you need to reset your motor actions, your planning, your visual skills, whatever it is you will have the ups and downs shown Fig. 2. Infact, if I periodically change the color of the background so I encourage you to hit one of the other color you will show performance similar to Fig. 2.
Fig. 2: What happens when you switch
Now I can refine my question. Imagine you are in case A (task A rewarded more than B, you are engaged in A so your performance are asymptotic) and now I change the color of the background so task B is now rewarded more, should you switch to task B or not? The response is, yes, if the net gain is positive - in other word - if the reward structure will be stable for long enough so the extra-gain of B/A will compensate for the cost of switching from A to B.
The idea is expressed in Fig. 3, below. Note that in this figure the y-axis is now the /reward/ rate, which is proportional to the hitting rate and the reward factor. basically, if the green area is bigger than the red one then you should switch. If not (for instance if the time remaining is too short) you should keep doing task A even if it is rewarded less.
Fig 3: the cost of task switching
In this experiment I want to study if we are able to estimate the cost associated to switching from task A to task B. Task A for instance would be writing a paper and task B analyzing the results from this new experiment I've just run. I know the paper is due like 3 days AGO but I keep working on analyzing those data. I do need to analyze those data some days so I'll get a cool paper later and then a real job but really, is that the best moment to do it? May be it would be better to finish this ##$&%$%$ paper (yes, the one that is now open in word and that I am trying not to look at) and THEN come back to my analysis.
In some cases, there may be a way to estimate the reward associated with each hour spent working on task A or B, and if we were rational we would pick up the task that has the highest reward rate at a given moment. For instance, if I finish this paper now, I can get it in this special issue of Journal Of Procrastination where it is likely to get published AND will be highly visible. If I wait on the other hand, it will get published in a lower profile journal (like Journal Of Hard Work), will get less visibility and I can forget about this Nobel price.
Experiment:
Task A = hitting green targets among red dots
Task B = hitting red targets among green dots
Reward structure:
Black Background = Task 1 - 1$ Task 2 - 0.5$
White Background = Task 1 - 0.5$ Task 1 - 1$
If you've been searching for a red object for a while you can now find it faster than at the beginning: there is a learning curve (Fig. 1: Learning Curve).
Now, each time you switch from Task A to task B you need to reset your motor actions, your planning, your visual skills, whatever it is you will have the ups and downs shown Fig. 2. Infact, if I periodically change the color of the background so I encourage you to hit one of the other color you will show performance similar to Fig. 2.
Fig. 2: What happens when you switch
Now I can refine my question. Imagine you are in case A (task A rewarded more than B, you are engaged in A so your performance are asymptotic) and now I change the color of the background so task B is now rewarded more, should you switch to task B or not? The response is, yes, if the net gain is positive - in other word - if the reward structure will be stable for long enough so the extra-gain of B/A will compensate for the cost of switching from A to B.
The idea is expressed in Fig. 3, below. Note that in this figure the y-axis is now the /reward/ rate, which is proportional to the hitting rate and the reward factor. basically, if the green area is bigger than the red one then you should switch. If not (for instance if the time remaining is too short) you should keep doing task A even if it is rewarded less.
Fig 3: the cost of task switching
