That was the title of a very interesting talk at the Biology and Economics conference I attended over the weekend at USC.  The authors are Juan Carillo, Isabelle Brocas and Ricardo Alonso.  It’s basically a model of how multitasking is accomplished when different modules in the brain are responsible for specialized tasks and those modules require scarce resources like oxygen in order to do their job.  (I cannot find a copy of the paper online.)

The brain is modeled as a kludgy organization.  Imagine that the listening-to-your-wife division and the watching-the-French-Open division of YourBrainINC operate independently of one another and care about nothing but completing their individual tasks.  What happens when both tasks are presented at the same time? In the model there is a central administrator in charge of deciding how to ration energy between the two divisions.  What makes this non-trivial is that only the individual divisions know how much juice they are going to need based on the level of difficulty of this particular instance of the task.

Here’s the key perspective of the model.  It is assumed that the divisions are greedy:  they want all the resources they need to accomplish their task and only the central administrator internalizes the tradeoffs across the two tasks.  This friction imposes limits on efficient resource allocation.  And these limits can be understood via a mechanism design problem which is novel in that there are no monetary transfers available.  (If only the brain had currency.)

The optimal scheme has a quota structure which has some rigidity.  There is a cap on the amount of resources a given division can utilize and that cap is determined solely by the needs of the other division.  (This is a familiar theme from economic incentive mechanisms.)  An implication is that there is too little flexibility in re-allocating resources to difficult tasks.  Holding fixed the difficulty of task A, as the difficulty of task B increases, eventually the cap binds.  The easy task is still accomplished perfectly but errors start to creep in on the difficult task.

(Drawing:  Our team is non-hierarchical from