Source
Shenhav, Botvinick and Cohen (2013), LIT-585, read in full in NOTE-446: the model (Eqs. 1–3; the retrieved text drops the equations themselves), §§4–7 and the open questions.
The proposal
A control signal has an identity, the parameter it targets, and an intensity, how far it displaces that parameter from its default. Its expected value is the sum over outcomes of their probability given the signal and the state, times their value, minus a cost that rises with intensity: “like physical effort, mental effort is assumed to carry intrinsic disutility”. Value is recursive, immediate reward plus discounted future EVC. dACC monitors control-relevant states and specifies the signal; lateral prefrontal cortex and subcortical structures implement it.
So the proposal places the allocation of control in a dedicated specifier above the structures it directs. That placement is what CLAIM-159 uses: it puts this account in the top-down cell of the drive profile.
What this does not say
It does not say the proposal is true. The paper “reports no new data” and
its model “is stated, not fitted”. A fit of the computation and a
dissociation of specifying from implementing are THEORY-070’s
promote_when, and THEORY-070 stays Proposed until the record holds them.