Howarth, Gleeson & Attwell (2012) revised the cortical signaling-energy budget, cutting the action-potential share from 47% to 21% and raising the postsynaptic share to 50%

Howarth, Gleeson & Attwell's "Updated Energy Budgets for Neural Computation in the Neocortex and Cerebellum" (J Cereb Blood Flow Metab, 2012) re-modeled the signaling-energy budget that Attwell & Laughlin first estimated in 2001 (claim-attwell-laughlin-2001-grey-matter-signaling-energy-split), after new data showed that mammalian action potentials are considerably more energy-efficient than the original model assumed. For the cerebral cortex, the abstract states the revised split: "most signaling energy (50%) is used on postsynaptic glutamate receptors, 21% is used on action potentials, 20% on resting potentials, 5% on presynaptic transmitter release, and 4% on transmitter recycling."

The change is driven almost entirely by the action-potential term, which falls from 47% (2001) to 21% (2012); the postsynaptic term rises from 34% to 50%, overtaking spikes as the single largest cost. Under the revision, action potentials plus postsynaptic effects sum to ~71% rather than the ~81% of the 2001 estimate — a materially different total, though still leaving signaling firmly dominant over baseline maintenance. The qualitative claim that the brain is signaling-dominated at the system level (claim-brain-inference-bound-like-ai-at-system-level) survives the revision; the specific ~81% figure does not, which is why it must be dated to 2001 rather than cited as the current best estimate from this research group.

The efficiency reappraisal that motivated the revision — spikes cost less than once thought because their sodium and potassium currents overlap less in time — is the same theme running through the vault's broader neuroenergetics cluster on how cheaply biological computation can be made to run (claim-competitive-plasticity-reduces-learning-energy).