We derive experimentally based estimates of the energy used by neural
mechanisms to code known quantities of information. Biophysical measur
ements from cells in the blowfly retina yield estimates of the ATP req
uired to generate graded (analog) electrical signals that transmit kno
wn amounts of information. Energy consumption is several orders of mag
nitude greater than the thermodynamic minimum. It costs 10(4) ATP mole
cules to transmit a bit at a chemical synapse, and 10(6) - 10(7) ATP f
or graded signals in an interneuron or a photoreceptor, or for spike c
oding. Therefore, in noise-limited signaling systems, a weak pathway o
f low capacity transmits information more economically, which promotes
the distribution of information among multiple pathways.