The idea that the nervous system operates on a mix of "binary" (action potentials as on/off functioning) and analog (dendritic and cell body summation of grade potentials) has been around since before I want to graduate school (1980's). So, Dr. Mehta is http://jamanetwork.com/journals/jama/article-abstract/336299!
The digital part is almost exclusively restricted to transmission over the distances between different neurons.
The
actual communication points between an axon and dendrite or cell body through a chemical synapse involves a lot non-digital-ish biochemical events: the action potential coming down the axon depolarizes its terminal (end), which triggers calcium influx, which triggers synaptic vesicles to release their contents (neurotransmitters), which diffuse across the space between the pre- and post-synaptic cell surfaces and bind receptors, which let ions flow in or stimulate intracellular biochemical changes, which can cause small local changes in membrane potential, which can add up (summation) to either trigger or not trigger an action potential. Each of these steps is subject to modulation due to its local biochemistry, the geometry of the cell it is part of, it location in the cell, the metabolic status of the cell (it previous history). It gets complex.
Each neuron has often been thought of as a little computer due to the complexity of these input/output relationships some cases, just single dendrites (a sub-part of the whole neuron) can act independently in a similar way.
In addition, the brain is massively parallel in its operation. This may or may not be considered digital (I think of it more as a computer architecture aspect), but it is a big part of its functioning.
All intensity coding in the brain is not based on rate of impulse production.
Sometimes more neurons in a group will be recruited to fire.
That could be considered adding additional transmission channels (more neurons in parallel), which I guess could also be digital.