Many belikeme posts are speculative, but this one is moreso. Imagine scare quotes everywhere to avoid assuming there are definitive hallmarks for what I speak of (see what I did there?, if so you win a belikeme know-prize).
Consciousness is, perhaps, “merely” the activation of existing memories, rather than copying data from long-term to working memory and processing it. Working memory is just that tiny subset of all memory currently being accessed. Activation itself changes the memory, often to reconcile it with memories that share similar information. It is seemingly more important to the brain that memories be reconciled than that they agree with extra-cranial facts. Alas that eyewitness accounts, demonstrably so unreliable, are counted on to convict accused persons.
Possibly, memory formation itself is sensed; if so this may feel different from activation, so that the first time a memory is formed, we sense it twice. Or, memory formation may not be sensed at all, and conscious awareness is wholly the activation of memories previously stored. Memory activation could be the physical basis for the notion that consciousness is “but an illusion”. It also might help erode cognitive traps that prompt so many cognitive scientists to pursue answers to “the hard problem”. It may also be useful in dismissing the evidently highly popular “thought” “experiments” that postulate “zombies” – actual human persons who act just as if they were conscious, but don’t actually experience the feelings that are under scrutiny.
Free will, then, would be the simultaneous activation of memories of both events and preferences, all of which stored previously, with the cloud of that activity producing action potentials coursing out to the neuromuscular junctions of interest. This activity forms new memories of having performed whatever action was “decided” upon, along with revising all the memories originally involved in the total decision-making event. A final lagniappe has the subconscious store a ginned-up memory of the emotion of having decided something via one’s “will”.
None of these notions refute the validity of guiding people’s behaviors, say via punishments for affronts (shunning shitstains, dunning dipshits, incarcerating incorrigibles). The offender’s own memories of punishments incurred, of reading or hearing about other people being punished, of being told about mystical beings who punish offenders, all get called up by the unconscious when planning next moves. The neural network that eventually transmits the action potentials and concocts the fictions of having “decided” on “behaviors” for “reasons” takes much of that stuff into account when creating the new memories and re-aligning the old.
The brain as a (von Neuman) computer is a pretty bad analogy, but perhaps it does not go too far to imagine talking about free will as something that utilizes the equivalent a framework when writing an app. The neophyte can build a product that satisfies specifications, but the savvy developer knows that directly calling into the DLL or whatever, while potentially fraught, provides the greatest flexibility and performance. What cognitive science needs to work out (in this highly speculative rant-context) is the equivalent of the entry points into the innards of the workings. Getting its hands on the steering wheel and pedals rather than pressing the “auto-drive” button.