The account I have sketched of the art of self-making shows it to include an unsettling amount of unconscious or subliminal manipulation along with the exercise of "pure reason." Doesn't this process itself undermine the concept of a responsible self? This question has been explored at length by Alfred Mele in Autonomous Agents (1995). He argues that beyond mere self-control there is autonomy, which he contrasts with heteronomy, in which a self-controlled agent is nevertheless also under the (partial) control of others. He proposes a Default Responsibility Principle: If no one else is responsible for your being in state A, you are. This nicely cuts off the infinite regress feared by Kane; it permits us to pass the buck to brainwashers (if such there be in your past) but not to "society" in general or to the agentless environment. Only if foresighted, purposeful agents have been manipulating you for their own ends are you absolved from personal responsibility for the actions undertaken by your body; those are not your deeds but your brainwashers' deeds in such a case. Fair enough, but educators certainly design their interactions with us in order to further their own ends, in particular the end of turning us into reliable moral agents. How are we to distinguish between good education, dubious propaganda, and bad brainwashing? When are you benefiting from a little help from your friends, and when are you being taken for a ride?
(Freedom Evolves, p.281)

Mele corrected Dennett in his 2006 book, Free Will and Luck

I do not accept the "Default Responsibility Principle" that Dennett attributes to me: "If no one else is responsible for your being in state A, you are" (2003, p. 281). If I am right, sometimes an agent is in state A and no one is responsible for that: the agent is not responsible for that, and no one else is either. In a review of Dennett 2003, Ronald Bailey (2003) reports that one of Dennett's theses "is nicely summed up by philosopher Alfred Mele's notion of a Default Responsibility Principle." Let me just say that I am not responsible for that principle and someone else is.
(Free Will and Luck, p.177)

Harry Frankfurt developed sophisticated (and sophistical) arguments to prove that alternative possibilities and the chance to do otherwise were not required for moral responsibility. He used hypothetical agents, now called "Frankfurt controllers," to show that an agent could be "free," at least in the sense of compatibilist freedom of action (if not libertarian free will), despite the fact that the controllers ensure that the agent could not have done otherwise.

The problem of moral responsibility is intimately connected historically with the problem of free will. But we must be careful to separate responsibility from moral responsibility.

If our actions are causally determined by other entities who manipulate our minds, including neuroscientists that directly manipulate or who implant manipulating devices, as well as more modest manipulators like hypnotists and brainwashers, libertarians do not see how we can feel responsible for such actions.

It presents the same problem as actions are directly caused by chance. They are simply random and determinists do not see how we can feel responsibility for them.

To be responsible for our actions, they must have been caused by something within us, they must "depend on us" (the Greeks called this ἐφ ἡμῖν). Modern "agent-causal" theorists demand that something in the agent's mind - perhaps a uniquely mental substance - gives us the power to cause our actions.

In our Cogito model, responsibility comes from an adequately determined will choosing from among randomly generated alternative possibilities.

Just because another agent (whether Frankfurt controller, nefarious neuroscientist, therapeutic psychiatrist, benevolent educator, or concerned parent) has successfully planted an idea as one of the alternative possibilities that "come to mind," we do not surrender responsibility for the second stage in our two-stage model, where we evaluate our options and decide the one best suited for our reasons, motives, and desires.

So for the default responsibility principle to produce mitigating circumstances or complete exculpation, the heteronomous agent must somehow directly cause an action, not merely suggest or implant actions in the early stage, for our reasoned consideration in the later stage of an adequately determined willed decision.