The appeal to respect for autonomy is not a single, tightly focused argument, but a loosely related family of defenses of neutralism. Their central contentions are, first, that each person must be treated as a rational agent, capable of shaping his own destiny, and, second, that a government violates this injunction whenever it tries to advance any particular conception of the good. Because all such arguments invoke a (near-)outright prohibition on government efforts to promote the good, they all sidestep the problem of balancing autonomy against competing values. As I shall argue, however, they win this advantage only at the cost of raising serious new problems.

One obvious problem is that the argument's central notion – respect for autonomy – is very vague. We clearly fail to respect somone's autonomy when we coerce him to do our bidding. Even if we coerce him “for his own good” we show disrespect by preventing him from choosing as he thinks best. But does the state similarly disrespect its citizens' autonomy when, for example, it tries to shape their preferences in nonrational ways? If so, exactly what is disrespectful about taking (benign) advantage of a causal process that would occur anyhow? And what about providing incentives to make certain options more attractive? Does this disrespect an agent's autonomy by preventing him from evaluating his options on their merits? Or does it respect his autonomy by giving him more attractive options to evaluate?