Here is among the most genuinely disturbing observations in political philosophy, and it is one of the capability approach's most important contributions.
People who have been systematically deprived of capabilities often do not report being unhappy or deprived. Women in deeply patriarchal societies often report being content with their restricted lives. People in conditions of severe poverty often report higher subjective wellbeing than you might expect. People who have never had access to education often do not report that access to education is what they most want.
This is the problem of adaptive preferences: human beings are remarkably good at adjusting their desires, aspirations, and felt satisfaction to fit their circumstances. This is a useful survival strategy. It would be psychologically overwhelming to constantly feel the gap between your situation and what your situation could be. So we adapt. We lower our sights. We stop wanting what we cannot have, or never knew was available.
The problem is that preference satisfaction, the standard measure of welfare in economics and utilitarian ethics, cannot detect this adaptation. If you ask a woman who has never been educated and has always been told she has no right to education whether she is satisfied with her life, she may well say yes. Her preferences are satisfied, even though she is deprived of a fundamental capability. Her satisfaction is not evidence of justice. It is evidence of successful adaptation to injustice.
Sen and Nussbaum argue that this is precisely why capability, what people are actually able to do and be, is the right measure of welfare, rather than subjective satisfaction or preference fulfillment. Capabilities are objective in the relevant sense: they can be assessed independently of whether the person who lacks them has learned to be okay with lacking them.
This has a troubling corollary that is worth sitting with. It implies that external observers, philosophers, development economists, policy makers, are sometimes in a better position to identify a person's deprivation than the person themselves. This is not paternalism in the traditional sense (overriding people's choices for their own good). It is something more subtle: recognizing that preferences themselves can be the product of injustice, and that respecting preferences uncritically is not the same as respecting persons.
But this also creates a genuine risk. The history of development policy is full of external experts deciding that they know better than local populations what local populations need, and imposing their judgments in ways that have been disastrous. The capability approach tries to navigate this by grounding its list in cross-cultural deliberation, by including political participation and practical reason as central capabilities (so people can deliberate about their own conditions), and by maintaining a distinction between expanding opportunities (which is always good) and restricting choices (which is paternalism). But the tension between respecting autonomy and identifying adaptive preferences does not fully disappear.