The M-and-D Example

In the AI Story

The example was constructed against the mid-century behaviorist tendency in moral philosophy, which located morality in observable conduct and treated inner states as epiphenomenal. Murdoch's insistence that M's private perceptual revision is morally significant — perhaps more significant than any outward act she performs — was a direct challenge to this orthodoxy. It has since become a load-bearing claim of contemporary virtue ethics and a standard reference point in discussions of moral attention.

The example's structural features matter. M's initial perception felt accurate to her. She did not experience herself as projecting; she experienced herself as seeing D justly. Only through effort did she come to recognize that her perception was distorted. This recognition required a standard outside the self — a sense of what it would mean to see D fairly — that is not reducible to M's feelings or preferences. The standard is what Murdoch calls Good, and without it M would have had no leverage against her own perception.

The revision was slow, private, and often painful. It required M to acknowledge that her ego had been doing work she would have preferred not to see — protecting her self-image by denigrating D. The acknowledgment was not a single moment but an ongoing discipline, because the ego returns to its distortions unless it is continuously resisted. Murdoch insists this ongoing inner combat is where most of the moral life actually occurs.

Applied to the AI context, the M-and-D structure becomes a diagnostic tool. When a person uses Claude to produce a report on a subject, the inner process — the quality of her perception of the subject, the accuracy of her mental picture, the ego's distortions and their correction — may or may not be happening. From the outside, the report is either produced or not. The productivity metrics will show the same output either way. But the M-and-D framework insists that whether real moral-intellectual work has occurred depends on what happened inside the person, and that this inner question matters — perhaps matters more than the output question.

Origin

The example first appears in Murdoch's 1964 essay 'The Idea of Perfection,' commissioned for the Leslie Stephen Lecture at Cambridge and later collected in The Sovereignty of Good. The essay was explicitly written as a response to Stuart Hampshire and R. M. Hare, who had argued that moral philosophy should focus on action and choice rather than on perception and inner state.

Key Ideas

Behavior can be perfect while perception is corrupt. Outward compliance is not moral adequacy if the inner picture is distorted by the ego.

Inner revision is the primary moral act. What M achieves is a more accurate perception of D — a perceptual correction that is slow, private, and transformative.

Moral work requires an external standard. Without Good as a reference point, M would have no leverage against her own ego.

Invisibility is not insignificance. The most important moral work is often the most invisible, and the AI age, which rewards visible output, makes it newly difficult to take seriously.

Debates & Critiques

Some critics argue that if M's revision produces no behavioral difference, the claim that it is morally significant is empty. Murdoch's defenders reply that the revision produces behavioral differences over time — the way M treats D across years of interaction, the warmth or coldness of her attention, the quality of her presence — and that these long-term consequences cannot be isolated into single observable acts.