Roses are, famously, red. Someone who knows that a rose is red, but cannot see its redness, lacks an important epistemic good. Marshall describes an epistemic good that he calls ‘being in touch’, roughly the good of ‘perceiving or experiencing something as it really is’, which he contrasts with the ‘test-passing propositional knowledge’ that a colour-blind person might have about colours (p. 48). In Frank Jackson's well-known story of the neuroscientist Mary, for example, the property of redness is revealed to Mary when she leaves her black-and-white laboratory and sees something red for the first time (p. 55). She already had propositional knowledge about redness; afterwards, she was also in touch with the redness of the object she perceived. Marshall provides a characterization of being in touch that distinguishes it from other epistemic states, and he argues that its value is irreplaceable by epistemic goods of other sorts, in particular mere propositional knowledge. His focus, however, is not on being in touch per se but rather on the implications of certain kinds of being in touch for the nature of morality and the motivation and justification of moral action.

Pain is, famously, painful. Someone who knows that a wombat struggling to free its injured leg from a trap is in pain, but who is not in touch with its pain, lacks an important epistemic good. Marshall argues that being in touch with another's pain requires having a mental state that resembles that pain. He glosses part of the content of pain in imperatival terms: ‘hey you, change this state in your foot!’ (p. 78). A well-functioning or internally coherent agent who is in pain will be motivated to relieve that pain, all else being equal. It follows that a well-functioning agent in touch with another's pain will be motivated to relieve the other's pain as well. Marshall uses this conclusion to provide a novel epistemic answer to the ‘why be moral?’ question; the same question that Plato, in the Republic, tried to answer by forging a connection between justice and the health of the soul (pp. 22–4). Marshall argues that the feelings and actions motivated by being in touch with others’ pain align with at least one compelling moral ideal, that of the compassionate person. A well-functioning agent in touch with others’ pain will thus be a morally good agent. And that person will possess an irreplaceable epistemic good – perceiving or experiencing this aspect of reality as it really is – that no well-functioning but non-compassionate person can possess. There is no corresponding epistemic good to which only morally bad or indifferent people have access, so the resulting epistemic asymmetry provides a reason to be moral that could in principle persuade an amoralist and should at least reassure compassionate people that they are not ‘dupes’ for accepting moral constraints (p. 21).

The central arguments for these conclusions are laid out in Part I of the book. Part II extends and reinforces them with chapters exploring ‘long-range compassion’ for spatially or temporally distant beings, the possibility and implications of being in touch with others’ pleasures and desires as well as their pains, the implications of being simultaneously in touch with the attitudes of multiple other beings, the ultimate scope of long-range compassion and its implications for permissible partiality, and how a well-functioning compassionate agent would respond to difficult cases involving such phenomena as sadism, masochism, hallucinations and the experience machine. A theme of these chapters is that the motivational implications of being in touch with others’ pains and other attitudes conform to what we would expect from an ideally compassionate moral agent. As Marshall notes, the arguments of Parts I and II are largely metaethically neutral, but in Part III, he uses the account so far developed to defend a naturalistic metaethical theory dubbed ‘Compassionate Moral Realism’. He argues that it satisfies several plausibly sufficient criteria for moral realism, supports the conclusion that pain is objectively bad, can explain the apparently ‘internal’ connection between at least some kinds of moral judgement and motivation, and can explain the possibility of moral knowledge in the face of sceptical arguments such as those based on pervasive moral disagreement and on the evolutionary history of our species. The book concludes with several short appendices that clarify the commitments of Marshall's argument with regard to well-functioning agency, the relationship between mind and body, and the content of pain.

I will now make three critical comments about Marshall's argument. All concern his central project of using the epistemic good of being in touch to answer the ‘why be moral?’ question. His answer is that compassionate people have access to an irreplaceable epistemic good – being in touch with others’ pains, pleasures, and desires – to which non-compassionate people lack access, while the latter lack distinctive access to any irreplaceable epistemic good of their own. My first concern is that being in touch with others’ pains and other attitudes, even if it is an irreplaceable epistemic good so far as it goes, seems of too little value to contribute much to answering the ‘why be moral?’ question. Let us distinguish (1) the epistemic good of knowing what it is like to be in touch with a certain property and (2) the epistemic good of being in touch with some particular property on some particular occasion. It seems valuable to know what it is like to see red or to experience pain or pleasure; someone who had test-passing propositional knowledge about these subjects but no first-hand experience would be lacking an important epistemic good. If Mary asks why she should step outside her lab, the fact that she will learn what it is like to see red is a compelling answer (p. 32). But it does not follow, and seems to me doubtful, that being in touch with particular instances of redness has much value. I would regret never having seen the redness of a rose, but not to see the redness of some particular rose is of the scarcest importance. Similarly, one might think, there is substantial value in knowing what pain is like, but scarcely any in being in touch with the particular pains of particular people.

To be sure, if one thought that each person (and wombat) were particularly important, then one might take being in touch with each person's pains to be of great value. But that thought comes perilously close to presupposing the moral value of persons; the amoralist will simply reject the premise. As a supplement to his argument, Marshall considers the possibility that we can be in touch only with our own and others’ pains, pleasures and desires, and not with the properties of ordinary physical objects (pp. 163–4). If that were so, he suggests, then the former would gain value through scarcity. That move would not help here, however. If only roses were red, then having seen a rose would matter to me, but seeing some particular rose would not, unless perhaps roses themselves were very rare – as pain (unfortunately), pleasure and desire are not.

Suppose that being in touch with others’ pains, pleasures and desires is indeed an important epistemic good. My second concern is about Marshall's thesis that the motives and actions of a well-functioning agent in touch with others’ attitudes will align with our moral ideal of a compassionate person. He argues plausibly, for example, that both will be more strongly motivated to relieve great pain than minor pain (p. 119), and to relieve pain than to prolong pleasure (pp. 103–8). But consider two facts and a dilemma. First, as Marshall acknowledges, human beings are finite beings who cannot possibly be in touch with the attitudes of even all presently existing beings, let alone all past, present and future beings (p. 129). Second, at least assuming the sort of broadly consequentialist moral theory to which Marshall is sympathetic, an ideally compassionate person would be motivationally responsive to the pain and other attitudes of at least all present and future beings, at least in so far as she can have any effect on them, and at least sometimes in an impartial way that treats each person's pains, pleasures and desires as equally important (pp. 115–16, 129–30).

Now, the dilemma is that either a well-functioning agent is motivated by her being in touch with others’ pain (etc.) or not. If not, then she might not be motivated to undertake morally compassionate actions such as relieving their pain. Marshall rejects this possibility. But if being in touch with others’ pain does motivate efforts to relieve it, then a well-functioning agent will be more strongly motivated to benefit the beings with whose pain she is in touch than to benefit the beings with whose pain she is not in touch. It seems to follow that the actions of a well-functioning but finite agent in touch with the pains of some other beings will not align with those of the ideally compassionate moral agent. The alignment evident in small-scale cases in which one can be in touch with the pain of everyone affected by one's actions unavoidably breaks down in large-scale cases in which this is impossible.

My third concern is also about Marshall's thesis that the actions of a well-functioning agent in touch with others’ pains, pleasures and desires will align with our moral ideal of a compassionate person. He appears to grant that well-functioning agents are motivated by pain and other attitudes only in a prima facie way: ‘given some affect and a lack of competing motivations, they will act in accordance with that affect’ (p. 234). He focuses throughout the book on cases in which agents lack competing motivations, but of course we do often have them, and they sometimes lead us to refrain from relieving our pains, indulging our pleasures and satisfying our desires. Some such cases may represent a compassionate response to future attitudes (e.g. getting an injection now to prevent greater pain later), but by all appearances people also value things besides pain, pleasure, and felt desire-satisfaction, and they act accordingly. But once we appreciate the prima facie character of the motivation in question, Marshall's convergence thesis breaks down. It is not true that among well-functioning agents only morally good people can have the epistemic good in question, since morally bad or indifferent people might be in touch with others’ attitudes, have some corresponding motivation to benefit them, but choose not to do so as a result of stronger motivations in favour of competing ends.

To be in touch with others’ pains and frustrated desires and make no (or little) effort to relieve them is often unpleasant. But if being in touch is as important an epistemic good as Marshall's argument requires, then it may well be worth that cost to possess it. Seen in this light, Marshall's epistemic strategy may provide reason for an amoralist to become the sort of everyday moral mediocrity who cares about others’ pain to some degree but cares about other things much more, but not reason to become a morally good person who gives substantial and morally appropriate weight to relieving others’ pain. For the same reason, but now in the opposite direction, his argument may not adequately answer the ‘why be moral?’ question of morally good people who are tempted to give somewhat greater motivational weight to interests that compete with morality, but without abandoning morality entirely.

These doubts notwithstanding, Compassionate Moral Realism is both intellectually rich and enjoyable to read. The arguments are clearly presented, with sensitive discussion of their assumptions, limitations and the possibilities for further extension, and well situated in the contemporary and historical literature; the discussions of Locke and Schopenhauer are particularly helpful. The resources of the account are exploited in imaginative ways to explain diverse moral phenomena, such as the apparent asymmetry in the moral significance of pain and pleasure (pp. 103–8) and the conflict between deontological and consequentialist intuitions (pp. 126–7). The illustrative examples are frequent and often charming. This is a good book to read.