The context here is that Craig had demonstrated decisively the imaginative block that faces us when we try to conceive, in proper detail, of a counterarithmetical reality. The projectivist is then poised to see this imaginative block as something expressed when we insist upon the necessity of arithmetic. But Wright commented, 'If as Craig makes plausible, we are unable to conceive of how any alternative determination might be viable, then that is how things are with us; it is a further, tendentious step to inflate our imaginative limitations into a metaphysical discovery'. And Craig, acknowledging that he and Wright agree that we should not ask the imagination to do too much, concedes immediately: 'It certainly is a further step'. Is it so clear that there is a further step? Only if claims of necessity are 'metaphysical discoveries', and this the projectivist will query. (Essays in Quasi-Realism, p.60)
He clarifies this by analogy to his meta-ethical position (p.70):
We do not find it trivial to cross from a sentiment to a moral judgment. Only certain sentiments -- those of a certain strength, or with certain objects, or those accompanied by sentiments about others who do not share them -- form a jumping-off point. We are also conscious that there are doubtless flaws and failures in our sentiments, which are perhaps capable of explanation in the same way that we explain the defects of those who are worse than ourselves. But when the sentiments are strong and nothing on the cards explains them by the presence of defects, we go ahead and moralize. We may be aware that our opinion is fallible, but that is because we can do something with the thought of an improved perspective, even when we are fairly certain that one will not be found, and here as elsewhere commitment can coexist with knowledge that we may be wrong. The 'step' from a fully integrated sentiment of sufficient strength to the moral expression now becomes no step at all: the moral is just the vocabulary in which to express that state. Avoiding it would not be an exercise in modesty, but an impoverishing idiosyncracy of expression.
Why should it not be like this with logical necessity? We have arrived at the residual class of propositions of whose truth we can make nothing. We cannot see our failure to make anything of them as the result of a contingent limitation in our own experience, nor of a misapprehension making us think that their truth should be open to display in a way in which it need not be. We express ourselves by saying that they cannot be true -- that their negations are necessary. There is the bare possibility of being shown wrong -- perhaps our search into the causes of our imaginative block was inadequate, or perhaps we were under a misapprehension of what it might be for the proposition to be true. We may be uncomfortably aware of even great philosophers who mistakenly projected what turned out to be rectifiable limitations of imagination -- the a priori has a bad history. But as Wright notices, we should have no wish to make ourselves infallible when deeming things a priori. We make the commitment in light of the best we can do. There is no step, and no illusion.
Yet I think I can make something of the idea that ideal conceivability and metaphysical possibility might come apart. Talk of how the world could have been, and talk of what can be coherently imagined (with idealized cognitive powers), are not obviously synonymous. There's plausibly a link of sorts: we typically take conceivability as at least a guide to possibility. There may even be a perfect coincidence between them, so that all and only logical possibilities are ideally conceivable. But does that really mean that apriority and necessity are one and the same thing? Or can we somehow separate them, even without any metaphysical divergence that we can latch on to? (Might there be a sense in which one holds "in virtue of" the other, for instance? Or are they the same thing just under different "modes of presentation"? How else might we make sense of this?)