I used to assume that to conceive experiences as actualisations of conceptual capacities, we would need to credit experiences with propositional content, the sort of content judgements have. And I used to assume that the content of an experience would need to include everything the experiences enables its subject to know noninferentially. But both these assumptions now strike me as wrong. 
As to the first, he still holds that experience is conceptually structured but it need not include all the claims that one might noninferentially make because of it.
Suppose I have a bird in plane view, and that puts me in a position to know noninferentially that what I see is a cardinal. It is not that I infer that what I see is a cardinal from the way it looks, as when I identify a bird’s species by comparing what I see with a photograph in a field guide. I can immediately recognise cardinals if the viewing conditions are good enough.
On my old assumption, since my experience puts me in a position to know noninferentially that what I see is a cardinal, its content would have to include a proposition in which the concept of a cardinal figures… But what seems right is this: my experience makes the bird visually present to me, and my recognitional capacity enables me to know noninferentially that what I see is a cardinal. Even if we go on assuming my experience has content, there is no need to suppose that the concept under which my recognitional capacity enables me to bring what I see figures in that content. 
If experience does not include all the conceptual contents I might judge, what subset does it include?
A natural stopping point, for visual experiences, would be proper sensibles of sight and common sensibles accessible to sight. We should conceive experience as drawing on conceptual capacities associated with concepts of proper and common sensibles. So should we suppose my experience when I see a cardinal has propositional content involving proper and common sensibles? That would preserve the other of those two assumptions I used to make. But I think this assumption is wrong too. What we need is an idea of content that is not propositional but intuitional…
So this connects to the second change. Experience is conceptually structured but not propositionally structured. It is intuitionally conceptually structured.
Propositional unity comes in various forms. Kant takes a classification of forms of judgement, and thus forms of propositional unity, from the logic of his day, and works to describe a corresponding form of intuitional unity for each... [But] [i]t is not obvious why Kant thinks the idea requires that to every form of propositional unity there must correspond a form of intuitional unity. And anyway we need not follow Kant in his inventory of forms of propositional unity. [4-5]
The concept of a bird, like the concept of a cardinal, need not be part of the content of the experience; the same considerations would apply. But perhaps we can say it is given to me in such an experience, not something I know by bringing a conceptual capacity to bear on what I anyway see, that what I see is an animal… because ‘animal’ captures the intuition’s categorical form, the distinctive kind of unity it has…
In an intuition unified by a form capturable by ‘animal’, we might recognise content, under the head of modes of space occupancy, that could not figure in intuitions of inanimate objects. We might think of common sensibles accessible to sight as including, for instance, postures such as perching and modes of locomotion such as hopping or flying. 
This leaves me wondering just what intuitional conceptual structure might be and how it rationalises - noninferentially - judgements based on it. On the old account, it was easy to see how the very same content could be present in experience and in judgement and hence how the world could rationally constrain judgement. Now, there is supposed to be some sort of shared content but I am not sure what it is. Both the changes make this harder to grasp. Propositional judgements are justfied by intuitional experiences. And recognitional capacities have to be exercised on experiences whilst not being grounded in inferences from them. Hmm.
(Later thoughts on Travis’ ‘Reason’s Reach’ are here.)