What kind of content must visual states have if they are to offer direct (non-inferential) justification for our external world beliefs? How must they present that content if the degree of justification they provide is to reflect the nuance of our changing visual experiences? This paper offers an argument for the view that visual states comprise not only a content, but a confidence relation to that content. That confidence relation explains how visual states can offer direct perceptual justification of differing degrees for external world beliefs. These confidence relations allow that visual states justify beliefs in a way that is sensitive to subtle differences in the character of our visual experiences, whilst still allowing that visual states give us non-inferential access to the external world in virtue of their content.

It is an apparent truism about visual perception that we can see only what is visible to us. It is also frequently accepted that visual perception is dynamic: our visual experiences are extended through, and can evolve over time. I argue that taking the dynamism of visual experience seriously renders certain simplistic interpretations of the first claim, that a subject at a given time can see only what is visible to her at that time, false: we can be meaningfully said to see invisible objects. This counterintuitive result in turn focuses our attention on the relationship between perception and memory. I show that it is difficult to draw a clear or simple distinction between the two. Memory and perception rely on, and blend with, one another. Together, these claims point us away from understanding visual perception as a simple reflection of the environment, and instead as closer to a process of dynamic modelling that draws together occurrent stimulation and stored information.

What, if anything, is epistemically wrong with beliefs involving accurate statistical generalizations about demographic groups? This paper argues that there is a perfectly general, underappreciated epistemic flaw which affects both ethically charged and uncharged statistical generalizations. Though common to both, this flaw can also explain why demographic statistical generalizations give rise to the concerns they do. To identify this flaw, we need to distinguish between the accuracy and the projectability of statistical beliefs. Statistical beliefs are accompanied by an implicit representation of the statistic's modal profile. Their modal profile determines the circumstances in which they can legitimately be projected to unobserved instances. Errors in that implicit content can be compatible with the accuracy of the "bare" statistic, whilst systematically leading to downstream errors in reasoning, in a manner which reveals an epistemic flaw with an important aspect of the belief state itself.There are some generalizations about demographic groups which we may hesitate to endorse, even as they are borne out by empirical data. These include claims like the following: "Black Americans are almost eight times more likely to have a homicide conviction than white Americans;" 2 "teenage girls perform less well at mathematics than boys;" 3 "gay men have far higher rates of sexually transmitted diseases than straight men;" 4 "conservative political views are associated with lower IQ." 5 These sound like classic racist, sexist or homophobic claims, and yet they are statistically supported. Nonetheless, we tend to feel uneasy with such beliefs, and to suspect that someone who drew on them in their reasoning about the people they met would be rationally remiss. 6 This gives rise to the following puzzle: what explains that unease? Is it due merely to a socially commendable but epistemically 228

I identify three aspects to the puzzle of the speckled hen: A general puzzle, an epistemic puzzle, and a puzzle for the representationalist. These puzzles rely on an underlying "pictorialist" assumption, that we visually perceive general, determinable properties only in virtue of determinate properties or more specific, local features of our visual experience. This assumption is mistaken: Visual perception frequently starts from a position of uncertainty, and is routinely able to acquire information about general properties in the absence of more specific information. Acknowledging that visual indeterminacy is structured this way resolves all three puzzles of the speckled hen.