Our knowledge of the external world is subject to much variation in type and degree of access. We don’t always perceive accurately or clearly or with the same amount of revelation. There are illusions, occlusions, blurring, darkness, variations in appearance, constancy effects, blindness (partial or total), stimulus overload, perspectival disparities, squinting, habituation, priming, etc. Some things are too small to see, some too large. One sound can drown out another. It can thus be hard to know what is going on around you and mistakes are common. But this kind of variation doesn’t apply to our knowledge of the internal world: here we know everything to the same degree with no variation of access. I know my pains as well as I know my intentions and beliefs; and I know individual instances of these types with the same degree of transparency. There are no analogues of perceptual illusions or occlusions or absences of light. Traditionally, it is supposed that such knowledge is certain and incorrigible; but it is also uniform, as if each mental state is bathed in an equal amount of illumination and appears quite unimpeded. There is epistemic invariance in introspection, unlike perception.
This should strike us as remarkable, because mental states themselves are very various. Sensations, thoughts, emotions, intentions, beliefs, and acts of will differ widely among themselves—as physical objects do. Yet they are all presented in the same uniform manner to introspection; it isn’t that some are more difficult to introspect than others in the way physical objects vary in their ease of perceptibility. There are no mental analogues of remote galaxies or invisible germs or atoms or things buried underground. Everything seems presented just as it is without any impediment to knowledge. So in addition to the traditional attributes of infallibility, incorrigibility, first-person authority, and certainty, we have epistemic invariance—the property of being always equally accessible. The contents of the mind don’t vary in their degree of availability to introspection. But that seems odd and inexplicable, since the mind is not homogeneous in itself; and one would expect some variation of access depending on the prevailing conditions of introspection. Why isn’t introspection more like perception in this respect? Surely there could be a mind that exhibited introspective variance: the different types of mental state are variously known, with the possibility of error, and analogues of blurring, darkness, blindness, and so on. Isn’t that what we would expect given the realities of knowledge in an imperfect world? Why is our knowledge of our own mind like God’s omniscient knowledge of everything? It seems nothing short of miraculous.
It may be replied that the traditional picture is wrong and epistemic variation is the way things really are. That picture of introspective knowledge is a Cartesian myth: we are not infallible and incorrigible with respect to our own minds, and there is variation in quality and degree of access from case to case. Thus we have unconscious mental states, unattended pains, being unsure what you really believe or desire, not knowing whether you are in love. So there is variance in degree and type of epistemic access with respect to one’s own mind. But these points, though not mistaken in themselves, don’t really restore the analogy to perception: we still don’t have the kind of variance that characterizes perceptual knowledge. Ordinary occurrent conscious mental states are all apparently known in the same way with the same degree of clarity and certainty. They are laid out before the introspective eye in equal measure, whether they are sensations, thoughts, acts of will, etc. A pain in the toe is as present to introspection as a thought in the head, despite its relative remoteness. No matter what your beliefs are about they are equally present to you. Sensations of touch are not more introspectively available than sensations of sight. Here it may be said that this is not really as surprising or remarkable as I am making out, for all of these mental states are really in the same place—the brain. The pain in my toe is really located in my brain, just like my thoughts; there is no difference of epistemic proximity. But this just raises another puzzle: why do different parts of the brain produce the same kind of introspective access? Suppose the introspective faculty is located in a certain part of the brain, say the prefrontal cortex, while the pain and thought centers are located in other parts: won’t those other parts be differently hooked up to the prefrontal cortex, more or less distant from it and employing different nerve fibers? If so, shouldn’t we expect a difference in degree of access, with signals from one brain part taking longer to reach the introspection center than signals from another brain part? How is introspective invariance consistent with cerebral variance? Situating all mental states in the brain doesn’t support introspective invariance; it undermines it. We still have the puzzle of why different compartments of the mind converge in their introspective accessibility.
Here is another way to put the point. You can selectively lose a sense but you can’t selectively lose the ability to detect the sensations delivered by a sense. You can go blind but you can’t go “blind” to your visual sensations. I have never heard of a case of someone losing their entire introspective faculty (they go “mind-blind”), still less of someone ceasing to detect their own visual sensations while still being aware of their auditory and tactual sensations. There are no such introspective breakdowns or pathologies. But they seem like logically conceivable scenarios—couldn’t they occur in some imaginary creature? Then there would a very distinct kind of introspective variance—knowledge of some sensations but not of others (which nevertheless exist). Suppose we adopt a biological perspective, always a salutary procedure, and consider the evolution of introspection. First consider sensations and introspective knowledge of these sensations: that is one possible kind of species psychology. Then consider thoughts and emotions along with their own introspective faculty. Why should that faculty be just like the faculty directed at sensations? The faculties could exist in different species, arising at different times, and with different objects—why should they function identically? If we put both faculties together in a single species, why should the result be epistemic invariance? These are different biological adaptations, so why should there be such strong convergence? Yet in our case the entire contents of our mind present themselves with exactly the same transparency. There is a uniformity here that is at odds with biological reality as well as mental heterogeneity. To put it simply: why shouldn’t thoughts be better known than pains (or vice versa)?
It might be retorted that the puzzle arises only under a misguided perceptual model of introspection (the term itself might be contested). If we insist on viewing so-called self-knowledge as a type of inner vision, then we shall feel puzzled about why it doesn’t have the characteristics of vision; but that picture isn’t compulsory, so the puzzle dissolves. I don’t think we need to be committed to an inner vision model to feel the force of the puzzle, but anyway this response doesn’t really advance the discussion, because the same puzzle arises under other conceptions of reports of one’s own mental states. Why should all mental phenomena be expressively identical? I express my pains with the same alacrity and finesse as I express my thoughts or emotions—there isn’t some sort of temporal delay or potential for selective breakdown. Intuitively, I have the mental state and I am aware of it, so I express it at will: it isn’t that in some cases the expression is thwarted or compromised. Logically speaking, a creature could exhibit selective expression, but we don’t do that—why? That is, why does our (conscious) mind always present itself to us with the kind of uniform availability that it does? The objects around me present themselves to my senses in all sorts of different ways, with great differences of accessibility, but the mental states inside me don’t do that—they just sit there with an equal degree of accessibility, like peas in a pod (or rather notlike that). This is a fact so familiar that it takes work even to notice it, but once noticed it cannot but appear puzzling. The physical world varies enormously in its degree of perceptual accessibility, but the mental world is unvarying in its degree of introspective accessibility (with the qualifications made earlier).  It’s as if we always have 2020-vision as far as the contents of our own (conscious) minds are concerned.
Consider animal minds before introspection ever evolved. At some point it did evolve and mental states began to be known by their bearers. Did it operate over all mental states initially or only a subset of them? Was it equally adept for all existing mental states? Did it go through a phase of epistemic variance? Do other animals have the same invariance that we have? What is the explanation of this invariance? These are puzzling questions indeed.
 Compare knowledge of one’s own body: here too we have marked epistemic variance, since some parts of the body are better known than other parts, even in the case of proprioception. I can’t see my back or feel my brain, for example—yet these body parts are as much parts of my body as any. But the interior of my mind isn’t like that: it is the analogue of a completely visible body. The mind is thus epistemically anomalous, puzzlingly so.