How brains build higher order representations of uncertainty

Abstract: Higher-order representations (HORs) are neural or computational states that are "about" first-order representations (FORs), encoding information not about the external world per se but about the agent's own representational processes -- such as the reliability, source, or structure of a FOR. These HORs appear critical to metacognition, learning, and even consciousness by some accounts, yet their dimensionality, construction, and neural substrates remain poorly understood. Here, we propose that metacognitive estimates of uncertainty or noise reflect a read-out of "posterior-like" HORs from a Bayesian perspective. We then discuss how these posterior-like HORs reflect a combination of "likelihood-like" estimates of current FOR uncertainty and "prior-like" learned distributions over expected FOR uncertainty, and how various emerging engineering and theory-based analytical approaches may be employed to examine the estimation processes and neural correlates associated with these highly under-explored components of our experienced uncertainty.