ABSTRACT
Prediction error minimization (PEM) is becoming increasingly influential as an explanatory framework in computational neuroscience, building on formal developments in machine learning and statistical physics (for recent research, introductions, and philosophical discussion, see Friston, 2010; Hohwy, 2013; Clark, 2016a; Metzinger and Wiese, 2017). This chapter reviews the main motivation for working within the framework, and then explains PEM in more detail, tying it in with fundamental considerations in theoretical neurobiology. I shall argue that PEM provides a unified account of brain processes for mind, cognition, and action.
