Review Early electrophysiological indicators for predictive processing in audition: A review ☆ , ☆☆ Alexandra Bendixen ⁎, Iria SanMiguel, Erich Schröger Institute for Psychology, University of Leipzig, Germany abstract article info Article history: Received 1 June 2011 Received in revised form 28 July 2011 Accepted 8 August 2011 Available online 23 August 2011 Keywords: Auditory information processing Predictive modeling Match detection Mismatch detection Event-related potential (ERP) Oscillatory activity The auditory system essentially deals with sequential type of input and thus requires processing that is particularly suited to extract stimulus relations within a sequence. Evidence from a variety of paradigms converges to show that the auditory system automatically uses stimulus predictability for facilitating its sequential processing. This type of predictive processing does not require attentional processing of the sounds or cognitive control of the predictions, nor does it involve the preparation of motor responses to the auditory stimuli. We will present a taxonomy of paradigms and resulting electrophysiological indicators for such automatic predictive processing in terms of event- related potential components and oscillatory activity. These indicators will include signals of fulfilled predictions (match signals such as N1 attenuation, repetition positivity, and early evoked gamma band response enhancement) as well as signals of violated predictions (mismatch signals such as the mismatch negativity and stimulus omission responses). We will show how recent approaches have revealed particularly early indicators of predictive processing down to the level of the auditory middle-latency responses. We will discuss the strength of the various indicators in terms of a truly predictive account of auditory processing (as opposed to, e.g., a retrospective verification of predictions). Finally, we will discuss the benefits of a predictive system within and beyond auditory processing. In conclusion, we argue in favor of the overwhelming evidence for predictions in audition, flexibly instantiated on different levels and timescales, and we aim to provide guidance along a variety of research paradigms illustrating the existence of these predictions. © 2011 Elsevier B.V. All rights reserved. 1. Introduction The concept of predictive processing has a long tradition in psychology, albeit under different terms. For instance, cognitive psychology has long been investigating the concept of mental models (Craik, 1943). Mental models can be conceived as internal representations of reality and play an important role in making inferences and decisions (Johnson-Laird, 1983). By means of mental models, we simulate our reality. This allows us to anticipate, for instance, the consequences of a given action. It also helps us to prepare for future events, such as the rain that is to be expected, or the bad mood of our spouse. We can then take appropriate action to prevent undesirable effects (e.g., getting wet by the rain) or even to change the future in such a way that the bad mood of our spouse will not occur in the first place. In other words, the ability of generating predictions enables goal-directed behavior that goes much beyond fix stimulus–response associations. This is the basis of an enormous flexibility in interacting with our physical and social environment. The concept of mental models is often associated with higher cognitive functions such as problem solving or language production. Yet the same notion is widespread in research on perception as a seemingly lower cognitive function. Examples are the theory of unconscious inferences of Hermann von Helmholtz (1867) or Irvin Rock (1983), according to which perception rests on logical inferences based on internal representations. However, these perceptual inferences are seen as retrospective rather than prospective in nature. More recently, the proactive or predictive aspect of mental models is increasingly discussed in perception as well (Friston, 2005; Baldeweg, 2006; Bar, 2007; Winkler et al., 2009; see also Gregory, 1980). Perception often needs to make extrapolations to facts International Journal of Psychophysiology 83 (2012) 120–131 ☆ Contribution to the Special Issue entitled “Predictive information processing in the brain: Principles, neural mechanisms and models” edited by J. Todd, E. Schröger, and I. Winkler. ☆☆ This paper is based on a German textbook chapter (Schröger E., SanMiguel, I., & Bendixen A., in press, Prädiktive Modellierung in der auditiven Wahrnehmung. In. E. Schröger & S. Koelsch (Eds.). Kognitive und Affektive Neurowissenschaften. Enzyklopädie der Psychologie (Serie II: Kognition, Band 9). Göttingen: Hogrefe). ⁎ Corresponding author at: Institute for Psychology, University of Leipzig, Seeburgstraße 14-20, D-04103 Leipzig, Germany. Tel.: + 49 341 9735907; fax: + 49 341 9735969. E-mail addresses: alexandra.bendixen@uni-leipzig.de (A. Bendixen), iria.sanmiguel@uni-leipzig.de (I. SanMiguel), schroger@uni-leipzig.de (E. Schröger). 0167-8760/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.ijpsycho.2011.08.003 Contents lists available at SciVerse ScienceDirect International Journal of Psychophysiology journal homepage: www.elsevier.com/locate/ijpsycho