Cronicon OPEN ACCESS EC NEUROLOGY EC NEUROLOGY Review Article A Dynamic Model Based on Bayesian Inference for Continuous Musical Expectation Citation: Keyvan Yahya. “A Dynamic Model Based on Bayesian Inference for Continuous Musical Expectation”. EC Neurology 12.12 (2020): 93-98. Abstract This paper aims to give an insight about continuous expectation in music based on a relatively broad range of the relevant litera- ture. After providing a preliminary framework that holds on to the established liaison of music expectation and active inference, we put forth a set of hypothesis that altogether suggest the possibility of a biologically plausible dynamic model that could fully account for the inference of expectations in monophonic music. We also put a special emphasis on the Bayesian theory of music perception that underpins these types of models that could infer what happens next during the successive time windows of a given melody on a basis of prior context or technically speaking, generative models denoted by ∑〖P(xn) 〗 containing a normal distribution and will be updated over time. Finally, in regard with the active actions one must take to infer the hidden parameters of a melody to perform the melodic expectation tasks, we point out to the significance of the concept of m-grams in a more precise and wider building and improvement of the hidden Markov chains representing dynamic expectations. That aside, we will propose that the predictive cod- ing models of monophonic melodies could reveal some certain patterns that would be listened on a regular basis and construct the distinctive quintessence of various local music. Keywords: Dynamic Model; Bayesian Inference; Continuous Musical Expectation Keyvan Yahya* Chemnitz University of Technology, Chemnitz, Germany *Corresponding Author: Keyvan Yahya, Chemnitz University of Technology, Chemnitz, Germany. Received: July 30, 2020; Published: November 28, 2020 Introduction Expectation and prediction of events plays a key role in music perception and cognition [1] and more generally is regarded as a fun- damental property of the human brain [2]. Music expectation has been extensively studied from different viewpoints including cogni- tive neuroscience [3], information theory [4-7], or neuropsychology of music [8-10]. However, there are only few works concerned with dynamic and continuous ratings of expectation in monophonic melodies [10]. In particular, it was shown that probability-based models, despite some limitations, are more sensitive to capture the variability in continuous ratings of humans for monophonic melodies [6]. There exist strong behavioral and physiological evidence that the nervous system maintains probabilistic distributions that are up- dated by sensory information using probabilistic, in particular Bayesian, inference [11,12]. Bayesian inference has been either implicitly or explicitly applied to model several aspects of musical processes such as key-finding [13], melodic and rhythmic expectation [14], har- monic analysis (Raphael and Stoddard, 2004), improvisation (Mavromatis 2005), and learning of melodic patterns [15]. In addition, neural theories derived from Bayesian updating schemes, such as predictive coding (Rao and Ballard 1999) have been used to providean account for musical functions such as pleasure (Gebauer., et al. 2012) and rhythm perception [16]. In this work, we propose a biologically inspired Bayesian model to account for melodic expectation in monophonic music. More pre- cisely, this model is based on encoding the probabilities of perceived musical notes as internal belief with a number of generative models and incorporates Bayesian inference to capture the continuous variations in dynamic musical expectation. Ideally, we intend to show that given optimal parameters, this model can predict the responses of the subjects for the dynamic ratings of the melodies in our experimental data.