{"id":55,"date":"2026-05-08T23:11:39","date_gmt":"2026-05-08T21:11:39","guid":{"rendered":"https:\/\/www.lovefrequencies.org\/microtonalia.live\/paf-section\/02\/"},"modified":"2026-09-19T19:29:21","modified_gmt":"2026-09-19T17:29:21","slug":"02","status":"publish","type":"paf_section","link":"https:\/\/www.lovefrequencies.org\/microtonalia.live\/paf-section\/02\/","title":{"rendered":"The Brain as a Predictive System"},"content":{"rendered":"

2.1 Predictive Processing: A Non-Technical Overview<\/strong><\/h3>\n

Predictive processing is a family of theories in cognitive neuroscience holding that the brain does not primarily build perception up from incoming sensory signals, but instead continuously generates predictions about what those signals should be and compares the predictions against what actually arrives. Where the two match, little happens; the prediction is confirmed and perception proceeds smoothly. Where they diverge, a prediction error is generated, and that error is what drives the system to update its internal model of the world.<\/p>\n

Friston (2010) formalized this idea under the free-energy principle, proposing that the brain, and living systems generally, act to minimize the long-run discrepancy between their internal models and the sensory evidence they receive, either by updating the model or by acting on the world to bring sensory evidence in line with the model’s predictions. Hohwy (2013) and Clark (2016) developed the philosophical and cognitive-scientific implications of this view at length, and Parr, Pezzulo, and Friston (2022) provide the current comprehensive technical treatment under the closely related framework of active inference. None of these accounts is uncontroversial; Section 7.2 already addresses the specific critique, raised by Bowers and Davis (2012), that Bayesian and predictive frameworks can be fitted too flexibly to be genuinely falsifiable, and that critique bears directly on how this paper’s own model is built, not just on the frameworks it borrows from.<\/p>\n

2.2 Perception as Model Updating<\/strong><\/h3>\n

A central consequence of the predictive processing view is that perception is not a passive readout of the world but an active, ongoing act of model updating. The brain is understood to hold a hierarchical generative model, a layered set of expectations running from very general, slowly changing beliefs about the kind of environment one is in, down to very specific, rapidly changing predictions about the next moment of sensory input. Each layer predicts the layer below it, and prediction error passed upward drives revision of the layers above.<\/p>\n

Crucially for everything that follows in this study, not all prediction errors are treated equally. Feldman and Friston (2010) showed that the brain assigns an estimated precision, a measure of how reliable or informative a given prediction error is judged to be, and that this precision estimate determines how strongly the error is allowed to revise the model. A highly precise, confidently expected signal that turns out wrong drives a large update; a signal already treated as noisy or uncertain drives comparatively little update even when it deviates from prediction. Precision weighting, introduced here in its general form, is the single mechanism this study’s three acoustic axes are argued, in Parts II and III, each to modulate in their own domain.<\/p>\n

2.3 Prediction Error and Environmental Structure<\/strong><\/h3>\n

Prediction error is not only a mechanism for fine-tuning perception moment to moment; it is also, on several accounts, the basis on which perception carves the world into structured units in the first place. Zacks, Speer, Swallow, Braver, and Reynolds (2007) proposed exactly this for the perception of ongoing activity: an event boundary, the perceived edge between one meaningful episode and the next, is registered precisely when the perceptual system’s moment-to-moment predictions suffer a transient failure, so that segmentation into discrete events is itself a signature of local prediction error rather than a separate perceptual operation applied afterward.<\/p>\n

This account matters for the present study for a specific reason, stated here so it does not need restating at every later use: it means that segmentation and prediction are not two different topics that this study is choosing to connect, but two descriptions of the same underlying process, one at the level of subjective structure and one at the level of information-processing mechanism. Section 6.2 relies on exactly this equivalence when it treats the softening of segmentation as evidence of a change in the underlying pattern of prediction error, and it is worth the reader knowing, from this point in the paper onward, that this equivalence has its own dedicated empirical literature rather than being an assumption specific to this study’s argument.<\/p>\n

2.4 Why Sound Is a Privileged Domain for Predictive Inquiry<\/strong><\/h3>\n

Predictive processing is proposed as a general account of perception across sensory modalities, but audition offers a specific methodological advantage for studying it that this study relies on throughout. Auditory perception is intrinsically and inescapably temporal: unlike a visual scene, which can be fixated, re-inspected, and held roughly stable while the visual system samples it, sound exists only as it unfolds, one moment replacing the last, with no equivalent of holding still to look again (Large & Jones, 1999). Prediction is therefore not an optional strategy the auditory system might use; it is built into the structure of the task, since a system that could not anticipate what comes next would have no way to make sense of a signal that never stops moving.<\/p>\n

Music intensifies this feature of ordinary listening rather than introducing a new one. Vuust, Heggli, Friston, and Kringelbach (2022) and, before them, Koelsch, Vuust, and Friston (2019) argue that music is best understood as a domain in which composers and performers deliberately structure the precision and timing of listeners’ predictions, building up confident expectations in order to satisfy, delay, or violate them for expressive effect. This is precisely the property this study depends on. If music is, among other things, a technology for shaping predictive precision in a listener, then a sound practice that systematically modulates frequency, time, and space, the three axes examined in Part II, is a plausible and directly testable place to look for systematic effects on that same predictive precision, and Part III of this study formalizes exactly that claim.<\/p>\n","protected":false},"menu_order":2,"template":"","class_list":["post-55","paf_section","type-paf_section","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.lovefrequencies.org\/microtonalia.live\/wp-json\/wp\/v2\/paf_section\/55","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.lovefrequencies.org\/microtonalia.live\/wp-json\/wp\/v2\/paf_section"}],"about":[{"href":"https:\/\/www.lovefrequencies.org\/microtonalia.live\/wp-json\/wp\/v2\/types\/paf_section"}],"wp:attachment":[{"href":"https:\/\/www.lovefrequencies.org\/microtonalia.live\/wp-json\/wp\/v2\/media?parent=55"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}