Predictive coding postulates that our brain develop an internal model of the external world to predict incoming information. However, these internal models need to be constantly updated, since the...Show morePredictive coding postulates that our brain develop an internal model of the external world to predict incoming information. However, these internal models need to be constantly updated, since the expected stimulus and actual sensory input could not coincide. This discrepancy generates activations in the brain, the so-called prediction errors. There are two types of prediction errors: the positive prediction error is triggered when an unpredicted stimulus appears, whilst negative prediction error is triggered by the omission of a predicted stimulus. This fMRI study explores the involvement of auditory subcortical nuclei, specifically the inferior colliculus (IC) and the medial geniculate body (MGB), in eliciting and encoding negative prediction errors. Five healthy participants were involved in the study, employing a variation of the auditory oddball paradigm with abstract rules to manipulate subjects’ expectations. Notably, subjects were aware that within a 7-sound sequence, one sound would be omitted in one of three known positions. We conducted a two-level analysis: the first level used a General Linear Model (GLM) on single subjects, followed by a second analysis using a Linear Mixed Effect (LME) model at the group level. Both bilateral MGB and left IC showed activations in response to omitted sounds. Notably, only the bilateral MGB resulted in encoding negative prediction errors, showing a decreased activation as subjects’ expectations of sound omissions increased. Our findings suggest that the auditory brain dynamically updates the internal model of the external environment when the sensory input deviates from expectations, such as the omission of expected sounds. The results provide insights into the hierarchical organisation of the auditory pathway, paving the way for future research to explore the differential involvement of auditory lemniscal and non-lemniscal pathways in encoding negative prediction errors. Additionally, this study posits the first steps in comprehending the neural processes underlying auditory perception as well as in providing insights for understanding perceptual anomalies, such as hallucinations and delusions.Show less
Research master thesis | Psychology (research) (MSc)
open access
Identical sensory input may be perceived differently, based on expectations and goals. For example, object recognition is facilitated for expected or task-relevant objects. At the same time,...Show moreIdentical sensory input may be perceived differently, based on expectations and goals. For example, object recognition is facilitated for expected or task-relevant objects. At the same time, unexpected objects are found to elicit a stronger neural response. These effects can be explained by predictive coding accounts of visual processing, presenting perception as a process of minimizing the difference between predicted and observed sensory input. However, as expectation and task-relevance are often conflated, it is insufficiently clear how these factors influence sensory processing in conscious perception. The current study aimed to investigate the relative influence of expectation and task-relevance on behavioral and neural measures of perception. During two EEG sessions, participants performed a task in which they discriminated between masked face and house images while we independently manipulated expectation and task-relevance. We find that images were more often correctly recognized when they were expected or task-relevant. In addition, we used multivariate pattern analysis to show that a classifier trained on sensory representations of face and house stimuli is better able to distinguish between expected or task-relevant face and house images than between unexpected or task-irrelevant images. These results suggest that expectation and task-relevance have independent effects on sensory processing. Finally, our results show that cue-based manipulations may activate sensory templates even before stimulus onset. We therefore recommend that future studies manipulate expectation and task-relevance without the use of explicit cues.Show less
