Project Details
Projekt Print View

Rapid learning of unfamiliar sound patterns based on spectrotemporal matching mechanisms

Applicant Professor Dr. Erich Schröger, since 6/2021
Subject Area General, Cognitive and Mathematical Psychology
Term from 2014 to 2021
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 262460535
 
Final Report Year 2021

Final Report Abstract

Over two funding periods the project sought to investigate human ability to extract and learn recurring sound patterns from the acoustic environment. This ability is crucial to the recognition and identification of meaningful auditory events and needs to be robust against variability in the input, that is, when instances of the same sound pattern vary in their exact acoustics (e.g., when different speakers pronounce the same word, or when different forms of background noise overlap a recurring sound pattern). The project proposed an approach that might particularly shed light on how “object templates” emerge from auditory experience on the basis of repeated presentations. It contained a series of behavioural and EEG studies to address questions, such as, how fast sensory learning of complex auditory patterns takes place, which mechanisms of pattern matching underlie sensory learning, how tolerant the auditory system is to certain forms of input variability, and to what extent different parts of a complex pattern contribute to an “object template”. Overall, the work resulting from the two funding periods of the project revealed that sensory learning of spectrotemporal auditory patterns occurs rapidly and is robust against gestaltpreserving variations of the input or small amounts of distortions regarding the exact acoustics of single pattern presentations. Such variations modulate the internal representation’s precision rather than the time course of pattern learning. This leads to the conclusions that within-pattern feature relations (rather than the exact feature values) are crucial for the formation of object templates and that automatic mechanisms of (correlative) pattern matching exist which allow for an efficient segregation of distortions from coherent input patterns. Overall, we characterized a row of specific markers in the EEG that can be used to track pattern matching and pattern learning – I) the MMN as an indirect marker of sensory memory trace formation, II) the periodicity-related increase in the SR which is modulated by periodicity strength and augmented by attention, and III) periodicity-related transient ERPs, parts of which reflect pattern matching while others reflect entrainment to the periodicity. The COVID19 pandemic, which affected the last active period of the project, had detrimental effects on the project’s progress: one originally proposed EEG experiment could not be carried out and needed to be replaced by a strand of behavioral online experiments following an alternative objective.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung