Inhalt des Dokuments
Research
"Why do things look as they do?”
(K. Koffka, Principles of Gestalt Psychology, 1935).
Our group's main goal is to try and understand perception. We investigate spatial vision as well as audition from the level of simple artificial stimuli up to object and scene perception. Hermann von Helmholtz in 1867 in his Handbuch der Physiologischen Optik was perhaps the first to realize that perception (vision) is a form of inference: percepts are hypotheses about the world based on incomplete sensory evidence. We strongly believe that progress towards understanding this inference necessitates the combination of psychophysical experiments and computational modelling.
Currently, our group has the following main foci:
- How are visual stimuli initially encoded, what is the form of spatial filters in early vision? How is this code adapted to the characteristics of natural stimuli? What are the spatial and temporal dynamics of contrast gain-control?
- How can pure tones be perceived while being masked in noise? Detecting tones or sounds hidden in disturbing background noise is an everyday feat mastered by the human auditory system. Yet, until now it remains unanswered which mechanisms enable this feat. We adress this question by combining recent statistical tools and classical auditory experiments.
- Progressing beyond early-level perception, we need to infer the critical stimulus features human observers use when perceiving complex, visual and auditory scenes. To this end we are developing non-linear system identification methods based on machine learning. Aided by these techniques, we attempt to estimate perceptive fields.
- Having identified critical features, we want to address the fundamental problem of similarity and its relation to human categorization behavior: Can we formalize our intuitions about "similar" objects and scenes? Do we find a psychological space spanned by the critical features in which stimuli, that are categorized as similar, "live" close together?
