Servicenavigation

Detailseite

Zurück

Projekt Druckansicht

Visual Ecology of Motion

Fachliche Zuordnung Allgemeine, Kognitive und Mathematische Psychologie
Förderung Förderung von 2015 bis 2023
Projektkennung Deutsche Forschungsgemeinschaft (DFG) - Projektnummer 274361309
 
Erstellungsjahr 2025

Zusammenfassung der Projektergebnisse

The project aims at an ecological perspective on visual motion, one of the most important capacities of the visual system. Rather than focussing on the visual input, i.e. the spatio-temporal variations of luminance on the retina, this perspective starts earlier, namely at the ecological sources that give rise to retinal motion patterns. The motion of oneself and the motion of others are the two main generators of retinal motion. Past research has shown fundamental differences between their computational requirements and perceptual mechanisms, suggesting that they are performed in different cortical pathways. Surprisingly, however, ecologically relevant situations containing both types of motion, as when one walks amongst others, have never been thoroughly investigated. Computational considerations suggest, on the one hand, that the two types of motion will inflict problems onto each other since they violate respective computational prerequisites. On the other hand, since both are supported by distinct perceptual mechanisms there may be synergies and cross talk between the systems. The project investigates these synergies and conflicts with a combination of psychophysical experiments and computational modelling. Results from the project showed that human observers can easily separate biological motion from self–motion when both are present together. Heading estimation and biological motion perception are processed independently in dual task settings. Heading perception from optic flow does not take biological motion information into account in any specific way. Yet, heading estimation can benefit from the information provided by a translating walker’s limb motion. When a walker moves, its pattern of articulation transiently provides stable points, which briefly contribute rigid scene information to the heading estimation processes. These finding are consistent with the separation of pathways and computational mechanisms between the two forms of complex motion, and with biases on heading perception produced by other forms of independent object motion. These biases can be modelled from pure optic flow analysis, considering only self translation and rotation in an assumed rigid scene without speficying biological motion or object motion. On the other hand, biological motion processing can provide supplementary information for tasks usually relying on optic flow analysis. For example, biological motion can provide cues such as articulation and facing, that can be used in addition to flow parsing to estimate scene-relative motion of other people, as evidenced by the biased perceived directions of intact point-light walkers, compared to non-biological dot clouds. Likewise, path estimation of self-motion can include cues from articulation in addition to heading estimation from optic flow. Moreover, biological motion cues are effectively used in travel distance estimation in crowded environments, either to compensate for excessive optic flow (as in the approaching crowd condition) or to derive distance information when optic flow is absent (as in the leading crowd condition). Participants were able to infer their travel speed from the articulation of biological motions, showing that biological motion can replace or augment optic flow information for accurate distance perception during self-motion. In recent developments, non-rigid motion was identified as a third type of ecological motion, distinct from self-motion and object motion, each with its dedicated perceptual pathways. Non-rigid motion patterns, such as those seen in liquids, smoke, or fire, challenge conventional motion perception assumptions, since the parts move independently, forming a distinct emergent pattern. Yet, we found that non-rigid motion can be detected through the temporal changes in pattern evolution in the visual field, with local flow derivatives playing a crucial role.

Projektbezogene Publikationen (Auswahl)

 
 

Zusatzinformationen

© 2026 DFG Kontakt / Impressum / Datenschutz

Textvergrößerung und Kontrastanpassung