How does noise influence the estimation of speed?

Area of Science:

• Visual perception
• Neuroscience
• Computational vision

Background:

• Coherent motion perception integrates local signals in space and time.
• Pooling local signals involves a trade-off between spatio-temporal resolution and segmentation.
• Understanding noise effects on perceived motion speed is crucial for visual tasks.

Purpose of the Study:

• To investigate how varying proportions of noise affect the perceived speed of coherent motion.
• To explore if perceived speed reduction in noise follows patterns seen with low-contrast gratings.
• To relate findings to models of speed decoding from elementary motion detectors.

Main Methods:

• Utilized a two-interval forced-choice procedure.
• Assessed speed perception using random-dot kinematograms with varying noise levels.
• Analyzed data considering coherence levels relative to directional judgment thresholds.

Main Results:

• Perceived speed remained largely unaffected by increasing noise, provided coherence was above directional thresholds.
• This indicates effective, though not perfect, segregation of signal and noise motion components.
• The findings challenge the analogy with speed reduction observed for low-contrast gratings.

Conclusions:

• The visual system demonstrates robust speed estimation capabilities even in the presence of significant visual noise.
• A model decoding speed from elementary motion detectors (EMDs) can explain noise independence.
• Speed estimation strategies are largely independent of stimulus spatial structure and noise levels.