Aliasing in biological vision

2009/03/03

Moiré artifacts and other forms of aliasing often appear in electronic imaging systems due to insufficient sampling of the optical images. Spatial aliasing occurs when the optical image bandwidth exceeds that of the image detector. Spatial aliasing manifests itself when a TV host wears a fine-striped shirt, producing distracting moirés.

George's moirés

George's moirés

Temporal aliasing occurs when the temporal bandwidth of the imaged scene exceeds the frame rate of the camera. Temporal aliasing manifests itself in the wagon wheel illusion in a Western movie. A rotating wagon wheel will appear to freeze or even to reverse its rotation.

Why don’t biological systems experience aliasing — or do they? Does the eye or brain circuits exhibit aliasing and moiré phenomena?

The human eye appears to avoid spatial aliasing by supersampling the optical signal: the point spread function of the eye optics covers several photoreceptors. What limits the resolution of the eye is the optics. The cornea, aqueous humor, the lens, vitrous humor, and three layers of retinal cells in front of the photoreceptors blur the signal and efficiently remove the very high-frequency content in the image. In addition, the arrangement of photoreceptors is not strictly regular as happens in a CCD chip for example. Thus chances of producing a spatial moiré pattern in the eye are nil.

However, temporal or spatiotemporal aliasing may occur, probably in motion processing areas of the visual cortex. One manifestation of this glitch is the motion reversal illusion sometimes perceived when a spoked wheel accelerates and decelerates through a range of RPMs. Skeptical that this could be the result of the stroboscopic effect of electric lighting, I made a black-and-white pinwheel and spun it. Even in daylight, at some rotation rates, I perceived subtle and transient episodes of illusory reversal. You can try this for yourself.

I spun this wheel to reproduce the rotation reversal illusion.

Spinning wheel to reproduce the rotation reversal illusion.

Is this evidence that the visual system has a semi-regular ‘frame rate’ similar to video and film cameras? I found it hard to believe. Temporally subsampled signals indicate inefficiencies undercutting the quantum efficiency (ie overall sensitivity) of the system, a problem that’s too easily fixed by introducing additional blurring (low-pass filtering) to limit the bandwidth before sampling. Temporal blurring is something very easy for cells to do and evolution would have long found a way to blur the signal, even if vision did work in regular discontinuous frames. There has to be something else to account for the illusory motion reversal.

In a recent paper by Kline and Eagleman (2008) in the Journal of Vision titled “Evidence against the snapshot hypothesis of illusory motion reversal”, the authors staged a series of experiments that argued against the snapshot hypothesis (frame-rate hypothesis).
The experiments support the rivalry hypothesis of the illusory motion reversal, ie that “the continuous stimulation of one direction of motion eventually causes rivalry with the opposite direction of motion, and the balance of the rivalry eventually tips, engendering the perception of reversed motion”. In other words, motion detection is hypothesized to consist of multiple motion detectors that each have undersampled spatiotemporal responses, many of which produce aliasing. Normally, those motion detectors that detect correct motion are stimulated stronger than their aliased counterparts resulting in correct perception. However, in continuous exposure, the strongly stimulated detectors desensitize and the aliased detectors get their opinions heard. That’s the putative mechanism, but no specific circuits responsible for this illusion have yet been described.

Advertisements

One Response to “Aliasing in biological vision”

  1. James McArde Says:

    So are you saying that illusory motion reversal in this case is a version of the aftereffect seen when, after a period spent watching motion that fills the entire visual field, the visual stimulus is removed and we see an opposite motion on looking at a static field, such as when sitting looking out of the rear of a vehicle? In this case does the effect occur during not as an aftereffect?


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: