Miniature eye movements

fixation, and stabilized vision
----------------
"Miniature eye movements are incessant, and though routinely stay within the foveal size range, are much larger than the intercone size. The interrod size would be bigger, but there are no rods in the human foveal 'pit'. That fixation seems unmoving reminds one of the stabilized camcorders that discount the shakiness of the hand."

Reading:
Chapter 6 of RHSC2.

Susana Martinez-Conde, S. L. Macknik, D. H. Hubel, "The Role of Fixational eye movements in visual perception", Nature Reviews 5: 229-240 (2004).

Demo
Try the green laser and the Timoptic eyeball, to show corneal reflection for monitoring small movements. Notice that there are two "purkinje images" in the reflection.

Illusion in Fig. 6.1: What do you see and what does it mean? [Can you now see yourself make saccades? How big are the eye movements you observe?]

Why miniature eye movements?
see RHSC2 chapter 11, page 327: on stabilized images: Are miniature eye movements useful or just noise? "Miniature eye movements of a certain velocity will tend to improve the perception of some [spatial] frequencies and worsen that of others." And: "somewhat exaggerated movements [experimentally induced] generally improve vision beyond normal."

Stabilized images
Eliminate the fading of stabilized images. Under natural conditions, there is considerable neck muscle tremor of the head and therefore much movement of the eye w.r.t. the image even without microsaccades.

Tremor
The amplitude of tremor is extremely small, of the order of seconds of arc, and therefore near the intercone diameter. See Fig. 6.5 for a power spectrum. It falls off about like a low pass filter, 20 dB per decade (power of 10 loss in amplitude for each factor of 10 increase in frequency of tremor, but not that a second pole seems to come in at about 100 Hz). Is the spectrum dominated by the globe's inertia at high frequencies? With a frequency of 90 Hz, tremor is above flicker fusion frequency.

One binocular measurement of tremor indicates it is uncorrelated between the two eyes; two other studies (cited in Hubel) say tremor is correlated.

Drift
Velocity of drift is slow, around 4' arc per second, therefore each drift in Fig6.3 seems to last about a second. Each instance of drift is necessarily terminated by a microsaccade. p. 130: "Drifts in the two eyes are apparently uncorrelated. Some have actually seen negative correlations of drift in the two eyes." What could negative correlation mean? miniatures of vergence!

"The randomness of drift in the dark"
Is drift a random walk? Fig. 6.6 implies that it is, in the dark. Is it position or velocity that is drifting? In the dark we can eliminate "slow correction" of the errors built up by the noise of drift. In the dark the mean rate of drift increases (implying some drift in the light is for correction). Some researchers find that even in the dark drift tends to be toward the fixation point, implying a possible role for proprioception, to govern slow correction.

Microsaccades
Looking at Fig 6.3, it seems that the rate of microsaccades is about 1 per second, and the size is about 5' arc. Actually a microsaccade can be any rapid eye movement less than 15' of arc. See the first figure in the Hubel paper, for an afterimage that can demo microsaccades.

Are microsaccades the sample clock of the sampled data delay model?

Fig 6.7 shows that for a log range of 1 min of arc up to 1000' of arc (1000/60 = 16 deg) there is an underlying similar relationship for the the whole range. microsaccades are less than 10 min of arc. Notice we have not emphasized the increase in velocity as the saccade amplitude increases...
Microsaccades are correlated in the two eyes.

Mean intersaccade intervals range from 300 msec to 5 sec, in human individuals.

"Suppression of microsaccades is not something that requires special training..."

In the dark microsaccades do not tend to re-foveate the eyes, thus microscaccades seem to be visually driven, not influenced by the eye position [either OMN output or proprioception]. But in Fig. 6.3 microsaccades to seem to correct for the "noise" of drift.

Summary
¶ A tremor: small amplitude that falls off with frequency of tremor.
¶ Drift: uncorrelated in the two eyes, and quite slow.
¶ Microsaccades interrupt drift.
¶ Stabilized images and miniature eye movements