Saturday, 20 April 2013

Unsteady images / blurring /letter spacing effects. Possible reasons.

Reading performance and visual processing: What appears to happen in the first 50ms of you seeing a word?

Recently I took part in a pilot study looking at crowding and ‘colour’. Further study will start again this summer.
I learnt a great deal from this initial study.
It required the ‘subject’ to look at a computer screen and then a letter T would appear.
The T would be upright, upside down, laying to its right or laying to its left; as below.

The letter was displayed for 50ms and then had to decide which way round it was.
What I did not expect was that I would ‘see’ or ‘perceive’ four or five letters from that 50ms. As below.

During the 50ms, of course my eye was moving relative to the screen, as a result of muscle tremor in the eye muscles, neck muscles and back muscles.  I had seen this movement in the eyetracking, but not appreciated its importance.

My visual system was recording positions that I was conscious of, as an overlapping image about every 10ms.

So if we consider say a couple of words next to each other. The visual data my brain was trying to cope with would be similar to that below.

The two words are actually

The gap between the words is obviously important to stop the words over lapping allowing the system to identify it as two separate words...

Another experience was this. In setting up the experiment, before I was aware of this effect, we had to decide on how many milliseconds the letter T would appear on the screen for.  At longer exposures I was NOT aware of several images. I perceived a clear letter T.

There were four of us in the pilot and it looked like each person took longer for the brain to compute a clear steady image.

What has this got to do with the theories about the origin of developmental dyslexia?

This was very much a visual experience which would limit the phonological output.

It concerns the way in which the spatial information and timing information about the image are computed by the brain. 

That is there are several competing images, collected in sequence during a fixation, the relative position of which are a consequence of small movements on the retina of the eye. These relative movements will be caused by slight changes in muscle tone controlled by or as a result of signals from the cerebellum.

The cerebellum is closely linked to the magnocellular system and this phenomenon can easily be linked to motion processing efficiency. Ideas developed by such as John Stein and Al Galaburda

When the background settings were changed, the clarity/ singularity of the image appeared to change.

The nearby presence (and closeness) of other letters (crowding) changed that perception of ‘singularity’.

This fits closely into Facoetti’s ideas on visual crowding and reading and Valdois’ ideas on visual attention span.

It also fits some of the ideas associated with the Meares-Irlen syndrome.

More work needs to be done.


  1. Nice observations. I am not sure to have understood correctly. Did you actually ‘perceive’ four or five letters from that 50ms, or expected to see them on the bases of you eye-tracker experience? Indeed the image is usually not blurred, nothwithstanding our occular saccades also beacause of neurons in the LIP area of the intraparietal sulcus, which provide a steady retinal representation by combining visual inuts (the Ts) with collicular dicharges about eye movements. So was actually the machine providing you a visual feedback about your muscle tremor, or what?
    As for a possible cause of surface developmental dislexia, do you know this paper from our group? We provided evidence that VWFA might be the famous orthographic input lexicon described in the Coltheart's reading model.
    What do you think?
    Alice Proverbio

  2. Hi Alice. I was not expecting anything other than to see an individual image. I was actually surprised. Although I had previously 'worried' about why the number of frames per second needed to perceive smooth movement in films was so high. I had decided that the 'system'must be taking several discrete pictures ( edge detection events???) and then computing them together a bit like the visual system crates the illusion of visual space out of lots of fixations. So yes I dud 'see them'. The eye tracker data (binocular) suggests that there is considerable instability during the initial phase of fixation. Instability ion the fixation disparity which implies individual eye movement rather than VOR effects. Some of this movement may be due associated with elasticity issues in the occulomotor system after the ballistic saccade, but that would be more predictable than what I see.
    I am thinking that this first 50ms was being processed without the integrating effect you refer to. Certainly a more prolonged exposure seemed to prevent the perception.
    Just read the abstract for the paper you cited. It refers to a 250ms exposure. It would be interesting to look at how people respond with reduced exposures.
    Just reading be back later.

  3. Hi my initial thoughts after a quick read of the paper above go like this.
    1. the visual features unit and letter units stage interaction must occur very soon after fixation. and must be occurring during the first 50ms.
    2.The emails going around last year where it was possible to read with minimal difficulty when the letters in the words were jumbled other than the start finish letter,suggest that there has to be a lexical recognition system which is independent of letter sequence were very significant.
    3. The very high silent reading speeds with comprehension by people who are saccading every 3 or 4 words needs to be compared with the less than 160 pm typical of dyslexic undergraduates who feel a need to subvocalise or actually vocalise as they read silently there is a clue to mechanism here I think...
    Sorry I do not think I explained my thinking very well there!