If you have got this far, you should have looked at a display of an English sentence which
you have probably encountered previously.
The display should have been shut before you came here, so that you can no longer see the words.
If you saw something wrong when you first looked at the display, you are not a suitable
subject for the experiment.
However, a significant subset of those who look at the display see nothing wrong with
it, even if asked to check carefully, and even if they know that other people looking at
the display do see something wrong.
Some of those who thought there was nothing wrong change their mind if later given one of
these two tasks with the display out of sight:
This raises interesting questions about the similarities and differences between looking
for an error when the display is before your eyes and when you are looking after the
display has been removed. I suspect that mechanisms for parallel visual processing in
different domains in chapter 9 of The Computer Revolution in Philosophy, which are also
related to the mechanisms suggested by Arnold Trehub in The Cognitive Brain (1991) will
turn out to explain how you can retain unnoticed qualia and later notice them. (Compare
the blind spot.)
If you saw the superfluous word when you first looked, or if you failed to see it when
looking at the display, and continued not to be aware of anything wrong even when you
considered the two questions, then the experiment has failed for you.
When using this demonstration in lectures on vision, I have found that it works for
between about 30% and 70% the people attending. I have unfortunately not kept records
of numbers for whom the duplicate word was visible in the first stage, or numbers who did
not notice the duplication while looking at the display and remained unaware of it when
asked the two questions. If any psychologist is willing to perform this experiment in a
systematic way, perhaps varying the words used and the format for presenting them, and
the order in which the questions are tried, I'll be interested to receive the results,
and, if appropriate, report them here.
Note added 6 Mar 2013
It occurs to me that the process of reading all this text in order to take part in the
online experiment may interfere with the contents of the visual memory required for
discovering that there were two occurrences of "will". If you did not see two occurrences
originally, and did not discover that you had seen two when asked the above questions you
may be able to try this experiment on acquaintances (or your students!).
Instead of asking them to go on to the next page after looking at the display, and
finding nothing wrong, ask them to shut their eyes, or look away from the screen, or cover
up the display, etc. Then ask them question (1) and see if that leads to discovery that
there were two "will"s. If it fails, try question (2). If that fails you may be able to
think of a different probe. Also you could try changing the order of the questions, in
case question (2) is more effective for some people, but only if asked first.
Note added 13 Mar 2013
There is now a video presentation of this demonstration that does not
require participants to read a lot of text. It can be found on Youtube here:
I have not attempted to find any other differences linked to either the inability to
notice the duplicate word during the initial presentation or the ability/inability to
notice it when asked the questions with the display out of sight.
If you did not see anything wrong with the display, please look back at it now, to see
what you did not notice previously: here.
Added: 30 Apr 2012
The following presentation on the functions of vision
What's vision for, and how does it work?
From Marr (and earlier) to Gibson and Beyond
This is a small part of the Meta-Morphogenesis project
School of Computer Science
The University of Birmingham