- seCtIon one: tHe PRoBLem
course, be a massive computational
task, and although it was not clear how
it could be achieved, most researchers
had assumed that somehow it must be –
otherwise how could we have such a sta-
ble and detailed view of the world in con-
sciousness? Change blindness challenged
that assumption. Perhaps we do not have
a stable and detailed view of the world at
all, in which case massive integration of
successive views is not necessary.
In fact, expensive eye trackers are not
needed to induce change blindness. In
the first experiment to use pictures, the
effect was obtained by moving the image
slightly (Blackmore et al., 1995). This
forces the eyes to move, resembling what
happens in a natural saccade. Subse-
quently, other less direct methods were
developed, such as using image flicker,
cuts in movies or during blinks, or a brief
blank flash between pictures (Simons,
2000), and all seem to have similar effects
(Domhoefer, Unema, and Velichkovsky,
2002). Motion detectors are also defeated
by changes that are too slow to produce
transients, and this provides another
method for eliciting change blindness
(Simons, Franconeri, and Reimer, 2000).
That the findings are genuinely surpris-
ing was confirmed by experiments asking
people to predict whether they or others
would notice the changes under various conditions. Typically a large metacog-
nitive error or ‘change blindness blindness’ was found – that is, people grossly
overestimated their ability to detect change (Levin, 2002). It is this ‘discrepancy
between what we see and what we think we see’ that justifies using the term
‘illusion’: ‘our awareness of our visual surroundings is far more sparse than most
people intuitively believe’ (Simons and Ambinder, 2005, pp. 48, 44).
One of the simplest methods for demonstrating change blindness is the flicker
method, developed by psychologists Rensink, O’Regan, and Clark (1997). They
showed an original image alternating with a modified image (each shown for
240 msecs), with blank grey screens (for 80 msecs) in between, and counted the
number of presentations before the participant noticed the change. With blanks
in between, it took many alternations to detect the change; without the blanks it
took only one or two.
They used this same method to investigate the effects of attention. When changes
were made in areas of greater interest, an average of seven alternations was
needed for participants to notice the change, whereas changes in areas of lesserFIGURE 3.8 • When these two pictures are
alternated with brief flashes of
grey in between, or moved slightly
when they are swapped, people
rarely notice the change. This is
one way to demonstrate change
blindness.