Lecture Dec 2
Triesmans theory: there are aspects of visual percepton that are unlimited in
capacity (you see pop out effects), other aspects do require attention (conjoining
things- a couple of features)
Perceptual glue: attention glues the features together that need to be conjoined, you
need attention to "finish the job" and "put things together", attention is "perceptual
If you have something that pops out at you, you have a flat line for reaction time. For
conjoining features you have an increasing libear function with twice the slope.
Popouts happen when you're searching for a single feature. Conjoining happens
when you have to conjoin the features.
3. Determination of Location and Identity: if you have to find something on the basis
of a conjunction of features, you actually have to find it (doesn't pop oput at you),
location and identity have to come together (you have to know where it is). When
you have a pop out effect you don’t have to know where it is you just have to know it
was there ("was there an O" "yes").
Run an exp where you give people a display, in the popout condition theres either an
H or an orange letter, you have to tell the person if theres either an orange letter or
an H (background of pink Os and Blue x's), the two stimuli look different than the
background so its based on one feature, so it should pop out (it does). Display for
100ms and people do this perfectly. Other situation, target is either a pink X or a
blue O (in a background of blue X's and pink O's), so you have to conjoin the features
to find it, you have to use attentional processes (have to attend to locations in the
display). You have to display these stimuli to like half a second to get correct
answers. This is hard. For both conditions you have to tell the experimenter where
the stimuli was. With popouts you have a hard time saying where it was because it
just popped out at you, but with the hard condition you had to search and find it so
you know ehrre it is.
To find this, you look at the wrong trials. For the easy condition people were about
70% correct, for the hard condition people knew where it was all the time.
The visual search (#1 experiment from last time) and this next one (#4) are
4. Illusory conjunctions: who's to say that if you have all these features that have ti
be conjoined, that sometimes they wont be conjoined wrong (see a red T or
something when there are red stimuli and T's). Treisman and Schmidt (1982).
Exp: row of characters, black digit on both ends. First thing to do was to say what
the two black digits are (to make sure they saw the whole display), then there were
3 letters in the middle, after you reported the two outside digits you had to report as
much as you could about the middle (each one was a letter and a colour). The
display was up for 100ms (something pretty brief). Subjects were told ahead of time "we only use the letters T, S, N, O, and X, and we only use the colours pink, green,
yellow, blue and brown". Subjects were under to pressure to report anything, aside
from the two outside digits. We're looking for evidence that the colours and letter
move around, and that you see conjunctions of things that are there.
The fact that the use 5 letters and 5 colours are important, in each trial 3 letters are
in the display (2 arent), and 3 colours are in the display (2 arent), no repetitions.
When people get an answer wrong, that leaves the other two letters in the display,
and the two letters that weren’t in the display as the wrong answers you gave. The
experimenter can then classify the wrong answer as letters that were included in
the display (in the other 2 positions), or letters that weren't in the d