A Theoretical and ERP Study into the Role of Reafferent Connections in Visual Attention and Object Recognition

Attention serves an important purpose in the human mind by selecting which
input signals, that enter our brain via our nerves, will be fully processed and
which are not. The question however is how attention is produced by mechanisms
and processes at the neuronal level.
Some computational models, like those based on feedback disinhibition of
activation, offer a possible solution to this problem. This study aims at
finding evidence that supports these computational models by means of ERPs. A
task is designed that predicts certain processes to take place in the brain.
Based on previous research it is expected that these processes can result in
measurable ERP effects. The experimental task designed to do this, is a task in
which figure-ground segregation and a spatial attention task are combined into
one.
The figure-ground segregation part of the task is a task in which the shape of
a figure has to be detected. To do this, a stimulus screen is presented, with a
background covering the whole screen area and a foreground figure that has to
be detected and recognized. The background consists of a slanted striped noise
pattern in one direction, in a 450 direction compared to horizontal. The
foreground figure consists of a similar striped noise pattern, but in which the
stripes are orthogonally oriented. The foreground figure can vary in size
(large or small), shape (normal or deviant, respectively square or round) and
location (left or right). The spatial attention part of the task is that the
subject only has to attend to one hemifield per block. The subject has to
report deviant shapes in the attended hemifield.

The hypothesis is: ERP-effects of object size can be measured during a visual
attention task.

The primary question in this study is whether effects of figure size can be
measured in ERPs during figure ground segregation tasks. The secondary question
is whether this applies for attended as well as unattended stimuli. This
secondary question partially serves as a control for the found results.

This study is aimed at finding ERP effects caused by feedback connections from
higher areas in the visual processing pathway to lower areas. This effect is a
filtering effect, that is caused by attention and underlying feedback
processes. The central hypothesis to be used, is that during recognition of
larger objects and the focussing of attention on larger objects, more neurons
in the feedback stream will become active, which in their turn will disinhibit
or enhance the activations of more neurons in the feedforward stream. An
electrophysiological result of this could possibly be measured in the EEG of
humans.
To carry out the ERP experiment, stimuli, similar to the stimuli used by Lamme
(1995), will be used. These stimuli consist of images with a background and an
object, like a filled square or circle, in the foreground. Foreground as well
as background consist of slanted grey stripes in various intensities of
brightness, under an angle of 450 to horizontal. Foreground and background are
of the same overall colour and brightness. The only feature that separates the
foreground from the background is that the angle of the lines in the foreground
figure is orthogonal (900) to the lines in the background.
On the screen, a centred fixation cross will be presented, towards which the
test subjects have to keep their eyes directed. The figures will be presented
randomly on one of two locations in the screen; left and right. The size of the
figure will also be randomly chosen from two sizes. Occasionally a figure with
a deviant shape (e.g. circle instead of square) will be shown. Then the test
subject has to react. In this way the attention of the subject can be
maintained towards the perception of the figure. To measure the effect of
spatial attention, the location to which the subject should attend, will be
varied in blocks (left versus right). Only when the deviant figure is on the
attended side, the subject should react. Accordingly we get a 2 × 2 design with
large versus small on one dimension and spatial attention versus no spatial
attention on the other.

For what is known, E.E.G. measures pose no risks.