Research Interests, Projects, and Results
General Research Interests:
Like many others, I am fascinated by the brain, and how it governs
our perceptions, actions, and emotions. I am trying to learn something
about general principles of information processing in the brain by
studying simple visual tasks in which performance is presumably
limited by the first few processing stages, rather than by cognitive
and intellectual constraints.
In a theoretical approach I try
to understand how simple mechanisms (such as image-filtering by
well-defined units, interactions between these units, and simple
decision processes) can interact to produce apparently complex
behavioral patterns. In the experimental approach I measure
performance of human observers in simple visual tasks, focusing on
those results that deviate from the predictions of current models and
thus reveal new information about visual
processing.
Projects and
Results:
An example of a simple visual task is the visual
search paradigm. In visual search experiments, a specific target item
is presented among distractor items and observers have to respond as
fast as possible, whether the target item is present or not. The
required reaction time serves as an estimate of task difficulty.
In a typical search experiment, one can distinguish two
different types of errors: Missed-target errors (where the observer
fails to detect a presented target) and false alarms (where the
observer claims to have found a target even though none was
presented). Typically, error rates are low (< 10%) and increase
with increasing display size (number of elements). Missed target
errors are more frequent than false alarms. We have shown that this
behavior can be explained by a simple signal detection model if we
assume that observers shift their decision criterion as a function of
display size according to certain (optimized) strategies (
Zenger & Fahle 1997a
).
In
an experimental study we found significant performance differences in
tasks where target-distractor similarities and distractor-distractor
similarities are exactly matched, indicating that search difficulty
cannot be explained entirely in terms of local stimulus properties (
Zenger and Fahle 1996a
/
b
) [A summary of this work is available
as postcript-file
].
More recently,
we have started experiments on texture segmentation. Specifically, we
are investigating quantitatively how spatial contrast differences can
be used for figure-ground segmentation. We have found some surprising
practice effects (Zenger & Fahle1997b
).
For further information,
comments, or questions please contact me by
email
.
back to my homepage